Introduction
The holy grail of exoplanetary science is ultimately the detection and atmospheric characterization of an Earth analogue. Recent observational surveys have already detected transiting exoplanets with terrestrial-like masses and/or radii (e.g. Leger et al. Reference Leger2009; Batalha et al. Reference Batalha2011; Barclay et al. Reference Barclay2013; Borucki et al. Reference Borucki2013), and a wide range of equilibrium temperatures, including some in the habitable zones of their host stars (e.g. Borucki et al. Reference Borucki2013; Quintana et al. Reference Quintana2014). To date, masses and radii have both been measured for about 20 transiting super-Earths, defined as planets with masses between one and ten Earth Masses (Valencia et al. Reference Valencia, O'Connell and Sasselov2006; Seager et al. Reference Seager2007). Furthermore, exoplanet occurrence rates derived from surveys are revealing that sub-Neptune size planets are the most numerous class of planets in the solar neighbourhood (Howard et al. Reference Howard2012; Fressin et al. Reference Fressin2013). Currently, characterizing the atmospheres of such low-mass planets is one of the most active frontiers of exoplanetary science.
Atmospheric characterization of super-Earths with current and upcoming facilities requires a focused assessment of objectives. Observational surveys increasingly desire to find super-Earths in the habitable zones of their host stars, and, if the planets happen to be transiting, to characterize their atmospheres. Our notions of habitability are commonly based on equilibrium temperatures (T eq) where liquid water can sustain, assuming that H2O is indeed abundant in the planets in the first place (Kasting Reference Kasting1993; Selsis Reference Selsis2007; Abe et al. Reference Abe, Abe-Ouchi, Sleep and Zahnle2011; Kaltenegger et al. Reference Kaltenegger, Sasselov and Rugheimer2013; Kopparapu et al. Reference Kopparapu2013a, Reference Kopparapub). However, testing this assumption by observationally detecting H2O in the atmosphere of a habitable super-Earth is beyond the reach of current observational facilities, and would be challenging even with larger forthcoming facilities within this decade (Kaltenegger & Traub Reference Kaltenegger and Traub2009; Belu et al. Reference Belu2011, Reference Belu2013; Hedelt et al. Reference Hedelt2013; Snellen et al. Reference Snellen2013). Therefore, currently there is no plausible means for directly assessing the atmospheric chemical compositions, and hence the true habitability, of habitable-zone super-Earths and terrestrial analogues.
A more achievable goal at the present time is to answer the more basic question of what is the frequency of H2O-rich super-Earths irrespective of whether their temperatures are habitable or not. Such a question opens up the sample space to short-period transiting super-Earths whose atmospheres can potentially be characterized with existing facilities. While the short periods increase the probability and frequency of transits, the higher atmospheric temperatures make them more favourable for detecting H2O in their atmospheric spectra. The outlook for characterization of such super-Earths is promising given that upcoming surveys from space, such as TESS (Ricker et al. Reference Ricker2014), CHEOPS (Broeg et al. Reference Broeg and Roberto2013) and PLATO (Rauer et al. Reference Rauer2013), and on ground (e.g. Snellen et al. Reference Snellen, Stuik and Navarro2012; Gillon et al. Reference Gillon, Jehin, Fumel, Magain, Queloz and Saglia2013) are expected to find large numbers of short-period super-Earths orbiting bright and low-mass stars.
In the present work, we develop a framework for identification and characterization of H2O-rich super-Earths with existing observational facilities. We present analytic prescriptions and theoretical results that are useful for planning and interpretation of super-Earth observations. We demonstrate how upper-limits on the atmospheric mean molecular mass can be derived for certain super-Earths based only on their masses and radii, and that only radii measured in certain bandpasses (‘opacity windows’) can be used to derive unbiased constraints on their interior compositions. We also explore the dependence of super-Earth spectra on chemical composition and temperature. Finally we investigate the detectability of H2O-rich super-Earth atmospheres with the Hubble Space Telescope Wide Field Camera 3 (HST WFC3) Spectrograph, and derive sensitivity estimates for super-Earths orbiting two broad stellar prototypes, a G dwarf and an M dwarf, over a wide range of planetary equilibrium temperatures (T eq) and stellar brightnesses.
Constraints from mass and radius
The observed mass and radius (M p, R p) of a super-Earth can be used to place nominal constraints on its interior and atmospheric composition using internal structure models. Figure 1 shows mass – radius relations for homogeneous planets of various compositions, along with the masses and radii of several transiting super-Earths. The internal structure model and mass – radius curves are described in Madhusudhan et al. (Reference Madhusudhan2012). The compositions, shown in Fig. 1 include the most common minerals typically invoked for super-Earth interiors, namely, Fe, silicates and H2O (Valencia et al. Reference Valencia, O'Connell and Sasselov2006; Seager et al. Reference Seager2007; Fortney et al. Reference Fortney, Marley and Barnes2007; Sotin et al. Reference Sotin, Grasset and Mocquet2007). Also shown are curves for SiC and C which, although rarely expected, can nevertheless be abundant in C-rich environments (Madhusudhan et al. Reference Madhusudhan2012). Given the mass of a super-Earth, its radius can be explained by an often degenerate set of solutions comprising of various proportions of the different minerals listed above (e.g. Rogers & Seager Reference Rogers and Seager2010a; Valencia et al. Reference Valencia, Ikoma, Guillot and Nettelmann2010; Wagner et al. Reference Wagner, Tosi, Sohl, Rauer and Spohn2012; Gong & Zhou Reference Gong and Zhou2012; Madhusudhan et al. Reference Madhusudhan2012).
Masses and radii of super-Earths and theoretical mass – radius relations. The coloured solid curves show mass – radius relations predicted by internal structure models of planets with different uniform compositions shown in the legend in the same order as the curves, i.e. Fe (bottom-most curve) to H2O (top-most curve). The model curves are from Madhusudhan et al. (Reference Madhusudhan2012). The blue dotted curves show maximum-radius curves (see subsection ‘Optimal mass – radius space for H2O detectability’) for pure-H2O planets with H2O-rich atmospheres of different temperatures (500, 1000, 1500, 2000 and 2500 K) as observed in absorption bands of H2O in transit; curves with larger radii correspond to higher temperatures. The black circles with error bars show measured masses and radii of known transiting super-Earths (M p = 1 − 10 M ⊕), adopted from the exoplanet orbit database (Wright et al. Reference Wright2011). Two values of radii are shown for 55 Cancri e. The red data point shows the radius measured in the visible, and the blue data point shows a grey radius obtained by combining visible and IR measurements (Winn et al. Reference Winn2011; Gillon Reference Gillon2012).
Despite the degeneracy in solutions, the likelihood of a H2O-rich atmosphere in a super-Earth can still be assessed from M p and R p. We define three super-Earth types in this regard (SE1, SE2 and SE3). For type SE1, planets with M p and R p lying between the iron and silicate curves, a wide range of compositions are possible, including H2O-rich and H2O-poor conditions. On the other hand, for type SE2 planets with M p and R p lying between the silicate and H2O curves, a volatile-rich envelope (e.g. made of H/He and H2O) is required to explain the density, unless, in rare cases, where a C-rich composition can be invoked based on the stellar abundances (e.g. Madhusudhan et al. Reference Madhusudhan2012; Moriarty et al. Reference Moriarty, Madhusudhan and Fischer2014). For typical O-rich host stars, therefore, type SE2 planets are good candidates for hosting H2O-rich envelopes and atmospheres. Finally, type SE3 super-Earths with M p and R p lying above the H2O curve, necessarily require an envelope composition lighter than H2O, most likely an H/He envelope and an atmosphere (e.g. Seager et al. Reference Seager2007). Using these attributes, the atmospheric mean molecular mass (μ) of super-Earths of different types can be constrained in different ways as discussed below.
