Julianne I. Moses C. Visscher J. J. Fortney A. P. Showman ...showman/publications/moses-etal...J. Preprint typeset using LATEX style emulateapj v. 8/13/10 DISEQUILIBRIUM CARBON, OXYGEN,

submitted to Astrophys. J.Preprint typeset using LATEX style emulateapj v. 8/13/10

DISEQUILIBRIUM CARBON, OXYGEN, AND NITROGEN CHEMISTRY IN THE ATMOSPHERES OFHD 189733b AND HD 209458b

Julianne I. MosesSpace Science Institute, 4750 Walnut Street, Suite 205, Boulder, CO, 80301, USA

C. VisscherLunar and Planetary Institute, Houston, TX, 77058, USA

J. J. FortneyDepartment of Astronomy and Astrophysics, University of California, Santa Cruz, CA, 95064, USA

A. P. Showman, N. K. Lewis, and C. A. GriffithDepartment of Planetary Sciences and Lunar and Planetary Laboratory, The University of Arizona, Tucson, AZ, 85721, USA

M. ShabramDepartment of Astronomy, University of Florida, Gainesville, FL, 32611, USA

A. J. FriedsonEarth and Space Sciences Division, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, 91109, USA

and

M. S. Marley, R. S. FreedmanNASA Ames Research Center, Moffett Field, CA, 94035, USA

submitted to Astrophys. J.

ABSTRACTWe have developed one-dimensional photochemical and thermochemical kinetics and diffusion mod-

els for the transiting exoplanets HD 189733b and HD 209458b to study the effects of disequilibriumchemistry on the atmospheric composition of “hot Jupiters.” The models transition smoothly fromthe thermochemical equilibrium regime in the deep troposphere to the transport-induced quenchingand photochemistry regimes in the upper atmosphere, allowing us to avoid some of the potential pit-falls associated with the time-scale arguments typically used to estimate the abundance of quencheddisequilibrium species in extrasolar planet atmospheres. As such, the models can be used to predictvertical profiles of atmospheric constituents over the entire spectrally active region of the atmosphere.Here we investigate the coupled chemistry of neutral carbon, hydrogen, oxygen, and nitrogen species,and we compare the model results with existing transit and eclipse observations. We find that thevertical profiles of molecular constituents are significantly affected by transport-induced quenchingand photochemistry, particularly on cooler HD 189733b; however, the presence of a stratosphericthermal inversion (such has been deduced for HD 209458b) can help maintain thermochemical equi-librium and reduce the effects of disequilibrium chemistry. For both HD 189733b and HD 209458b,the methane and ammonia mole fractions are found to be enhanced over their equilibrium values atpressures of a few bar to less than a millibar due to transport-induced quenching, but the CH4 andNH3 mole fractions drop off sharply at higher altitudes due to photochemical processes. Becausethe atmosphere of HD 189733b is cooler than that of HD 209458b, the disequilibrium abundancesof ammonia and methane are predicted to be greater on HD 189733b than on HD 209458b for oth-erwise similar assumptions about metallicity or diffusion coefficients. Atomic species, unsaturatedhydrocarbons (particularly C2H2), some nitriles (particularly HCN), and radicals like OH, CH3, andNH2 are enhanced over equilibrium predictions because of quenching and photochemistry. In con-trast, CO, H2O, N2, and CO2 more closely follow their equilibrium profiles, except at pressures ∼<1 microbar, where CO, H2O, and N2 are photochemically destroyed and CO2 is produced before itseventual high-altitude destruction. Synthetic spectra created from our photochemical model resultsshow that equilibrium assumptions are reasonable for predicting the spectrum of HD 209458b, butdisequilibrium chemistry alters HD 189733b transmission and emission spectra over what would bepredicted from thermochemical-equilibrium models. The enhanced abundances of HCN, CH4, andNH3 in particular are expected to affect the spectral signatures and thermal profiles HD 189733b andother, relatively cool, close-in transiting exoplanets. We examine the sensitivity of our results to theassumed temperature structure and eddy diffusion coefficient profiles and discuss further observational

2 Moses et al.

consequences of these models.Subject headings: planetary systems — planets and satellites: atmospheres — planets and satellites:

composition — planets and satellites: individual (HD 189733b, HD 209458b) —stars: individual (HD 189733, HD 209458)

1. INTRODUCTION

The detection and characterization of extrasolar plan-ets is one of the most exciting and fastest growing fieldsin astronomy (e.g., Charbonneau et al. 2007; Marley etal. 2007; Udry et al. 2007; Deming & Seager 2009; Baraffeet al. 2010). Of the handful of tools available for charac-terization of exoplanet atmospheres, transit and eclipseobservations have proven to be invaluable for inferringproperties of both the planet and its host star (e.g., Char-bonneau et al. 2000, 2002, 2005; Henry et al. 2000; Sea-ger & Sasselov 2000; Brown 2001; Brown et al. 2001;Burrows et al. 2001; Hubbard et al. 2001; Deming etal. 2005b; Knutson et al. 2007, 2009; Madhusudhan &Seager 2009). The wavelength dependence of the signalduring transit and eclipse observations can allow identi-fication of atmospheric constituents on “hot Jupiter” ex-oplanets (e.g., Charbonneau et al. 2002; Vidal-Madjar etal. 2003; Tinetti et al. 2007; Grillmair et al. 2008; Swainet al. 2008b, 2009a; Madhusudhan & Seager 2009), allow-ing a better understanding of the chemical and physicalprocesses operating in these exotic atmospheres.

Hot Jupiters are often divided into two classes, thosethat appear to have stratospheric thermal inversions, andthose that do not (e.g., Fortney et al. 2008a; Hubeny etal. 2003; Harrington et al. 2007; Burrows et al. 2007,2008; Knutson et al. 2008; Spiegel et al. 2009; Mad-husudhan & Seager 2009). Although the ultimate rea-sons for these differences are still under investigation(see also Spiegel et al. 2009; Zahnle et al. 2009; Knut-son et al. 2010; Youdin & Mitchell 2010; Fressin et al.2010; Machalek et al. 2008, 2010; Madhusudhan & Sea-ger 2010a), exoplanets with thermal inversions appear tohave some source of high-altitude absorption, whereas ex-oplanets without thermal inversions are apparently miss-ing this stratospheric absorber, perhaps because it is tiedup in condensed phases (e.g., Hubeny et al. 2003; Fort-ney et al. 2008a; Burrows et al. 2008) or is destroyed byphotolysis and/or other mechanisms related to chromo-spherically active host stars (Knutson et al. 2010). Theobservations discussed in the the above citations sug-gest that HD 209458b has a strong stratospheric ther-mal inversion, whereas HD 189733b does not. These twotransiting planets, which have been the most extensivelystudied and characterized to date because of their brightparent stars and large transit depths, become convenientendmembers for our study of the chemistry of the twomain classes of hot Jupiters.

Constraints on the composition and thermal structureof HD 189733b and HD 209458b have been providedthrough (1) the identification of Na, H, and other atomicneutral and ionic species in the planets’ upper atmo-spheres (Charbonneau et al. 2002; Vidal-Madjar et al.2003, 2004, 2008; Ben-Jaffel 2007, 2008; Ballester et al.2007; Snellen et al. 2008; Redfield et al. 2008; Sing etal. 2008a; Ehrenreich et al. 2008; Langland-Shula et al.

Electronic address: [email protected]

2009; Linsky et al. 2010; Ben-Jaffel 2010; Lecavelier desEtangs et al. 2010), (2) the identification of molecularspecies such as H2O, CH4, CO2, and CO in the ∼1 barto 0.1 mbar atmospheric region (Tinetti et al. 2007; Fort-ney & Marley 2007; Burrows et al. 2007, 2008; Grillmairet al. 2008; Barman 2008; Swain et al. 2008a,b, 2009a,b,2010; Beaulieu et al. 2008, 2009; Désert et al. 2009; Mad-husudhan & Seager 2009), (3) observational and theo-retical inferences regarding the vertical and horizontaltemperature structure, including the suggested presenceof a stratospheric temperature inversion on HD 209458bbut not HD 189733b, and the identification of longitudi-nal temperature variations and phase lags (e.g., Knutsonet al. 2007, 2009; Cooper & Showman 2006; Fortney etal. 2006a, 2008a, 2010; Richardson et al. 2007; Burrowset al. 2007, 2008; Cowan et al. 2007; Showman et al.2008, 2009; Madhusudhan & Seager 2009, 2010a), and(4) the inferred presence of hazes, clouds, and Rayleighscatterers that can affect atmospheric spectral behavior(e.g., Charbonneau et al. 2002; Fortney et al. 2003; Fort-ney 2005; Iro et al. 2005; Pont et al. 2008; Lecavelier desEtangs et al. 2008a,b; Sing et al. 2008a,b, 2009; Shabramet al. 2011). Note, however, that the data reductionand observational analyses are often difficult due to lowsignal-to-noise, potential variable stellar activity of thehost star (including star spots), and/or poorly under-stood systematic instrument effects (e.g., Gibson et al.2010); spectral models are also poorly constrained, mak-ing analysis of the observations difficult. Various groupsdo not always agree, even when examining the same orsimilar datasets (cf. Désert et al. 2009 vs. Beaulieu etal. 2008 vs. Ehrenreich et al. 2007; Swain et al. 2008bvs. Sing et al. 2009 vs. Gibson et al. 2010; Stevensonet al. 2010 vs. Beaulieu et al. 2010). An additionalproblem with analyses is the lack of reliable molecularline parameters for hot bands of gases other than H2O,CO, and CO2. Even if the data reduction and analysiswere not inherently difficult, interpretation of thermalinfrared data can be hampered by the fact that molecu-lar bands can manifest as either absorption or emissionfeatures, depending on the atmospheric thermal struc-ture, creating a degeneracy between temperature profilesand compositions (e.g., Burrows et al. 2008; Madhusud-han & Seager 2009, 2010a; Tinetti et al. 2010; Fortneyet al. 2010; O’Donovan et al. 2010). Despite these com-plications, transit and eclipse observations represent ourbest current means for characterizing the atmospheres ofhot Jupiters. Theoretical models are needed to interpretthese observations.

Existing chemical models for transiting exoplanetstend to fall into two groups: thermochemical-equilibriummodels and photochemical models. Thermochemicalequilibrium is assumed for most models (e.g., Burrows &Sharp 1999; Seager et al. 2000; Lodders & Fegley 2002;Sharp & Burrows 2007; Burrows et al. 2007, 2008; Fort-ney et al. 2008a; Visscher et al. 2010a), although poten-tial departures from equilibrium are sometimes exploredthrough evaluations of simple time-scale arguments (e.g.,

Disequilibrium Chemistry on HD 189733b and HD 209458b 3

Lodders & Fegley 2002; Fortney et al. 2006a, 2008b; Viss-cher et al. 2006, 2010a; Madhusudhan & Seager 2010b)or more complex models with parameterized chemicalkinetics (Cooper & Showman 2006). Thermochemicalequilibrium is a reasonable starting point for exoplanetcomposition predictions, but the strong ultraviolet fluxincident on close-in transiting planets like HD 189733band HD 209458b can drive the observable upper regionsof the atmosphere out of equilibrium. Moreover, some ofthe existing transport and kinetics time-scale argumentshave known flaws (for further discussion, see Yung etal. 1988; Smith 1998; Bézard et al. 2002; Visscher et al.2010b; Moses et al. 2010; Visscher & Moses 2011) andare often misused, leading to questionable predictions ofquenched disequilibrium abundances.

