Exoplanet Atmospheres at High Spectral Resolution

Jayne Birkby Sagan Fellow, Harvard-Smithsonian Centre for Astrophysics [email protected] http://www.cfa.harvard.edu/~jbirkby @jaynebirkby Detecting molecules with transmission spectroscopy Starlight filtering through an atmosphere during transit is imprinted with the planet’s spectrum 2 Transit depth ~ (Rp/Rs) Relative flux Relative

Time

Credit: Deming (Nature) Transit depth Transit Wavelength Transmission spectra with HST/WFC3 reveal water in hot Jupiters

Deming et al. 2013 Huitson et al. 2013

H2O

Wakeford et al. 2013 Deming et al. 2013

McCullough et al. 2014 Kreidberg et al. 2015

Mandell et al. 2013 Kreidberg et al. 2014

Benneke 2015 Transmission spectra with HST/WFC3 reveal water and clouds in warm Neptunes

Fraine et al. 2014

Water detected (5σ) in dust-free atmosphere HAT-P-11 b Transit depth (ppm) Transit Warm Neptune

Flux variations -4 at 10 level Solar-metallicity, cloud-free Knutson et al. 2014 Hydrogen-poor solar model are important Solar-metallicity, thick clouds

GJ 436 b Warm Neptune Flat featureless spectrum in cloudy/hazy atmosphere (cloudy) spectrum for super-Earth GJ 1214 b (cloudy) for GJ super-Earth 1214 spectrum orbits60 with HST/WFC3 Relative transit depth (x10-6) revealed a featureless a featureless revealed Kreidberg et al. 2014 Kreidberg Even space telescopes suffer systematics >> planet signal Even space telescopes suffer systematics >> planet signal

XO-1 b HST/NICMOS Hot Jupiter HST/WFC3 Deming et al. (2013) Even space telescopes suffer systematics >> planet signal

XO-1 b HST/NICMOS Hot Jupiter HST/WFC3 Deming et al. (2013)

Instruments must be stable Ground-based observations and well-characterized with require similarly bright repeatable results reference stars ! “The nearest transiting potentially habitable planet is ~11 pc away.” ! “The nearest [non-transiting] potentially habitable planet is ~2.5 pc away.” ! (Dressing & Charbonneau 2015) Detecting molecules with High Dispersion Spectroscopy (HDS) A CRIRES/VLT survey of hot Jupiter atmospheres

• CRIRES: CRyogenic high-resolution InfraRed Echelle Spectrograph • R=100,000 spectrograph, 8.2 m mirror • 155hrs • 5 brightest host stars visible from Paranal, Chile (K ~ 4 – 6 mag):

HD 209458 b, HD 189733 b, 51 Peg b, ! Boo b, HD 179499 b HDS detects the radial velocity shift of the planetary spectrum

Dayside

Nightside

5 Relativeline depth10 x τ Boötis b ModelModel COCO lineslines R~100,000 of the the radial detects velocity shiftHDS Nightside Blue-shifted R e l a tive li n e de p th x 105 planetary Model Model CO lines CO spectrum τ Boötis b lines Dayside Red-shifted

Ph a se -0.2 0.2 0.4 0.0 0.8 0.6 Toy model of CO lines Toy model of CO lines 2.309 Wavelength / R~100,000 2.310 Telluric 2.311 2.312 μ Planet 2.313 m Telluric features eliminated by removing common modes in time

SNR~200 per resolution element Phase Wavelength CRIRES pipeline output after alignment

SYSREM - Treats pixel channels as light curves Mean-subtracted spectra as input to Sysrem (1024 light curves per detector) - PCA-like algorithm identifies trends as a function of time After first Sysrem iteration, airmass-like trend removed - Data are self-calibrating

Standard deviation ~ 5x10-4 After optimal Sysrem iterations Combine signal from individual lines via

Birkby et al. 2013 10x nominal model injected cross-correlation HDS detects carbon monoxide RV trail in a transiting hot Jupiter atmosphere 0.8 Toy model of CO lines

0.6

Secondary eclipse

0.4

OrbitalPhase Phase

0.2

RV - Vsys (km/s) Snellen et al. 2010

0.0 Transit CO detected at 5.6σ at 2.3 µm in HD 209458 b during transit with 5 hrs on CRIRES/VLT

-0.2 2.309 2.310 2.311 2.312 2.313 Wavelength / μm HDS detects carbon monoxide RV trail in a transiting hot Jupiter atmosphere 0.8 Toy model of CO lines

