Biosignature Gases in H₂-Dominated Atmospheres on Rocky Exoplanets

BIOSIGNATURE GASES IN H₂-DOMINATED ATMOSPHERES ON ROCKY EXOPLANETS

The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

Citation Seager, S., W. Bains, and R. Hu. "BIOSIGNATURE GASES IN H₂- DOMINATED ATMOSPHERES ON ROCKY EXOPLANETS." The Astrophysical Journal, 777:2 (2013). pp.1-19.

As Published http://dx.doi.org/10.1088/0004-637x/777/2/95

Publisher Institute of Physics Publishing

Version Author's final manuscript

Citable link http://hdl.handle.net/1721.1/85944

Detailed Terms http://creativecommons.org/licenses/by-nc-sa/4.0/ arXiv:1309.6016v1 [astro-ph.EP] 24 Sep 2013 n ag nuhpo fptnilyhbtbeexo- habitable study potentially atmosphere of followup to for pool is accessible enough challenges challenge planets large near-term extreme biggest a of The number find overcome. a be unless can one futuristic biosignature a of study certainty. near and a detection is gases eventual the that ntermaue otsa ye ri,adms rsize or mass num- and (based orbit, the type, planets in star habitable host difference measured potentially large their on seemingly a needed is of be planets such bers could of there pool large molecules because The complex envelopes for life. massive hot for with too sur- required surface those with planet not any planets and making rocky water, liquid mean face we habitable potentially aelnTlsoe(tp/wwgt.r/,adteThirt the and (http://www.tmt.org/). (http://www.gmto.org/), Telescope Meter Telescope Magellan (http://www.eso.org/public/teles-instr/e-elt.html), ade ta.(06)adtegat2-t 0mtrclass 40-meter to 20- giant telescopes the and ground-based (2006)) the al. In as et Gardner (such year. Telescope telescopes Space larger each Webb sophisticated of planets development more to mass and evolution lower num- natural increasing and the of sentimentaddition, smaller discovery This the of with bers exoplanets. out borne of as being future taken is the usually is for spectroscopy inevitable sensing remote by gases xpae lntSre aelt,P ereRce)hsbe has la Ricker) for Program George 2017. Explorer PI Astrophysics in NASA’s Satellite, under Survey selected Planet Exoplanet 1 2 h oi fbointr ae,hwvr a remain may however, gases, biosignature of topic The h eeto feolntamshrcbiosignature atmospheric exoplanet of detection The rpittpstuigL using 2013 typeset 25, Preprint September version Draft o xml h l-k pc-ae ESmsin(Transtin mission TESS space-based all-sky the example For ae.W n htpoohmclypoue tm r h most the ra are CO high atoms a with H of atmospheres produced plausibilty In photochemically the atmospheres. that evaluating find mod We for a framework dates. using estimate by biomass temperatures, and surface habitable and atmospheres rdcdO stemjrdsrcieseis otptnilbios potential Most species. destructive major the CH is OH produced h aoaiiyo o-Vrdainevrnet oacmlto of accumulation to environments som H enabling an radiation U scenario, in low-UV from same the of fluxes for favorability UV O) lower The (or The H so of interest. in concentration of lower (and gases H biosignature environments, all radiation nearly UV sun-Earth-like In molecules. hmsr ilgnrt uhgssailgcly hog photoche through H abiologically, in gases gases such biosignature generate will chemistry ilyb eetdi rnmsinsetawt the with spectra transmission in detected be tially eansal H stable retain rmsn isgauegscniae,icuigNH including candidates, gas biosignature promising nigi htms ae rdcdb ieta r ul hydrogenat fully are that life by produced gases photolys most CH by directly that destroyed is are finding gases these as environments, h xrml ag Telescope Large Extremely The ISGAUEGSSI H IN GASES BIOSIGNATURE ue at xpaesaebigdsoee ihicesn frequ increasing with discovered being are exoplanets Earth Super 3 4 laeteeoemr aoal nlwU,a oprdt solar-like to compared as UV, low in favorable more therefore are Cl H , 2 2 ,aenteetv in flf na H an in life of signs effective not are S, topeei lsl nlgu otecs foiie atmosphere oxidized of case the to analogous closely is atmosphere 1. A INTRODUCTION T JS)sae o anhi 2018, in launch for slated (JWST) E 2 tl mltajv 5/2/11 v. emulateapj style X 1 dmntdamshrs esuybointr ae nexoplan on gases biosignature study We atmospheres. -dominated otne ofe h concept the fuel to continues 2 rc topee o ue at xpaestastn stars M transiting exoplanets Earth super for atmospheres -rich 2 .Seager S. DMNTDAMSHRSO OK EXOPLANETS ROCKY ON ATMOSPHERES -DOMINATED rf eso etme 5 2013 25, September version Draft h Giant the 2 James By . 2 unch ABSTRACT , ees tmcOi h ao etutv pce o some for species destructive major the is O atomic levels, .Bains W. , en y g 2 rc topee eas h oiatatmospheric dominant the because atmosphere, -rich ae ebSaeTelescope Space Webb James opei oeua eetosbsdo aafo the at- from many Telescope data Space on where based Hubble 2010), at- detections Seager Jupiter molecular hot mospheric (e.g., today’s studies in de- mosphere faced robustly continously This to is atmospheres. telescopes challenge exoplanet terrestrial of in capability molecules the tect is gases ture 2013). Seager by review gas the greenhouse (see concomitant potency the hence atmospheric and the the including composition, pressure), known, surface (and or the mass observed that atmosphere yet temperatures is not surface reason themselves planet’s are The given for a known. suitable controlling the yet factors conditions even not surface water—is that potentially liquid with is are that planets is planets of low-mass habitable—that iden- or pool small on of enough point fraction large related geophysical a contemporary, by tifying A contaminated overly positives. not false are spectroscopically ac- and are can atmosphere, active, that planet those the means in gases cumulate spec- gases biosignature atmospheric biosignature Useful measured from useful tra). inferred produces be in- will that are (which that life those by and habited temperature) surface inferred and t nErhlk risi n ftems substantial most the While of astronomers. one plan- face ever is Earth-sized to orbits to challenges signals technological down imaged Earth-like reach small directly in to the For ets ability reach the atmospheres. to planets, planet necessity terrestrial a dif- of is from events epochs level transit to precise ferent numerous ability highly together ref- a the adding to while and planets, systematics remove transiting 2013, and For identify al. et therein). Deming erences (see controversial remain 3 eodmjrcalnefrtesuyo biosigna- of study the for challenge major second A n N and 1 , .Hu R. , 2 ,aefvrbeee nslrlk UV solar-like in even favorable are O, sadntb o ) oesubtle more A O). (or H by not and is 1 lamshr ihphotochemistry, with atmosphere el , isgauegsst accumulate. to gases biosignature e ityo eceity Suitable geochemistry. or mistry gauegss uha M and DMS as such gases, ignature ue tr ol rdc a produce would stars M quiet V dfrso neeet uhas such element, an of forms ed ecssO ilrpdydestroy rapidly will O) cases me bnatratv pce nH in species reactive abundant nyadsm ilb beto able be will some and ency g fbointr a candi- gas biosignature of nge eetbebointr gases biosignature detectable V niomns few A environments. UV, ,weephotochemically where s, rthe or . t ihti H thin with ets pte pc Telescope Space Spitzer ol poten- could 2 2 2 technology development is ongoing, there are as yet no the mean molecular weight µ (e.g., Seager 2010), solid plans to launch a space telescope capable of directly kT imaging terrestrial-size planets. H = , (1) A third major challenge in the study of biosignature µmH g gases has to do with the geological false positive signa- where k is Boltzmann’s constant, T is temperature, mH tures. These false positives are gases that are produced is the mass of the hydrogen atom, and g is the surface geologically and emitted by volcanoes or vents in the gravity. The point is that when H2 dominates the atmo- crust or ocean. Geochemistry has the same chemicals to spheric composition over terrestrial planet atmosphere work with that life does, and therefore false positives are gases CO2 and N2, the mean molecular weight is ∼20 inevitable. While early theoretical studies favored detec- times smaller and hence the scale height is ∼20 times tion of redox disequilibrium (such as O2 and CH4) that larger. The observational imprint of an atmosphere is should not both exist in an atmosphere in photochemical usually taken as about 5H. steady state, often one molecule of the set is too weak Turning back to biosignature gases, they have been spectroscopically for potential detection. The conven- studied theoretically as indicators of life on planets with tionally adopted approach (at least in theoretical studies) oxidized atmospheres for over half a century, beginning is therefore to identify a biosignature gas that is many with Lederberg (1965) and Lovelock (1965). One high- orders of magnitude out of thermodynamic equilibrium light from the last decade is the realization that low UV with the expected gas composition of the atmosphere and radiation environments compared to solar lead to a much to study the gas in the context of the planet atmosphere higher concentration of biosignature gases, as studied for environment via atmospheric spectra that cover a wide Earth-like planet atmospheres. This is because the stel- wavelength range. A more likely outcome to the field of lar UV creates the radical OH (in some cases O) which biosignature gases will be to develop probabilistic assess- destroys many gases in the atmosphere and thus reduces ment of the likelihood a molecule in a given atmosphere the gas lifetime. A low UV radiation environment is can be attributed to life, because spectroscopic data and taken to be that of a planet orbiting a UV quiet M dwarf the information for a complete assessment of the plane- star (see Figure 1 and discussion in §5.6). tary environment will be limited. A second highlight in biosignature gas research in the To increase the chances of detecting exoplanet atmo- last decade is the theoretical exploration of potential spheric biosignature gases we are motivated to widen the biosignature gases beyond the conventionally considered parameter space of types of planets where biosignature dominant Earth or early Earth based ones of O2,O3, gases can accumulate and should be sought out obser- N2O, and CH4. The variations studied include dimethyl- vationally. We here describe, for the first time to our sulfide (DMS; Pilcher 2003), methyl chloride (CH3Cl; knowledge, the case for and against biosignature gases Segura et al. 2005), and other sulfur compounds includ- in hydrogen-rich atmospheres. Some massive enough or ing CS2 and COS (Domagal-Goldman et al. 2011). See cold enough super Earths (loosely defined as planets with Seager et al. (2012) for a review, Seager et al. (2013) for up to 10 Earth masses) will be able to retain hydrogen a biosignature gas classification scheme, and Seager et al. in their atmospheres (see the discussion in §5.2). In gen- (2013) for a biomass model estimate intended as a plaus- eral, planets are expected to outgas or capture hydro- ability check to consider biosignature gas surface fluxes gen from the nebula during planet formation. Here we different from Earth values. are concerned with super Earths with relatively thin hy- We begin with a description of our atmosphere and drogen atmospheres and not planets with massive atmo- biomass estimate model in §2. We present general results spheres or envelopes (as in mini-Neptunes) which will in §3 and specific results for a number of potential and have surfaces too hot for liquid water (Rogers and Sea- unlikely biosignature gases in §4. A discussion in §5 is ger, in prep.) or may not even have a surface. A thin followed by a summary and conclusion in §6. hydrogen atmosphere does not add much to either the 2. mass or the size of the planet (Adams et al. 2008), so MODEL that an H2-rich atmosphere itself does not aid in planet The model goal is to computationally generate atmo- discovery or detection. spheric spectra for exoplanets with H2-rich atmospheres Super Earths with H2-rich atmospheres are nonetheless with biosignature gases. The model consistes of a photo- in some ways more favorable for detection and study than chemistry code which takes biosignature gases as surface their terrestrial planet counterparts with N2- or CO2- fluxes, an approximate temperature profile calculation, dominated atmospheres. A more massive planet than and a line-by-line spectral calculation (Seager et al. 2013; Earth (i.e., more likely to retain atmospheric H2 than Hu et al. 2012). The model also uses a biomass model es- Earth) is easier to discover than an Earth-mass planet timate, to check whether or not a biosignature gas could via the radial velocity technique. A more massive planet be the result of a plausible surface ecology. than Earth is also larger and so easier to discover or de- 2.1. tect by the transit technique than a lower-mass planet. Model Atmosphere For example a 10 M⊕ planet of Earth-like composition Photochemistry Model The focus on chemistry is criti- would have a radius 1.75 times larger than Earth (e.g., cal for biosignature gases, because sinks control a biosig- Seager et al. 2007). The larger planet area is more fa- nature gas lifetime and hence the gas’ potential to accu- vorable for atmosphere study in reflected or thermally mulate in the planetary atmosphere. A model for atmo- emitted radiation than an Earth-size planet. For transit spheric chemistry is required to connect the amount of transmission spectra, planets with H2-rich atmospheres biosignature gas in the atmosphere (as required for de- have a much larger signal compared to H-poor atmo- tection) to the biosignature source flux at the planetary spheres because of the larger scale height H, based on surface. 3

]

−1 0 10 nm −2

−5 10

−10 10

−15 10 Sun GJ 1214 M star with T = 3000 K without chromospheric emission

Flux inFluxHabitablethe m Zone[W −20 eff

10 2 3 4 10 10 10 Wavelength [nm]

Figure 1. Comparison of stellar fluxes. The radiative flux received by a planet in the habitable zone of a solar-like star, a weakly active M dwarf star (like GJ 1214), and a theoretically simulated quiet M dwarf star with an effective temperature of 3000 K with no chromospheric emission. The flux is scaled so that the planet has a surface temperature of 290 K. The spectrum of the sun-like star is from the Air Mass Zero reference spectrum during a solar quiet period (http://rredc.nrel.gov/solar/spectra/am0/). The spectrum of GJ 1214 contains two parts: for wavelengths shorter than 300 nm we take the most recent HST measurement (France et al. 2013), for wavelengths longer than 300 nm we take the NextGen simulated spectrum for an M dwarf star having parameters closest to those of GJ 1214 (i.e., effective temperature of 3000 K, surface gravity log(g) = 5.0, and metallicity = 0.5). The spectrum with no chromospheric emission is also from the NextGen model (Allard et al. 1997). Under the common definition of weakly active, or relatively quiet M dwarf, the UV environment in its habitable zone can differ by more than six orders of magnitude.