Upper-limit on μ of SE3-type atmospheres
In SE3-type planets, which are expected to host volatile-rich envelopes, nominal constraints can be placed on the μ of the atmosphere from the mass and a monochromatic radius. For such super-Earths, a minimum thickness of the atmosphere can be defined as the difference between the observed radius (R p) and the radius of a 100% water planet 
$({R_{{{\rm H}_2}{\rm O}}})$ with the observed mass (M p) (Kipping et al. Reference Kipping, Spiegel and Sasselov2013). Expressing the atmospheric thickness (H) in a molecular spectral band in terms of the scale height (H sc) of the atmosphere gives
$${R_{\rm p}} - {R_{{{\rm H}_2}{\rm O}}}({M_{\rm p}}) = {N_{{\rm sc}}}{H_{{\rm sc}}} = {N_{{\rm sc}}}{k_{\rm B}}T/{\rm \mu g},$$where N sc is the number of scale heights of the atmosphere contributing to the effective radius of the planet at the observed wavelength, and is typically of the order 5–10 (see subsection ‘Transmission spectra’). Here, μ is the mean molecular mass of the atmosphere, g is the acceleration due to gravity, k B is the Boltzmann constant, and T is a characteristic temperature of the atmosphere at the day – night terminator of the planet. Equation (1) can be used to derive a nominal upper-limit on the atmospheric μ as
$${\rm \mu} {\rm \lesssim} {{\rm \mu} _{{\rm max}}} = \displaystyle{{10\; \,{k_{\rm B}}\,{T_{{\rm eq}}}} \over {{g_0}({M_{\rm p}})[{R_{\rm p}} - {R_{{{\rm H}_{\rm 2}}{\rm O}}}({M_{\rm p}})]}}.$$
Here a characteristic T of T eq and a maximal N sc of 10 are assumed, where T eq is the equilibrium temperature of the planet assuming full redistribution (see e.g. Madhusudhan Reference Madhusudhan2012), and g 0 (M p) is the acceleration due to gravity of a pure-H2O planet with a mass M p and is given by GM p/
${R_{{{\rm H}_2}{\rm O}}}$, where G is the gravitational constant.
Upper-limit on μ of SE1- and SE2-type planets
For planets in the SE1 and SE2 types, hardly any constraints can be placed on the atmospheric μ by a monochromatic radius measurement, as their masses and radii can be explained by various interior compositions, as discussed above. For planets in these classes, multi-wavelength observations of the radius (i.e. transmission spectra) are required to discern the presence of an atmosphere. The vertical extent of the atmosphere (N scH sc in equation (1)) for such planets is derived from the difference between radius measured in a molecular band (R p,molec) and that measured in a spectral band with no strong opacity (R p,0), i.e. an ‘opacity window’ (see subsection ‘Atmospheric thickness and transit depth’). The constraint on μ is then given by
$${\rm \mu} {\rm \lesssim} {{\rm \mu} _{{\rm max}}} = \displaystyle{{10\; \,{k_{\rm B}}\,{T_{{\rm eq}}}} \over {{g_0}({M_{\rm p}},{R_{{\rm p},0}})[{R_{{\rm p,molec}}} - {R_{{\rm p,0}}}]}},$$where g 0 (M p,R p,0) = GM p/R p,02. This approach which is applicable to all super-Earth types, is discussed in more detail section ‘Constraints from atmospheric spectra’.
Optimal mass–radius space for H2O detectability
Given the mass and temperature of a super-Earth, equation (2) can be used to define an upper-limit on the observable radius in a H2O absorption band for super-Earths with H2O-rich atmospheres. The maximum possible radius of a H2O-rich super-Earth of a given mass can be approximated by the sum of the radius of a pure-H2O planet of the same mass and the maximum possible atmospheric thickness for a given planetary temperature:
$${R_{\rm p}}{\rm \lesssim} {R_{{\rm p},{\rm max}}}({M_{\rm p}}) = {R_{{{\rm H}_2}{\rm O}}}({M_{\rm p}}) + 10{\rm \;} {k_{\rm B}}{T_{{\rm eq}}}/18\,{\rm \mu} {g_0}.$$Here a pure H2O atmosphere is assumed with a μ of 18 amu; μ is the atomic mass unit (amu).
Therefore, given the M p and T eq of a super-Earth, its radius should lie below R p,max for it to host a H2O-rich atmosphere. Figure 1 shows curves of R p,max for H2O-rich planets for a wide range of temperatures (T eq = 500 − 2500 K) encompassing those of currently known super-Earths. These curves define the limits of M p − R p space for detecting H2O-rich super-Earths; i.e. planets with R p above a curve with the corresponding T eq are unlikely to host H2O-rich atmospheres. As an example, the super-Earth GJ 1214b with T eq~550 K is less likely to host a H2O-rich atmosphere (see subsection ‘Constraints from internal structure models’ for the detailed discussion). Additionally, as shown in Fig. 1, the available M p − R p space for detecting H2O-rich super-Earths is larger for hotter planets.
Constraints from atmospheric spectra
In this section, we attempt to answer the question of which super-Earths are best targets for atmospheric characterization. Our goal here is to identify factors which influence the observable signal of a H2O-rich atmosphere, so as to aid in selecting optimal super-Earths for follow-up observations and detailed atmospheric characterization. In order to model the spectra, we use the one-dimensional (1D) approach developed in Madhusudhan & Seager (Reference Madhusudhan and Seager2009) which allows parametric prescriptions for the compositions and temperature structures, and is applicable over a wide range of temperatures and compositions (e.g. Madhusudhan & Seager Reference Madhusudhan and Seager2009; Madhusudhan Reference Madhusudhan2012).
Transmission spectra
A transmission spectrum, observed when the planet is in transit, probes the atmosphere near the day – night terminator of the planet. Several recent studies have investigated methods to use transmission spectra of super-Earths to constrain their various atmospheric properties, including mean-molecular masses, temperature profiles, and the presence of scatterers (e.g. Miller-Ricci et al. Reference Miller-Ricci, Seager and Sasselov2009; Benneke & Seager Reference Bennekke and Seager2012, Reference Bennekke and Seager2013; Howe & Burrows Reference Howe and Burrows2012). Figures 2 and 3 show model transmission spectra with the bulk parameters (planet radius and gravity, and stellar radius) of the GJ 1214b system, but exploring different chemical compositions and temperatures. For ease of illustration, in these models we assume isothermal temperature profiles, also because infrared (IR) transmission spectra are not strongly sensitive to the temperature profile at the terminator (Miller-Ricci & Fortney Reference Miller-Ricci and Fortney2010; Howe & Burrows Reference Howe and Burrows2012). The H2O-rich models comprise 100% H2O, and the H2-rich atmospheres comprise Solar abundance composition (Madhusudhan & Seager Reference Madhusudhan and Seager2011), i.e. H2 and He constitute ~99.9% of the composition by volume but contribute minimal spectral features, and the rest in molecules such as H2O, CH4, CO and CO2 which contribute the prominent spectral features.