Kp = 150 km/s

0.6

Secondary eclipse

0.4

OrbitalPhase Phase

0.2

RV - Vsys (km/s) Snellen et al. 2010

0.0 Transit CO detected at 5.6σ at 2.3 µm in HD 209458 b during transit with 5 hrs on CRIRES/VLT

-0.2 2.309 2.310 2.311 2.312 2.313 Non-transiting planet = spectroscopic binary Wavelength / μm in a carbon monoxidedetects HDS non-transiting on the CO detected at CO detected Observed dayside hemisphere (non-transiting) (non-transiting) hemisphere dayside hot Jupiter atmosphere 6 σ at Planet rest frame Planet (K with 2.3 µm p =110 km/s) CRIRES/VLT in Brogi et al. 2012 Brogi τ Boo b Boo RV trail

Ph a se -0.2 0.2 0.4 0.0 0.8 0.6 Toy model of CO lines Toy model of CO lines 2.309 Wavelength / 2.310 2.311 2.312 μ 2.313 m HDS provides unambiguous detections of CO in hot Jupiter atmospheres

51 Peg b τ Boo b HD 189733 b Non-transiting Non-transiting Transiting [email protected]µm [email protected]µm Brogi et al. 2013 Brogi et al. 2012 [email protected]µm de Kok et al. 2013 Other molecules are more complex and appear at different wavelengths Model CO lines

Model H2O lines 5

Relativeline depth10 x Model CO lines

Model CO2 lines Model CH4 lines Other molecules are more complex and appear at different wavelengths Model CO lines

Model H2O lines 5

Greater uncertainty Relativeline depth10 x Model CO lines in line positions at high temperatures

Model CO2! lines Model CH4 lines More telluric contamination

Spectral time-series @ 2.3μm Time

Wavelength

Spectral time-series @ 3.2μm Time 2000K Wavelength HDS provides unambiguous detections of

Boo b - CO complex51 Peg b - COmolecules + H O in hot Jupiter atmospheres 2 90 140 70 90 140 70 60 120 ) )

° 60 ° 120 50 ) ) -1 -1 50 100

(km s 40 (km s

P 100 P K K 40 Orbital inclination ( Orbital inclination ( 80 30 80 51 Peg b τ Boo b HD 189733 b Non-transiting Non-transiting 30 Transiting 60 [email protected]µm 20 60 [email protected]µm [email protected]µm -40 -20 0 20 40 -80 -60 -40 -20 0 20 Brogi et al. 2013 -1 -1Brogi et al. 2012 de Kok et al. 2013 Vsys (km s ) Vsys (km s ) Significance () Significance ()

-2.7 -1.6 -0.5 0.6 1.7 2.8 3.9 5.0 -3.8 -2.6 -1.4 -0.2 1.0 2.2 3.4 4.6

HD 189733b - CO HD 179949b - CO + H2O 180 160 160 90 90 70 ) ) 70 ° 140 ° 140 60 ) )

-1 60 -1 120 50

(km s 120 (km s

P 50 P K K HD 189733 b 51 Peg b 100 Orbital inclination ( HD 179949 b 40 Orbital inclination ( 100 40 Transiting Non-transiting Non-transiting [email protected]µm [email protected]µm 80 80 [email protected]µm Birkby et al. 2013 30 30 Birkby et al. in prep Brogi et al. 2014 -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 -1 -1 Vsys (km s ) Vsys (km s ) Significance () Significance ()

-4.5 -3.0 -1.5 0.0 1.5 3.0 4.5 -3.0 -1.5 0.0 1.5 3.0 4.5 6.0

See also Rodler et al. 2012; 2013 (CO in τ Boo b & HD 189733 b); Lockwood et al. 2014 (H2O in τ Boo b) Exoplanet atmospheres as fossil records? C/O ratio could reveal where and how a planet formed in its protoplanetary disk

Oberg et al. (2011) Helling et al. (2014) C/O ratio could reveal where and how a planet formed in its protoplanetary disk

Oberg et al. (2011) Helling et al. (2014)

Marois et al. (2010) For HR 8799 planets: i) Super-stellar C/O: core accretion at location ii) Stellar C/O: gas collapse at location Barman et al. 2015, Teske et al. 2014 HR 8799 bcde Simulations identify 3.5µm as spectral ‘sweet spot’ for measuring C/O ratio

Simulation of CRIRES sensitivity de Kok, Birkby et al. (2014) RelativeCorrelation

Wavelength / μm Simulations identify 3.5µm as spectral ‘sweet spot’ for measuring C/O ratio

Simulation of CRIRES sensitivity de Kok, Birkby et al. (2014)

H2O! HD 209458 b CH4! τ Boo b CO2 (Birkby et al. in prep.)