Our photochemical model is presented in Hu et al. tion, Rayleigh scattering and aerosol Mie scattering con- (2012). The photochemical code computes the steady- tributing to the optical depth. The model was developed state chemical composition of an exoplanetary atmo- from the ground-up and has been tested and validated by sphere. The system can be described by a set of time- reproducing the atmospheric composition of Earth and dependent continuity equations, one equation for species Mars (Hu et al. 2012; Hu 2013). at each altitude. Each equation describes: chemical pro- For biosignature gases that are minor chemical per- duction; chemical loss; eddy diffusion and molecular dif- turbers in the atmosphere, the biosignature lifetime can fusion (contributing to production or loss); sedimenta- be estimated based on the abundance of the major chem- tion (for aerosols only); emission and dry deposition at ical sink. For this paper, the values of NH3 and N2O the lower boundary; and diffusion-limited atmospheric surface source fluxes are calculated from the full pho- escape for light species at the upper boundary. The code tochemistry model, whereas the calculations of surface includes 111 species, 824 chemical reactions, and 71 pho- source fluxes for other biosignature gases are simplified tochemical reactions. Starting from an initial state, the estimates. One more point to note is that photochem- system is numerically evolved to the steady state in which istry is relevant high in the atmosphere typically above the number densities no longer change. mbar levels to which stellar UV radiation can penetrate The generic model computes chemical and photochem- from above. ical reactions among all O, H, N, C, S species, and forma- Temperature-Pressure Profile The precise temperature- tion of sulfur and sulfate aerosols. The numerical code pressure structure of the atmosphere is less impor- is designed to have the flexibility of choosing a subset tant than photochemistry for a first-order description of of species and reactions in the computation. The code biosignatures in H2-rich atmospheres. The reason is that therefore has the ability to treat both oxidized and re- most biosignature gases of interest have sources and sinks duced conditions, by allowing selection of “fast species”. not signficantly affected by minor deviations in the tem- For the chemical and photochemical reactions, we use perature pressure profile. Morever the biosignature gases the most up-to-date reaction rates data from both the themselves are secondary players in governing the heat- NIST database (http://kinetics.nist.gov) and the JPL ing structure of the atmosphere. publication (Sander et al. 2011). Ultraviolet and visi- We therefore justify using the photochemistry model in ble radiation in the atmosphere is computed by the δ- a stand-alone mode, with a pre-calculated temperature- Eddington two-stream method with molecular absorp- pressure profile. The calculated temperature-pressure 4 profile is approximate and is one which assumes a sur- surface biofluxes. face temperature (e.g., 290 K), an appropriate adiabatic Based on life on Earth, a summary overview is that a −2 −2 lapse rate for H2-rich compositions, and a constant tem- biomass surface density of 10 g m is sensible, 100 g m perature above the convective layer of the atmosphere. is plausible, and 5000 g m−2 is possible. In real sit- Such assumed temperature profiles are consistent with uations, the total biomass is nearly always limited by greenhouse warming in the troposphere and lack of ul- energy, bulk nutrients (carbon, nitrogen), trace nutri- traviolet absorber in the stratosphere. The semi-major ents (iron, etc.) or all three. Regarding global surface axis of the planet is then derived based on the assumed biofluxes we provide values and references where needed temperature profile by balancing the energy flux of in- in our results and discussion. coming stellar radiation and outgoing planetary thermal The biomass model estimates are tied to the type emission. The details of this procedure are described in of biosignature gas and so we briefly summarize our Hu et al. (2012). biosignature classification scheme before discussion each Synthetic Spectra To generate exoplanet transmis- biomass model estimate. sion and thermal emission spectra, we use a line- Type I biosignature gases are generated as byproduct by-line radiative transfer code (Seager et al. 2000; gases from microbial energy extraction. For example, Madhusudhan & Seager 2009; Hu et al. 2012). Opaci- on Earth many microbes extract energy from chemical ties are based on molecular absorption with cross sec- energy gradients using the abundant atmospheric O2 for tions computed based on data from the HITRAN 2008 aerobic oxidation, database (Rothman et al. 2009), molecular collision- induced absorption when necessary (e.g., Borysow 2002), X+O2 → oxidized X. (2) Rayleigh scattering, and aerosol extinction computed For example: H2O is generated from H2; CO2 from or- based on Mie theory. The atmospheric transmission is 2− ganics; SO2 or SO4 from H2S; rust from iron sulfide computed for each wavelength by integrating the op- (FeS); NO− and NO− from NH ; etc. tical depth along the limb path (as outlined in, e.g., 2 3 3 On an exoplanet with an H2-rich atmosphere, the Seager & Sasselov 2000; Miller-Ricci et al. 2009). The abundant reductant would now atmospheric H such that planetary thermal emission is computed by integrating 2 the radiative transfer equation without scattering for H2 + X → reduced X. (3) each wavelength (e.g., Seager 2010). The oxidant must come from the interior. We consider clouds in the emergent spectra for thermal In other words, for chemical potential energy gradi- emission by considering 50% cloud coverage by averag- ents to exist on a planet with an H2-rich atmosphere, ing a cloudy and cloud-free spectra. We omit clouds for the planetary crust must (in part) be oxidized in order the transmission spectra model because the clouds are at to enable a redox couple with the reduced atmosphere. low altitudes whereas the spectral features form at high The byproduct is always a reduced gas, because in a re- altitudes. ducing environment H2-rich compounds are the available 2.2. Biomass Model Estimates reductants. To be more specific, oxidants would include gases such as CO2 and SO2. A biomass model estimate has been developed by The Type I biosignature gas biomass model is based Seager et al. (2013) that ties biomass surface density to on thermodynamics and is derived from conservation of a given biosignature gas surface source flux. The moti- energy and discussed in detail in Seager et al. (2013). vating rationale is that with a biomass estimate, biosig- The biomass model estimate is nature gas source fluxes can be free parameters in model predictions, by giving a physical plausibility check in Fsource ΣB ≃ ∆G . (4) terms of reasonable biomass. The approach aims to en- Pme able consideration of a wide variety of both gas species −2 and their atmospheric concentration to be considered in Here ΣB is the biomass surface density in g m , ∆G biosignature model predictions. The biomass model es- is the Gibbs Free energy of the chemical redox reac- timates are valid to one or two orders of magnitude. We tion from which energy is extracted (e.g., equation (2)). provide a summary of the biomass model here with the ∆G depends on the standard free energy of reaction full details available in Seager et al. (2013). (∆G0), and the concentration of the reactants and products. Reactant and product concentrations can include The biomass model is used in the following algorithm. + First we calculate the amount of biosignature gas re- ocean pH (concentration of H ) in reactions that gen- quired to be present at “detectable” levels in an exo- erate or consume protons. ∆G0 values are taken from planet atmosphere from a theoretical spectrum (we de- Amend & Shock (2001). fine a detection metric in §2.3). Second, we determine The term Pme is an empirically determined microbial the gas source flux necessary to produce the atmospheric maintenance energy consumption rate, that is, the mini- biosignature gas in the required atmospheric concentra- mum amount of energy an organism needs per unit time tion. The biosignature gas atmospheric concentration is to survive in an active state (i.e., a state in which the a function not only of the gas surface source flux, but organism is ready to grow). An empirical relation has also of other atmospheric and surface sources and sinks. been identified by Tijhuis et al. (1993) that follows an Third, we estimate the biomass that could produce the Arrhenius law necessary biosignature gas source flux. Fourth, we con- −EA Pme = A exp . (5) sider whether the estimated biomass surface density is RT physically plausible, by comparison to maximum terres- − trial biomass surface density values and total plausible Here EA =6.94 × 104 kJ mol 1 is the activation energy, 5

R = 8.314 kJ mol−1 K−1 is the universal gas constant, on thermodynamics (Seager et al. 2013). The order of and T in units of K is the temperature. The constant magnitude nature of the Type III biosignatures derives A is 4.3 × 107 kJ g−1 s−1 for aerobic growth and 2.5 × the reliance on laboratory rates for microbial production 107 kJ g−1 s−1 for anaerobic growth (Tijhuis et al. 1993). rates, and this is also possibly a terracentricity issue. Here per g refers to per g of wet weight of the organism. The lack of ecosystem context is a major limitation −1 −1 Pme is in units of kJ g s . for the biomass estimate. An ecosystem contains not The free parameter in this biomass model estimate only the producers (i.e., the biomass estimate derived (equation (4)) is the biosignature gas source flux Fsource ultimately from the bioflux Fsource) but also consumers, −2 −1 (in units of mole m s ). Fsource is the flux of the whereas the biomass model estimate considers only the metabolic byproduct and is also the surface bioflux re- producers. In this sense the biomass estimate in equa- quired to generate a given biosignature gas concentration tion (4) is a minimum. We can fairly say that in the case in the atmosphere. of a very small or very large biomass estimate, the as- Type II biosignature gases are byproduct gases pro- sessment of biosignature gas plausibility is valid: a small duced by the metabolic reactions for biomass building, biomass estimate gives room for consumers even as a and require energy. On Earth these are reactions that minimum biomass and a large biomass estimate as a min- capture environmental carbon (and to a lesser extent imum will remain large regardless of the consumers. For other elements) in biomass. Type II biosignature re- the intermediate case where a large but not unreason- actions are energy-consuming, and on Earth the energy able biomass is needed to generate a detectable biosig- comes from sunlight via photosynthesis. nature, the decision on whether the gas is a plausible There is no useful biomass model for Type II biosig- biosignature is more complicated, and will depend on nature gases because once the biomass is built a Type II the planetary context: geochemistry, surface conditions, biosignature gas is no longer generated. atmospheric composition and other factors. Type III biosignature gases are produced by life but not Again, we do not argue the biosignature biomass model as byproducts of their central chemical functions. Type estimates are an accurate prediction of an extraterrestrial III biosignature gases appear to be special to particu- ecology, rather we emphasize the goal of the biomass lar species or groups of organisms, and require energy model estimates is the order of magnitude nature for for their production. Because the chemical nature and a first order asssessment of the plausability of a given amount released for Type III biosignature gases are not biosignature gas candidate. linked to the local chemistry and thermodynamics, the Type III biosignature gas biomass model is an estimate 2.3. Detection Metric based on lab culture production rates. We now describe our metric for a “detection” that leads We estimate the biomass surface density by taking the −2 to a required biosignature gas concentration. For now, biosignature gas source flux Fsource (in units of mole m the detection has to be a theoretical exercise using syn- −1 s ) divided by the mean gas production rate in the lab thetic data. We determine the required biosignature gas −1 −1 Rlab (in units of mole g s ), concentration based on a spectral feature detection with a SNR=10. Specifically, we describe the SNR of the spec- Fsource ΣB ≃ . (6) tral feature as the difference between the flux in the ab- Rlab sorption feature and the flux in the surrounding continuum (on either side of the feature) taking into account We take the maximum observed for the Type III Rlab the uncertainties on the data, rates Ffield values from different studies (Seager et al. 2013). The caveat of the Type III biomass estimate ex- |Fout − Fin| plicitly assumes that the range of R for life on exoplan- SNR = , (7) 2 2 ets is similar to that for life in Earth’s lab environment. σFout + σFin Nonetheless we have showed based on Earth’s values that q the Type III biomass model is valid to one or two orders where Fin ± σFin is the flux density inside the absorp- of magnitude. The goal, again, is to use the biomass esti- tion feature and Fout ± σFout is the flux density in the mate to argue for or against plausibilty of a biosignature surrounding continuum, and σ is the uncertainty on the gas based on Earth’s biomass surface density values and measurement. not for any prediction of quantitative values. The uncertainties of the in-feature flux and continuum Bioindicators are defined as the end product of chem- flux are calculated for limiting scenarios. For thermal ical reactions of a biosignature gas. emission we consider a futuristic space telescope able to Model caveats are related to the order of magnitude block out the light of the host star. The uncertainties nature of the biomass estimates, the possible terracen- of the in-feature flux and continuum flux are calculated tricity of the biomass model estimates, and the lack of for a limiting scenario: an 1.75 times Earth-sized planet ecosystem context (see Seager et al. (2013) §6.1, 6.2, 6.3 orbiting a star3 at 10 pc observed (via direct imaging) for a detailed discussion). Here we provide a summary with a 6 m-diameter telescope mirror operating within overview. 50% of the shot noise limit and a quantum efficiency of The order of magnitude nature of the Type I biomass 20%. The integration time is assumed to be 20 hours. estimate derives from the dependency of the estimate We note that collecting area, observational integration on Pme , itself very sensitive to temperature. The pos- time, and source distance are interchangeable depending sible terracentricity of our estimates is related to use of on the time-dependent observational systematics. This m P e , which is derived from observations of terrestrial mi- 3 croorganisms, but we argue the dependency is largely Assuming perfect removal of starlight. 6