The second group of chemical models exploits knowl-edge of Jovian photochemistry to predict the conse-quences of the large ultraviolet flux incident on exoplanetatmospheres (Liang et al. 2003, 2004; Yelle 2004; GarćıaMuñoz 2007; Koskinen et al. 2007, 2009; Zahnle et al.2009, 2011; Line et al. 2010). Of these latter models,Yelle (2004), Garćıa Muñoz (2007), and Koskinen et al.(2007, 2009) focus on the uppermost regions of the at-mosphere, i.e., the thermosphere, ionosphere, and exo-sphere and do not consider what goes on deeper than1 microbar. In contrast, Liang et al. (2003, 2004) andLine et al. (2010) explore the photochemistry of thedeeper stratosphere and troposphere using thermochem-ical equilibrium and/or transport-quenched disequilib-rium as a lower boundary condition for their models;however, their focus on low-temperature reactions (forthe former two studies) and their lack of fully reversedreactions (all three studies) prevents them from extend-ing the kinetic models into the thermochemical regime attemperatures ∼> 1000 K, and their assumptions can resultin artificial constant-with-altitude profiles for the the ma-jor species in the troposphere. Zahnle et al. (2009, 2011)have published the only models to date that have at-tempted to bridge the thermochemistry and photochem-istry regimes with a single model; however, their relianceon isothermal atmospheric models can result in someunrealistic vertical profiles and species abundances thatwill not be present with more realistic atmospheric struc-tures, and it is not clear whether their reaction schemewill reproduce equilibrium at high temperatures.

We therefore have developed a unique one-dimensional(1-D) thermochemical and photochemical kinetics andtransport model to circumvent some of the problemsdescribed above. The model has the ability to tran-sition seamlessly from the thermochemical equilibriumregime in the deep troposphere of giant planets to the“transport-quenched” disequilibrium and photochemicalregimes in the upper troposphere and stratosphere to il-luminate how chemical processes in each regime combineto influence the observational properties of giant-planetatmospheres. In an initial set of studies, we applied themodel to the deep troposphere of Jupiter to ensure thatthe model satisfies observational constraints on a well-studied planet within our own solar system (Visscher etal. 2010b; Moses et al. 2010). We now apply the modelto the atmospheres of HD 189733b and HD 209458b,as representatives of the two classes of hot Jupiters —those with and without stratospheric thermal inversions.Our main goal is to more realistically predict the ver-

tical profiles of the major observable carbon-, oxygen-,and nitrogen-bearing species on these close-in transitingplanets to aid the analysis and interpretation of primarytransit and secondary eclipse observations. By includ-ing both thermochemical and photochemical kinetics in atraditional 1-D chemistry/diffusion model, we can avoidsome of the shortcomings and pitfalls of previous chem-ical models — we do not need to correctly identify inadvance the mechanism and rate-limiting steps for trans-port quenching of species like CO, CH4, HCN, and NH3,we can avoid other uncertainties in the time-scale ar-guments such as the correct length scale to use for thetransport time scale (e.g., Smith 1998), we do not need torestrict the model to any particular pressure region, andwe can consider realistic atmospheric structures ratherthan isothermal atmospheres.

As we will show in the following sections, disequi-librium processes like transport-induced quenching andphotochemistry can perturb the composition of HD189733b and HD 209458b away from equilibrium. Theeffects of these disequilibrium processes are more sig-nificant for cooler exoplanets like HD 189733b than forwarmer exoplanets like HD 209458b, but there are po-tential observational consequences for both planets. Inparticular, transport-induced quenching and photochem-istry can enhance the abundance of atmospheric con-stituents like CH4, HCN, NH3, and C2H2, and the pres-ence of these disequilibrium molecules will affect spec-tral and photometric behavior. We compare the modelresults with existing transit and secondary eclipse ob-servations and discuss in detail the major disequilibriummechanisms affecting the abundance of carbon, nitrogen,and oxygen species on HD 189733b and HD 209458b.

2. MODEL

We use KINETICS, the Caltech/JPL 1-D photo-chemistry/diffusion code (Allen et al. 1981; see also Yunget al. 1984; Gladstone et al. 1996; Moses et al. 2000a,b,2005; Liang et al. 2003, 2004; Line et al. 2010) to calcu-late the vertical distributions of tropospheric and strato-spheric constituents on extrasolar giant planets. Thisfully implicit, finite-difference code solves the coupledmass-continuity equations as a function of pressure foreach species:

∂ni∂t

+∂Φi∂z

= Pi − Li (1)

where ni is the number density ( cm−3), Φi is the flux( cm−2 s−1), Pi is the chemical production rate ( cm−3s−1), and Li is the chemical loss rate ( cm−3 s−1) ofspecies i, all of which are explicit functions of time tand altitude z. The flux transport terms include bothmolecular and eddy diffusion, the latter parameterizedby an eddy diffusion coefficient Kzz. The calculationsare allowed to evolve until steady state is achieved, anda convergence criterion of one part in a thousand is used.We consider 198 atmospheric levels with a vertical reso-lution ranging from 30 grid points per atmospheric scaleheight in the deep atmosphere to 3 grid points per scaleheight in the upper atmosphere. A finer grid is neededin the deep atmosphere to accurately track the quenchpoints for some of the disequilibrium species. The modelextends from the deep tropospheres of HD 209458b (bot-

4 Moses et al.

tom boundary at 2754 K, 1 kbar) and HD 189733b (2600K, 1.67 kbar) to well past the homopause region intothe thermosphere (top boundary at pressures less than10−10 bar, exospheric temperature assumed to be 12,000K). The upper boundary is chosen at a high-enough al-titude such that we encompass the photochemical pro-duction and destruction regions for the main molecularspecies H2, H2O, CO, CH4, CO2, N2, NH3, and HCN forplausible assumptions about Kzz, and such that assump-tions about upper boundary conditions do not affect theresults. Although the models extend into the thermo-sphere, our intent is not to specifically model thermo-spheric chemistry. Instead, the thermospheric tempera-ture profile and escape boundary condition are includedin an ad hoc manner based on the results of Yelle (2004),Garćıa Muñoz (2007), and Koskinen et al. (2010) (seebelow for further details).

We use free-convection and mixing-length theories(e.g., Stone 1976; Flasar & Gierasch 1977) to define theeddy diffusion coefficients Kzz in the deep, adiabatic por-tion of the troposphere and estimate Kzz profiles throughthe bulk of the rest of the atmosphere from the root-mean-square (rms) vertical velocities derived from globalhorizontal averages at a given pressure level from the3-D GCM simulations of Showman et al. (2009), i.e.,by assuming Kzz = w(z)L(z), where w(z) is the hori-zontally averaged global rms vertical velocity from theGCM simulations and L(z) is assumed to be the atmo-spheric pressure scale height H(z) (but could be somefraction of H(z), see Smith 1998). Note that this proce-dure provides only a crude estimate of Kzz, and it wouldhave been preferable to have calculated Kzz from theeddy vertical velocity times the eddy displacement, butthis information was not available from the GCMs. Inparticular, the procedure may overestimate Kzz in the∼10-200 bar radiative region, where the vertical motiontends to consist of small-scale wave oscillations. In thefuture, it would be useful to include passive tracers intothe GCMs to directly and rigorously calculate the rate ofvertical mixing. Figure 1 shows the eddy diffusion coeffi-cient profiles adopted for our nominal models. However,given the above caveats, we consider Kzz profiles to beessentially free parameters in the models, and we exam-ine the sensitivity of our results to reasonable variationsin the adopted Kzz profiles. The molecular diffusion co-efficients are described in Moses et al. (2000a).

Temperature profiles are not calculated self-consistently within the KINETICS code but areinstead required as inputs to the model. The tempera-tures adopted in our model are shown in Fig. 2. Mostof these profiles derive from the 3-D GCM simulationsof Showman et al. (2009) for the region from 170 bar toa few µbar or the 1-D models of Fortney et al. (2006a,2010) for the region from 1000 to 1.0 × 10−6 bar.Because the nitrogen species in particular quench atvery high temperatures, we needed to extend the GCMprofiles to deeper levels (i.e., to reach temperaturesof at least 2500 K). The 1-D models of Fortney et al.(2006a, 2010) already include the deep adiabatic region,but the GCM profiles do not. For HD 189733b, the1-D “4π” profiles of Fortney et al. (2006a, 2010) thatassume the incident stellar energy is redistributed overthe entire planet compare very well with the GCM

Fig. 1.— Eddy diffusion coefficient profiles adopted for our nom-inal models, as estimated from rms vertical velocity profiles ob-tained from the GCMs of Showman et al. (2009) (see text). Forboth planets, the black line represents the profile for the dayside-average model, and the gray line represents the profile for theterminator-average model. The high-pressure value of Kzz = 1010

cm2 s−1 is an estimate of the eddy diffusion coefficient in the adia-batic region from free-convection and mixing-length theory (Stone1976), and Kzz ’s are assumed to be constant at altitudes abovethe top level of the GCMs.

theoretical thermal profiles in the 1–100-bar region, sowe have extended the GCM profiles to deeper pressuresby simply using the 1-D profiles. For HD 209458b,the 1-D “4π” profiles are somewhat warmer than theGCM profiles in the 1–100-bar region, and we shiftthe 1-D profiles by -60 K before connecting the GCMprofiles to this deeper adiabatic region. At low pressures(high altitudes), we extend the profiles by assumingeither an isothermal atmosphere or a high-temperaturethermosphere, using the thermospheric profiles of Yelle(2004) and Garćıa Muñoz (2007) as a guide for thelatter. Once the pressure-temperature profiles areestablished, the rest of the background atmospheric grid(e.g., densities, altitudes) is derived through solving thehydrostatic equilibrium equation.