Model transmission spectra with system parameters of GJ1214b. Top panel: The blue and red curves show model transmission spectra for an atmosphere with H2-rich and H2O-rich composition. For the H2-rich case, a solar abundance composition in thermochemical equilibrium was assumed. The horizontal grey line shows the transit depth corresponding to the bulk radius of the planet, i.e. the ‘surface’, which can be observed in the opacity windows (see subsection ‘‘Surface radius’ of a super-Earth’). The grey vertical bands show opacity windows for a H2O-rich atmosphere in the near-IR, shown in 0.05 μm-wide bands centred at 1.05, 1.26, 1.62, 2.20 and 3.85 μm. The black curves at the bottom show commonly used photometric bandpasses: Y, J, H and Ks, which are accessible from the ground and Spitzer IRAC bands at 3.6 and 4.5 μm. Bottom panel: The blue and red curves show the same spectra as in the top panel but in units of number of scale heights (N sc,λ; see equation (5)) in the atmosphere above the ‘surface’, i.e. the grey line in the top panel.
Model transmission spectra of H2O-rich super-Earths as a function of atmospheric temperature. Top panel: All the spectra assume a H2O-rich composition and isothermal temperature profiles with the specified temperatures. Bottom panel: Spectra in units of N sc,λ corresponding to spectra in top panel (see the caption of Fig. 2 for description). The models show that spectral features are enhanced with increasing temperature due to the increasing scale height (equation (5)), but the number of scale heights probed by the spectra is relatively independent of temperature.
‘Surface radius’ of a super-Earth
Knowing the surface radius of a super-Earth is important to constrain its interior composition. Transit depths, or equivalently radii, of super-Earths have been reported in multiple bandpasses in the visible and IR wavelengths. Radius measurements of a super-Earth at different wavelengths cannot be combined to improve upon the measurement uncertainties, because different spectral bandpasses encode information from different depths of the planetary atmosphere. On the other hand, radii of super-Earths are routinely used to constrain their interior compositions, irrespective of the observed spectral bandpass (e.g. Demory et al. Reference Demory2011; Winn et al. Reference Winn2011; Gillon et al. Reference Gillon2012). Such an exercise assumes that the radius used represents the bulk radius of the planet, without any contribution from an overlying atmosphere. Therefore, the resulting constraints on the interior composition can be erroneous if the adopted radius was measured at wavelengths where significant absorption from an atmosphere is possible. Thus, it is important to identify spectral bandpasses in which the radii measured represent the bulk ‘surface’ radius (R ps) of the planet and those in which the radii may include significant contribution from the atmosphere. Here, ‘surface’ is nominally defined as the altitude where τ~1 when observed at wavelengths with minimal atmospheric opacity. In ‘opacity windows’, where atmospheric molecular line absorption is minimal, the τ~1 surface may imply (a) a physical solid/liquid surface; (b) the deeper, high pressure, regions of a gaseous atmosphere where collision-induced absorption (CIA) may contribute significant opacity; or (c) the presence of a cloud deck aloft in the atmosphere. In the presence of CIA or cloud opacity, R ps represents only an upper-limit on the bulk radius of the planet.
We identify spectral ranges of several opacity windows in Fig. 2, which provide ideal bandpasses at which to measure R ps of super-Earths (e.g. narrow bands at 1.05, 1.26 and 1.62 μm). As shown in the figure, all the spectra approach R ps of the planet at these wavelengths where there is minimal H2O absorption and hence the starlight passes largely unimpeded through the planetary atmosphere. These bandpasses also coincide with the bands in which telluric H2O contamination is minimal. As such, the conventional near-IR bandpasses of Y (1.0 μm), J (1.2 μm), H (1.6 μm) and K (2.1 μm), which are accessible with ground-based facilities and partly with the HST WFC3 spectrograph, provide good bandpasses to measure surface radii of H2O-rich super-Earths, which can be used to constrain their bulk compositions. Ideally, however, narrower bands identified in Fig. 2 would provide the best estimates of R ps.
Atmospheric thickness and transit depth
The wavelength-dependent super-Earth radius (R pλ) measured outside the opacity windows can include contributions from molecular opacity in the planet's atmosphere and, hence, can be larger than the surface radius. The λ-dependent thickness of the atmosphere (H λ), as alluded to in equation (1), is given by
$${H_{\rm \lambda}} = {R_{{\rm p\lambda}}} - {R_{{\rm ps}}} = {N_{{\rm sc},{\rm \lambda}}} {k_{\rm B}}{T_{{\rm eq}}}/\mu {\rm g}.$$For a transiting super-Earth, the transit depth at primary eclipse is given by
$${\delta _{\rm \lambda}} = R_{{\rm p\lambda}} ^2 /R_{{\rm s\lambda}} ^2 = {({R_{{\rm ps}}} + {H_{\rm \lambda}} )^2}/R_{{\rm s\lambda}} ^2, $$where R sλ is the stellar radius. Therefore, the λ-dependent contribution of the atmosphere to the transit depth is given by
$${{\rm \Delta} _{\rm \lambda}} \sim 2{H_{\rm \lambda}} {R_{{\rm ps}}}/R_{{\rm s\lambda}} ^2 = 2{N_{{\rm sc},{\rm \lambda}}} {k_{\rm B}}{T_{{\rm eq}}}{R_{{\rm ps}}}/(\mu gR_{{\rm s\lambda}} ^2 ).$$This quantity, Δλ, constitutes the ‘signal’ when planning observations of a super-Earth atmosphere. For a transiting super-Earth, T eq, R ps and R s are expected to be known from the system parameters. On the other hand, N sc,λ and μ which depend on the chemical composition of the planetary atmosphere, are not known a priori and, hence, need to be estimated based on theoretical models. For an assumed molecular composition, μ is known, e.g. 2.37 for a solar-composition (dominated by H2 and He) atmosphere, 18 for 100% H2O, 44 for 100% CO2, etc.
We use model spectra to estimate N sc,λ for a representative range of atmospheric compositions and temperatures. The lower panel of Fig. 2 shows N sc,λ for solar as well as H2O-rich atmospheric compositions for model transmission spectra of super-Earth GJ 1214b. For both models, in the opacity windows N sc,λ is close to zero, leading to H λ~0 and Δλ~0 and hence R p,λ = R ps. On the other hand, at wavelengths corresponding to the molecular absorption features, N sc,λ is non-zero and can be as high as 8 at the centre of the absorption bands, depending on the strength of the feature. Therefore, in estimating transit depths of super-Earth atmospheres using equations (5)–(7), the following values of N sc,λ are recommended for a given chemical composition:
$$\eqalign{&{N_{{\rm sc,\lambda}}} \sim 0{\rm \; \;} ({\rm in}\,{\rm opacity}\,{\rm windows}), \cr & {N_{{\rm sc,\lambda}}} \sim 5 - 8{\rm \; \;} ({\rm in}\,{\rm molecular}\,{\rm bands}).}$$These values of N sc,λ hold generally true irrespective of the atmospheric temperatures, as discussed in the following subsection ‘Effect of temperature’ and in Fig. 3.
Effect of temperature
For a super-Earth with a given radius and composition, hotter atmospheres are more conducive to atmospheric observations and characterization. As shown in Fig. 3, for a H2O-rich atmosphere, the transit depth in the H2O absorption bands increases linearly with temperature, since the atmospheric scale height, and hence the atmospheric thickness (H λ), depend linearly on temperature, as discussed in equations (5) and (7) (also see Howe & Burrows Reference Howe and Burrows2012). On the other hand, the transit depths in the opacity windows remain almost unaffected. Therefore, given a host star, a close-in super-Earth orbiting it offers better chances for atmospheric molecular detections compared to the same planet orbiting farther out. In this regard, even though cool super-Earths represent a natural progression towards finding habitable planets, they are less optimal for atmospheric characterization via transmission spectroscopy as exemplified by several recent studies of GJ 1214b (e.g. Bean et al. Reference Bean2011; Berta et al. Reference Berta2012; Désert et al. Reference Désert2011). It is to be noted, however, as shown in the bottom panel of Fig. 3, that the atmospheric thickness in scale heights (N sc,λ) for a H2O-rich atmosphere is independent of the temperature, so equation (8) is still applicable.