0 telecope scenario is based on a TPF-I type telescope 10 −1 CO (Lawson et al. 2008). 10 2 For transit transmission spectra, we use the same equa- −2 10 tion as above, but with the denominator replaced by −3 2 10 the noise in the stellar flux (F∗), as in (4σF ), be- −4 cause transmission observations measure the difference 10 −5 between the in-transit stellar flux and out-of-transitp stel- 10 −6 lar flux. For transmission spectra we consider a 6.5-m 10 space telescope, having quantum efficiency of 0.25 ob- −7 10 serving with 50% photon noise limit, with integration −8 time of 60 hours in-transit and 60 hours out-of-transit 10 H −9 (assuming observing of multiple transits). Again, we 10 OH −10 O note that collecting area, observational integration time, 10

and source distance are interchangeable depending on the MixingRatio −11 10 time-dependent observational systematics. This scenario −12 is based on the JWST. 10 −13 3. 10 PHOTOCHEMISTRY RESULTS: H IS THE DOMINANT −14 PHOTOCHEMICALLY-PRODUCED REACTIVE SPECIES 10 −15 IN H2-RICH ATMOSPHERES 10 −16 In an H2-rich terrestrial exoplanet atmosphere, atomic 10 H is the largest sink for most atmospheric molecules in- −17 10 cluding biosignature gases. This is in contrast to oxi- −18 10 dizing atmospheres (atmospheres with substantial O2 or −3 −2 −1 0 1 2 3 CO and H O and without H ) where the OH radical 10 10 10 10 10 10 10 2 2 2 UV Flux [Solar UV Flux] (and in some cases O) plays the role of the dominant sink. We note that for H2-rich atmospheres with high Figure 2. Mixing ratio dependence of the reactive species (H, CO2 levels, atomic O will be abundant (Figure 2) and OH, and O) on UV flux in a H2-rich atmosphere with some CO2. Shown are different CO2 levels. The curves correspond to a CO2 for some molecules will dominate the removal chemistry. surface emission flux of Earth’s volcanic emission rate (3×1015 m2 −1 To explain the high H concentration we review the pro- s ; solid lines), CO2 emission rate 100 times higher than Earth’s duction of H, OH and O in H2-rich atmospheres. To rate (dashed lines) and CO2 emission rate 100 times lower than qualitatively outline the main points we use a simplified Earth’s rate (dotted lines). The planet has 10 M⊕ and 1.75 R⊕ description of the main chemical pathways. This dis- and is in a 1.6-AU orbit of a sun-like star (with UV adjusted). The fiducial atmosphere is 90% H2 and 10% N2 by volume, in a 1 bar cussion serves for illustration only, and is later backed atmosphere with a 290 K surface temperature. The main point is up with a more detailed computational photochemistry that the H concentration does not depend on the amount of CO2 model. in the atmosphere, whereas the amount of O is critically controlled To derive atmospheric concentrations of a species [A] by the level of CO2 in the atmosphere. Compared to H, OH is always a minor constituent in the atmosphere (by a few orders of we take photochemical equilibrium, magnitude). As the UV flux increases, more of the destructive, d[A] reactive species are generated. = P − [A]L =0, (8) dt forming H from H2 (equation (11)) illustrates the impor- P tance of water vapor. Water is needed to form H in the [A]= . (9) first place, in this case. L The second major point is that the reason H can ac- where [A] is the mixing ratio of species A, and P and L cumulate to high concentrations is because the H + H + are the production and loss rates respectively of species M reaction rate that removes the H atoms is relatively A. Below, the K are reaction constants and J is the slow. The rates are photodestruction rate associated with a stated reaction. K =2.8 × 10−18 exp(−1800.0/T ) [m3s−1], (15) We consider an H2 atmosphere with some H2O. −39 −1 3 −1 H2O + hν → H+OH J, (10) Km =6.64 × 10 (T/298.0) × N [m s ], (16) −6 −1 OH+H2 → H2O + H K, (11) J ≃ 10 [s ], (17)

H+H+M → H2 + M Km. (12) where N is the number density of species M in units of −3 Combining the above two equations we have molecules m . The rates are from Sander et al. (2011). The third major point is that in the H2-rich atmosphere the OH concentration is low, because [OH] reacts K[OH][H2] J[H2O] [H] = = , (13) with H2 to recombine to H2O. s Km[M] s Km H is produced by photodissociation of water vapor and and not predominantly by the direct photodissociation of H2. J[H2O] The reason is that the photons with high enough energy [OH] = . (14) (λ< 85 nm; Mentall & Gentieu 1970) to photodissociate K[H] H2 are not available. The high-energy photons that could The simpliﬁed atmosphere reveals a number of relevant dissociate H2 directly are absorbed at the pressure lev- points. The ﬁrst major point is that the role of OH in els of nanobars by H2 itself, and the photons that could 7

−6 penetrate down to pressure levels relevant to observations 10 (0.1 mbar to 1 bar) are those that can dissociate water. −7 H 10 O −8 Photodissociation of H2O, in comparison, is caused by 10 OH photons of lower energy (λ ≤ 240 nm; Banks & Kockarts −9 10 1973), photons which can penetrate more deeply in the −10 atmosphere than the ones that photodissociate H2. We 10 −11 note that like other photochemical products, H is formed 10 −12 primarily above the mbar level, before all of the photodis- 10 sociating stellar photons are absorbed. −13 10 MixingRatio The H concentration is dependent on the stellar UV −14 levels and presence of H O. Low-UV environments are 10 2 −15 favorable for biosignature build up, since the initial pho- 10 −16 tolysis that starts the OH formation chain will be weaker. 10 −17 A similar situation is described for oxidized atmospheres 10 in Segura et al. (2005). −18 10 We must beware that for some molecules, in some 1 situations, atomic O will be the dominant destructive ] 10 −1 0 species. There is no simple model (as in the above equa- 10 H −1 O tions) but with our full simulation we find in atmospheres 10 OH −2 with high CO atomic O will abundant (see Figure 2). s Cl [m 2 3 10 −3 The key point is that reaction rates with O are faster 10 than reaction rates with H for some molecules (see Ta- −4 10 ble 1). −5 10 There is a very important point of comparison between −6 the dominant reactive species, H in H -rich atmospheres 10 2 −7 and OH in oxidized atmospheres. The concentrations of 10 −8 H and OH in the two different types of atmospheres vary 10 −9 (see Figure 3 and Table 1, as well as more generally Table 10 −10 4 in Hu et al. (2012)). This can be understood qualita- 10 tively because OH is much more reactive than H. OH −11 10 will react faster with any atmospheric component than −12 H, and so, for the same impinging stellar UV flux, OH ChemicalRemoval CH Rate for 10 −3 −2 −1 0 1 2 3 10 10 10 10 10 10 10 will build up to a lower atmospheric concentration than UV Flux [Solar UV Flux] H. The rate of removal of a biosignature gas by H or OH is a product of the concentration and the reactivity. OH, Figure 3. Destructive power of reactive species (H, OH, and O) in with lower concentration but greater reactivity will re- a reduced atmosphere. The atmosphere considered has 90% H2 and 10% N2 by volume, with CO2, CH4, SO2, and H2S emission from move a biosignature gas at a similar rate to H, which has its surface, for a 1 bar atmosphere on a planet with 10 M⊕ and a greater concentration but a lower reactivity. In other 1.75 R⊕. Shown for comparison, are cases for an N2-dominated words, while the mechanism of chemistry clearance and atmosphere (diamond markers) and Earth’s current atmosphere the end products are different, the loss rate is fairly sim- (circular markers). Top panel: Mixing ratios of H, OH, and O as a function of UV flux. The mixing ratio of H exceeds the other ilar. For more details of the formation and destruction reactive species OH and O. Bottom panel: The column-integrated of the reactive species H, OH, and O in reduced and ox- chemical removal rates as a result of reactions with H, O, and OH, idized atmospheres, see Hu et al. (2012). for which we have used CH3Cl as an example. The removal rates are scaled by the steady-state mixing ratio of CH3Cl to have a 4. dimension of velocity. This panel shows that removal by H is the RESULTS: POTENTIAL AND UNLIKELY dominant loss rate, and that the loss rates scale approximately BIOSIGNATURE GASES linearly with UV flux incoming to the exoplanet atmosphere. We now turn to describe the potential and unlikely biosignature gases in an H2 atmosphere by their biosignature category. The biosignature categories developed composition and the required surface temperature. The in Seager et al. (2013) and summarized in §2.2 are an es- eddy diffusion coefficients are scaled up by a factor of 6.3 sential aide for calculations because of the common for- from those measured in Earths atmosphere, in order to mation pathways that belong to each biosignature class. account for the difference in the mean molecular mass. We consider a planet with 10M⊕, 1.75 R⊕, an atmo- Important minor gases considered are H2O (evaporated sphere with 90% H2 and 10% N2 by volume. The atmo- from a liquid water ocean), CO2 (about 100 ppm), and sphere scenario is the hydrogen-rich case among the exo- CH4 and H2S (emitted from surface). Deposition veloc- planet benchmark scenarios detailed in Hu et al. (2012), ities of H2, CH4 are assumed to be zero, the deposition and we here outline the key specifics. The planet surface velocity of CO is 10−10 ms−1, and deposition velocities of pressure is 1 bar and the planet surface temperature is oxidants including O2,O3, and H2O2 and sulfur species 290 K. The temperature drops with increasing altitude are assumed to be the values as on Earth. See Hu et al. according to an adiabatic lapse rate, until reaching 160 (2012) for the rationale for these specifics and for the de- K and is prescribed as constant above. The semi-major scription of carbon, oxygen, and sulfur chemistry in such axis of the planets orbit is 1.4 AU if orbiting a sun-like an H2-dominated atmosphere. star, 0.037 AU if orbiting an M5V dwarf star, this is the The amount of UV flux on the planet from the star consistent planet-star separation given the atmospheric is critical to destroying biosignature gases and so we 8

Reaction A nE T = 270 K T = 370 K T = 470 K −18 −24 −22 −21 DMS + H → CH3SH + CH3 4.81 × 10 1.70 9.00 2.63 × 10 2.11 × 10 2.91 × 10 −17 −24 −22 −21 CH3Cl + H → CH3 + HCl 1.83 × 10 0 19.29 1.97 × 10 1.46 × 10 1.92 × 10 −17 −21 −20 −19 CH3Br + H → CH3 + HBr 8.49 × 10 0 24.44 1.59 × 10 3.01 × 10 1.63 × 10 −17 −18 −17 −17 CH3I+H → CH3 + HI 2.74 × 10 1.66 2.49 7.67 × 10 1.75 × 10 3.09 × 10 −17 −18 −18 −18 DMS + OH → CH3SCH2 + H2O 1.13 × 10 0 2.10 4.43 × 10 5.71 × 10 6.60 × 10 −18 −20 −19 −19 CH3Cl + OH → CH2Cl + H2O 1.40 × 10 1.60 8.65 2.54 × 10 1.89 × 10 3.17 × 10 −19 −20 −20 −19 CH3Br + OH → CH2Br + H2O 2.08 × 10 1.30 4.16 2.87 × 10 7.13 × 10 1.30 × 10 −18 −20 −19 −19 CH3I + OH → CH2I+H2O 3.10 × 10 0 9.31 4.90 × 10 1.50 × 10 2.86 × 10 −17 −17 −17 −17 DMS + O → CH3SO + CH3 1.30 × 10 0 -3.40 5.91 × 10 3.93 × 10 3.10 × 10 −17 −23 −22 −21 CH3Cl + O → CH2Cl + OH 1.74 × 10 0 28.68 1.77 × 10 8.77 × 10 8.26 × 10 −17 −23 −22 −21 CH3Br + O → CH2Br + OH 2.21 × 10 0 30.76 1.77 × 10 8.77 × 10 8.26 × 10 −18 −17 −17 −17 CH3I+O → CH3 + IO 6.19 × 10 0 -2.84 2.19 × 10 1.56 × 10 1.28 × 10

