Exoplanet atmospheres

Size: px

Start display at page:

Download "Exoplanet atmospheres"

Augusta Cummings
5 years ago
Views:

Exoplanet atmospheres at high spectral resolution Matteo Brogi Hubble Fellow, CU-Boulder I. Snellen, R. de Kok, H. Schwarz (Leiden, NL) J. Birkby (CfA, USA), S. Albrecht (Aarhus, DK)! J. Bean (Chicago, USA), A.

1 Exoplanet atmospheres at high spectral resolution Matteo Brogi Hubble Fellow, CU-Boulder I. Snellen, R. de Kok, H. Schwarz (Leiden, NL) J. Birkby (CfA, USA), S. Albrecht (Aarhus, DK)! J. Bean (Chicago, USA), A. Chiavassa (Nice, FR), J.-M. Désert (Amsterdam, NL), E. Kempton (Grinnell, USA), M. Line (NASA Ames), A. Sozzetti (Torino, IT) March 31, 216 Observatoire de la Côte d Azur, Nice, France

2 Outline From finding to characterizing exoplanets by studying their atmospheres Detecting molecular species at high spectral resolution and measuring masses and inclinations in the meanwhile Constraining the atmospheric properties thermal inversion layers, rotation, winds Future applications Characterizing TESS targets, combining high-resolution spectroscopy and direct imaging

3 How do we find exoplanets? RADIAL VELOCITIES Periodic shift in stellar lines Fit to determine P, a, e Lower limit on M P (i unknown) TRANSITS Planet orbit seen edge-on Periodic dip in the light curve Depth ~ star/planet area E. de Mooij ESO

4 The golden era of exoplanet discoveries ,+ confirmed exoplanets This plot is misleading! Mass (M Jupiter ) Detection threshold Transits Radial velocities Direct imaging Other Solar System Semi-major axis (AU)

Statistics of exoplanet population Mostly from Kepler on transiting planets (Fressin et al. 213) Planets per star (P < 85 days).2.15.1.5 Earths Super-! Earths Small! Neptunes Large!

5 Statistics of exoplanet population Mostly from Kepler on transiting planets (Fressin et al. 213) Planets per star (P < 85 days) Earths Super-! Earths Small! Neptunes Large! Neptunes FGK stars P =.8-85 days η any = (52 ± 4) % η = (16.6 ± 3.6) % The most common planets have no analogs in the Solar System Giant! Planets Planet radius (R )

6 Measuring the planet bulk density Planets with measurements of both radii and masses Lighter gray = bigger error bars 2. Planet radius (R Jup ) Degeneracies in the planet bulk composition Inflated radii of giant planets.5 Pure H 2 Pure rock Pure H 2 O Pure iron Planet mass (M Jup )

7 The structure and composition of small planets Plot from Berta et al. (215) Lighter gray = bigger error bars The smallest planets seem to be compatible with Earth-Venus composition

8 Observing exoplanet atmospheres From statistical to individual properties of exoplanets Molecular/atomic species, inversion layers, clouds/hazes Solving degeneracies between planet interior-envelope Super-Earths or mini-giants (or something unexpected?) Linking composition to origin & evolution Reliable estimates of C,N,O elemental abundances vs. stellar values Determining surface conditions and hence habitability Atmospheric circulation, T/p profile, composition, biomarkers

9 Atmospheric characterization: transiting planets Star and planet are not spatially resolved Monitoring of the total light from the star+planet system, at various wavelengths Eclipse (Dayside spectra) Planet occulted by the stellar disk Measuring the missing planet flux Transit (Transmission spectra) Starlight filtering through the planet atmosphere Measuring the planet radius Phase curve Relative contribution of planet day/night side Constraints on molecular species & abundances, T/p profile, longitudinal energy balance

10 Atmospheric composition and transmission signals Transit depth D = (R P /R S ) 2 Jupiter-Sun ~ 1% Earth-Sun ~.8% Change in transit depth due to atmospheric opacity D = nhr P /(R S ) 2 H = kt/gµ R S n log( κ) R P

11 Atmospheric composition and transmission signals Transit depth D = (R P /R S ) 2 Jupiter-Sun ~ 1% Earth-Sun ~.8% Change in transit depth due to atmospheric opacity D = nhr P /(R S ) 2 Find and target planets orbiting small stars (late-k and M-dwarfs) R S R P n log( κ) H = kt/gµ Target warmer planets with lower density Maximize the change in opacity between spectral channels How do we maximize the signal?