HD 189733b orbits its host star at a distance of ∼0.031AU (e.g., Southworth 2010). The cool but chromospher-ically active HD 189733 has an effective temperature inthe range ∼5050–5090 K (Bouchy et al. 2005; Knutsonet al. 2010) and has typically been classified in the litera-ture as a K1V-to-K2V main-sequence star (e.g., Barneset al. 2010; Shkolnik et al. 2008), although the Hippar-cos catalog lists it as a G5V (Perryman et al. 1997; seealso Montes et al. 2001). In any case, HD 189733 is lessluminous than the Sun, and HD 189733b is not one ofthe hottest of the transiting exoplanets that have beendiscovered to date, despite its small orbital semimajoraxis. HD 189733 is part of a binary star system, but theM-dwarf companion to HD 189733 is ∼200 AU away andshould not significantly affect the flux incident on HD189733b. Because well-calibrated ultraviolet spectra forHD 189733 are available for only a specific range of wave-lengths (e.g., Lecavelier des Etangs et al. 2010), we setup a normalized 1-AU spectrum using the Hubble SpaceTelescope (HST ) STIS spectra of epsilon Eridani (a K2Vstar) from the CoolCAT database (Ayers 2005) for the1150–2830 Å region, after correcting for stellar distance.


Fig. 2.— Temperature profiles for our models of HD 189733b (left) and HD 209458b (right), based on the general circulation modelsof Showman et al. (2009) (solid lines) and on the 1-D radiative-convective models of Fortney et al. (2006a, 2010) (dotted lines), for anassumed solar-composition atmosphere. For HD 189733b, the models include a projected-area-weighted profile averaged over the daysidehemisphere (to best represent the planet’s apparent disk during the secondary eclipse), profiles averaged over the eastern and westernterminators (to best represent the ingress and egress regions during the transits), a terminator-average profile, and 1-D profiles calculatedassuming the incident stellar energy is distributed over the entire planet (curve labeled “4pi”) or over the dayside hemisphere only (curvelabeled “2pi”). For HD 209458b, the models include a straight (i.e., not projected-area) average over the dayside hemisphere, along witheastern and western terminator averages, a terminator average, and an average over the nightside hemisphere. Most profiles have beengiven an ad hoc high-temperature thermosphere at high altitudes based on the models of Yelle (2004) and Garćıa Muñoz (2007). Alsoshown are pressure-temperature curves (dashed gray lines) at which the major nitrogen species, N2 and NH3, and the major carbonspecies, CO and CH4, have equal abundances in thermochemical equilibrium. To the left-hand side of these CH4 = CO and NH3 = N2curves, the reduced species (CH4 and NH3) dominate, while the more oxidized species (CO and N2) dominate to the right. Note thatboth planets are within the CO and N2-dominated regimes over much of the atmosphere for all temperature profiles except for the coolestHD 189733b western-terminator-average profile. The gray dot-dashed lines represent the approximate condensation curves for the majorsilicates enstatite (MgSiO3) and forsterite (Mg2SiO4), based on Visscher et al. (2010a). A color version of this figure is available in theonline journal.

At the upper end of this wavelength range, the flux fromepsilon Eridani is much less than for typical G stars likeour Sun at the same stellar distance, whereas the fluxat the shorter EUV wavelength end of this range com-pares well to main-sequence G stars. We therefore usethe 1-AU solar flux divided by 10 for wavelengths longerthan 2830 Å, and we use the 1-AU solar flux for low-to-average solar conditions as an analog for our 1-AU-normalized HD 189733 flux for wavelengths below 1150Å. The photochemical calculations are then performedassuming that HD 189733b is located 0.031 AU fromthis K2V-analog star. The host star for HD 209458bis a G0V stellar type (Hipparcos, Perryman et al. 1997),and we simply use the solar flux as an analog for HD290458, and assume HD 209458b orbits at a distance of0.047 AU (e.g., Southworth 2010). The stellar zenithangle θ is fixed at 48◦ for the dayside-average models— where 〈cos θ〉 = 2/3 (θ ≈ 48◦) is the projected-area-weighted average of the cosine of the stellar zenith angleover the planetary disk at secondary eclipse conditions— and θ is fixed at ∼84◦–86◦ for the terminator models.Multiple Rayleigh scattering by H2 and He is consideredin the model using a Feautrier radiative-transfer method(Michelangeli et al. 1992).

We consider the neutral chemistry of 90 carbon-,hydrogen-, oxygen-, and nitrogen-bearing species: themodel contains H, H2, He, C, CH, excited singlet 1CH2,ground-state triplet 3CH2, CH3, CH4, C2, C2H, C2H2,C2H3, C2H4, C2H5, C2H6, C3H2, C3H3, CH3C2H,CH2CCH2, C3H5, C3H6, C3H7, C3H8, C4H, C4H2,C4H3, C4H4, C4H5, C4H6, C4H8, C4H9, C4H10, C5H3,C5H4, C6H2, C6H3, C6H4, C6H5, non-cyclic C6H6,

C6H6 (benzene), O, O(1D), O2, OH, H2O, CO, CO2,HCO, H2CO, CH2OH, CH3O, CH3OH, HCCO, H2CCO,C2H3O, CH3CHO, C2H4OH, HO2, H2O2, N, N2, NH,NH2, NH3, NNH, N2H2, singlet biradical H2NN, N2H3,N2H4, CN, HCN, H2CN, CH2NH, CH3NH, CH2NH2,CH3NH2, CH2CN, CH3CN, C3N, HC3N, C2H2CN,C2H3CN, NO, NO2, N2O, HNO, HNO2, NCO, HNCO,and CH3NO. Metals, rock-forming elements, and sul-fur and phosphorus species are not considered, nor ision chemistry. See Zahnle et al. (2009) for a discus-sion of the effects of sulfur photochemistry on exo-planet composition. The reaction list derives originallyfrom the Jupiter and Saturn models of Gladstone et al.(1996), Moses (1996), and Moses et al. (1995a,b, 2000a,b,2005), although extensive updates that account for high-temperature kinetics have been included based largelyon combustion-chemistry literature (e.g., Baulch et al.1992, 1994, 2005; Atkinson et al. 1997, 2006; Smith etal. 2000; Tsang 1987, 1991; Dean & Bozzelli 2000). SeeVisscher et al. (2010b) for further discussion of the hydro-carbon and oxygen kinetics and Moses et al. (2010) forfurther discussion of the nitrogen kinetics. The modelcontains ∼1600 reactions, with the rate coefficients of∼800 of the reactions being taken from literature values,and the remaining ∼800 reactions being the reverse ofthese “forward” reactions, with the rate coefficients ofthe reverse reactions being calculated internally at eachpressure-temperature point along the grid using the ther-modynamic principle of microscopic reversibility (e.g.,kfor/krev = Keq, where kfor is the rate coefficient for theforward reaction, krev is the rate coefficient for the re-verse reaction, and Keq is the equilibrium constant of the

6 Moses et al.

reaction; see Visscher & Moses 2011 for further details).The rate-coefficient expressions for the forward reactions(and their literature sources) are provided in Table S1 ofthe Supplementary Material. The rate coefficients for thereverse reactions for the pressure-temperature conditionsof certain of our exoplanet models can be found in thefull model outputs also presented in the SupplementaryMaterial.

We assume a 1× protosolar composition atmosphere(Lodders 2003, 2009; Lodders et al. 2009) in thermo-chemical equilibrium for our initial conditions for allthe constituents, using the NASA CEA code (Gordon& McBride 1994) for the equilibrium calculations. Ther-modynamic parameters are taken from Gurvich et al.(1989-1994), the JANAF tables (Chase 1998), Burcat &Ruscic (2005), or other literature sources. Note that al-though we do not consider rock-forming elements andtheir resulting effects on the chemistry of oxygen, we doassume that ∼20% of the oxygen has been sequesteredalong with silicates and metals (see Lodders 2004) in thedeep atmospheres of HD 189733b and HD 209458b, thusdepleting the available oxygen above the silicate clouds.As such, our “solar” abundance of oxygen is slightly lessthan that used by many other groups, and the speciesprofiles themselves should not be considered accuratewithin and below the silicate cloud condensation regionsdue to our neglect of the effects of rock-forming elements(see Fig. 2).

For boundary conditions, we assume zero flux at thetop and bottom boundaries so that no mass enters orleaves the system. The lower boundary in our models isat a high-enough temperature that thermochemical equi-librium will prevail, and a zero-flux boundary condition isreasonable. We also checked cases with a fixed-mixing-ratio lower boundary condition for all constituents, setat their equilibrium values, rather than a zero-flux lowerboundary condition and found no difference in the re-sults. The upper boundary in our model is at a pressureof a few times 10−5 µbar, which typically corresponds toa planetary radius less than 3 times the 1-bar radius Rp inexoplanet hydrodynamic models (and in some cases muchless than 3Rp, depending on the model, see Yelle 2004,Tian et al. 2005, Garćıa Muñoz 2007). At these high al-titudes, the hydrodynamic models and model-data com-parisons show the atmosphere of HD 209458b to be es-caping, with an overall mass loss rate that is low enoughthat exoplanets like HD 209458b would lose less than∼1% of their mass over their lifetime (e.g., Lecavelierdes Etangs et al. 2004; Tian et al. 2005; Murray-Clay etal. 2009). As such, we also test cases with an upper es-cape boundary condition for species like atomic H, witha typical upward flux of ∼ (6 − 60) × 1011 cm−2 s−1(e.g., 107–108 g s−1 for HD 209458b at a radius of 3Rp,see Yelle 2004, Garćıa Muñoz 2007, Tian et al. 2005, andKoskinen et al. 2010), or we investigate a range of possi-ble upper boundary escape fluxes from 1010 to 1016 cm−2s−1, but we find that this upper boundary condition hasno effect on the stratospheric or lower-thermospheric re-sults. However, we have not included a hydrodynamicwind connecting to this escaping region, which is themain reason the escape flux has little effect on our modelstratosphere. Hydrodynamic models (Yelle 2004; Tian etal. 2005; Garćıa Muñoz 2007) show that the atmosphere

should be close to hydrostatic equilibrium below 2-3Rp,with the base of the planetary wind usually being locatednear the 1-nbar level (see also the discussion in Koskinenet al. 2010). Thus, escape from a hydrodynamic or ther-mal wind is not expected to substantially influence thecomposition of the stratosphere, and hydrostatic equilib-rium is a reasonable assumption in our models. However,if an atmospheric bulk wind actually dominates down tothe base of the thermosphere (or deeper) for any particu-lar exoplanet, then hydrodynamic flow and atmosphericescape could conceivably affect the stratospheric compo-sition on such exoplanets.

3. RESULTS

Our model results suggest that the three main chemi-cal processes — thermochemical equilibrium, transport-induced quenching, and photochemistry — all oper-ate effectively in the atmospheres of HD 189733b andHD 209458b. Thermochemical equilibrium dominatesat pressures greater than a few bars, transport-inducedquenching can dominate for some species in the ∼1–10−3bar region, and photochemistry can dominate at pres-sures less than ∼10−3 bar, except when strong strato-spheric thermal inversions exist. All three processes com-bine to influence the vertical profiles of the important ob-servable constituents on HD 189733b and HD 209458b.As such, all three processes should be considered in inves-tigations of the chemistry and composition of extrasolargiant planets. Below we describe how the different dis-equilibrium processes affect the results, and we discussthe dominant mechanisms that control the abundancesof neutral carbon, oxygen, and nitrogen species in ourthermochemical and photochemical kinetics and trans-port models. The column abundances for some of theinteresting species in our models are shown in Table 1,and full model outputs are supplied in the Supplemen-tary Material.