Table 1 Reaction rates with H, OH, and O of select Type III biosignature gases. Second order reaction rates in units of m3 molecule−1 s−1 are computed from the formula k(T )= A(T/298)n exp(−E/RT ) where T is the temperature in K and R is the gas constant (R = 8.314472 × 10−3 kJ mole−1). The reactions rate are compiled from the NIST Chemical Kinetics Database. consider the same planet orbiting three different star We start by focusing on the Type I biosignature gases, types. The first star type is a sun-like star. The sec- gases generated by reactions that extract energy from ex- ond star type is a weakly active 0.2 R⊙ M5V dwarf ternal, environmental redox gradients. The most likely star, with EUV taken as that expected for GJ 1214b Type I metabolic product in an H2-rich atmosphere (France et al. 2013). The third star type is a quiescent would be those in which non-hydrogen elements are in M star with no chromospheric and only photospheric ra- their most hydrogenated form4. In a reducing environ- diation, again an 0.2 R⊙ M5V dwarf star (see §5.6). UV ment, life captures chemical energy by reducing envi- radiation received by the planet is scaled according to the ronmental chemicals. In the presence of excess hydro- semi-major axis, and the stellar UV spectra are from: gen, the most energy life could extract from chemical the Air Mass Zero reference spectrum for the sun-like potential energy gradients would be from converting ele- star (http://rredc.nrel.gov/solar/spectra/am0/); the UV ments from relatively oxidized compounds to their fully fluxes are from France et al. (2013) for the weakly active reduced form. An additional reason for focusing on Type M dwarf star (using the values for GJ 1214b), and from I biosignature gas products that are in their most hydro- simulated spectra of cool stars (Allard et al. 1997) for genated from is that they are likely long lived in an H2- the UV quiet M dwarf star (see Figure 1). dominated atmosphere, because molecules in their most Whether or not a biosignature gas is detectable can be hydrogenated form cannot undergo any further reactions technique and spectral feature dependent. The required with H. atmospheric concentration depends on the strength of a given absorption feature, and different techniques are 4.1.1. Type I Biosignature Gas Overview sensitive to different wavelength ranges. For example, The most reduced form of the most abundant non- thermal emission detection sensitivity follows the plane- metal elements, C, N, O, P, S, H, Si, F, and Cl are tary thermal emission flux (approximately a black body CH , NH , H O, PH , H S, H , SiH , HCl, and HF. The peaking in the mid-IR), whereas the transmission spectra 4 3 2 3 2 2 4 most promising Type I biosignature gas is NH3, which sensitivity in the infrared follows the thermal emission is further described below (§4.1.2). The other reduced flux of the star (approximately a black body). An illus- molecules are unlikely biosignature gases for a variety of trative example is NH3 with a strong absorption feature reasons. Some (PH3, SiH4), require energy input to make at 10.3-10.7 µm suitable for planetary thermal emission the reduced product from geologically available materi- but for transmission a weaker absorption feature at 2.8- als, and so would not be produced by Type I biosignature 3.2 µm is more easily detected than the 10 µm feature gas reactions. Some are always present in their most re- because of overall photon fluxes of the star. For trans- duced form and so life cannot reduce them further (F, mission spectra we avoid consideration of sun-like stars Cl). H2S and CH4 are not viable for reasons discussed because the observational signal (the planet atmosphere- below, largely because geological and biologically sources star area ratio) is too low (e.g., Kaltenegger & Traub cannot be discriminated between. In the case of H and 2009). 2 H2O, they are naturally present in an H2-dominated at- Biosignature gas results are summarized for thermal mosphere at relevant potentially-habitable planet tem- emission detectability (for sun-like and M dwarf stars) peratures. in Table 2 and for transmission spectra detectability (for An aside about PH3. We note that trace amounts M dwarf stars only) in Table 3. Select promising biosig- of phosphine is produced by some anaerobic ecologies nature gases are shown via their thermal emission spectra on Earth (Glindemann et al. 2005). It is controver- for a variety of atmospheres for intercomparison: CH3Cl sial whether the microorganisms in these environments (Figure 4); DMS (Figure 5); N2O (Figure 6); NH3 (see are making PH3, or whether the bacteria are making Seager et al. (2013) Figure 2), and via their transmission acid which is attacking environmental iron that con- spectra for H2-dominated atmospheres (Figure 7). 4 Life might produce molecules with elements in intermediate 4.1. redox states as life does on Earth. In an H2 atmosphere such Type I Biosignature Gases: Fully Reduced Forms molecules are likely to be photochemically hydrogenated. 9

Molecule Mixing Wave- Surf. Flux Biomass Surf. Flux Biomass Surf. Flux Biomass Dominant Ratio length Sun-Like Estimate Active M Estimate Quiet M Estimate Removal [ppm] [µm] [m−2 s−1] [gm−2] [m−2 s−1] [gm−2] [m−2 s−1] [gm−2] Path Type I 15 −4 14 −6 5 −6 NH3 0.10 10.3-10.8 2.4×10 4.0×10 5.1×10 8.0×10 8.2×10 9.5×10 photolysis Type III 17 3 15 11 CH3Cl 9.0 13.0-14.2 1.0×10 2.8×10 2.9×10 77 4.7×10 0.013 H DMS 0.10 2.2-2.8 4.2×1019 190 1.8×1019 82 2.4×1013 1.1×10−4 O 17 7 17 7 11 CS2 0.59 6.3-6.9 8.7×10 5.5×10 3.6×10 2.3×10 5.9×10 37 O OCS 0.10 4.7-5.1 2.5×1015 1.3×105 1.0×1014 5.5×103 1.3×1010 0.67 H 15 14 11 N2O 0.38 7.5-9.0 3.8×10 – 5.4×10 – 1.3×10 – photolysis

Table 2 Results for thermal emission spectra. Potential biosignature gas required concentrations, related required biosignature gas surface ﬂuxes (in units of molecules m−2 s−1), estimated biomass surface densities, and the dominant removal path or destructive species. Results are given for three cases: for a planet orbiting a sun-like star, a weakly active M5V dwarf star (denoted“Active M”) and a quiescent M5V dwarf star (denoted “Quiet M”). The planet considered has 10 M⊕, 1.75 R⊕, an atmosphere with 90% H2 and 10% N2 by volume, with a surface temperature of 290 K and a surface pressure of 1 bar. Note that compounds with the removal path dominated by O, the required surface ﬂux sensitively depends on the CO2 emission/deposition.

Molecule Mixing Wave- Surf. Flux Biomass Surf. Flux Biomass Dominant Ratio length Active M Estimate Quiet M Estimate Removal [ppm] [µm] [m−2 s−1] [gm−2] [m−2 s−1] [gm−2] Path Type I 16 7 −9 NH3 11 2.8-3.2 5.5×10 1.1 8.8×10 1.8×10 photolysis Type III 15 2 11 −2 CH3Cl 10 3.2-3.4 3.2×10 8.6×10 5.2×10 1.4×10 H DMS 0.32 3.1-3.6 5.8×1019 2.6×102 7.8×1013 3.6×10−4 O 17 7 11 CS2 0.38 6.4-6.9 2.3×10 1.5×10 3.8×10 24 O OCS 1.8 4.7-5.1 1.9×1015 9.9×104 2.3×1011 12 H 15 12 N2O 11 3.8-4.1 4.8×10 – 3.7×10 – photolysis

Table 3 Results for transmission spectra. Potential biosignature gas required concentrations, related required biosignature gas surface fluxes (in units of molecules m−2 s−1), estimated biomass surface densities, and the dominant removal path or destructive species. Results are given both for two cases, a planet orbiting a weakly active M5V dwarf star (denoted “Active M”) and a quiescent M5V dwarf star (denoted “Quiet M”). The planet considered has 10 M⊕, 1.75 R⊕, an atmosphere with 90% H2 and 10% N2 by volume, with a surface temperature of 290 K and a surface pressure of 1 bar. Note that a planet orbiting a sun-like star is not considered for transmission spectra because the overall detection signal is too low because of the small planet atmosphere annulus area to sun-like star area. Note that compounds with the removal path dominated by O, the required surface flux sensitively depends on the CO2 emission/deposition. tains traces of phosphide, and this attack is making the We have therefore proposed NH3 as a biosignature gas phosphine gas (Roels & Verstraete 2001). Phosphine is in an H2-rich atmosphere (Seager et al. 2013). NH3 is a potential biosignature in other highly reduced envi- a good biosignature gas candidate for any thin H2-rich ronments. Phosphine is reactive and thermodynami- exoplanet atmosphere because of its short lifetime and cally disfavored over elemental phosphorus and hydrogen lack of geological production sources. NH3 as a biosig- at Earth surface pressure and temperature. Phosphine nature gas is a new idea, and one that is specific to a might be a Type I biosignature gas under conditions of non-Earth-like planet. On Earth, NH3 is not a useful very high H2 pressure, which would favour production biosignature gas because, as a highly valuable molecule of PH3 over elemental phosphorus. Phosphine could be for life that is produced in only small quantities, it is also be produced as a Type III biosignature gas, analo- rapidly depleted by life and is unable to accumulate in gous to reactive signaling molecules such as NO or C2H4 the atmosphere. NH3 is also a very poor biosignature gas on Earth. on Earth because it is very soluble, so the trace amounts produced will stay dissolved in water and not escape to 4.1.2. NH3 as the Strongest Candidate Biosignature Gas in the atmosphere. an H2 Atmosphere The summary of the biosignature gas idea is that NH3 NH3 is the strongest candidate biosignature gas in a would be produced from hydrogen and nitrogen, in an thin, H2 atmosphere because, like O2 in Earth’s atmo- atmosphere rich in both, sphere, there is no plausible geological or photochemical 3H + N → 2NH . (18) mechanism for producing high concentrations on rocky 2 2 3 planets with thin atmospheres (but c.f. the false posi- This is an exothermic reaction which could be used to tive discussion below). NH3 is readily photolyzed in the capture energy. The industrial version of this reaction is upper atmosphere to yield N2 and in volcanic gases is called the Haber process for ammonia production at high thermally broken down at high temperatures. The triple temperatures; hence we call such a planet a cold Haber bond of N2 makes it extremely kinetically stable and so world. We proposed that in an H2-rich atmosphere, life any N in the atmosphere ends up being trapped as N2. could find a way to catalyze the breaking of the N2 triple 10 bond and the H2 bond to produce NH3, and capture the the same weakly active M dwarf star, the optimal wave- energy released. In contrast, life on Earth solely fixes length range for detection is 2.8–3.2 µm, the required nitrogen in an energy-requiring process. Energy capture concentration is 11 ppm, and the required surface source 16 −2 −1 would yield an excess of NH3 over that needed by life flux is 5.5 × 10 molecule m s , resulting in a sur- 2 to build biomass, and so the excess would accumulate in face biomass of 1 g m . For this particular example, NH3 the atmosphere. Is a cold Haber World possible? We be- in transmission vs. thermal emission, it is more difficult lieve yes, based on synthetic chemistry on Earth that can to detect NH3 and hence a higher biomass is actually catalyze the breakage of each of H2 (Nishibayashi et al. required for the same hypothetical type of life. 1998) and N2 bonds (Yandulov & Schrock 2003; Schrock We emphasize that the NH3 biosignature gas concept is 2011) at Earth’s surface pressure and temperature; what not changed for a planet with a massive (yet still “thin”) is not yet known is a catalytic system that can break atmosphere with high surface pressure. As long as the both at once. surface conditions are suitable for liquid water, NH3 will We showed in Seager et al. (2013) that for an Earth- not be created by uncatalyzed chemical reactions. size, Earth-mass planet with a 1 bar atmosphere of 75% NH3 is not immune to false postives. Although a rocky by volume N2 and 25% by volume H2 (including carbon planet with a thin H2-dominated atmosphere is unlikely species via a CO2 emission flux), a potentially detectable to have an NH3 false positive, the challenge is in iden- NH3 atmosphere concentration of 0.1 ppm is sustainable tifying the planetary (and stellar) characteristics. We by a very reasonable biomass surface density of 9 × 10−2 describe three scenarios that could lead to the nonbio- −2 g m . This modest surface density corresponds to a logical production of NH3. layer less than one bacterial cell thick. For compari- A rocky world with a hot surface of ∼820 K could son, the phytoplankton that are the major contributor to generated NH3 by the conventional Haber process if there Earth’s oxygen atmosphere are present in Earth’s oceans is surface iron. Such a hot surface temperature could at around 10 g m−2. For interest, we note that standard presumably be ruled out from other observations. −2 printer paper is between 80 and 100 g m . A second scenario where NH3 is naturally occuring is in For an H2-dominated atmosphere with 90% H2 and the atmospheres of gas giant planets or the so-called mini 10% N2 on a planet with 10 M⊕, 1.75 R⊕ orbiting a sun- Neptunes. The deep atmosphere may reach conditions like star, but all other parameters the same as the above, where NH3 can be formed kinetically at the extremely the viability of NH3 as a biosignature gas in a thermal high pressures necessary for NH3 formation to be pos- emission spectrum still holds based on a physically rea- sible thermodynamically. On Jupiter, for example, the sonable biomass surface density. We now describe the es- H2 + N2 → NH3 reaction becomes significant in compar- timate for the biomass surface density, using the Type I ison with vertical transport at about 1500 K, 1400 bar biomass equation (equation (4)). We use the NH3 source (Prinn & Olaguer 1981). The only way we can discrim- flux of 2.4×1015 molecule m−2 s−1 (see Table 2). To com- inate between planets with a massive envelope and a pute ∆G we used T = 290 K, and reactant and product rocky planet with a thin atmosphere where the pres- concentrations at the surface in terms of partial pressures for the thermodynamic formation of NH3 are not −7 sures of N2 = 0.1, H2 = 0.9, NH3 = 1.4 × 10 , giving reached, is with high-resolution spectra to assess the sur- −1 −6 −1 −1 ∆G = 85.6 kJ mole . Pme = 7.0 × 10 kJ g s , face pressure (Benneke & Seager 2012, 2013). − − we find a biomass surface density of 4.9 × 10 2 g m 2. A third scenario for an NH3 false positive is for planets Based on a reasonable biomass surface density, we there- with outgassed NH3 during evolution. The importance of fore consider the NH3 production flux to be viable in ammonia for the atmospheric evolution of Titan relates our Haber World scenario. The global annual biogenic to primordial ammonia which accreted with the ices of NH3 surface emission in the Haber World would be about the moon and has not subsequently been broken down 1100 Tg yr−1. This is much higher than the Earth’s either by internal heat (likely on a rocky planet) or by −1 natural NH3 emission at 10 Tg yr (Seinfeld & Pandis external UV photolysis (which will rapidly break down 2000). Comparing NH3 production on the Haber world any NH3 in the atmosphere) (Shin et al. 2012). In this and on Earth, however, is not valid. We are postulating case, ammonia is therefore present as ice in the interior. that production of NH3 on the Haber world is a major This would be a challenging case to ascertain, and illus- source of metabolic energy for life. A better emission rate trates of how an assignment of any gas as a biosignature comparison is to the biosignature gas O2 from Earth’s gas candidate has to be given a detailed probabilistic as- principle energy metabolism, photosynthesis. Earth’s sessment based on what we know about the planet con- global oxygen flux is 200 times larger than the Haber cerned. 5 −1 World’s NH3 surface emission, at about 2 × 10 Tg yr For any case, a quiescent M star with no chromospheric (Friend et al. 2009). UV emission—hence a planet with little to no destruc- Turning to a weakly active M5V dwarf star, for the tive UV flux—NH3 can easily accumulate in the planet same fiducial planet, the NH3 surface flux required to atmosphere and act as a signficant false positive. NH3 sustain a detectable level of atmospheric NH3 in a ther- is destroyed by photolysis and is very sensitive to the mal emission spectrum is 5.1 × 1014 molecule m−2 s−1 amount of UV radiation. (see Table 2). This value is about 5 times lower than the 4.1.3. CH4 and H2S as Unlikely Biosignature Gases sun-like star example above and therefore converts into a biomass estimate of about 5 times smaller than the CH4 has been described at length as a possible biosig- sun-like star example above, or about 1 × 10−2 g m−2, nature gas on early Earth and on exoplanets in ox- due to the linear scaling of the problem. For a transmis- idized atmospheres (e.g., Hitchcock & Lovelock 1967; sion spectra measurement for the same planet orbiting Des Marais et al. 2002). This is despite the risk of a geologically derived false positive, because it is believed that 11 in an oxidized environment geological production of CH4 in the familiar Earth environment. Photosynthesis must will be small, and so if enough CH4 is produced it may convert carbon from its environmental form, which is the be attributed to life. This is the case on Earth where at form most thermodynamically stable at surface tempera- least 99% of the atmospheric CH4 derives directly from tures and pressures, into biomass. Biomass is of interme- life or from industrial destruction of fossil hydrocarbons diate redox state (Bains & Seager 2012). The key point formed from past life (Wang et al. 2004). However, the therefore is that in an oxidized environment like Earth, 1775 ppb concentration of CH4 in Earth’s atmosphere photosynthesis must reduce oxidized carbon (CO2) and (Solomon et al. 2007) is not enough to be detected re- will generate an oxidized byproduct. On Earth environ- motely with envisioned space telescope capabilities. mental carbon is captured in photosynthesis, producing CH4 is a poor biosignature gas in an H2-rich atmo- O2 as a byproduct, sphere because it is both produced volcancially and is an H2O+CO2 → CH2O+O2. (20) end product of CO2 photochemistry in the atmosphere. Terrestrial volcanic emission rates of CH4 and CO2 would Here CH2O represents biomass. lead to substantial build-up of CH4 in H2 dominated at- Photosynthesis, by definition, will have the same goal mospheres. Even small amounts of outgassed CO2 will in an H2-rich environment as in an oxidized environ- lead to an accumulation of CH4 in the atmosphere, be- ment; to harvest light energy and to build carbon-based cause CH4 has a very long lifetime in an H2-rich atmo- biomass. Because CH4 is the most thermodynamically sphere and CH4 would be produced by stable gaseous form of carbon in this environment, photosynthesis would oxidize the carbon in CH4 and produce 4H2 + CO2 → CH4 + 2H2O. (19) a reduced byproduct. The lowest energy route is to di- More specifically, considering Earth’s volcanic emission rectly split the CH4, as, rates of CH4 and CO2, and with deposition velocities of −6 −1 −2 −1 CH4 + H2O + X → CH2O + XH, (21) 10 ms for CO2 and zero m s for CH4, CH4 will accumulate up to ∼ 10 ppm in a 1 bar 90% H2, 10% N2 where X is an atom that is oxidized in the environment, atmosphere with a temperature profile similar to Earth’s and has been reduced to XH, consuming energy in the (Hu et al. 2012). Even in the case of no surface CH4 process, and CH2O again represents biomass. (We note emission, CO2 emission into the same atmosphere would that the oxidation state of the oxygen is not changed in lead to the atmospheric production and accumulation of this process, unlike oxygenic photosynthesis on Earth, so CH4 up to 5 ppm. This example is intended to show that formally this is not splitting water even though water is the false positive risk of CH4 is so high in a H2-dominated involved.) atmosphere as to make CH4 an implausible biosignature The null result for biosignature photosynthetic biosig- gas. natures on an H2-dominated atmosphere is based on the H2S is even more unfavorable than CH4 as a biosig- point that most non-metals (C, O, S, the halogens) are nature gas in an H2-rich atmosphere because of the likely to be in their most reduced state already on the same geological false positive issues as with carbon gases. surface of this world, and so cannot play the role of X in An added problem is the generation of aerosols which the above described photosynthesis process. may blanket any spectral features and the fact that the One exception might have been hydrogen, which is ox- H2S spectral features are heavily contaminated by atmo- idized in water and methane, and so a possible photospheric water vapor making them potentially difficult to synthetic reaction is detect (Hu et al. 2013). CH4 + H2O → CH2O+2H2, (22)