Effect of clouds
The presence of clouds in a super-Earth atmosphere can critically influence the interpretation of its transmission spectrum. Recent observations have suggested the possibility of clouds and/or hazes in a wide range of exoplanetary atmospheres (Pont et al. Reference Pont, Knutson, Gilliland, Moutou and Charbonneau2008; Madhusudhan et al. Reference Madhusudhan2012; Marley et al. Reference Marley, Ackerman, Cuzzi, Kitzmann, Mackwell, Simon-Miller, Harder and Bullock2013; Morley et al. Reference Morley2013). Clouds are particularly important in a super-Earth atmosphere because they can obscure the spectral features of the atmosphere leading to a nearly featureless ‘flat’ transmission spectrum in the IR. A similar spectrum with subsided spectral features can also be caused due to a high mean-molecular mass, e.g. rich in H2O, CO2, etc., as shown in Fig. 2. Owing to this degenerate set of solutions, observations of featureless spectra can be challenging to interpret, requiring very high-precision observations to break the degeneracy. The super-Earth GJ 1214b is a classic example in this regard (discussed in detail in subsection ‘The atmosphere of super-Earth GJ 1214b’).
High-temperature super-Earths are better candidates for atmospheric characterization due to the lower probability of clouds in their atmospheres. The presence of clouds, and their chemical composition, is a strong function of the temperature. Several recent studies have investigated the compositions of clouds in exoplanetary atmospheres (e.g. Howe & Burrows Reference Howe and Burrows2012; Kempton et al. Reference Kempton, Zahnle and Fortney2012; Marley et al. 2013; Morley et al. Reference Morley2013). Figure 4 shows condensation temperatures for several compounds expected in planetary atmospheres. At very low temperatures (T ≲ 300 K), H2O itself condenses out of the upper atmosphere, making spectroscopic observations of H2O in super-Earths extremely challenging, similar to the challenges in measuring H2O abundances in giant planets in the solar system (see e.g. Atreya Reference Atreya and Barbieri2010). On the other hand, even for warmer atmospheres (T~300–1000 K), several other volatile species, such as NaCl, KCl, Na2S, etc., can condense out leading to cloudy super-Earth atmospheres in this temperature range, as is likely the case with the super-Earth GJ 1214b (Bean et al. Reference Bean2011; Morley et al. Reference Morley2013). Furthermore, even very high-temperature atmospheres (T~1000–2000 K) can host clouds made of refractory species (e.g. MgSiO3, Fe, etc.).
Sensitivity simulations of the detection of the atmospheric H2O feature at 1.4 μm in H2O-rich super-Earths using the HST WFC3 spectrograph in spatial scanning mode. Calculations were made for transits of a G8 V star (e.g. 55 Cnc) in the top panel (blue) and transits of an M4 V star (e.g. GJ 1214) in the bottom panel (red). S/N contours of 0.03, 0.1, 0.3, 1 and 3 are shown in top panel, and contours of 1, 3, 10 and 30 are shown in the bottom panel. In ten transits, H2O would be detected in 55 Cnc e at an S/N~6.4 and GJ 1214b an S/N~21.5. Parameter space in the GJ 1214 panel is limited to stars with V ≥ 7.5, set by the brightest known M star (GJ 411). Other known exoplanetary systems with similar host star spectral type are shown with filled circles, and transiting exoplanets indicated with large open circles. Condensation curves of various compounds at 1 bar are shown. Sensitivities are calculated assuming no obscuring clouds. In scenarios where clouds are present, a definitive detection of the atmosphere may not be possible, even at high S/N, e.g. in the case of GJ 1214b.
Refractory condensates which form in high-T atmospheres tend to be heavier than low-T volatile condensates, and, hence, would likely lead to less extended cloud altitudes due to efficient gravitational settling of the condensates (see e.g. Spiegel et al. Reference Spiegel, Silverio and Burrows2009). Consequently, higher-T super-Earth atmospheres could be expected to have lesser cloud covers, leading to strong spectral signatures. Furthermore, extremely irradiated super-Earths with T ≳ 2000 K, such as 55 Cancri e, form a potentially ideal sample with ‘cloud free’ atmospheres which present the best chances for detecting H2O-features in super-Earth spectra.
Thermal emission spectra
Contrary to transmission spectra, thermal emission spectra of a transiting planet observed at secondary eclipse probe its dayside atmosphere. Such spectra allow constraints on both the chemical composition as well as temperature profile of the planet's dayside atmosphere. Observations of thermal emission have been reported for several dozens of giant exoplanets, but are only beginning for transiting super-Earths (e.g. Demory et al. Reference Demory2012; Gillon et al. Reference Gillon, Jehin, Fumel, Magain, Queloz and Saglia2014). The primary challenge is that the eclipse depth, which is a measure of the planet – star flux ratio, depends on both the radius and temperature of the planet and star, and can be significantly smaller than the transit depth. The planet – star flux ratio is given by
$${{{\,f_{{\rm p\lambda}}}} \over {{\,f_{{\rm s\lambda}}}}} = {{{B_{\rm \lambda}} ({T_{\rm p}})R_{\rm p}^2} \over {{B_{\rm \lambda}} ({T_{\rm s}})R_{\rm s}^2}},$$where f pλ and f sλ are the fluxes from the planet and star, respectively, B λ(T) is the Planck function, and T p (T s) and R p (R s) are the brightness temperature and radius of the planet (star), respectively. As evident from equation (9), the eclipse depth is lower than the transit depth, equation (6), by a factor of B λ(T p)/B λ(T s), and is therefore harder to detect, especially for cooler planets. It is clear from equation (9) that for a given star, bigger and hotter planets lead to higher planet-star flux contrasts. Consequently, the only robust detection of thermal emission from a super-Earth has been reported for 55 Cancri e with T eq~2000 − 2400 K (Demory et al. Reference Demory2012). Despite the apparent challenge in observing thermal emission from super-Earths, emission spectra provide unique constraints on the vertical temperature profile and chemical composition on the dayside atmosphere which is inaccessible from transmission spectra (see section ‘Case studies’).
‘Surface temperature’ and atmospheric constraints
Observations of thermal emission from super-Earths in opacity windows allow determination of their ‘surface’ temperatures (T sf). As discussed in subsection ‘Surface radius of a super-Earth’, opacity windows are wavelengths where the opacity in the planetary atmosphere is minimal. Radiation emitted from the planetary surface in these spectral bandpasses traverse unimpeded through the planetary atmosphere before reaching the observer. Consequently, brightness temperatures measured in such bandpasses constrain the ‘surface temperatures’ of the super-Earths. For planets with gaseous atmospheres, such observations can constrain the temperature in the lower atmosphere (Madhusudhan Reference Madhusudhan2012), irrespective of the atmospheric composition. We discuss these aspects in subsection ‘Constraints from thermal emission spectra’ and Fig. 7. For example, Fig. 7 shows model emission spectra for super-Earth 55 Cancri e with different atmospheric compositions (solar composition and H2O-rich), and a blackbody spectrum corresponding to the temperature of the lower atmosphere, i.e. the surface temperature. As shown in the figure, the different spectra converge to the blackbody spectrum in the opacity windows in the Y, J, H and, to some extent, the K bands.