3.1. Equilibrium vs. transport-induced quenchingvs. photochemistry

Fig. 3 illustrates how the individual disequilibrium pro-cesses affect the mole fractions of some species in ourdayside-average HD 189733b and HD 209458b models.The dashed curves show the thermochemical-equilibriumsolutions. Note that CH4 dominates over CO, H2O dom-inates over CO, and NH3 dominates over N2 at equilib-rium in the deepest regions of both atmospheres, butthe dominance reverses at higher altitudes such that COand N2 are the dominant carbon- and nitrogen-bearingspecies at pressures less than ∼10 bar (i.e., over mostof the observable portion of the atmosphere; see alsoFig. 2). At the highest altitudes in the HD 209458bmodel, atomic species C, N, and O are the dominantcarbon, nitrogen, and oxygen carriers. The same is nottrue of the HD 189733b model because the particularmodel shown in Fig. 3 assumes a cold, isothermal ther-mosphere (see Fig. 2); for the HD 189733b models with ahot thermosphere, atomic species also dominate at highaltitudes. Although H2O remains an important carrierof oxygen at all pressures on these planets, thermochem-ical equilibrium models predict that the mixing ratios ofCH4 and NH3 will drop off dramatically with increasingaltitude, such that the ratio of the column abundancesof CO/CH4 and N2/NH3 at 1 bar will be a couple orders


TABLE 1Column abundances above 1 bar

nominal Kzz nominal Kzz Kzz = 107 Kzz = 1011 nominal Kzz nominal Kzzterminator dayside dayside dayside dayside west term.

Species HD 189733b HD 189733b HD 189733b HD 189733b HD 209458b HD 209458b

H 2.1 × 1020 9.3 × 1020 1.2 × 1021 7.8 × 1020 2.2 × 1022 3.5 × 1021H2 1.0 × 1026 1.0 × 1026 1.0 × 1026 1.0 × 1026 2.3 × 1026 2.3 × 1026He 1.9 × 1025 1.9 × 1025 1.9 × 1025 1.9 × 1025 4.5 × 1025 4.5 × 1025C 1.6 × 1015 1.5 × 1016 8.2 × 1014 1.2 × 1014 7.8 × 1015 2.5 × 1015CH3 2.2 × 1017 2.9 × 1017 7.5 × 1016 4.3 × 1018 2.2 × 1017 1.4 × 1017CH4 5.9 × 1021 1.8 × 1021 4.6 × 1020 2.8 × 1022 1.4 × 1020 4.8 × 1020C2H2 7.7 × 1016 8.1 × 1016 1.6 × 1016 6.2 × 1018 4.6 × 1015 2.4 × 1015C2H4 2.0 × 1016 9.8 × 1015 6.2 × 1014 2.1 × 1018 2.1 × 1014 4.2 × 1014C2H6 7.1 × 1014 8.2 × 1013 5.3 × 1012 1.8 × 1016 4.2 × 1011 3.3 × 1012C3H2 2.0 × 1015 5.1 × 1015 1.5 × 1015 3.2 × 1015 1.6 × 107 1.1 × 1011C4H2 5.3 × 1012 3.4 × 1011 1.2 × 1010 8.3 × 1013 7.7 × 104 7.5 × 107C6H6 7.2 × 1012 2.7 × 1012 2.5 × 1012 3.0 × 1011 5.0 × 10−4 3.5O 2.8 × 1014 9.5 × 1014 6.9 × 1013 5.6 × 1014 6.0 × 1016 3.0 × 1015OH 1.6 × 1015 8.5 × 1015 7.9 × 1015 1.4 × 1016 5.4 × 1017 4.7 × 1016H2O 4.8 × 1022 4.3 × 1022 4.1 × 1022 7.3 × 1022 9.3 × 1022 9.4 × 1022CO 4.9 × 1022 5.4 × 1022 5.5 × 1022 2.4 × 1022 1.3 × 1023 1.3 × 1023CO2 1.6 × 1019 1.2 × 1019 1.2 × 1019 8.7 × 1018 1.9 × 1019 2.8 × 1019HCO 1.6 × 1014 4.7 × 1014 4.8 × 1014 2.1 × 1014 5.0 × 1015 1.8 × 1015H2CO 9.7 × 1015 1.0 × 1016 1.1 × 1016 4.5 × 1015 2.3 × 1016 2.4 × 1016CH3OH 5.2 × 1013 3.4 × 1013 9.5 × 1012 8.4 × 1014 9.4 × 1012 1.3 × 1013N 2.7 × 1014 3.7 × 1014 1.3 × 1012 4.6 × 1014 2.7 × 1015 1.1 × 1015N2 7.4 × 1021 7.6 × 1021 8.2 × 1021 4.9 × 1021 1.9 × 1022 1.9 × 1022NH2 7.1 × 1015 2.4 × 1016 2.9 × 1015 8.6 × 1016 4.1 × 1016 9.7 × 1015NH3 1.5 × 1021 1.1 × 1021 1.3 × 1020 3.8 × 1021 1.7 × 1020 2.1 × 1020HCN 1.6 × 1020 1.4 × 1020 6.6 × 1018 2.8 × 1021 1.5 × 1019 1.5 × 1019H2CN 1.2 × 1013 1.8 × 1013 9.3 × 1011 3.9 × 1014 5.9 × 1012 2.9 × 1012CH2NH 1.7 × 1015 9.3 × 1014 4.6 × 1013 2.0 × 1016 6.0 × 1013 9.8 × 1013CH3CN 5.8 × 1014 1.9 × 1014 2.3 × 1012 6.1 × 1016 9.5 × 1011 2.5 × 1012NO 1.4 × 1014 3.1 × 1014 6.5 × 1012 4.8 × 1014 6.5 × 1015 2.9 × 1014HNCO 3.7 × 1015 4.2 × 1015 5.3 × 1014 6.9 × 1015 1.2 × 1015 9.4 × 1014

Note. — Model column abundances have units of cm−2; Kzz ’s have units of cm2 s−1. “Terminator” refers to a terminator-

average, “dayside” refers to a dayside-average, and “west-term” refers to the western (colder) terminator temperature profile forthe planet in question from the GCMs of Showman et al. (2009). Our “nominal” Kzz profiles are shown in Fig. 1.

of magnitude. Thermochemical equilibrium predicts aneven smaller column abundance for species like HCN andC2H2, such that these species would be at most minorconstituents on HD 189733b and HD 209458b if theiratmospheres were in equilibrium.

The situation changes significantly when vertical trans-port is considered. The dotted curves in Fig. 3 showthe results of models in which thermochemical kineticsand vertical transport are operating but in which theplanet receives no photolyzing ultraviolet radiation fromthe host star, i.e., so that we can examine the influenceof transport-induced quenching separately from that ofphotochemistry. At deep pressure levels in these atmo-spheres, temperatures are high enough that kinetic en-ergy barriers to reactions can be overcome; reactions pro-ceed in both the forward and reverse directions, and equi-librium is maintained. Our kinetics models, which havefully reversed reactions, reproduce equilibrium predic-tions in these high-temperature, high-pressure regions.However, as a gas parcel is lifted to cooler, higher alti-tudes, kinetic energy barriers can begin to become sig-nificant, and exothermic “forward” reactions can becomefavored over their endothermic reverses. When the trans-port time scale drops below the chemical kinetic con-version time scale between different molecular species,the species can become quenched such that their mole

fractions become “frozen in” at values representative ofthe quench point (i.e., where the two time constants areequal). Prinn & Barshay (1977) first developed this con-cept analytically to explain the unexpectedly large COabundance in Jupiter’s upper troposphere. Based on COand other observations, transport-induced quenching ap-pears to be ubiquitous in substellar objects, operatingon our solar-system giant planets (e.g., Prinn & Barshay1977; Lewis & Prinn 1980; Prinn & Olaguer 1981; Lewis& Fegley 1984; Fegley & Prinn 1985; Fegley & Lodders1994; Lodders & Fegley 2002; Bézard et al. 2002; Viss-cher & Fegley 2005; Visscher et al. 2010b; Moses et al.2010), on brown dwarfs (e.g., Fegley & Lodders 1996;Noll et al. 1997; Griffith & Yelle 1999; Saumon et al.2000, 2003, 2006; Leggett et al. 2007; Hubeny & Burrows2007; Geballe et al. 2009; Yamamura et al. 2010; King etal. 2010; Visscher & Moses 2011), and most likely on ex-trasolar planets, as well (e.g., Cooper & Showman 2006;Fortney et al. 2006a; Burrows et al. 2008; Line et al. 2010;Madhusudhan & Seager 2010b and references therein).

The dotted curves in Fig. 3 show that transport-induced quenching can greatly increase the predictedabundances of several key species, including CH4, NH3,and HCN, over expectations based on thermochemicalequilibrium (dashed curves). The quench point for a ma-jor species like CH4 is caused by the breakdown of effi-

8 Moses et al.

Fig. 3.— Mole fractions for several species in our models for HD 189733b (left) and HD 209458b (right) for different assumptions about thechemistry and transport. All HD 189733b models assume a projected-area-weighted dayside average thermal structure with an isothermalextension at high altitudes (i.e., with a cold thermosphere), and all HD 209458b models assume a dayside-average thermal structurewith a hot thermosphere (see Fig. 2 for the temperature profiles for these models). The dashed lines are for a purely thermochemicalequilibrium model (no kinetics or transport), the dotted lines are for a model that has thermochemical kinetics and transport but noincident ultraviolet photons (i.e., no photochemistry), and the solid lines represent our full model with thermochemical and photochemicalkinetics and transport all operating. The eddy diffusion coefficient Kzz is fixed at a constant value of 109 cm2 s−1 in the transport models.A color version of this figure is available in the online journal.


Fig. 3. (continued).

10 Moses et al.

Fig. 3. (continued).

cient kinetic interconversion between itself and anothermajor species (CO in this case) and results in an ob-vious departure from the predicted equilibrium curve.Once an abundant species like CH4 becomes quenched,its quenched abundance can affect other minor specieslike C2Hx that might not have reached their own quenchpoints yet; these species may continue to react withthe CH4 at higher altitudes in an equilibrium with thequenched CH4 until they too reach their own quenchpoints. Thus, minor species may depart from equilibriumwhen a major parent molecule quenches, but their molefractions may not become constant until they themselvesquench at higher, cooler altitudes. As such, the molefractions from the full kinetics and transport model canhave complicated behavior that is not easily predictedfrom simple time-constant arguments using a single rate-limiting kinetic reaction. Ammonia, for example, has twoquench points in the HD 189733b model. Ammonia firstdiverges from the equilibrium curve at the N2-NH3 inter-conversion quench point, which is at relatively high tem-peratures and pressures due to the strong N N triplebond. However, NH3 remains in a net equilibrium withCH4 (i.e., CH4 + NH3 ↔ HCN + 3H2) that continuesuntil cooler temperatures initiate a final quench point.