4.2. Type II Biosignature Gases: No Viable but again H2 is not a useful biosignature gas because it Biosignature Gases is already present in the H2-dominated atmosphere. For completeness, we describe some other unlikely but Type II biosignature gases are those produced by interesting possibilites for X and XH. Silicon, phosphorus metabolic reactions for biomass building. Biomass build- and boron are likely to be present as the oxidized min- ing on Earth primarily occurs by photosynthesis, which erals silicates, phosphates and borates respectively, but has the dual goal of harvesting light energy to use for using these as a sink for the electrons in photosynthesis, metabolism and also for capturing carbon for biomass for example in the reaction with silica to generate silane, building. We have not identified any useful biosignature gases of 1 1 CH4 + SiO2 → CH2O+ SiH4, (23) Type II in an H2-rich, 1 bar atmosphere. Photosynthesis 2 2 in a reduced environment such as an H2-dominated atmosphere would generate reduced byproduct gases, which requires more energy than the reaction in equation (22) are not useful as biosignature gases because those species under a range of conditions, and so would represent a are already expected to be present in their most reduced very inefficient way of generating biomass. Reduction of a metal with a positive electrochemical forms in the H2-dominated atmosphere. The concept of photosynthesis on a planet with an potential would be more energetically efficient, as for example in the reduction of copper oxide to copper, H2-dominated atmosphere is nonetheless worth some dis- 5 cussion , starting with a brief review of photosynthesis CH4 + 2Cu(O) → CH2O+2Cu+H2O, (24)

5 see AbSciCon 2008 abstract by N. Sleep but produces no volatile product, and is dependent on http://online.liebertpub.com/doi/pdf/10.1089/ast.2008.1246, a supply of oxidized metal. (There are clear parallels and Pierrehumbert & Gaidos (2011) for a discussion of photosynthetic active radiation that reaches the surface under thick H2 atmospheres. 12 with anoxygenic photosynthesis on Earth for this type of trial bioflux production rates. We now show why CH3Cl reaction.) By contrast, the reaction in equation (22) is is a potential biosignature gas in H2 rich atmospheres in limited only by the supply of methane, as life in water low UV environments—because the amount of biomass is not limited by the chemical availability of water. In to generate a detectable concentration of CH3Cl is phys- summary, photosynthesis in the reducing environment ically plausible. We use our biomass estimate framework will either generate H2, which will not be detectable in (§2.2 and Seager et al. (2013)). a hydrogen-dominated atmosphere, or will produce non- Considering the thermal emission spectrum for our volatile products, i.e. not products not in gas form which fiducial planet with a 1 bar atmosphere of 90% H2 and by definition will not be detectable as atmospheric gases. 10% N2, a spectral signature of 9 ppm is required for 4.3. spectral detection using our detection metric. This state- Type III Biosignature Gases are Most Viable in ment is for a spectral band feature in absorption at 13.0- Low-UV Environments 14.2 µm (see Figure 4); this is the band accessible in Life produces many molecules for reasons that are not an H2 atmosphere, weaker than the 6.6-7.6 µm band related to the generation of energy, which we refer to as that would be masked by H2-H2 collision-induced absorp- Type III biosignature gases. The gases are produced for tion. In order to sustain an atmospheric concentration reasons such as stress, signaling, and other physiological of 9 ppm of CH3Cl on our model planet in the habitable functions, and some of these have already been discussed zone for a sun-like, weakly active M5V dwarf star, and quantitatively in detail as biosignature gases in oxidized UV quiet M5V dwarf star, the surface bioflux produc- 17 −2 −1 atmospheres, (e.g., CH3Cl (Segura et al. 2005) and DMS tion rate would need to be 1.0 × 10 molecule m s and other sulfur compounds (Domagal-Goldman et al. (1.7 × 10−7 mole m−2 s−1), 2.9 × 1015 molecule m−2 s−1 2011)). (4.8 × 10−9 mole m−2 s−1), 4.7 × 1011 molecule m−2 s−1 The fate of Type III biosignature gas molecules de- (7.8 × 10−13 mole m−2 s−1), respectively. Estimating pends on the level of relevant reactive species in the at- the biomass with equation (6) and with the lab rate mosphere, and hence on stellar UV flux. In low UV en- at 6.17 × 10−11 mole g−1 s−1 (see Seager et al. 2013), vironments, some Type III gases can accumulate to de- the biomass surface density would need to be about tectable levels. In the relatively high UV environments of 3000 g m−2,80gm−2, 0.001 g m−2 for each star-type re- sun-like stars, in an H2-rich atmosphere, many Type III spectively. A globally averaged density of 3000 g m−2 is gases could be rapidly driven to their most hydrogenated likely too high, one of 80 g m−2 is high but not impossi- form, and in some cases will not accumulate to detectable ble, according to terrestrial biodensities (see Seager et al. levels unless we assume unrealistic production rates. In 2013). these extreme cases in a high UV environment, we would The results show that CH3Cl is a more viable biosig- only be able to infer the presence of the biosignature gas nature gas in low-UV as compared to high-UV environ- by detecting the end-product of photochemical attack, ments. We emphasize that although our estimates of which we call a bioindicator. Only in a few cases might biomass surface density for Type III biosignature pro- bioindicators be useful, because many are not spectro- duction are approximate, the resulting trend is robust. scopically active (and hence not detectable) and others For a spectral detection in transmission for our fidu- are indistinguishable from geological cases as well (e.g., cial Earth transiting an M5V star, the required con- DMS will end up as CH4 and H2S, and N2O will end up centration is about 10 ppm in the wavelength range as N2 and H2O.) 3.2-3.4 µm. The surface bioflux and biomass esti- We now show that Type III biosignature gas survival mates for a weakly active and quiet star respectively and hence plausibility depends highly on the UV flux are 3.2 × 1015 molecule m−2 s−1 and 900 g m−2 and level of the host star. We consider the three fiducial 5.2 × 1011 molecule m−2 s−1 and 0.001 g m−2. The re- star types that differ in UV radiation levels: the sun-like quired biomass surface density for the weakly active M star; the weakly active M5V dwarf star; and the quies- star is higher than the average surface biomass in Earth’s cent M5V dwarf star (Figure 1). We consider the same oceans and the biomass surface density for the quiet M5V model planet as above, a 10 M⊕, 1.75 R⊕ planet with an dwarf star is much lower than Earth’s and very plausible, atmosphere with 90% H2 and 10% N2 by volume, with again emphasizing the trend that low-UV radiation en- a surface temperature of 290 K and a surface pressure vironments are more favorable for Type III biosignature of 1 bar. Results for the cases we modeled are listed for gas accumulation. thermal emission spectra in Table 2 and for transmission The different values for biosignature gas surface flux spectra in Table 3. and for the biomass estimates for transmission spectra Our first example of a Type III biosignature gas is as compared to thermal emission spectra are in general methyl chloride (CH3Cl). CH3Cl is produced in trace due to either or both of longer atmospheric pathlengths amounts by many microorganisms on Earth. The de- and different favorable wavelengths (depending on the tectability of CH3Cl in Earth-like atmospheres in the molecule of interest). low UV environment of UV quiet M stars has already The fate of CH3Cl in its destruction by H, is to end been studied by Segura et al. (2005) and later as a po- up in its fully hydrogenated form, HCl, with the overall tential biosignature gas in more generalized oxidized at- reaction as, mospheres by Seager et al. (2013). Here for the first time we study CH3Cl as a poten- CH3Cl+H2 → CH4 + HCl. (25) tial biosignature gas in a thin H2-rich atmosphere. For this, we go beyond previous work not only by consid- HCl could be a bioindicator. The HCl molecule is stable ering an H2 atmosphere but also by using our biomass to further photochemistry, because if it is photolyzed, estimate framework so as not to be constrained by terres- the Cl atoms generated will be predominately react with 13