Observations and model transmission spectra of GJ1214b. The solid curves show model spectra of GJ 1214b with the different chemical compositions described in the legend. The black (green) curve corresponds to a H2-rich atmosphere model assuming solar abundances and chemical equilibrium (non-equilibrium). In the green model, CH4 is depleted by a factor of 100 relative to chemical equilibrium, motivated by similar requirements for the hot Neptune GJ 436b (Madhusudhan & Seager Reference Madhusudhan and Seager2011). The magenta model corresponds to a H2O-rich atmosphere of GJ 1214b. The various symbols with error bars show the observations using different instruments from space and ground, as described in the legend and reported by Bean et al. (Reference Bean2011), Désert et al. (Reference Désert2011) and Kreidberg et al. (Reference Kreidberg2014). Only a subset of all available observations in the literature is shown here for clarity. All the observations in the literature to date are consistent with a flat spectrum, shown as a grey horizontal line, but are also consistent with a cloudy atmosphere of unknown composition. The observations, however, rule out a cloud-free H2-rich or H2O-rich atmosphere (also see Kriedberg et al. Reference Kreidberg2014). In the present work, we also suggest that a H2O-rich atmosphere is unlikely based on internal structure models (see subsection ‘Constraints from internal structure models’).
Model transmission spectra of super-Earth 55 Cancri e. The green and red solid curves show two model spectra of the planet with H2- and H2O-rich compositions, respectively. The grey horizontal curve shows a model spectrum with no atmosphere. The large black circles with uncertainties show photometric measurements of the transit depth in the visible, using the MOST telescope (Winn et al. Reference Winn2011), and in the IR at 4.5 μm using the Spitzer space telescope (Gillon Reference Gillon2012). The small blue circles with uncertainties show our simulated observations with WFC3 in spatial scanning mode for both the model atmospheres. HST WFC3 observations would be able to provide an unambiguous constraint on the chemical composition of the atmosphere, particularly on the possibility of a H2O-rich atmosphere. The high temperatures on the planet (T~2000 − 2400 K) also imply that the atmosphere is very likely cloud-free, thereby removing ambiguities in interpretation of the spectra.
Thermal emission spectra for 55 Cancri e predicted for different atmospheric compositions. The red curve indicates a H2O-rich atmosphere, whereas the green curve is representative of an H2-rich atmosphere. The blue curve shows a blackbody curve corresponding to the bottom optically thick ‘surface’ of the atmosphere, if present. Red data points indicate our simulated WFC3 secondary eclipse observations taken using spatial scanning (see subsection ‘Constraints from thermal emission spectra’). The scatter and error bars are shown assuming a H2O-rich atmosphere. The large black circle with uncertainties shows the Spitzer 4.5 μm detection (Demory et al. Reference Demory2012). The inset shows a close-up of the WFC3 IR red spectral region. The curves show the H2O-rich and H2-rich models relative to the blackbody continuum, and the blue squares show the predicted sensitivities. HST WFC3 observations will be capable of detecting thermal emission from the planet in all the modelled scenarios and will be able to discriminate between the different models and detect a H2O-rich atmosphere if present, as discussed in subsection ‘Constraints from thermal emission spectra’
On the other hand, for wavelengths corresponding to molecular absorption bands, the τ~1 surface lies higher up in the atmosphere. Therefore, brightness temperatures measured in the molecular bands allow joint constraints on the temperature profiles and molecular composition of the dayside atmosphere, as has been extensively demonstrated for hot Jupiter atmospheres (Madhusudhan et al. Reference Madhusudhan2011). We discuss model atmospheres and spectral features in thermal emission for specific super-Earths in section ‘Case studies’.
Prospects with HST and optimal discovery space
In this section, we investigate the following fundamental question: What properties of super-Earths and their host stars allow the best chances to detect H2O in their atmospheres with existing facilities? To this end, we consider HST WFC3 spectrograph as our instrument of choice, since this is the only space-based instrument, which currently has the spectroscopic capability to detect H2O in exoplanetary atmospheres (e.g. Deming et al. Reference Deming2013). In our study, we consider two archetypes for stellar hosts and planetary sizes, GJ 1214b and 55 Cancri e, and estimate the detectability of H2O features in each case but over a wide range in planetary temperatures and stellar brightnesses.
Our sensitivity estimates, shown in Fig. 4 and described below, were derived in the context of a detection of the 1.4 μm H2O feature. However, the results are applicable to any feature in the IR passband from 0.9 to 1.65 μm. The recently implemented spatial scanning capability of HST WFC3 makes it possible to observe bright targets with a dense temporal sampling. This technique has proved very successful for exoplanet transmission spectroscopy (e.g. McCullough & MacKenty 2012; Deming et al. Reference Deming2013). The efficiency and precision of the spectrophotometry are greatly improved by scanning the point source over the detector such that the counts are distributed over a wide range of pixels.
While the current Exposure Time Calculator (ETC) for WFC3 does not calculate the implementation of this new scanning mode, it is straightforward to estimate the exposure time and expected S/N, as detailed in McCullough & MacKenty (2012). We assume a 256 pixel subarray to ensure parallel buffer dumps so that there is no interruption in the data acquisition. We use model spectra of a G8 V and an M4 V spectral class (Pickles Reference Pickles1998), scaled to the appropriate brightness, to estimate the count rate in the passband of interest as a function of H magnitude, which is the band close to the 1.4 μm H2O line of interest. The count contributions of the sky, dark current, thermal background and readout noise derived using the ETC are minimal given the brightnesses of targets in question.
Instead of the light being concentrated as a point source, it is distributed over scores of pixels. However, the scan rate (R in arcsec s−1) is limited to 5 arcsec s−1. The maximum number of scanned pixels is 200, although fewer are used if the count rate permits and a lower scan rate is needed. In order to avoid approaching the non-linearity regime of the detector, we limit the exposure to <60% of the maximum well depth. A nominal HST orbit length of 2700 s is assumed. Together with the optimal exposure time, we include the nominal overhead time (60 s), scan return time (5 s) and the readout time (0.3 s).
The S/N estimates are based on the expected atmospheric signal as described in equation (7). The noise estimates are derived by binning over the entire 1.4 μm H2O feature using a passband of 0.2 μm. Sensitivity estimates are made for two scenarios, one corresponding to circumstances similar to 55 Cnc e, a super-Earth in orbit around a G8 V star (top panel in Fig. 4), and one similar to GJ 1214b, a super-Earth in orbit around an M star (bottom panel in Fig. 4). Our sensitivity simulations match the actual noise levels of the first observations of transiting exoplanets using the spatial scanning mode (Deming et al. Reference Deming2013).
In calculating the signal for each scenario, we adopt the planetary radius and gravity and the stellar spectrum of the prototype, and we vary the equilibrium temperature (i.e. semi-major axis) of the planet and the stellar brightness. Both planets, GJ 1214b and 55 Cnc e, are identified, and shown with the sample of known super-Earth (i.e. M sin i < 10) exoplanets with host stars that have the same approximate spectral type (4900 K < T eff < 5600 K, roughly G5 to K2 for the 55 Cnc example, and T eff < 3800 K, roughly any M star for the GJ 1214 example). Those exoplanets that are known to transit are shown with a large open circle surrounding the filled symbol. In the case of GJ 1214, it is the only known transiting exoplanet that fits these criteria.