The details of the particular quenching mechanisms,

and their sensitivity to temperature and transport as-sumptions, will be discussed below. At this point, wesimply want to illustrate that transport-induced quench-ing can be an important process on extrasolar giant plan-ets that can allow disequilibrium species to be carriedto relatively high altitudes to potentially affect spec-tral behavior. Note that transport-induced quenchingis much more important on the cooler HD 189733b thanon HD 209458b. Although Fig. 3 shows that quenchingdoes occur in our dayside HD 209458b kinetics/transportmodels, the effectiveness of transport-induced quenchingon HD 209458b is limited by the presence of a strongstratospheric thermal inversion — as temperatures be-come hot enough in the stratosphere, thermochemicalkinetics drives the composition back toward equilibrium.When the inversion is weak or absent on HD 209458b(i.e., at the terminators or nightside), transport-inducedquenching becomes more important.

The stratospheric thermal inversion also affects theheight to which the molecules can be carried. The modelsfor both planets in Fig. 3 have the same eddy diffusioncoefficient, yet a comparison of the dotted-line modelsfor both planets shows that the species reach differentpressure levels on each planet. The difference is onlypartly due to the higher temperatures creating a lower


homopause level on HD 209458b (where the term “ho-mopause” refers to the pressure level at which the molec-ular and eddy diffusion coefficient are equal; above thehomopause, molecular diffusion acts to limit the abun-dance of heavier species, whereas below the homopausethe atmosphere can be well mixed). This point is demon-strated by a close comparison of CO and N2 on eachplanet, as these thermochemically stable species reachtheir homopause levels, yet they are carried to onlyslightly different pressures on each planet. The maincause of the limited penetration of molecular species tohigh altitudes on dayside HD 209458b is the high strato-spheric temperatures — species like CH4 and NH3 can-not survive the thermochemical kinetics at these temper-atures and are converted in to atomic C, N, and H, aswell as to the more stable species CO and N2.

Another interesting point to note from Fig. 3 is thatsome molecules like CO, H2O, N2, and CO2 are rela-tively unaffected by transport-induced quenching on ei-ther planet. Although CO, H2O, and N2 technicallyalso quench in the thermochemical kinetics and trans-port models, they are already the dominant species attheir quench points, and the quenching does not greatlyaffect their abundances. As we will show below, thisresult is somewhat dependent on assumptions aboutthe eddy diffusion coefficient and temperature structure.On HD 189733b, CO2 quenches eventually (see wheredotted curve diverges from dashed curve), but only atvery high altitudes, and its abundance follows equilib-rium at the pressure levels at which the observationsare the most sensitive (∼1–10−4 bar, see Fortney et al.2008a). Other species like hydrocarbons, nitriles, andmost nitrogen-bearing species are strongly affected bytransport-induced quenching.

A comparison of the solid curves with the dotted curvesin Fig. 3 shows the influence of photochemistry. Whenultraviolet photons are present, molecules like H2O, CO,and NH3 are photolyzed, initiating a slew of photochem-ical reactions that alter the chemistry and composition.As discussed by Liang et al. (2003), one main result ofthe photochemistry in the atmospheres of close-in extra-solar giant planets is the production of huge quantitiesof atomic hydrogen from photolysis of H2O, followed bycatalytic destruction of H2 from the reaction of molecularhydrogen with the photolytic product OH. Other similarcatalytic cycles are initiated deeper in the atmospherefrom NH3 photolysis and from reaction of H with CH4(see also Zahnle et al. 2011).

Because many molecules are vulnerable to attack byatomic hydrogen at these temperatures, the productionof atomic H sets the stage for the subsequent photo-chemistry. Some molecules like H2O, N2, and CO areefficiently recycled and are relatively stable against pho-tochemistry. As can be seen from Fig. 3, photochemistryonly affects the profiles of these species at very high alti-tudes, where visible and infrared transit and eclipse ob-servations have little sensitivity. Other molecules likeCH4 and NH3 have weaker bonds and/or are more re-active and less efficiently recycled. These molecules canbe removed from portions of the visible atmosphere, tobe replaced by photochemical products. The big win-ners in terms of photochemistry are atomic species suchas H (by far the dominant photochemical product), C,O, and N; some radicals like CH3, NH2, and OH; un-

saturated hydrocarbons like C2H2; and some nitriles likeHCN. Note that photochemistry (and thermochemistry,if the atmosphere is warm) can keep molecular speciesfrom reaching their homopause levels, as species can be-come kinetically destroyed faster than eddy diffusion cantransport them to high altitudes. Below we discuss thekinetic mechanisms in more detail.

3.2. Quench kineticsAs shown in Fig. 3 and discussed above, transport-

induced quenching is an important process in the atmo-spheres of extrasolar giant planets. For relatively warmplanets like HD 189733b and HD 209458b, the thermalstructure lies within the CO and N2 stability fields atthe temperatures and pressures where quenching occurs,so that the less abundant CH4 and NH3 are strongly af-fected by quenching. This situation differs from that ofcooler planets like GJ 436b (which is in the CH4 and N2stability fields, so that quenching of CO and NH3 canbe important) or from cold planets like Jupiter (which isin the CH4 and NH3 stability fields, so that quenchingof CO and N2 are important) or from brown dwarfs likeGliese 229b whose thermal structure crosses the CO =CH4 and N2 = NH3 boundaries in such a location thatquenching of all four molecules — CO, CH4, N2, andNH3 — could be important.

Methane in our exoplanet models is quenched when itcan no longer be kinetically converted to CO effectively.The dominant reaction scheme controlling CH4 → COconversion and methane quenching in our HD 189733band HD 209458b models turns out to be

H + CH4 → CH3 + H2CH3 + H2O → CH3OH + HCH3OH + H → CH3O + H2CH3O + M → H + H2CO + MH2CO + H → HCO + H2HCO + M → H + CO + M

Net : CH4 + H2O → CO + 3H2,(2)

where M represents any third body. Note that thisscheme is exactly the reverse of the dominant CO → CH4conversion scheme found to control quenching of CO onJupiter (Visscher et al. 2010b) and Gliese 229B (Visscher& Moses 2011), where that CO-quenching process beginswith H + CO + M → HCO + M and continues throughCH4 production via CH3 + H2 → CH4 + H.

The rate-limiting step for the CH4 → CO conver-sion process is the slowest reaction in the fastest overallconversion scheme; for the above scheme (2), the rate-limiting reaction is CH3OH + H → CH3O + H2. Therate coefficient for the total reaction CH3OH + H →products has been measured in the laboratory (Aders& Wagner 1971; Hoyermann et al. 1981; Vandooren &Van Tiggelen 1981; Cribb et al. 1992), and the productbranching has been studied through theoretical calcula-tions (e.g., Lendvay et al. 1997; Jodkowski et al. 1999;Carvalho et al. 2008). We adopt a rate coefficient of1.1 × 10−22 T 3.4 exp(−3640/T ) cm3 s−1 (T in K) forCH3OH + H → CH3O + H2 based on the ab initio cal-culations of Jodkowski et al. (1999). The overall scheme

12 Moses et al.

(2) has some uncertainty attached to it, largely due touncertainties in the individual rate coefficients for thedifferent product branches for the CH3OH + H reaction.In particular, although the CH3OH + H → CH3 + H2Opathway is much less important than the → CH2OH +H2 (dominant) and → CH3O + H2 pathways (see Nor-ton & Dryer 1989; Lendvay et al. 1997), the → CH3 +H2O pathway may still be a critical reaction that breaksC-O bonds bonds under extrasolar-planet conditions, asother possibilities such as CH3OH thermal decomposi-tion may be even slower (e.g., Visscher et al. 2010b). Weuse the Visscher et al. (2010b) estimate for the rate co-efficient for the reaction CH3OH + H → CH3 + H2Oof 9.4 × 10−9 exp(−12400/T ) cm3 s−1, based on Nor-ton & Dryer (1989) and Lendvay et al. (1997), and weassume microscopic reversibility (see Visscher & Moses2011 for details) to derive the rate coefficient of the re-verse reaction that appears in scheme (2). If we haveoverestimated the rate coefficient for CH3OH + H →CH3 + H2O, then CH3 + OH + M → CH3OH + M mayreplace the reaction CH3 + H2O → CH3OH + H as theC-O bond-forming reaction in scheme (2), but CH3OH+ H → CH3O + H2 may still be the rate-limiting stepof the overall process.

In any case, despite these uncertainties, the CH3OH+ H → CH3O + H2 reaction is much more likely to bethe rate-limiting step in the conversion of CH4 to CO onexoplanets than the reaction OH + CH3 → H2 + H2CO(i.e., the reverse of the CO quench reaction originallyproposed by Prinn & Barshay 1977), as the latter reac-tion is about three orders of magnitude too slow to bea likely player in any dominant conversion scheme (seeDean & Westmoreland 1987; Yung et al. 1988; Griffith& Yelle 1999; Bézard et al. 2002; Visscher et al. 2010b).Note also that the CH3O + M → H2CO + H + M re-action in scheme (2) — the reverse of which has beensuggested as the rate-limiting step in CO quenching onJupiter (Yung et al. 1988; Bézard et al. 2002) — is abouta factor of six faster than our rate-limiting step at thequench level in our HD 189733b and HD 209458b models.

For a species like CH4 with a single dominant quenchpoint, the transport time scale arguments of Prinn &Barshay (1977) can be used to predict the quenchedabundance of CH4 on exoplanets (see also the updatesproposed by Visscher & Moses 2011). With this pro-cedure, the quench point is located where the chemicalkinetic conversion time scale τchem is equal to the ver-tical transport time scale τdyn. Then, the mole fractionof the quenched species above the quench level is sim-ply the equilibrium mole fraction of the constituent atthat quench point. In our models, the chemical kineticconversion time scale for CH4 is

τchem =[CH4]

d[CH4]/dt

=[CH4]

1.1 × 10−22 T 3.4e(−3640/T )[CH3OH] [H](3)

where [X] represents the equilibrium concentration ofspecies X.