300 H Atmospheres CO 2 2 2.8 µm band (see Figure 5). Via photochemistry, this H - H 0 ppm 2 2 0.5 ppm mixing ratio corresponds to a surface ﬂux in our ﬁdu- 5 ppm 50 ppm cial H2-dominated atmosphere for a sun-like star, weakly 500 ppm 250 H O H O 2 active M5V dwarf star, and quiet M5V dwarf star as 2 4.2×1019 molecules m−2 s−1 (6.9×10−5 moles m−2 s−1), 1.8×1019 molecules m−2 s−1 (3.0×10−5 moles m−2 s−1), 200 CO 13 −2 −1 −11 −2 −1 2 2.4×10 molecules m s (4.1×10 moles m s ), CH 4 respectively. Using a DMS lab production rate of 3.64 × CH Cl −7 −1 −1

BrightnessTemperature [K] 3 CH 3Cl 10 moles g s (Seager et al. 2013), we come up with 150 am implied biomass surface density estimate for the three −2 −2 −4 −2 300 star types of about 200 g m , 100 g m , 10 g m , CO CO N Atmospheres 2 2 2 respectively. The ﬁrst two values are high, but physically 280 H O 2 plausible as compared to Earth biomass surface density 260 ranges. For transmission spectra the numbers are about

H2O a factor of two higher for the weakly active and quiet 240 M5V dwarf star (see Figure 7 and Table 3).

220 300 H2 Atmospheres CO 0 ppm CH 2 CH Cl 4 H - H 0.5 ppm 200 3 2 2 BrightnessTemperature [K] CH 3Cl 5 ppm 50 ppm 180 250 500 ppm H2O 300 H2O CO 2 Atmospheres CH 4 280 H2O CO 2 CO 2 H O CO 2 260 2 200 CO 2 240 CH 4 BrightnessTemperature [K] 220 150 DMS DMS DMS 200 300 CO N Atmospheres CH Cl 2 2 180 3 BrightnessTemperature [K] 280 CO 2 H2O 160 260 2 3 4 5 8 10 20 30 50 80 100 H O Wavelength [microns] 2 240 CH Figure 4. Theoretical infrared thermal emission spectra of a super 4 Earth exoplanet with various levels of atmospheric CH3Cl in a 220 1 bar atmosphere with a surface temperature of 290 K for a planet with 10 M⊕ and 1.75 R⊕. From top to bottom, the panels show the 200 spectra of H2-, N2-, and CO2-dominated atmospheres, respectively, BrightnessTemperature [K] DMS DMS DMS and the detailed compositions of these reference atmospheres are 180 described in §4 for the H2-dominated planet and in Hu et al. (2012) 300 CO Atmospheres for the N2 and CO2 dominated cases. We find that over 5 ppm of 2 280 CO CO H O CH3Cl is required for detection via thermal emission for H2-, N2-, CO 2 2 2 and CO2-dominated atmospheres. 2 260 H2O 240 CH H to reform HCl. HCl would not be expected to be 4 present in significant levels at atmospheric altitudes for 220 spectral detection without life taking non-volatile forms 200 of Cl and putting it into the atmosphere, because all ge- 180 BrightnessTemperature [K] ological sources are non-volatile chlorides (such as NaCl) DMS DMS and any HCl that is volcanically released would be effi- 160 2 3 4 5 8 10 20 30 50 80 100 ciently rained out of the troposphere. The limiting prob- Wavelength [microns] lem is that the HCl spectral features are too weak to be Figure 5. Theoretical infrared thermal emission spectra of a super detectable and are likely to be contaminated by CH4 in Earth exoplanet with various levels of atmospheric DMS in a 1 bar the 3 to 4 µm range. atmosphere with a surface temperature of 290 K for a planet with As a second Type III biosignature gas exam- 10 M⊕ and 1.75 R⊕. From top to bottom, the panels show the ple we consider dimethyl sulfide (DMS). DMS has spectra of H2-, N2-, and CO2-dominated atmospheres, respectively, and the detailed compositions of these reference atmospheres are been studied before in oxidizing atmospheres by described in §4 for the H2-dominated planet and in Hu et al. (2012) Domagal-Goldman et al. (2011) who concluded that for the N2 and CO2 dominated cases. We find that 0.1 ppm of DMS DMS itself is not a potentially detectable biosignature is required for future detection via thermal emission for H2, N2, gas in oxidized atmospheres under sun-like UV radiation, and CO2-dominated atmospheres. but one of its photolytic breakdown products ethane is detectable (we call this a “bioindicator” gas). Using the The DMS results show again that the lowest UV en- same atmosphere and framework as the above CH3Cl vironment is most favorable. There are two other other example, for thermal emission spectra we find a mix- relevant points related to DMS appearing to be a favor- ing ratio required for detection of 0.1 ppm in the 2.2- able biosignature gas in each of the three UV radiation 14 environments studied. The first point is that gases de- rich in H2. Ammonia oxidation requires a strong oxidiz- stroyed by reaction with O (as opposed to gases destroyed ing agent, which again is likely to be missing from the by reactions with H) show a similar surface flux require- environment. ment between the sun-like and weakly active M dwarf N2O as a Type I biosignature gas, therefore, seems star. This is because the release of O from CO2 photoly- unlikely, although not impossible, from very rare envi- sis is largely driven by Lyman-alpha emission which are ronments in which there are oxidized nitrogen species similar at the habitable zones for the sun-like and weakly generated geochemically. active M dwarf star used in this study (Figure 1). N2O, however, could be a Type III biosignature gas The second point is that the high Rlab values used for as NO is for some organisms on Earth. We have calcu- DMS, and hence the low biomass surface density esti- lated the surface fluxes for a detectable amount of N2O mates, are a result of the biology of DMS production. in a thin, H2-dominated atmosphere and find relatively On Earth, DMS is the waste product of consumption of low required surface fluxes. The reason is that N2O is DMSP by marine organisms consuming marine plankton. destroyed by photodissociation, a slower rate than by DMSP is accumulated in large amounts by some marine reaction with H. N2O may therfore be a plausible biosig- species. Thus organisms that generate DMS do not have nature gas candidate, even in an atmosphere subject to to invest their own resources to make DMS, and so are strong UV radiation (see Figures 6 and 7 and Tables 2) not limited to how much they can make. Maximal pro- and 3). A biomass estimate (as a plausibility check) is duction rates are therefore very high. This is discussed not possible for N2O, as it is only known as a Type I further in Seager et al. (2013). biosignature on Earth (and so therefore Type III Rlab In terms of a bioindicator, DMS will react with H2 to rates are not available). generate CH4 and H2S. Neither is a useful bioindicator 5. as CH4 and H2S are expected to be present in the atmo- DISCUSSION sphere naturally. This is in contrast to oxidized atmo- 5.1. What Constitutes an H2-Dominated Atmosphere? spheres, where ethane may be a bioindicator gas, as the expected to be the end product of DMS photodestruc- We have calculated biosignature gas accumulation in tion by combination of methyl radicals generated from an atmosphere with 90% H2 and 10% N2 by volume. A attack of O on DMS (Domagal-Goldman et al. 2011). super Earth exoplanet atmosphere can have many other As a third and fourth Type III biosignature gas ex- gas species. The concentration of the major destructive species, H, O, and OH will depend on the amounts of ample, we used CS2 and OCS. For these two gases we find the same trend as the other Type III biosignature these other gas species. gases, that is in a low UV environment the biosignature As an example, we explore the changing effect of the gases can accumulate (see Tables 2 and 3). The biomass reactive species in an H2-dominated atmosphere for dif- estimates (as a plausibility check) are too high for the ferent UV flux levels, based on the surface flux levels of sun-like and weakly active M dwarf star environments CO2 (Figure 2). A few key points are as follows. The H abundance is almost not affected by the CO2 mixing to be plausible as compared to terrestrial biomass sur- −8 −2 face density values. OCS in the UV environment of a ratios ranging from 10 to 10 . The O abundance de- weakly active M dwarf star may be an exception with an pends on both CO2 and UV levels, such that both a high estimate at the upper limit of plausibility. CO2 level and a high UV flux lead to high atmospheric O. Only in extreme cases (e.g., H2-dominated atmospheres As a fifth example we describe N2O. On Earth N2O is a Type I biosignature gas produced by nitrifying bacteria. with >1% CO2, shown by dashed lines), the abundance of O may be very close to the abundance of H. The OH N2O is not likely to be produced in a thin H2-rich atmosphere because there is unlikely to be much nitrate avail- abundance depends on a complex source-sink network, ultimately driven by H2O and CO2 photolysis. Notably, able. Here we explain further. N2O has been suggested as a biosignature gas in Earths atmosphere (Segura et al. the amount of H is always at least 4 orders of magnitude higher than the amount of OH. 2005). N2O is a Type I biosignature gas formed by two processes on Earth—the oxidation of ammonia by at- The effect of changing the H2 mixing ratio and the mospheric oxygen and the reduction of nitrate in anoxic addition and variation of other active gases on the H environments, concentration will need to be considered on a case-by- case basis as they will react not only with H and OH but 2NH3 + 2O2 → N2O+3H2O, (26) also with other gas species. − NO3 + H → N2O + H2O. (27) 5.2. Can Super Earths Retain H2-Dominated Analogous reactions on a hydrogen-dominated world Atmospheres? would be the reduction of nitrate by atmospheric hy- Whether or not a super Earth planet can retain H2 drogen stably from atmospheric escape is not known. Although NO− + H → N O + H O+OH−, (28) many models and studies for exoplanet atmospheric es- 3 2 2 2 cape exist (see e.g., Lammer et al. 2012, and references or the oxidation of ammonia by a geologically derived therein), the permanent limitation is that there are too oxidant. many unknowns to provide a definitive and quantitative Nitrate is formed on Earth by oxidation of NO gener- statement on which planets will retain H2. One of the ated by lightning in Earths oxygen-rich atmosphere, or challenges is the unknown history and present state of by biological processes—neither are likely in an H2-rich the host star’s EUV flux. Another major challenge is environment, so it is not clear whether nitrate reduction the defining the mechanism for atmospheric escape for a is a useful energy source in a world with an atmosphere given exoplanet, for example whether or not the regime 15

300 H Atmospheres tional detection of H -rich atmospheres will ultimately CO 2 2 2 H - H 2 2 be needed to conﬁrm the scenario of thin H2-dominated atmospheres on super Earths. 250 H O H O 2 2 5.3. Upper Temperatures for Life

N2O Super Earths with H2-dominated atmospheres can 200 have surface temperatures hotter than Earth due to CO 2 0 ppm an H2 greenhouse eﬀect from H2-H2 collision-induced 0.5 ppm 5 ppm