Ostensibly, our estimates agree with the common notion that brighter and/or smaller host stars greatly enhance the detectability of super-Earth atmospheres. While a higher planetary temperature around a larger star, as in the case of 55 Cnc e, can also enhance the transit signal, the dependence of the signal on the stellar radius is stronger, as shown in equation (7). For example, in these calculations, we estimate an S/N ~ 21.5 in the 1.4 μm H2O feature to be obtained in observations of ten transits of GJ 1214b (S/N = 1, 3, 10 and 30 contours are shown), whereas for 55 Cnc e, ten transits results in an estimated S/N of ~6.4 (S/N = 0.03, 0.1, 0.3, 1 and 3 contours are shown).
However, there are two additional important factors, beyond the nominal sensitivity estimates discussed above, that need to be considered while planning atmospheric observations of super-Earths. First, as discussed in subsection ‘Effect of clouds’, clouds can play a critical role in low-temperature atmospheres of super-Earths. In our sensitivity estimates discuss above, we have assumed spectral features of H2O as observed in a cloud-free atmosphere. However, the strength of the actual atmospheric signal is highly dependent on the presence or absence of clouds. Figure 4 shows condensation temperatures of some common species at a nominal 1-bar pressure (Lodders Reference Lodders2002; Sudarsky et al. Reference Sudarsky, Burrows and Hubeny2003; Morley et al. Reference Morley2013). As shown in the figure, a wide range of condensates are possible in super-Earth atmospheres, volatile species for temperatures below ~1000 K and refractory species for temperatures as high as ~2000 K. The presence of resulting clouds, if present at high-enough altitudes, can even completely mask the spectral features in a transmission spectrum leading to a featureless spectrum.
Consequently, atmospheres of low-temperature super-Earths (T eq ≲ 1000) can be challenging to observe and interpret even if the planet-star radius ratios and sensitivity estimates are highly favourable, as is the case for GJ 1214b, as discussed in subsection ‘The atmosphere of super-Earth GJ 1214b’. While super-Earths with T eq ~1000 − 2000 can still host clouds made of refractory condensates (silicates, Fe, etc.), the cloud-altitude would likely be lower compared to volatile condensates (e.g. NaCl, H2O, etc.) owing to more efficient gravitational settling due to heavier molecules. On the other hand, very high-T super-Earths with T eq ≳ 2000, such as 55 Cancri e, likely present the clearest atmospheres with the best potential for observations of their spectra. The super-Earth 55 Cancri e is particularly favourable given its extremely bright (V = 5.95; H = 4.27) host star.
The second factor is that the short periods of high-T eq super-Earths provide a much higher number of transit opportunities. Atmospheric characterization of any super-Earth with current instruments, such as the HST WFC3, require co-adding multiple transits to be able to make definitive detections of molecular features. Consequently, it is important to be able to schedule multiple transits for a dedicated programme to characterize super-Earth atmospheres. Short period planets are advantageous in this regard by offering a higher frequency of transit opportunities. For example, for 55 Cnc e, ~495 transits occur in a year, compared to 230 transits of GJ 1214b in a year.
Case studies
The small sizes of super-Earths mean that observing their atmospheres requires particularly favourable conditions that either enhance the signal or improve the precision. For a given planet size, the transit and eclipse depths are larger for smaller host stars and hotter planets. On the other hand, a brighter host star yields better precision in the observations. While GJ 1214b orbits a small star (an M dwarf), 55 Cancri e orbits an extremely bright (V = 6) Sun-like star in a very close (18 h) orbit, because of which both these super-Earths are great candidates for atmospheric studies. In what follows, we discuss the atmospheric constraints for these two planets possible with observations using existing facilities.
The atmosphere of super-Earth GJ 1214b
The super-Earth GJ 1214b (Charbonneau et al. Reference Charbonneau2009) is one of the most observed exoplanets to date. The planet, with a mass (M p) of 6.47M ⊕ and a radius (R p) of 2.68R ⊕ orbits a late M dwarf (R s = 0.21
${R_ \odot} $, T eff = 3030 K), resulting in a large transit depth. Consequently, despite the relatively faint host star (V = 15.1) and a relatively low equilibrium temperature (T eq~550 K), the planet is particularly suitable for transit observations and atmospheric characterization.
Constraints from transmission spectroscopy
Transmission spectroscopy and photometry of the planet's atmosphere at the day – night terminator have been reported over a wide spectral baseline ranging from the visible to mid-IR (~ 0.6 − 5 μm) (e.g. Bean et al. Reference Bean, Miller-Ricci Kempton and Homeier2010, Reference Bean2011; Croll et al. Reference Croll2011; Désert et al. Reference Désert2011; Berta et al. Reference Berta2012; de Mooij et al. Reference de Mooij2012; Kriedberg et al. Reference Kreidberg2014). The observed bandpasses, from a wide array of facilities from ground and space, encompass spectral features of prominent molecules such as H2O and CH2, as well as continuum bandpasses or opacity windows (see subsection ‘Surface radius of a super-Earth’). Several modelling efforts in the recent past have aided in the interpretation of the observed spectra (see e.g. Miller-Ricci & Fortney Reference Miller-Ricci and Fortney2010; Bean et al. Reference Bean2011; Benneke & Seager Reference Bennekke and Seager2012; Kepmton et al. Reference Kempton, Zahnle and Fortney2012; Howe & Burrows Reference Howe and Burrows2012; Morley et al. Reference Morley2013).
The sum total of data shows that the transmission spectrum of the planet is consistent with a flat horizontal line over the entire wavelength range observed to date, as shown in Fig. 5. The flat transmission spectrum of GJ 1214b rules out a cloud-free atmosphere for several plausible compositions, e.g. dominated by H2, H2O, CO2 or CH4 (Kreidberg et al. Reference Kreidberg2014). However, a featureless transmission spectrum is also consistent with a cloudy atmosphere, of unconstrained composition including a solar abundance H2-rich composition. In this scenario, the presence of clouds aloft in the atmosphere could be blocking the starlight thereby masking out any molecular absorption features due to the planetary atmosphere. Currently available data are, therefore, inconclusive about the true atmospheric composition of GJ 1214b.
Constraints from internal structure models
The observed mass and radius of GJ 1214b, together with internal structure models, suggest that a cloud-free H2O-rich atmospheric composition is unlikely for this planet. As discussed in section ‘Constraints from mass and radius’, GJ 1214b can be classified as an SE3-type super-Earth with its radius, R p, being larger the radius 
$({R_{{{\rm H}_2}{\rm O}}})$ of a pure-H2O planet with the same mass. As such, the radius differential between R p and 
${R_{{{\rm H}_2}{\rm O}}}$ can be used to place an upper limit on the mean molecular mass of the planetary atmosphere, as given by equation (2). Considering the parameters of GJ 1214b, we find a μmax of 2.0 ± 0.7 amu, consistent with a H2-rich atmosphere. Consequently, it is less likely that the observed radius of GJ 1214b can be explained solely by a cloud-free atmosphere with a high μ, e.g. H2O-rich, in agreement with what is already known from the transmission spectrum as discussed above.
A low atmospheric μ for GJ 1214b would also be consistent with internal structure models which suggest a light-element (H/He) composition for the planetary envelope (Rogers & Seager Reference Rogers and Seager2010b; Valencia et al. Reference Valencia, Guillot, Parmentier and Freedman2013). As shown in Fig. 1, the radius of the planet is higher than the R p,max for a pure-H2O planet with a H2O-rich atmosphere for the mass of GJ 1214b and its T eq of ~550 K. We find that a purely H2O – ice interior of GJ 1214b would require N sc ≳ 75 scale heights of a H2O-rich atmosphere to explain the radius, which is physically implausible, and an even higher N sc for other gases such as CO2 or N2. Consequently, as discussed above, a significantly lighter element than H2O, such as an H-rich atmosphere would be required to explain the radius. Our interpretation is consistent with the results of Rogers & Seager (Reference Rogers and Seager2010b) who also suggested the requirement of an H/He envelope in the planet to explain its mass and radius. Although our results rule out a cloud-free H2O-rich atmosphere in GJ 1214b, a H2O-rich lower atmosphere with a lighter species in the upper atmosphere together with a very high-altitude cloud/haze cover, cannot be conclusively ruled out in the present work.