The transport time scale is τdyn = L2/Kzz, where Lis an effective length scale over which dynamical mixingoperates. As discussed by Smith (1998), L is generally

some fraction of the scale height H. For quenching ofCO on cold planets like Jupiter, L ≈ 0.1H, whereas forquenching of CH4 on close-in transiting hot Jupiters, Lis closer to 0.5H (see Visscher & Moses 2011). When weuse the procedure suggested by Smith (1998) to deriveL and when we calculate τchem as described in Eq. (3)for our rate-limiting step, we derive a quenched molefraction within ∼15% of our kinetics/transport modelresults for HD 189733b and HD 209458b. Therefore,as is also discussed by Bézard et al. (2002), Cooper &Showman (2006), Visscher et al. (2010b); Visscher &Moses (2011), and Moses et al. (2010), the transporttime-scale approach is valid provided that the Smith(1998) formalism for τdyn is used and provided that thekinetic rate-limiting step is a reasonable one. We cau-tion that the time-scale approach has been misused fre-quently in exoplanet and brown-dwarf literature, whereerrors in the calculation of the rate coefficients for therate-limiting steps have arisen due to incorrect rever-sals of three-body reactions (Griffith & Yelle 1999; Lineet al. 2010), or where numerous other misconceptionsabout rate-limiting reactions and their rate coefficientsand transport time scales are ubiquitous (see Bézard etal. 2002, Visscher et al. 2010b; Visscher & Moses 2011,or Moses et al. 2010 for a more detailed discussion of theproblem). Some of these problems have serious implica-tions with respect to conclusions about the strength of at-mospheric mixing, particularly on brown dwarfs (Troyeret al. 2007; Visscher & Moses 2011), where Kzz is likelyseveral orders of magnitude greater than the sluggish Kzz≈ 102–104 cm2 s−1 that has often been reported (e.g.,Griffith & Yelle 1999; Saumon et al. 2003, 2006; Leggettet al. 2007; Geballe et al. 2009).

The time-scale approach cannot be used for speciesthat have more than one quench point or that otherwiseexhibit complicated kinetic behavior. As an example,NH3 quenches very deep in our model atmosphere whenconversion between NH3 and N2 first becomes inhibited.Two mechanisms closely compete as the dominant mech-anism involved in NH3 → N2 conversion, and several oth-ers are waiting in the wings should these mechanisms beinoperative. The fastest scheme is

2 (NH3 + H → NH2 + H2)2NH2 → H2NN + H2H2NN → N2H2

N2H2 + H → NNH + H2NNH + M → N2 + H + M

H2 + M → 2H + MNet : 2NH3 → N2 + 3H2,

(4)

followed closely by

NH3 + H → NH2 + H2NH2 + NH3 → N2H3 + H2N2H3 + M → N2H2 + H + MN2H2 + H → NNH + H2NNH + M → N2 + H + M

Net : 2NH3 → N2 + 3H2.(5)


The rate-limiting step in scheme (4) is 2NH2 → H2NN+ H2 — a reaction that is also believed to be impor-tant in the combustion chemistry of nitrogen compounds(e.g., Dean & Bozzelli 2000). We adopt a rate coef-ficient of 2 × 10−3 T−3.08 exp(−1695/T ) for this reac-tion based on the QRRK calculations of Dean & West-moreland (1987) for pressures of 1 atm, and we ignorethe very slight pressure dependence for the reaction. Inscheme (5), the rate-limiting step is N2H3 + M → N2H2+ H + M, where we again take the rate coefficient fromthe QRRK calculations of Dean & Westmoreland (1987).Note that these two reaction schemes are somewhat spec-ulative and rely on rate coefficients derived from theoret-ical calculations rather than laboratory measurements.As a whole, the kinetics of nitrogen compounds is lesswell understood than that of the C-H-O system (seeMoses et al. 2010 for further discussion).

After conversion of NH3 to N2 is quenched (at ∼1960K in the dayside HD 189733b model), the NH3 mole frac-tion does not become constant with altitude because am-monia and other NHx species continue to react with hy-drocarbons in a net equilibrium between NH3 and HCN.The simple time-scale approach therefore gives a good in-dication of where NH3 first leaves the equilibrium curveat depth, but it does not provide a good estimate for theNH3 mole fraction several scale heights above the NH3–N2 quench point. The dominant scheme involved in NH3→ HCN conversion is

NH3 + H → NH2 + H2CH4 + H → CH3 + H2

NH2 + CH3 + M→ CH3NH2 + MCH3NH2 + H → CH2NH2 + H2

CH2NH2 → CH2NH + HCH2NH + H → H2CN + H2H2CN + M → HCN + H + M

H2 + M → 2H + MNet : NH3 + CH4 → HCN + 3H2.

(6)

Again, this scheme is the reverse of the dominant HCN→ NH3 conversion scheme derived for Jupiter’s deep tro-posphere (Moses et al. 2010), although the rate-limitingstep in the above scheme (6) for our exoplanet model-ing is CH3NH2 + H → CH2NH2 + H2. We adopt theDean & Bozzelli (2000) theoretically-derived rate coeffi-cient of 9.3 × 10−16 T 1.5 exp (−2750/T ) cm3 s−1 for thisrate-limiting reaction.

Note that HCN quenches once CH4 and NH3 are fullyquenched by schemes (2) and (6); however, kinetic pro-cesses continue to affect molecules like C2H2 throughoutthe atmospheric column, such that these molecules arenever truly quenched, although the C2H2 profiles in thekinetics/transport models do depart significantly fromequilibrium at pressures ∼< 0.1 bar. The dominant mech-anism controlling the interchange of C2H2 and CH4 inour models is

C2H2 + H + M→ C2H3 + MC2H3 + H2 → C2H4 + H

C2H4 + H + M→ C2H5 + M

C2H5 + H2 → C2H6 + HC2H6 + M → 2CH3 + M

2 (CH3 + H2 → CH4 + H)2H + M → H2 + M

Net : C2H2 + 3H2 → 2CH4 .(7)

Thermal decomposition of C2H6 is the rate-limiting stepin this scheme.

In all, transport-induced quenching can have a signifi-cant effect on the predicted abundances of major specieslike CH4, NH3, HCN, and C2H2 on hot Jupiters. Giventhe overall uncertainties in the thermal structure, eddydiffusion coefficients, kinetic rate coefficients, and effectsfrom horizontal transport, the quantitative predictionsfrom our kinetics and transport models should not betaken too seriously. However, the general behavior inthe models should be heeded — the column abundanceof these species will likely be enhanced over equilibriumpredictions (and in some cases significantly enhanced),and transport-induced quenching is more significant forcooler planets or atmospheric regions.

We now move on to discuss the effects of the photo-chemistry that is initiated when ultraviolet photons areabsorbed by atmospheric constituents.

3.3. Photochemistry of oxygen speciesFigure 4 shows vertical profiles of the major oxygen-

bearing species in our nominal thermo/photochemical ki-netics and transport models. Photolysis of the majorparent molecules CO and H2O drives much of the oxy-gen chemistry. Water photolysis in particular is a keycomponent of exoplanet photochemistry, as it sets up acatalytic cycle that destroys H2 and produces H:

H2O + hν → H + OHOH + H2 → H2O + HNet : H2 → 2 H,

(8)

where hν represents an ultraviolet photon. Water is re-cycled with this efficient scheme, but H2 is destroyed,resulting in a huge net production rate for atomic H,which then replaces H2 as the dominant atmosphericconstituent at high altitudes (see Fig. 4). This transi-tion from H2 to H likely signals a transition into thehigh-temperature thermosphere on hot Jupiters, as nopotential molecular coolants survive when atomic H be-comes the dominant constituent at high altitudes (seealso Garćıa Muñoz 2007). Atomic species therefore areexpected to dominate in the thermospheres of these close-in transiting planets (e.g., Yelle 2004; Garćıa Muñoz2007; Koskinen et al. 2010), which might help explainthe lack of obvious H2 emission in far-ultraviolet day-glow observations of HD 209458b (France et al. 2010).From a column-integrated sense, scheme (8) is signifi-cantly more important as a source of atomic H than di-rect H2 photodissociation. Atomic H flows downwardfrom its production region and remains at abundancesgreatly in excess of equilibrium until in reaches altitudesmarked by its mole-fraction minimum shown in Fig. 4,near 0.01-0.1 bar on both planets, at which point atomicH again follows its equilibrium curve to lower altitudes.

14 Moses et al.

Fig. 4.— Mole-fraction profiles for several oxygen-bearing species in our models for HD 189733b (left) and HD 209458b (right) fordayside-average (top) and terminator-average (bottom) conditions, assuming the nominal Kzz profiles shown in Fig. 1. A color version ofthis figure is available in the online journal.

Photolysis of CO to produce C + O and photolysis ofH2O to produce 2H + O or O(1D) + H2 also contributeto H2 destruction through schemes such as

CO + hν → C + OO + H2 → OH + H

OH + H2 → H2O + HNet : CO + 2H2 → C + H2O + 2H

and

H2O + hν → 2H + OO + H2 → OH + H

OH + H2 → H2O + HNet : 2H2 → 4H

and

H2O + hν → H2 + O(1D)O(1D) + H2 → OH + H

OH + H2 → H2O + H

Net : H2 → 2 H.

Reactions of atomic H with H2O and OH can partiallyrecycle the H2, but key reactions like 2H + M → H2do not operate effectively at low pressures, and the netresult is that the loss rate for H2 exceeds the productionrate at high altitudes. The H atoms produced in thesemechanisms diffuse both upward and downward, wherethey strongly affect the abundance of other species.

Although H2O photolysis operates effectively down tomillibar regions in these models and the atomic H pro-duced from the photolysis continues to attack the H2O,water is not permanently removed from this atmosphericregion because of the efficient recycling by the reactionOH + H2 → H2O + H. Note that rapid replenishmentby diffusion from other altitudes can also play an impor-tant role in maintaining H2O for our models with coolerstratospheric temperature profiles (see also Line et al.2010). As an example, water is in approximate chem-ical equilibrium at pressures greater than ∼0.1 bar inthe HD 189733b dayside-average model shown in Fig. 4;however, the transport time scales are shorter than thenet photochemical loss time scales for H2O in the regionfrom ∼0.1 bar to a few µbar, so that atmospheric mixing


“quenches” the H2O mole fraction and keeps it roughlyconstant through this region. Since the H2O mole frac-tion quenches at a value similar to the equilibrium valueand since equilibrium would keep the H2O mole frac-tion constant in this region anyway, the water profile forthe thermo/photochemical kinetics and transport modelis quite similar to the thermochemical equilibrium pro-file for our HD 189733b models. For our much warmerdayside HD 209458b model, equilibrium chemistry dom-inates the H2O profile over a much more expanded alti-tude range — transport is only important at high alti-tudes, and only if the eddy diffusion coefficient is large.This effect can be seen from detailed comparisons of theH2O profile from the HD 209458b dayside-average modelin Fig. 3 in which it is assumed that Kzz = 109 cm2 s−1,in contrast to the H2O profile in Fig. 4 from an otherwisesimilar HD 209458b dayside-average model in which it isassumed that mixing is more vigorous at high altitudes(see the nominal Kzz profile from Fig. 1). The Kzz =109 cm2 s−1 model has the H2O profile closely followingequilibrium, whereas our higher Kzz nominal model haswater being carried to higher altitudes before its even-tual chemical destruction. However, this additional H2Ofrom transport processes does not add much to the col-umn abundance of H2O on the planet. The net resulton both the cooler and warmer classes of planets is anH2O profile that closely follows the equilibrium profilethrough the observable regions of the atmosphere.