CO! HD 179949 b C/O<1 (Brogi et al. 2014) H2O! CH4 RelativeCorrelation

H2O! CO2! CH4

H2O! CO2

Wavelength / μm Accurate line lists are essential for HDS HD 209458 b shows no evidence of TiO that could potentially cause an inversion layer

Observed Injected -7 =10 TiO VMR -8 =10 TiO VMR -9 =10 TiO VMR -10 =10 TiO VMR -11 =10 TiO VMR

Hoeijmakers, de Kok, Snellen, Brogi, Birkby, Schwarz, 2014 HD 209458 b shows no evidence of TiO that could potentially cause an inversion layer

Observed Injected -7 =10 TiO VMR -8 =10 TiO VMR -9 =10 TiO VMR -10 =10 TiO VMR -11 =10 TiO VMR

Hoeijmakers, de Kok, Snellen, Brogi, Birkby, Schwarz, 2014 Stellar model M-dwarf atmospheres observations HD 209458 b shows no evidence of TiO that could potentially cause an inversion layer

Observed Injected -7 =10 TiO VMR -8

=10 Accurate line lists are crucial TiO for the high resolution technique VMR -9 =10 TiO VMR -10 =10

TiO Data cross-correlated with Barnard’s star (M-dwarf) VMR -11 =10 TiO VMR

Hoeijmakers, de Kok, Snellen, Brogi, Birkby, Schwarz, 2014 Stellar model M-dwarf atmospheres observations Monitoring atmospheric dynamics with HDS Nightside features are deeper and Dayside potentially easier to detect

Night side

Reveals heat circulation de Kok, Birkby et al. (2014) HDS is sensitive to line shape/shift from winds and rotation Winds only Winds+rotation

Pressure

Winds blue-shift entire upper atmosphere

Rotation creates double-peaked line profile

Showman et al. 2013 Measuring the length of an “exoday” Combining high contrast imaging (HCI) and HDS reveals planet angular momentum

Star-planet separation ~ 0.4”

Credit: ESO/A.M. Lagrange HDS currently reaches contrast ratios of 10-4

Hot Jupiters have large RV shifts (~10km/s) but small angular separation on the sky High contrast imaging (HCI) on 8m telescope can reach a raw contrast ratio of 10-3

Star-planet separation is large but RV shift is small

PSF of AO-assisted HCI observations with an 8m telescope at 0.5µm, with a Strehl ratio of 0.3 under 0.6 arcsecond seeing conditions (no SDI, ADI, etc) ∴ HDS+HCI can achieve contrast ratios of 10-7

HCI

HDS

Snellen, de Kok, Birkby et al. (2015) Spectra extracted at every position along the slit and stellar/telluric profile removed

Pixel counts on CRIRES/VLT detector

~0.4″

1 hr total exposure Spectra extracted at every position along the slit and stellar/telluric profile removed Dispersion Position Pixel counts on CRIRES/VLT detector

~0.4″

1 hr total exposure 1 pixel = 0.086 arcsec Long slit aka “Poor man’s” IFU

Residual spectra were cross-correlated with model atmospheres containing CO (and H2O) at different abundances for a range of temperature-pressure profiles. Spectra extracted at every position along the slit and stellar/telluric profile removed Dispersion Position Cross-correlation (CC) values

~0.4″

Long slit aka 1 hour integration “Poor man’s” IFU CO detected in β Pic b. Strongest CC at RV = -15.4±1.7 km/s at ~0.4″

Consistent with position from direct imaging and with a circular orbit. H2O only seen at SNR~2. No methane.

Snellen, Brandl, de Kok, Brogi, Birkby, Schwarz, 2014 CC at planet position is rotationally broadened

= instrument Vrot = 25 ± 3 km/s profile Prot ~ 8.1±1.0 hrs

Assumed:!

- Mp=11±5MJ!