BrightnessTemperature [K] opacities (Borysow 2002; Pierrehumbert & Gaidos 2011). 50 ppm N2O N O N O 150 500 ppm 2 2 While the hypothetical planets we have described in this 300 paper were constructed to have 1 bar atmospheres with N Atmospheres CO 2 2 CO 2 Earth-type surface temperatures, many H2-dominated 280 planet atmospheres are likely to have hotter surface tem- H O 2 peratures than Earth, even for planets orbiting beyond 260 1 AU of their host stars. H2O 240 An important question for understanding the poten- N O 2 tial of biosignature gases on a planet with an H2- 220 dominated atmospheres is therefore, “how hot a planet 0 ppm 0.5 ppm can be and still sustain life?” On Earth, organisms 200 5 ppm BrightnessTemperature [K] 50 ppm N O that grow at 395 K are known (Lovley & Kashefi 2003; 2 N2O N2O 500 ppm 180 Takai et al. 2008) and have been cultured in the lab 300 at elevated pressures equal to in situ pressures. Fur- CO Atmospheres 2 thermore, proteins can function at 410 K to 420 K 280 H2O CO 2 CO 2 (Tanaka et al. 2006; Sawano et al. 2007; Unsworth et al. H O CO 2 260 2 2007) motivating a consensus that life at 420 K is plau- 240 sible (Deming & Baross 1993; Cowan 2004). CH 4 N O Life might exist at temperatures even higher than 220 2 420 K. The main argument for a maximum tempera- 200 0 ppm ture for life involves the temperature at which the basic 0.5 ppm building blocks of life (DNA, proteins, carbohydrates, 180 5 ppm BrightnessTemperature [K] 50 ppm and lipids) break down. Many of the component chemi- 500 ppm N2O 160 cals of life, including DNA, many of the amino acids that 2 3 4 5 8 10 20 30 50 80 100 Wavelength [microns] make up proteins, and many of the key metabolites that allow lifes biochemistry to function are rapidly chemi- Figure 6. Theoretical infrared thermal emission spectra of a super cally broken down above 470 K (e.g., Cowan 2004) The Earth exoplanet with various levels of atmospheric N2O in a 1 bar atmosphere with a surface temperature of 290 K for a planet with maximum temperature at which life could exist therefore 10 M⊕ and 1.75 R⊕. From top to bottom, the panels show the may lie between 420 K and 470K. spectra of H2-, N2-, and CO2-dominated atmospheres, respectively, and the detailed compositions of these reference atmospheres are 5.4. What Surface Pressure is too High? described in §4 for the H2-dominated planet and in Hu et al. (2012) for the N2 and CO2 dominated cases. We find that about 0.4 ppm Many super Earth atmospheres will be much more of N2O is required for future detection via thermal emission for massive than the 1 bar atmosphere on Earth. For tem- H2, N2, and CO2-dominated atmospheres. peratures suitable for the existence of liquid water (see §5.3), the surface pressure could be as high as 1000 bar or higher (Wagner & Pruß 2002). There are three key of rapid hydrodynamic escape was reached in a planet’s points to show that the high surface pressures does not history or which non-thermal mechanism, if any, came destroy the biosignature gases before they can reach the into a dominant role (see Table 4.1 and references therein high atmosphere. in Seager 2010). With an unknown initial atmospheric “Can life generate potentially detectable biogsignature reservoir and an unknown present atmospheric composi- gases under a massive atmosphere?” The answer is yes, tion, the regime and type of atmospheric escape is diffi- provided the surface temperature is compatible with life, cult to impossible to identify. then in principle life can survive and generate biosigna- Some super Earths will have been formed with at- ture gases. The chemistry described in this paper still mospheres with H2, based on both theoretical and ob- holds under a massive atmosphere, because the photo- servational evidence. Theoretically, planetary building chemical destruction occurs above 1 mbar. Furthermore, blocks containing water-rich minerals that can release H we showed in Seager et al. (2012) that the biomass sur- (Elkins-Tanton & Seager 2008; Schaefer & Fegley 2010). face density estimates are unchanged under a massive Observationally, a large number and variety in radius of atmosphere as long as the photochemical loss rate dom- Kepler mini-Neptunes that must have H or an H/He en- inates. For biosignature gases whose loss is dominated velope to explain their radii. So either from outgassing by deposition at the surface (i.e. are absorbed by the or nebular capture of gases, some super Earths should surface), then the biosignature source flux and hence have started out with H2-rich atmospheres and those biomass surface density will scale linearly with planetary with high enough gravity and low enough temperatures atmosphere mass. and/or with magnetic fields should be able to retain The second key question is, “Can the high density and the H2 (e.g., Pierrehumbert & Gaidos 2011). Observa- pressures on the surface under a massive atmosphere gen- 16

1.77 CH Cl 0 ppm 3 0.5 ppm 0.0162 0.081 5 ppm CH 3Cl 50 ppm 1.765 500 ppm 0.0808 0.0161

CO 2 CO 2 1.76 0.0806 0.0161 H2O 1.755 0.0804 H2 - H 2 PlanetRadius Radius] [Earth H2O 0.016 PlanetRadius Stellar Radius / CH 1.75 4 0.0802 1.77 DMS DMS 0 ppm 0.5 ppm 0.0162 0.081 5 ppm 50 ppm 1.765 500 ppm 0.0808 0.0161

1.76 0.0806 0.0161 H2O 1.755 0.0804 H2 - H 2

PlanetRadius Radius] [Earth H O 2 0.016 PlanetRadius Stellar Radius / CH CO 0.0802 1.75 4 2 1.77 N O 0 ppm 2 0.0162 0.081 0.5 ppm 5 ppm 50 ppm 1.765 N O 500 ppm 2 0.0808 0.0161 CO 1.76 2 0.0806

H2O 0.0161 1.755 0.0804 H2 - H 2

PlanetRadius Radius] [Earth H O 2 0.016 PlanetRadius Stellar Radius / CH 4 CO 0.0802 1.75 2 1.77 NH 0 ppm 3 0.5 ppm 0.0162 0.081 5 ppm 50 ppm 1.765 NH 500 ppm 3 0.0808 0.0161 CO 1.76 2 0.0806 0.0161

1.755 0.0804 H2 - H 2 PlanetRadius Radius] [Earth H2O 0.016 PlanetRadius Stellar Radius / CH 4 CO 0.0802 1.75 2 0.2 0.5 1 2 5 10 20 50 100 Wavelength [microns]