Constraints from thermal emission spectra
Observations of thermal emission from GJ 1214b have led to only nominal constraints on its atmospheric composition. Given the low temperature of the planet, thermal emission from GJ 1214b is challenging to observe with existing instruments. Recently, Gillon (Reference Gillon2014) reported upper-limits on thermal emission in the Spitzer photometric bands at 3.6 and 4.5 μm. The data are consistent with conclusions derived from transmission spectroscopy of GJ 1214b, as discussed in subsection ‘Constraints from transmission spectroscopy’. The data rule out a cloud-free H2-rich composition in the dayside atmosphere of the planet. However, the data are consistent with a H2O-rich atmosphere as well as a cloudy H2-rich atmosphere. Additionally, the data are also consistent with a blackbody spectrum with a temperature of 500–600 K, indicating the possibility of an isothermal temperature structure with unconstrained chemical composition. Consequently, current observations of thermal emission from GJ 1214b do not provide any significant constraints beyond what is already known from extensive observations of transmission spectra of the planet.
New observations of thermal emission from GJ 1214b with existing facilities will be challenging, if not impossible. In the wavelength range (~1–2.3 μm) of current instruments in the near-IR, e.g. the HST WFC3 spectrograph and ground-based instruments, the predicted planet – star flux contrast in dayside thermal emission is below 40 ppm. Detecting such a signal, would require precisions better than 10 ppm, which current instruments are not likely to achieve, particularly given the faint host star (V = 15.1). In the future, however, the James Webb Space Telescope (JWST) will be able to detect such a weak signal.
The atmosphere of super-Earth 55 Cancri e
The super-Earth 55 Cancri e presents arguably the best chance for comprehensively characterizing a super-Earth atmosphere using both transmission as well as thermal emission spectroscopy. The planet has a mass of 8.4 M ⊕ and a visible radius of 2.0 R ⊕, and orbits a nearby G dwarf at a period of 18 h (Demory et al. Reference Demory2011; Winn et al. Reference Winn2011; Endl et al. Reference Endl2012). The parent star is the brightest star (V = 6) known to host a transiting exoplanet, and has led to measurements of the planet's radius at exquisite precision in the visible as well as in the Spitzer 4.5 μm IRAC band (Winn et al. Reference Winn2011; Demory et al. Reference Demory2012; Gillon Reference Gillon2012). Furthermore, due to its very short orbit, the planet has an equilibrium temperature of ~2000–2400 K, which leads to significant thermal emission, as has been observed in the Spitzer 4.5 μm IRAC photometric band, the first for any super-Earth (Demory et al. Reference Demory2012).
Constraints from internal structure models
55 Cancri e is an SE2-type super-Earth, as described in section ‘Constraints from mass and radius’, with its M p and R p lying between the mass – radius relations of a pure-silicate and a pure-H2O planet, as shown in Fig. 1. As such, the existence of a potential atmosphere cannot be conclusively constrained without multi-colour atmospheric observations (discussed in subsection ‘Constraints from transmission spectra’). However, the M p and R p of 55 Cancri e have led to two contrasting hypotheses for its interior composition, with different implications for its atmospheric composition. Considering a terrestrial-like O2-rich mineralogy, consisting of Fe, silicates and H2O, would require that the planet host a massive (≳10%) envelope of supercritical H2O (Valencia et al. Reference Valencia, Ikoma, Guillot and Nettelmann2010) in order to explain the R p, as suggested in several recent works (Winn et al. Reference Winn2011; Demory et al. Reference Demory2011, Reference Demory2012; Gillon . Reference Gillon2012). However, it is yet to be conclusively demonstrated if a massive H2O envelope of the planet would be stable against atmospheric escape (Valencia et al. Reference Valencia, Ikoma, Guillot and Nettelmann2010) and instability from nightside condensation (Castan & Menou Reference Castan and Menou2011; Heng & Kopparla Reference Heng and Kopparla2012) given the long age (10.2 ± 2.5 Gyr; von Braun et al. Reference von Braun2011) and extreme irradiation of the system. An alternate hypothesis suggests that the planet is C-rich, composed of Fe, C (as graphite + diamond), SiC, and silicates, without the requirement of any volatile envelope (Madhusudhan et al. Reference Madhusudhan2012).
The contrasting interpretations for the interior composition of 55 Cancri e suggest three possible compositions for its atmosphere. First, an O2-rich composition in the planet requires that the planet host a massive H2O envelope causing a H2O-rich atmosphere. Ehrenreich et al. (Reference Ehrenreich2012) place an upper-limit on the escape rate of hydrogen resulting from photodissociation of a H2O-rich atmosphere, and find it consistent with a stable H2O-rich atmosphere. Second, a C-rich composition would be unlikely to host a H2O-rich atmosphere, and may even host no atmosphere at all. Finally, the data can also be explained with a H2-rich atmosphere overlying an interior of any composition, O2-rich or C-rich. These three scenarios for the atmospheric composition of 55 Cancri e can be constrained using spectroscopic observations as discussed below.
Constraints from transmission spectra
Transmission spectroscopy of 55 Cancri e using HST WFC3 can constrain the atmospheric composition at the day – night terminator of the planet. Figure 6 shows our model transmission spectra of 55 Cancri e in the three possible atmospheric scenarios (H2-rich, H2O-rich and no atmosphere) and simulated observations assuming ten transits observed with the HST WFC3 as derived in section ‘Prospects with HST and optimal discovery space’. As shown in Fig. 6, the radii of 55 Cancri e previously measured in photometric bandpasses in the visible and Spitzer IRAC 4.5 μm band are consistent with any of the three possible compositions. On the other hand, WFC3 observations will be able to conclusively constrain all the three scenarios. As discussed in subsection ‘The atmosphere of super-Earth GJ 1214b’, previous studies have attempted to make a similar determination using HST WFC3 for the super-Earth GJ 1214b. However, due to the much lower temperature of GJ 1214b (~500 K) clouds of various compositions are possible (e.g. Morley et al. Reference Morley2013), making a conclusive determination of a H2O-rich atmosphere difficult for that planet. On the other hand, at the 2000–2400 K temperature of 55 Cancri e, clouds of any known composition are highly unlikely to exist, thereby making the interpretation of its transmission spectra substantially easier than that of GJ 1214b.
HST observations will be able to detect the presence of a H2O-rich atmosphere better than 6-σ in the water bands in the WFC3 bandpass, and the presence of a H2-rich atmosphere at better than 10-σ in the same bands. It is important to note that non-detection in the WFC3 bandpass, although highly unlikely, would be an extremely important result. Such a result would strongly imply the lack of an atmosphere in 55 Cancri e, for which the only known explanation to date is one of a C-rich interior, as discussed in subsection ‘Constraints from internal structure models’ Consequently, a flat spectrum across multi-wavelength observations, including WFC3, of 55 Cancri e can provide conclusive evidence for the lack of a H2- or H2O-rich atmosphere in the planet, without the ambiguities that plague similar efforts for GJ 1214b.