Carbon monoxide also survives in the atmospheres ofHD 189733b and HD 209458b despite rapid loss fromOH + CO → CO2 + H, from H + CO + M → HCO +M, and from photolysis by EUV photons that continuesdown to mbar levels. The first two reactions dominatethe column-integrated CO loss in the stratosphere; theCO2 produced in the first reaction efficiently reacts withH to reform CO, and the HCO produced in the secondreaction efficiently reacts with H to reform CO, so carbonmonoxide is readily recycled on both HD 189733b andHD 209458b. Even in the case of CO photolysis, somerecycling pathways such as the following can operate:

CO + hν → C + OC + H2 → CH + H

CH + H2O → H2CO + HH2CO + H → HCO + H2HCO + H → CO + H2Net : H2O → O + H2,

allowing CO to persist. However, these recycling schemesare not 100% effective, and some of the CO photodestruc-tion leads to the production of other species. Atomiccarbon and oxygen released from CO photolysis at highaltitudes can remain as C and O, for example, and someof the O can lead to H2O production through reactionslike O + H2 → OH + H followed by OH + H2 → H2O+ H, while some of the C can go toward producing hy-drocarbons and HCN (see below, and Liang et al. 2004;Line et al. 2010; Zahnle et al. 2011). As with H2O, theCO profiles are expected to closely follow the equilibriumprofiles, except at high altitudes, where the CO will beconverted to C, O, and H2O (i.e., for cooler-temperaturemodels) or where high eddy diffusion coefficients may

allow it to be transported to higher altitudes than equi-librium models predict (i.e., for our nominal dayside HD209458b model).

The photochemistry of CO2 is not particularly interest-ing, as CO2 remains in equilibrium with CO throughoutmost of the atmosphere, via the reaction scheme:

OH + CO → CO2 + HH + CO2 → OH + CO

Net : Nothing.

Some additional CO2 in excess of equilibrium is pro-duced at high altitudes from CO photochemistry on HD189733b and/or from transport on HD 209458b, but theCO2 profiles on both planets remain close to equilibriumpredictions.

Aside from atomic H, few photochemical productsbuild up from the chemistry of oxygen-bearing speciesin our hot-Jupiter models. Some O, OH, NO, and evenO2 molecules are produced at high altitudes and diffusedownward; however, the photochemical lifetimes of thesespecies are quite short, and they are not likely to achievecolumn abundances great enough to be detectable at vis-ible or infrared wavelengths with current technologiesunless the planet’s metallicity is enhanced greatly oversolar (see Fig. 4 and Table 1). Atomic O is producedfrom the reaction of H with OH to form O + H2, witha lesser contributions from H2O, OH, and CO photol-ysis. Atomic O is lost predominantly from O + H2 →OH + H. Because atomic O is not readily destroyed byH, it can be transported up to its homopause level inour HD 209458b models, unlike the situation for molec-ular species. Some fraction of the O column will thenmake it into the thermosphere on planets with vigor-ous vertical mixing, where high thermospheric temper-atures and hydrodynamic winds or other processes maythen distribute the atomic O over a large radial distance(Lecavelier des Etangs et al. 2004; Yelle 2004; Tian et al.2005; Garćıa Muñoz 2007; Penz et al. 2008; Murray-Clayet al. 2009; Lammer et al. 2009), which will increase thelikelihood of its detection at ultraviolet wavelengths (e.g.,see Vidal-Madjar et al. 2004; Ben-Jaffel 2010; Koskinenet al. 2010). Even in cases where Kzz is small in theupper atmosphere, the dynamical wind from the hydro-dynamic escape — which is not included in our models— may then dominate over molecular diffusion in con-trolling the behavior of heavy atmospheric constituents,so that the concept of a homopause may not actuallybe appropriate for close-in transiting planets. The dom-inant OH production mechanism is H + H2O → OH +H2, with the atomic H coming from the H2 catalytic de-struction scheme (8) described above. The reverse of thisreaction is its dominant loss mechanism. In general, thecooler the stratosphere, the larger the increase in the OHabundance over equilibrium predictions.

The main production mechanism for NO is N + OH→ NO + H. The OH derives from water, and the N cancome from either N2 or NH3 via schemes such as thefollowing:

H2O + hν → OH + HN2 + H → N + NHNH + H → N + H2

16 Moses et al.

N + OH → NO + HNet : H2O + N2 → N + NO + H2

and

H2O + hν → OH + HNH3 + H → NH2 + H2NH2 + H → NH + H2NH + H → N + H2N + OH → NO + H

Net : H2O + NH3 + H → NO + 3H2.

The NO is lost through reactions with C to form N +CO and/or CN + O, through photolysis to form N + O,and through reaction with atomic H. Molecular oxygenis produced through the reaction O + OH → O2 + H,and is lost through the reverse of this reaction.

As a general rule of thumb, photochemistry inthese warm, highly irradiated stratospheres favors smallmolecules and/or molecules with strong bonds. Our re-sults are consistent with other exoplanet photochemicalmodels (Liang et al. 2003, 2004; Line et al. 2010; Zahnleet al. 2011) in that we see little photochemical productionof molecules like H2CO, CH3OH, H2CCO, or CH3CHO.

3.4. Photochemistry of carbon speciesThe photochemistry of carbon species is more inter-

esting than that of oxygen species, as the primary par-ent molecule CH4 has weaker bonds. Several carbon-bearing photochemical products can build up in the at-mospheres of extrasolar giant planets. Figure 5 showsvertical profiles of the major carbon species in our nom-inal thermo/photochemical kinetics and transport mod-els. Carbon photochemistry is initiated by CO photol-ysis at high altitudes and by hydrogen abstraction frommethane at lower altitudes. At pressures less than ∼1µbar, much of the C that is released from CO photolysisremains as atomic C, although some acetylene and otherhydrocarbon production can occur through schemes suchas

2 (CO + hν → C + O)2 (C + H2 → CH + H)CH + H2 → 1CH2 + H

1CH2 + H2 → CH3 + HCH + CH3 → C2H3 + HC2H3 + H → C2H2 + H2

2 ( 2H + M → H2 + M)Net : 2CO + H2 → C2H2 + 2O.

(9)

At greater pressures (and in particular where CH4 orCH3 become more abundant than atomic C), acetyleneand other hydrocarbons are produced from methane de-struction through pathways such as

2 (H2O + hν → H + OH )2 (OH + H2 → H2O + H)2 (CH4 + H → CH3 + H2 )

CH3 + H → 1CH2 + H21CH2 + H2 → 3CH2 + H2

3CH2 + CH3 → C2H4 + HC2H4 + H → C2H3 + H2C2H3 + H → C2H2 + H2

Net : 2CH4 → C2H2 + 3H2.(10)

Note the importance of H2O photolysis in providing theatomic H used to destroy the CH4. These pathways areefficient enough that C2H2 becomes the dominant hydro-carbon at high altitudes under certain conditions in ourmodels (see Fig. 5).

Acetylene is more effectively hydrogenated with in-creasing pressure, such that the following scheme actsto remove the C2H2 and recycle the methane at middle-stratospheric pressures:

C2H2 + H + M → C2H3 + MC2H3 + H2 → C2H4 + H

C2H4 + H + M → C2H5 + MC2H5 + H → 2CH3

2 (CH3 + H2 → CH4 + H)Net : C2H2 + 3H2 → 2CH4.

(11)

This scheme, rather than photolysis, is responsible forremoving C2H2 from the middle and lower stratosphere,as the C2H2 photolysis products tend to simply recyclethe C2H2.

Other C2Hx hydrocarbons are less able to survive thelarge background H abundance. Ethane is produced inthe stratosphere through 2CH3 + M → C2H6 + M andthrough C2H5 + H2 → C2H6 + H. Ethane is lost throughhydrogen abstraction by atomic H (C2H6 + H → C2H5+ H2), followed by H + C2H5 → 2CH3, and eventualmethane production. The column abundance of photo-chemically produced ethane is quite small in our models.Ethylene (C2H4) fares a bit better: it is produced pre-dominantly from C2H3 + H2 → C2H4 + H (see someof the C2H2 production schemes above) and is lost fromthe reverse of this reaction (heading toward C2H2 pro-duction), as well as H + C2H4 + M → C2H5 + M, toeventually form methane. Although the mole fraction ofC2H4 can exceed that of C2H2 at the C2H2 mole-fractionminimum near 0.1–10−2 bar in some of our models, thepeak C2H4 mole fraction never approaches that of thepeak C2H2 mole fraction, and the column abundance ofC2H4 is intermediate between that of C2H2 and C2H6 inour HD 189733b and HD 209458b models (see Table 1).

Although our kinetics is far from complete for C3-to-C6species, we do attempt to track the production and lossof some complex hydrocarbons. The dominant C3 hydro-carbon in our models is actually the C3H2 radical. Thisresult may be due to our insufficient knowledge of C3Hxkinetics, and we encourage further laboratory and theo-retical studies of the fate of C3Hx species under the low-pressure, high-temperature, reducing conditions in therelevant regions of extrasolar-planet atmospheres. BothCH and C readily insert into C2H2 (e.g., Berman et al.1982; Canosa et al. 1997; Haider & Husain 1993; Kaiser et


Fig. 5.— Mole-fraction profiles for several carbon-bearing species in our models for HD 189733b (left) and HD 209458b (right) fordayside-average (top) and terminator-average (bottom) conditions, assuming the nominal Kzz profiles shown in Fig. 1. A color version ofthis figure is available in the online journal.

al. 2001; Casavecchia et al. 2001; Mebel et al. 2007), andthese insertion reactions likely dominate the productionof C3 hydrocarbons on extrasolar giant planets, but thefate of the C3H2, C3H, and C3 radicals produced fromthese reactions is unclear. We implicitly assume that theC3 and C3H radicals that are produced from the reactionof C with C2H2 quickly end up as C3H2 — an assump-tion that may only be correct at altitudes where H2 is thedominant atmospheric constituent. The main loss pro-cess for C3H2 in our models is H + C3H2 + M → C3H3+ M. The resulting propargyl radicals can be photolyzedto recycle the C3H2, can react with atomic H to reformC2H2, can react with H2 or H to form C3H4, or can reactwith other C3Hx radicals to form C6 hydrocarbons andeventually benzene.

Although benzene is not a major constituent in ourmodels, it does reach ppb mole fractions in the few-microbar region in our HD 189733b models; however,benzene does not survive in the warmer HD 209458bmodels. While ppb levels are by no means sufficient forC6H6 condensation on HD 189733b, it does suggest thatpolycyclic aromatic hydrocarbon (PAH) formation couldpossibly occur at high altitudes on cooler transiting exo-planets, which could ultimately lead to the formation ofa “soot” aerosol layer, as has been suggested by Zahnle et

al. (2011). In contrast, PAH formation will be stronglysuppressed on planets with strong stratospheric thermalinversions (see also Liang et al. 2004). Our neutral chem-istry is not particularly conducive for soot formation,even under HD 189733b conditions, but it is possiblethat photoionization of N2 or other species could lead tothe production of high-molecular-weight ions and photo-chemically produced aerosols through a Titan-like ion-chemistry process (e.g., Waite et al. 2007, 2009; Imanaka& Smith 2007; Vuitton et al. 2007, 2008), although itis also possible that the high abundance of H2O and/orCO and the low abundance of CH4 at high altitudes onhighly irradiated exoplanets could short-circuit this pro-cess. The potential for ion chemistry playing a role in theformation of complex species in extrasolar giant planetatmospheres deserves further study.