- Rp=1.65±0.06RJ! - Small obliquity! (Radius from Currie et al. 2013)

Snellen et al., 2014 HDS(+HCI) in the future Cross-correlation strength scales ~with the square root of the number of lines Cross-correlation strength scales ~with the square root of the number of lines

1.0 0.8 0.6 0.4 Radiance 0.2 High-resolution model spectrum of CO molecular bands 0.0 1.50 1.55 1.60 1.65 1.70 Wavelength (micron)

- region covered by CRIRES observations 1.0 0.8 0.6 0.4 Radiance 0.2 0.0 2.25 2.30 2.35 2.40 2.45 2.50 Wavelength (micron) Multiple IR high-resolution spectrographs are planned for the near future

Crossfield 2014 Multiple IR high-resolution spectrographs are planned for the near future

Crossfield 2014

Molecular spectra (CO, H2O, CO2, CH4) as function of orbital phase ➔ photochemistry, atmospheric structure versus longitude Multiple IR high-resolution spectrographs are planned for the near future R~25,000 with NIRSPEC/Keck is the lowest resolution so far proven to work for HDS

Crossfield 2014

Molecular spectra (CO, H2O, CO2, CH4) as function of orbital phase ➔ photochemistry, atmospheric structure versus longitude Biomarkers and maps Biomarkers are spectral signatures that indicate past or present life

Earth transmission spectrum (Kaltenegger & Truaub 2009))

Water (H2O) Oxygen (O2) needed for life produced by plants

Ozone (O3) Methane (CH4) tracer of O2 produced by bacteria M-dwarfs are ideal for hunting Earth-like planets

RV Earth-Sun ~0.1 m/s RV Earth-M5V ~1 m/s Transit Earth-Sun ~0.01% Transit Earth-M5V ~0.1% HZ ~ 1 year HZ ~ 12 days

HZ (habitable zone): where water is a liquid M-dwarfs are ideal for hunting Earth-like planets

RV Earth-Sun ~0.1 m/s RV Earth-M5V ~1 m/s Transit Earth-Sun ~0.01% Transit Earth-M5V ~0.1% HZ ~ 1 year HZ ~ 12 days

HZ (habitable zone): where water is a liquid

Hazard warning: Oxygen and ozone produced abiotically by water photolysis (Wordsworth & Pierrehumbert 2014) HDS on an ELT could detect oxygen in Earth-like planets in the HZ of M-dwarfs

Porb=11.8 days

M5V+Earth twin in habitable zone: - observe 12-30 transits (12 day orbit, 3 visible per year) - detect O2 at 5σ - takes 4-20 years for the nearest systems (I~10.5 mag, η⨁=1)

Uses velocity separation Snellen, de Kok, le Poole, Brogi, Birkby 2013; Rodler & Lopez-Morales 2014 HDS+HCI with an ELT could detect Earth-size planets orbiting the nearest stars E-ELT/METIS simulations (infrared IFU)

Contrast curve for planet thermal emission

SNR=8

Uses spatial separation

Snellen, de Kok, Birkby et al., 2015 HDS+HCI with an ELT could detect Earth-size planets orbiting the nearest stars

E-ELT/optical IFU simulations Cross-correlation map for reflected light around Proxima within the habitable zone SNR ~ 10 Contrast ~ 6.0 × 10−7

Snellen, de Kok, Birkby et al., 2015 Uses spatial separation Simulations of HDS+HCI show ELTs can map exoplanet atmospheric surfaces

Snellen et al. 2014

Assuming CRIRES-like+AO instrument on an ELT (39m), mapping of β Pic b would

be twice as efficient as VLT mapping of a brown dwarf (Crossfield et al. 2014) Thoughts for ATLAST:

• HDS is robust against (most) tellurics/variations in the Earth’s atmosphere • Ground-based HDS becomes background limited beyond ~4-5 μm • Ground-based HDS requires bright stars on an 8-m class telescope • HDS should work in optical and UV with enough photons • The future is bright - TESS will provide bright star targets • Passively cooled ATLAST would open up the >5μm regime to HDS [email protected] http://www.cfa.harvard.edu/~jbirkby Take home messages:

• HDS unambiguously identifies molecular features in exoplanet atmospheres and probes their thermal structure, but accurate line lists are crucial. ! • C/O ratios measured with HDS may reveal planet formation mechanism and birth location in protoplanetary disk. ! • HDS+HCI reveals the rotational velocity of giant planets at wide separations. ! • HDS in the era of giant segmented mirror telescopes can identify biomarkers and create maps of atmospheric surface features.

[email protected] http://www.cfa.harvard.edu/~jbirkby