Constraints from thermal emission spectra
55 Cancri e is the only super-Earth for which thermal emission from the planet can be detected using existing instruments, given the extremely high dayside temperature (2400 K) of the planet. Figure 7 shows model thermal emission spectra of 55 Cancri e for the two possible atmospheric compositions, i.e. H2O-rich versus H2-rich based on solar abundances. We find that a precision of ~5 ppm, which is attainable with ten eclipses observed with HST WFC3, would be able to detect thermal emission in the opacity windows, i.e. from the planetary ‘surface’ (see subsection ‘Surface temperature and atmospheric constraints’), at ≳5-σ. The H2O feature in either scenario, which is given by the flux differential between the blackbody continuum, in the opacity windows, and the emission within the water band, can be detected at ≳ 4-σ for both scenarios. A non-detection of an H2O feature, i.e. the observation of a blackbody spectrum, would imply either the lack of an atmosphere, or an isothermal temperature profile. The degeneracy between the two solutions can be lifted by complementary observations of transmission spectra as discussed in subsection ‘Constraints from transmission spectra’.
A thermal emission spectrum of 55 Cancri e obtained with HST WFC3 can allow three specific constraints that are unprecedented for a super-Earth. First, one will be able to determine the thermal profile of the dayside atmosphere of the planet. A thermal profile decreasing with altitude will give rise to molecular absorption features, as seen in the deep water features in Fig. 7. Second, the difference between the continuum regions and strong water absorption or emission features in the WFC3 bandpass can be used to place joint constraints on the temperature gradient as well as the H2O-abundance in the atmosphere using detailed retrieval algorithms (e.g. Madhusudhan & Seager Reference Madhusudhan and Seager2009; Madhusudhan et al. Reference Madhusudhan2011; Lee et al. Reference Lee, Fletcher and Irwin2012; Line et al. Reference Line2012). Third, the spectral energy distribution of thermal emission can be used to constrain the day – night energy redistribution in the planetary atmosphere (Madhusudhan & Seager Reference Madhusudhan and Seager2009).
Discussion and summary
In the present work, we find that detection and characterization of hot super-Earths orbiting nearby stars provide a greater strategic advantage in the coming decade than detecting cool super-Earths close to the habitable zones of their host stars. A habitable planet, with temperatures in the vicinity of ~300 K to sustain liquid water on the surface, is not necessarily inhabited. At a minimum, such a determination would require characterizing the planet's atmospheric composition. However, the atmospheres of such cool planets will be challenging to detect even using the best facilities expected in the coming decade, including the JWST. Even when detected, the likely presence of clouds will confound the interpretation of spectra, as is already the case for the super-Earths GJ 1214b (with T~500 K) even with the most extensive observations with current facilities, including HST, Spitzer and major ground-based facilities. In the future, however, JWST is expected to be able to characterize super-Earths like GJ 1214b to greater precision, and more so for hotter planets which will have stronger atmospheric signatures, as discussed in the present work.
At the present time, a more important and tractable question than characterizing habitable exoplanets, is whether H2O is indeed a dominant constituent of super-Earth atmospheres. Addressing this question is independent of whether the planet is habitable or not. In fact, the best super-Earths that would be able to help answer this question would be transiting super-Earths with very high temperatures and orbit host stars that are either very bright, to allow better spectroscopic precision, or are small in size, to allow larger planet – star radius ratio and flux ratio. Currently, three super-Earths are known to transit such favourable host stars, albeit with a wide range of temperatures: GJ 1214b (T eq ~550 K; Charbonneau et al. Reference Charbonneau2009), HD 976548 (T eq ~900 K; Dragomir et al. Reference Dragomir2013) and 55 Cancri e (Teq ~2000–2400 K; Winn et al. Reference Winn2011). Of these, 55 Cancri e offers the best chances for detecting H2O in both transmission and thermal emission as discussed in the present work. Such hot super-Earths form an ideal sample in which to investigate the possibility and chemical abundances of H2O-rich atmospheres in super-Earths.
Current and upcoming observational facilities are well suited to characterize H2O-rich atmospheres in hot transiting super-Earths. We demonstrate in the present work how multi-colour observations of H2O-rich super-Earths can constrain the surface radii and temperatures of super-Earths, as well as constrain their H2O abundances. Observations in the opacity windows, which probe the lower region (‘surface’) of a super-Earth atmosphere, are possible with a wide range of ground-based facilities in the near-IR spectral region, particularly the z (1.0 μm), J (1.2 μm), H (1.6 μm) and K (2.1 μm) bands. On the other hand, precise measurements in the H2O bands at 1.4 μm are possible with the HST WFC3 spectrograph. Similarly, molecular features of H2O and other possible molecules (e.g. CH4, CO and CO2) are also accessible with the warm Spitzer photometric bandpasses at 3.6 and 4.5 μm. In the future, JWST will have the capability to revolutionize atmospheric characterization of hot H2O-rich super-Earths with high-resolution spectroscopy over a much broader spectral range than is currently available. The prime super-Earths for characterization with JWST will be discovered in large numbers by upcoming surveys such as TESS (Ricker et al. Reference Ricker2014), CHEOPS (Broeg et al. Reference Broeg and Roberto2013), and PLATO (Rauer et al. Reference Rauer2013) from space, and by several ground-based efforts (e.g. Snellen et al. Reference Snellen, Stuik and Navarro2012; Gillon et al. Reference Gillon, Jehin, Fumel, Magain, Queloz and Saglia2013).
These new observational surveys would benefit from focusing on finding close-in transiting super-Earths orbiting bright stars. Very close-in super-Earths of SE2 and SE3 classes with T eq ≥ 2000 K would be particularly important, because the resulting planets could be successfully followed up with IR telescopes to search for H2O in their atmospheres as discussed above. In the present work, we simulated observations using the HST WFC3 spectrograph to identify parameters of super-Earth systems which would be most conducive for detecting their H2O-rich atmospheres. We consider two scenarios of super-Earth host stars, a G dwarf and an M dwarf. We generally conclude that hotter planets around brighter or smaller host stars are more conducive for detecting H2O-rich atmospheres. In principle, for M dwarf stellar hosts, super-Earths with T eq as low as 500 K, such as GJ 1214b, could still be favourable for H2O detection. However, the possibility of clouds in low-temperature super-Earth atmospheres, especially for T eq ≲ 1000 K, complicate the interpretation of H2O from spectra, as known from current observations of GJ 1214b. Consequently, very high-temperature super-Earths (T eq ≥ 2000 K) orbiting bright stars currently present the best chances for constraining H2O-rich atmospheres. Among currently known super-Earths, we find that the super-Earth 55 Cancri e (T eq ~2000 − 2400 K) presents the best chances for conclusively determining the presence of a H2O-rich atmosphere in a super-Earth using spectroscopy in both transmission and thermal emission using HST WFC3.
Ultimately, detecting molecules such as H2O and other bio-signatures in super-Earth atmospheres might be possible to detect for a subset of super-Earths with JWST, but for Earth-size planets the wait could be longer depending on the stellar hosts. In the meantime, however, a detailed survey of the abundances of H2O in the most favourable sample of the hot super-Earth population is possible with existing facilities as discussed above. The resulting constraints can help us estimate the likelihood of H2O-rich atmospheres in habitable planets that are being discovered in parallel but whose atmospheres cannot be characterized in the near future.
Acknowledgements
NM acknowledges support from the Yale Center for Astronomy and Astrophysics (YCAA) at Yale University for support through the YCAA postdoctoral prize fellowship. SR acknowledges support by the National Science Foundation through Astronomy and Astrophysics Research Grant AST-1313268. We thank Peter McCullough, Avi Mandell and Debra Fischer for helpful discussions. This research has made use of the Exoplanet Orbit Database and the Exoplanet Data Explorer at exoplanets.org.