Figure 7. Theoretical transmission spectra for potential biosignature gases in a 10 M⊕, 1.75 R⊕, 1 bar atmosphere composed of 90% H2 and 10% N2 and with a surface temperature of 290 K. Potential biosignature gases, including CH3Cl, DMS, N2O, and NH3, have spectral features in infrared wavelengths from 1 to 10 µm, making these gases detectable at various atmospheric mixing ratios (see Table 3). 17 erate false positives?” The answer is almost entirely of atmosphere masses will also exist. A challenge is pre- no, because we have shown that it is largely the Type sented in observational atmosphere studies because we III biosignature gases that are viable biosignature candi- can only “see” to an optical depth of a few and this lim- dates. Recall that Type III gases are those produced for iting optical depth can be reached well above any surface reasons other than energy extraction, and not from chem- for a thick atmosphere. In many cases surface conditions icals in the environment. Therefore they are unlikely to cannot be probed. Ideally, high-resolution spectra can be the product of nonbiological chemistry; a statement be used to tell whether or not the atmosphere is thick that holds even under a massive atmosphere. A com- or thin (i.e. whether or not one can observe down to ment related to NH3 potential false positive is in order. the planetary surface), based on the shape of the spec- For NH3 to be generated from N2 and H2 kinetically, the tral features, as described in detail in Benneke & Seager temperature has to be well above any temperature com- (2012, 2013). patible with life for any pressure where water is liquid, We support the search for biosignature gases regardless extrapolating from the known fact that at 300 bar and of being able to classify a planet as habitable, because 673 K N2 and H2 still need a catalyst to be converted identifying biosignature gas molecules may be more eas- to NH3, and higher pressures shouldn’t change this. The ily attainable than high-spectral resolution characteriza- false positive risk is instead in detecting NH3 without tion of a super Earth atmospheric spectrum. That said, being able to identify the surface as cold enough not to where possible, planetary radii can be used to discrim- possibly generate NH3 kinetically. inate planets worthy of followup since those with small The third key question is, “Will the high surface pres- enough radii can be identified to likely have thin atmo- sures enable fast chemical reactions that destroy the spheres and those with radii large enough to have mas- biosignature gases generated at the surface?” The an- sive H2 or H2/He envelopes are unsuitable (Adams et al. swer is no, for pressures under about 1000 bar. In prin- 2008). ciple, if the upward diffusion or convective motions bring the biosignature gas to higher altitudes faster than the 5.6. The M Star UV Radiation gas is destroyed by kinetic reactions, the higher surface Biosignature gases can more easily accumulate in a pressures will not interfere with biosignature gas acccu- low-UV radiation environment as compared to a high- lumation in the atmosphere. UV radiation environment because the UV creates the Up to 1000 bar, reaction rates extrapolated from low- destructive atmospheric species. We have shown this for pressure kinetic experiments should be valid to an or- H2 atmospheres in this paper and Segura et al. (2005) der of magnitude. The additional caveat is that low- have shown this for Earth-like planet atmospheres. pressure gas kinetics are usually measured at high tem- Whether or not truly UV quiet M dwarfs exist and if perature and we are extrapolating to high pressure and UV activity is correlated with photometric stability is low temperature. For example, at 1000 bar and 300 K, unclear. Recently France et al. (2013) observed a small we estimate the half life of hydrogenation of CH3Cl to be sample of six planet-hosting M dwarf stars with HST 6.0×1011 yr, the half life of DMS to be 3×1010 yr, and the observations at far-FUV and near-UV wavelengths and 20 half-life of N2Oat1 × 10 yr. These numbers are based found none to be UV quiet. Other studies with much on thermochemical equilibrium of H2 and H based on larger numbers of M dwarf stars are ongoing, including the relative Gibbs free energy of formation of atomic hy- some with UV emission from GALEX (A. West, 2013 drogen and T and P (e.g., Borgnakke & Sonntag 2009). private communication). A general understanding is that The overview is that there is very little free H at high magnetic activity, as traced by Hα in M dwarfs decreases pressures and low temperatures since in the absence of with age but that M dwarfs appear to have finite activity UV, H atoms will be generated almost entirely thermo- lifetimes such that the early-type M dwarfs (M0-M3) spin chemically. down quickly with an activity lifetime of about about At pressures above 1000 bar we are less confident that 1 to 2 Gyr whereas later-type M dwarf stars (M5-M7) chemistry can be extrapolated even qualitatively from continue to spin rapidly for billions of years (West et al. low pressure experiments. By 1000 bar, most gases will 2006, 2008). have densities approaching those of their liquids. In- For the time being UV (that is the relevant FUV and creases in pressure will force molecules closer than their EUV) radiation emitted by stars of interest is not usu- van der Waals radii, directly altering molecular orbitals ally measured or theoretically known and so we have and reaction pathways. Below ∼1000 bar we can con- worked with three different UV-radiation environments sider molecules separate entities and we can still con- (Figure 1). sider the molecules chemistry to be qualitatively simi- A relevant point for quiet M stars with extremely low lar to that of their dilute (ideal) gas state, and hence UV radiation (if they exist) is that false positives for order-of-magnitude extrapolations from low pressure gas biosignature gases destroyed by photolysis may also more chemistry is justifiable. easily accumulate. This is relevant for NH3, for which the lifetime in our fiducial H2 planet atmosphere for a planet 5.5. Can we Identify Exoplanets with H2-Dominated Atmospheres that are Potentially Habitable? orbiting a quiet M star is about 1.4 Gyr, according to our photochemistry models. This means that the false Given the argument that life can generate biosignature positive risk that comes from primordial NH3 would be gases on a planet with an H2-rich atmosphere, but that high6. Related issues with other gases that are primarily the surface must have the right temperatures, how can destroyed by photolysis (and not destruction by reactions we identify suitable planets for further study? The prob- 6 lem is that super Earths are observed with a wide range NH3 requires on average two photons for destruction hence its of masses and sizes, and we can anticipate that a range UV lifetime is particularly sensitive to UV levels. 18 with H and O) should be investigated. individual biosignature gases. We considered a fiducial super Earth of 10 M⊕ and 1.75 R⊕ with a 1 bar atmo- 5.7. Detection Prospects sphere predominantly composed of 90% H2 and 10% N2 Is there any hope that the next space telescope, the by volume, and semi-major axes compatible with hab- James Webb Space Telescope could be the first to provide itable surface temperatures. Although deviations from evidence of biosignature gases? Yes, if–and only if–every our fiducial model will yield different spectral features, single factor is in our favor. atmospheric concentrations, etc., the main findings sum- First, we need to discover a pool of transiting planets marized here will still hold. orbiting nearby (i.e., bright) M dwarf stars. Second, the Our major finding is that for H2-rich atmospheres, low- planet atmosphere should preferably have an atmosphere UV radiation environments are more favorable for biosig- rich in molecular hydrogen to increase the planetary at- nature gas accumulation than high-UV radiation envi- mosphere scale height. Third, the M dwarf star needs ronments. Specifically, H is the dominant reactive species to be a UV quiet M dwarf star with little EUV radia- generated by photochemistry in an H2-rich atmosphere. tion. Fourth, the planet must have life that produces In atmospheres with high levels of CO2 atomic O will biosignature gases that are spectroscopically active. be the dominant destructive species for some molecules. Several biosignature gases, if they exist, are detectable The low UV environments of UV quiet M stars are fa- with tens of hours of JWST time, based on our detection vorable for accumulation of biosignature gases in an H2- metric. Although our detection metric assumes photon dominated atmosphere. The high UV environment of noise is the limiting factor, many more detailed simula- sun-like and active M dwarf stars largely prevents biosig- tions of JWST detectability show that spectral features nature gas accumulation due to rapid photochemical de- of similar magnitude are detectable (e.g., Deming et al. struction via H (or sometimes O), its concentration con- 2009). trolled by UV photolysis. High UV radiation is also un- For detecting molecules using transmission spec- favorable for the accumulation of biosignature gases in troscopy, the background exoplanet atmosphere domi- oxidized atmospheres (Segura et al. 2005), although in nated by H2 or CO2 has little effect on the detectability of contrast OH is the main reactive species in oxidized at- the biosignature gases of interest we studied. This is be- mospheres. cause the transmission observations are better performed We investigated the plausibility of a number of biosig- in the near infrared than in the mid infrared because of nature gases, including H2,CH4, H2S, DMS, NH3, N2O, a higher stellar photon flux at near-infrared wavelengths, NO, CH3Cl, HCl. While not exhaustive, we came up and the contamination effects of either of the dominant with some plausible biosignature gases and others that CO2 absorption or collision-induced H2-H2 absorption are unsuitable as biosignature gases, as follows. are minimal in the near infrared. As long as all of the Our list of plausible biosignature gases is dominated biosignature gases of interest have features in the near in- by Type III biosignature gases in low-UV environments. frared (see Figure 7 for the spectral features), these gases These include CH3Cl, DMS, and N2O. Type III are gases may be detected for atmospheres with any level of CO2. produced for specialized functions and therefore could The key issue here, instead of spectral contamination, is well include small molecules as yet unknown. We there- the mean molecular mass. The depth of the transmission fore support the idea of searching for high concentrations spectral feature is 1 order-of-magnitude larger for H2- of gases that do not belong in chemical equilibrium. dominated atmospheres compared with CO2-dominated We also presented a new biosignature gas candidate, atmospheres (see the scale height in equation (1)). NH3, the only one we found reasonable as a Type I biosig- For detecting molecules via thermal emission with fu- nature gas candidate, and one unique to a hydrogen-rich ture direct imaging techniques, one may expect the CO2 environment. Type I are gas produced as byproducts or H2-H2 contamination to be important because thermal from energy extraction from the environment. emission of the planet peaks in the mid infrared where Our list of unlikely biosignature gases is dominated by CO2 and H2-H2 contamination is most substantial. For Type I biosignature gases, as any biosignature gases pro- individual gases, however, there are often multiple ab- duced from energy extraction (such as CH4 or H2S and sorption bands to mitigate this issue. Similarly, a variety numerous others), will be either be produced by geo- of wavelength ranges are usually available to choose from chemical or photochemical processes or likely rapidly de- for the other biosignature gases of interest studied in this stroyed by hydrogenation in a hydrogen-dominated envi- paper (as shown in Figures 4 to 7). ronment. At this point we conclude by emphasizing a related We have not identified any unique biosignature gas pro- point that the plausibility of a specific biosignature gas duced by any type of photosynthesis in a thin H2-rich at- depends on the planet surface gravity, atmospheric pres- mosphere comparable to O2 in oxidized atmospheres. In sure, and other characteristics, because such characteris- an H2-dominated environment the most likely photosyn- tics affect which atmospheric wavelength “windows” are thetic byproduct is molecular hydrogen, already preva- most favorable. Individual planets and their atmospheres lent in the H2-dominated atmosphere, or non-volatile should be considered on a case-by-case basis. mineral products. This is in contrast to O2 produced 6. by photosynthsis in oxidized environments that is quite SUMMARY AND CONCLUSION robust to most false positive scenarios. (We call biosig- We have provided a “proof of concept” that biosigna- nature gases from biomass building Type II.) ture gases can accumulate in exoplanets with thin H2- Bioindicators would be helpful, but aren’t easily or dominated atmospheres. We used a model atmosphere uniquely detectable. The examples we gave were the hy- including a detailed photochemistry code and also em- drogen halides. ployed a biomass model estimate to assess plausibility of Overall, the promise of biosignature gases in H2 atmo- 19 spheres is real. We have aimed to provide a conceptual Lammer, H., Güdel, M., Kulikov, Y., Ribas, I., Zaqarashvili, and quantitative framework to show that there are at T. V., Khodachenko, M. L., Kislyakova, K. G., Gröller, H., least some viable biosignature gases that could be pro- Odert, P., Leitzinger, M., Fichtinger, B., Krauss, S., Hausleitner, W., Holmström, M., Sanz-Forcada, J., duced either by life’s capture of environmental chemical Lichtenegger, H. I. M., Hanslmeier, A., Shematovich, V. I., energy or are in a category of gases produced by ter- Bisikalo, D., Rauer, H., & Fridlund, M. 2012, Earth, Planets, restrial life. We intend for the results here to fuel the and Space, 64, 179 motivation for discovery of habitable Earths and super Lawson, P. R., Lay, O. P., Martin, S. R., Peters, R. D., Earths orbiting M dwarf stars and their atmospheric fol- Gappinger, R. O., Ksendzov, A., Scharf, D. P., Booth, A. J., Beichman, C. A., Serabyn, E., Johnston, K. J., & Danchi, lowup with the JWST. W. C. 2008, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 7013, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series We thank Jean-Michel Desert and Kartik Sheth for Lederberg, J. 1965, Nature, 207, 9 motivating questions. We thank Foundational Questions Lovelock, J. E. 1965, Nature, 207, 568 Institute (FQXI) for funding the seeds of this work many Lovley, D. R. & Kashefi, K. 2003, Science, 301, 934 years ago. Madhusudhan, N. & Seager, S. 2009, ApJ, 707, 24 Mentall, J. E. & Gentieu, E. P. 1970, J. Chem. Phys., 52, 5641 Miller-Ricci, E., Seager, S., & Sasselov, D. 2009, ApJ, 690, 1056 Nishibayashi, Y., Iwai, S., & Hidai, M. 1998, Science, 279, 540 REFERENCES Pierrehumbert, R. & Gaidos, E. 2011, ApJ, 734, L13+ Pilcher, C. B. 2003, Astrobiology, 3, 471 Prinn, R. G. & Olaguer, E. P. 1981, J. Geophys. Res., 86, 9895 Roels, J. & Verstraete, W. 2001, Bioresource Technology, 79, 243 Adams, E. R., Seager, S., & Elkins-Tanton, L. 2008, ApJ, 673, Rothman, L. S., Gordon, I. E., Barbe, A., Benner, D. C., 1160 Bernath, P. F., Birk, M., Boudon, V., Brown, L. R., Allard, F., Hauschildt, P. H., Alexander, D. R., & Starrfield, S. Campargue, A., Champion, J.-P., Chance, K., Coudert, L. H., 1997, ARA&A, 35, 137 Dana, V., Devi, V. M., Fally, S., Flaud, J.-M., Gamache, R. R., Amend, J. P. & Shock, E. L. 2001, FEMS Microbiology Reviews, Goldman, A., Jacquemart, D., Kleiner, I., Lacome, N., Lafferty, 25, 175 W. J., Mandin, J.-Y., Massie, S. T., Mikhailenko, S. N., Miller, Bains, W. & Seager, S. 2012, Astrobiology, 12, 271 C. E., Moazzen-Ahmadi, N., Naumenko, O. V., Nikitin, A. V., Banks, P. M. & Kockarts, G. 1973, Aeronomy. (New York: Orphal, J., Perevalov, V. I., Perrin, A., Predoi-Cross, A., Academic Press) Rinsland, C. P., Rotger, M., Simeˇcková,ˇ M., Smith, M. A. H., Benneke, B. & Seager, S. 2012, ApJ, 753, 100 Sung, K., Tashkun, S. A., Tennyson, J., Toth, R. A., Vandaele, —. 2013, submitted to ApJ A. C., & Vander Auwera, J. 2009, Borgnakke, C. & Sonntag, R. E. 2009, Fundamentals of J. Quant. Spec. Radiat. Transf., 110, 533 Thermodynamics (New Jersey: John Wiley and Sons) Sander, S. P., Friedl, R. R., & Abbatt, J. P. D. e. a. 2011, Borysow, A. 2002, A&A, 390, 779 Chemical Kinetics and Photochemical Data for Use in Cowan, D. A. 2004, Trends in Microbiology, 12, 58 Atmospheric Studies, Evaluation Number 17 (Pasadena, CA: Deming, D., Seager, S., Winn, J., Miller-Ricci, E., Clampin,M., JPL), JPL Publication 10-6 Lindler, D., Greene, T., Charbonneau, D., Laughlin, G., Ricker, Sawano, M., Yamamoto, H., Ogasahara, K., Kidokoro, S.-i., G., Latham, D., & Ennico, K. 2009, PASP, 121, 952 Katoh, S., Ohnuma, T., Katoh, E., Yokoyama, S., & Yutani, K. Deming, D., Wilkins, A., McCullough, P., Burrows, A., Fortney, 2007, Biochemistry, 47, 721 J. J., Agol, E., Dobbs-Dixon, I., Madhusudhan, N., Crouzet, Schaefer, L. & Fegley, B. 2010, Icarus, 208, 438 N., Desert, J.-M., Gilliland, R. L., Haynes, K., Knutson, H. A., Schrock, R. R. 2011, Nature Chem., 3, 95 Line, M., Magic, Z., Mandell, A. M., Ranjan, S., Charbonneau, Seager, S. 2010, Exoplanet Atmospheres: Physical Processes D., Clampin, M., Seager, S., & Showman, A. P. 2013, ArXiv (Princeton: Princeton University Press) e-prints Seager, S. 2013, Science, 340, 577 Deming, J. W. & Baross, J. A. 1993, Geo. Cosmo. Acta, 57, 3219 Seager, S., Bains, W., & Hu, R. 2013, ApJ, in press Des Marais, D. J., Harwit, M. O., Jucks, K. W., Kasting, J. F., Seager, S., Kuchner, M., Hier-Majumder, C. A., & Militzer, B. Lin, D. N. C., Lunine, J. I., Schneider, J., Seager, S., Traub, 2007, ApJ, 669, 1279 W. A., & Woolf, N. J. 2002, Astrobiology, 2, 153 Seager, S. & Sasselov, D. D. 2000, ApJ, 537, 916 Domagal-Goldman, S. D., Meadows, V. S., Claire, M. W., & Seager, S., Schrenk, M., & Bains, W. 2012, Astrobiology, 12, 61 Kasting, J. F. 2011, Astrobiology, 11, 419 Seager, S., Whitney, B. A., & Sasselov, D. D. 2000, ApJ, 540, 504 Elkins-Tanton, L. T. & Seager, S. 2008, ApJ, 685, 1237 Segura, A., Kasting, J. F., Meadows, V., Cohen, M., Scalo, J., France, K., Froning, C. S., Linsky, J. L., Roberge, A., Stocke, Crisp, D., Butler, R. A. H., & Tinetti, G. 2005, Astrobiology, 5, J. T., Tian, F., Bushinsky, R., Désert, J.-M., Mauas, P., 706 Vieytes, M., & Walkowicz, L. M. 2013, ApJ, 763, 149 Seinfeld, J. H. & Pandis, S. N. 2000, Atmospheric Chemistry and Friend, A. D., Geider, R. J., Behrenfeld, M. J., & Still, C. J. Physics: From Air Pollution to Climate Change (Hoboken, NJ: 2009, in Photosynthesis in silico: Understanding Complexity John Wiley and Sons, Inc.) from Molecules to Ecosystem., ed. A. Laisk, L. Nedbal, & B. V. Shin, K., Kumar, R., Udachin, K. A., Alavi S., & Ripmeester, Govindjee (Springer), 465–497 J. A. 2012, Proceedings of the National Academy of Sciences, Gardner, J. P., Mather, J. C., Clampin, M., Doyon, R., 109, 14785 Greenhouse, M. A., Hammel, H. B., Hutchings, J. B., Jakobsen, Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., P., Lilly, S. J., Long, K. S., Lunine, J. I., McCaughrean, M. J., Averyt, K. B., Tignor, M., & Miller, H. L. 2007, Contribution Mountain, M., Nella, J., Rieke, G. H., Rieke, M. J., Rix, H.-W., of Working Group I to the Fourth Assessment Report of the Smith, E. P., Sonneborn, G., Stiavelli, M., Stockman, H. S., Intergovernmental Panel on Climate Change, 2007 (Cambridge: Windhorst, R. A., & Wright, G. S. 2006, Space Sci. Rev., 123, Cambridge University Press) 485 Takai, K., Nakamura, K., Toki, T., Tsunogai, U., Miyazaki, M., Glindemann, D., Edwards, M., Liu, J. A., & Kuschk, P. 2005, Miyazaki, J., Hirayama, H., Nakagawa, S., Nunoura, T., & Ecological Engineering, 24, 457 Horikoshi, K. 2008, PNAS, 105, 10949 Hitchcock, D. R. & Lovelock, J. E. 1967, Icarus, 7, 149 Tanaka, T., Sawano, M., Ogasahara, K., Sakaguchi, Y., Hu, R. 2013, PhD thesis, MIT Bagautdinov, B., Katoh, E., Kuroishi, C., Shinkai, A., Hu, R., Seager, S., & Bains, W. 2012, ApJ, 761, 166 Yokoyama, S., & Yutani, K. 2006, FEBS Letters, 580, 4224 —. 2013, Submitted to ApJ Tijhuis, L., van Loosdrecht, M. C. M., & Heijnen, J. J. 1993, Kaltenegger, L. & Traub, W. A. 2009, ApJ, 698, 519 Biotechnology and Bioengineering, 42, 509 20

Unsworth, L. D., van der Oost, J., & Koutsopoulos, S. 2007, West, A. A., Bochanski, J. J., Hawley, S. L., Cruz, K. L., Covey, FEBS Journal, 274, 4044 K. R., Silvestri, N. M., Reid, I. N., & Liebert, J. 2006, AJ, 132, Wagner, W. & Pruß, A. 2002, Journal of Physical and Chemical 2507 Reference Data, 31, 387 West, A. A., Hawley, S. L., Bochanski, J. J., Covey, K. R., Reid, Wang, J. S., Logan, J. A., McElroy, M. B., Duncan, B. N., I. N., Dhital, S., Hilton, E. J., & Masuda, M. 2008, AJ, 135, 785 Megretskaia, I. A., & Yantosca, R. M. 2004, Global Yandulov, D. V. & Schrock, R. R. 2003, Science, 301, 76 Biogeochem. Cycles, 18, GB3011