Unlike the case for CO and H2O, methane can be ef-ficiently destroyed and not recycled in the upper atmo-spheres of HD 189733b and HD 209458b. Some of thecarbon ends up in C2H2 and other hydrocarbons, as wellas CO, but cross reactions with nitrogen species becomea key mechanism for removing CH4 from the upper at-mosphere. Hydrogen cyanide is the dominant product ofthis coupled carbon-nitrogen chemistry. The dominantphotochemical scheme that permanently converts CH4

18 Moses et al.

Fig. 6.— Mole-fraction profiles for several nitrogen-bearing species in our models for HD 189733b (left) and HD 209458b (right) fordayside-average (top) and terminator-average (bottom) conditions, assuming the nominal Kzz profiles shown in Fig. 1. A color version ofthis figure is available in the online journal.

into HCN in the stratospheres of our exoplanet modelsis

2 (H2O + hν → H + OH )2 (OH + H2 → H2O + H)

NH3 + H → NH2 + H2CH4 + H → CH3 + H2NH2 + H → NH + H2NH + H → N + H2

N + CH3 → H2CN + HH2CN + H → HCN + H2

Net : CH4 + NH3 → HCN + 3H2.(12)

Hydrogen cyanide then becomes a major disequilib-rium product in exoplanet atmospheres and can replacemethane as the second most abundant carbon-bearingspecies at high altitudes.

As is the case for oxygen compounds, the photochem-istry of carbon compounds is more effective for cooler ex-oplanets, both because of the larger abundance of CH4from transport-induced quenching and because the lowertemperatures allow disequilibrium products to survive

more readily. The dominant carbon-bearing photochem-ical products are HCN, C2H2, C, and CH3, with someCO2 being produced at high altitudes.

3.5. Photochemistry of nitrogen speciesNitrogen photochemistry parallels that of carbon

chemistry: N2, like CO, is a high-altitude source,and NH3, like CH4, is a critical transport-quenchedspecies that drives photochemistry throughout the bulkof the stratosphere. Figure 6 shows vertical profilesof the major nitrogen-bearing species in our nominalthermo/photochemical kinetics and transport models.At high altitudes, nitrogen photochemistry is initiatedby N2 destruction through schemes such as

CO + hν → C + OC + N2 → CN + N

CN + H2 → HCN + HNet : CO + N2 + H2 → HCN + N + O + H,

where the resulting HCN and N are significant high-altitude products.

Molecular nitrogen is fairly stable on HD 189733b andHD 209458b. Although N2 can combine with atomic H


to form NNH, thermal decomposition of this unstablespecies acts to recycle the N2. At thermospheric alti-tudes, the strong N N bond is attacked by atomic C,H, and O to form atomic N, which is then the domi-nant nitrogen species at high altitudes. Photolysis ofN2 should also occur in the thermosphere, where the N2would no longer be shielded by a larger surrounding col-umn of H2; however, we have left direct N2 photolysisout of our models because it is unlikely to be importantin the stratospheres of giant planets due to shielding byH2, and the net result of the thermospheric chemistry issimilar to that of photolysis.

Ammonia has much weaker bonds and is less stablethan N2, so much of the interesting nitrogen chemistryin our models involves NH3. Throughout the bulk ofthe stratosphere, NH3 destruction leads to HCN produc-tion through scheme (12) discussed in section 3.4. Thisscheme is efficient enough — and HCN is recycled read-ily enough — that HCN can take over from NH3 andCH4 as important nitrogen and carbon carriers in theupper stratosphere. The main loss process for HCN inhot-Jupiter stratospheres is the reaction H + HCN →CN + H2, with a much lesser contribution from HCNphotolysis to form CN + H2. The CN produced by theseprocesses reacts with H2 to reform HCN, so HCN recy-cling is prevalent.

Complex nitriles are also formed from photochemistryon cooler extrasolar planets like HD 189733b. For exam-ple, some of the CN produced in the H + HCN reactioncan react with C2H2 to form HC3N, or some of the CH3produced from hydrogen abstraction of methane can re-act with HCN to form CH3CN. These nitriles tend to beunstable in a background bath of H, and their destruc-tion pathways tend to reproduce the HCN, so they arenowhere near as abundant as HCN in our HD 189733bmodels. However, as with benzene and potential PAHs,further nitrile chemistry may produce condensible prod-ucts, particularly on colder exoplanets. On the otherhand, complex nitriles will be very unstable on planetslike HD 209458b that have a strong stratospheric thermalinversion. Other carbon-nitrogen bonded species, includ-ing imines and amines, are relatively unimportant in ourmodels, although the H2CN that is produced during thecourse of scheme (12) can undergo three-body recombi-nation with H to form some CH2NH that persists in ourcooler atmospheric models. Note that nitrogen hydrideslike hydrazine (N2H4) are unimportant in our models,unlike the situation on the much colder Jupiter.

Cross reactions between nitrogen and oxygen speciescan also occur in exoplanet stratospheres, leading to theproduction of species like NO and HNCO. The dominantschemes producing and destroying NO were discussed insection 3.3. HNCO is produced predominantly throughthe reaction NH2 + CO → HNCO + H, where the NH2derives from hydrogen abstraction of ammonia by atomicH. The dominant loss for HNCO is reaction with atomicH in the reverse of the main production reaction.

The dominant nitrogen-bearing photochemical prod-ucts in our exoplanet models are HCN, N, NO, and NH2.As with oxygen and carbon photochemistry, cooler plan-ets tend to have more abundant and more varied dise-quilibrium nitrogen species than warmer planets. Thistendency results from the larger abundance of photo-chemically active, transport-quenched species like NH3

on cooler planets and from the greater stability of thephotochemical products at low temperatures.

3.6. Sensitivity to temperatureThe atmospheric thermal structure can strongly affect

the abundance of disequilibrium constituents. As shownin Fig. 3, high temperatures in the stratospheres of tran-siting hot Jupiters can allow photochemically producedspecies to chemically react with other atmospheric con-stituents to drive the composition back toward equilib-rium. Disequilibrium compositions are thus difficult tomaintain on planets like HD 209458b that have strongthermal inversions. In contrast, thermochemical kineticsis inhibited when stratospheric temperatures are cooler,so that photochemically produced species are more likelyto survive on HD 189733b and other exoplanets withoutstratospheric thermal inversions. Temperatures in thedeep atmosphere also affect the abundance of disequilib-rium constituents through controlling where transport-induced quenching begins to operate. Fig. 7 illustratesthis importance of this effect for key transport-quenchedspecies like CH4, NH3, and HCN.

In Fig. 7, all models have the same constant-with-altitude eddy Kzz = 109 cm2 s−1 profile, but the thermalstructure is different for each model (see Fig. 2 for theadopted temperature profiles). In general, colder tem-peratures at depth imply that transport will be able tocompete with thermochemical kinetics at greater pres-sures; the quench point will therefore be deeper oncolder planets, and the mole fraction of the quenchedspecies will be larger. This effect can be seen with theHD 189733b models shown in Fig. 7, where the cold-est western-terminator-average model has a CH4 quenchpoint near the ∼7 bar, ∼1485 K level, whereas thewarmest projected-area-weighted dayside-average mod-els have quench points near 2.2 bar, ∼1533 K. The re-sulting CH4 quenched mole fraction is a factor of 10greater for the cooler western-terminator model than forthe warmer dayside-average model. Ammonia quenchesat even greater pressures and temperatures, at whichpoint the temperature profiles are more uniform (seeFig. 2), so that the differences between models are lesspronounced. The HCN profiles are more complicated be-cause the quenched abundances of both NH3 and CH4affect the HCN abundance, and HCN is photochemi-cally produced at high altitudes, but Fig. 7 suggeststhat HCN first quenches in the 10-30 bar region, wherethe GCM simulations of Showman et al. (2009) exhibitwestern-terminator-average temperatures that are actu-ally warmer than the dayside-average temperatures atthese pressures. As such, the western-terminator-averagemodel has a smaller quenched HCN abundance than thedayside-average models. The 1-D “4pi” profile from Fort-ney et al. (2006a, 2010) is the coolest of all our models at10-30 bars, resulting in the largest quenched HCN molefraction in this model. Given the lack of constraints ontemperatures and eddy diffusion coefficients at depth,the quantitative results for the abundance of quenchedspecies have significant uncertainty in our models; more-over, horizontal advection time scales may be shorterthan chemical time scales at pressures greater than afew mbar, acting to homogenize the composition longi-tudinally (see section 3.8).

For HD 209458b, the presence of high stratospheric

20 Moses et al.

Fig. 7.— Mole-fraction profiles for CH4 (top), NH3 (middle), and HCN (bottom) in our models for HD 189733b (left) and HD 209458b(right) for different assumptions about the atmospheric thermal structure (as labeled, and see Fig. 2). All models assume a constant Kzz= 109 cm2 s−1. A color version of this figure is available in the online journal.


temperatures, particularly on the dayside, is the biggestfactor in controlling the composition, as the disequi-librium molecular constituents do not survive in themiddle and upper stratosphere. The cooler the atmo-sphere is at high altitudes, the more likely disequilib-rium species are to survive to these altitudes. Note thatthe lack of UV photons in the nightside-average modelshown in Fig. 7 also helps keep disequilibrium moleculespresent at high altitudes. Even without a strong ther-mal inversion at the terminators of HD 209458b, strato-spheric temperatures are considerably higher than onHD 189733b, enabling thermochemical kinetic processesto destroy disequilibrium constituents. Temperatures atdepth are also much warmer on HD 209458b than onHD 189733b, due to the different luminosities of the hoststars. Transport-induced quenching is therefore less ef-fective on HD 209458b than on HD 189733b, and thepredicted quenched abundances of CH4, NH3, and HCNare correspondingly reduced.

The abundances of HCN and CH4 are particularly sen-sitive to temperature, whereas species like H2O, CO, andCO2 tend to more closely follow their equilibrium profilessuch that these species are more sensitive to metallicitythan to temperature on HD 189733b and HD 209458b.This conclusion will differ for colder planets like GJ 436bthat are in the CH4 rather than CO stability field.

3.7. Sensitivity to eddy diffusion coefficientsThe model results are also sensitive to the assumed

eddy

Documents

Julianne I. Moses C. Visscher J. J. Fortney A. P. Showman ...showman/publications/moses-etal...J. Preprint typeset using LATEX style emulateapj v. 8/13/10 DISEQUILIBRIUM CARBON, OXYGEN,