12 From low- to high-resolution spectra Pont+ 213; Burrows 214 HST (low-res spectroscopy) + Spitzer (photometry) HD b 1.14 M Jup 1.18 R Jup K1-2V star.157 Radius ratio Rayleigh at 1,3 K Settling dust Scale height Cloud deck Wavelength (Å) 11

13 From low- to high-resolution spectra Pont+ 213; Burrows 214 HST (low-res spectroscopy) + Spitzer (photometry) µm R = 1, HD b 1.14 M Jup 1.18 R Jup K1-2V star.157 Radius ratio Rayleigh at 1,3 K Settling dust Scale height Cloud deck Wavelength (Å) 11

14 Molecular fingerprints at high spectral resolution Models by Remco de Kok.159 CO Radius ratio (+ arbitrary shift) H 2 O CH 4 Molecules resolved into individual lines Wavelength (nm)

15 Hot Jupiters at high spectral resolution HJs sensibly move along the orbit during few hours of observations Planet motion resolved Carbon monoxide µm.8.6 blue-shifted.4 red-shifted Orbital phase Wavelength (μm)

16 Hot Jupiters at high spectral resolution HJs sensibly move along the orbit during few hours of observations Planet motion resolved Carbon monoxide µm.8.6 blue-shifted.4 red-shifted Orbital phase.2 Telluric and planet signal disentangled Planet radial velocity can be measured Planet: 1/1 km/s Star: 1/1 m/s Wavelength (μm)

17 Exoplanets at high-spectral resolution Transiting planets are observable: During transit Before/after secondary eclipse Carbon monoxide µm.8.6 Dayside spectroscopy.4 Orbital phase.2 Transmission spectroscopy Wavelength (μm)

18 Exoplanets at high-spectral resolution The thermal spectrum of the planet is targeted directly Dayside Spectroscopy Carbon monoxide µm Orbital phase.2 Dayside spectroscopy applicable to non-transiting planets! Wavelength (μm)

19 The data analysis: a two-step process 1 Removing telluric lines The Earth s atmospheric absorption is stationary in wavelength The planet moves along the orbit and it is Doppler-shifted 5 hours of real data + 2x planet signal (CO) Time Wavelength 16

20 The data analysis: a two-step process 1 Removing telluric lines The Earth s atmospheric absorption is stationary in wavelength The planet moves along the orbit and it is Doppler-shifted 5 hours of real data + 2x planet signal (CO) Time Wavelength 16

21 The data analysis: a two-step process 1 Removing telluric lines The Earth s atmospheric absorption is stationary in wavelength The planet moves along the orbit and it is Doppler-shifted 5 hours of real data + 2x planet signal (CO) Time Wavelength 2 Cross-correlating with model spectra (Models by Remco de Kok) Cross-correlation matrix: CCF(RV, t) Time / orbital phase Planet radial velocity 16

orbit and it is Doppler-shifted 5 hours of real data + 2x planet signal (CO)

Cross-correlation matrix: CCF(RV, t) Shifting and co-adding to planet

22 The data analysis: a two-step process 1 Removing telluric lines The Earth s atmospheric absorption is stationary in wavelength The planet moves along the orbit and it is Doppler-shifted 5 hours of real data + 2x planet signal (CO) Time Wavelength 2 Cross-correlating with model spectra (Models by Remco de Kok) Cross-correlation matrix: CCF(RV, t) Shifting and co-adding to planet rest-frame (Requires knowledge of planet V orb ) Time / orbital phase Planet radial velocity 16

The orbital motion of τ Boo b Brogi+ 212 Night 3 Night 1 3 6

3µm Carbon monoxide detected (6σ) Night 2 Planet radial

velocity (km s -1 ) 9 7 6 5 4 3 Orbital Inclination ( o ) +6

Inferred: Orbital inclination i Planet mass M P = ƒ(m S )!

23 The orbital motion of τ Boo b Brogi+ 212 Night 3 Night hours of VLT/CRIRES, 2.3µm Carbon monoxide detected (6σ) Night 2 Planet radial velocity KP (km s -1 ) Systemic velocity (km s -1 ) Orbital Inclination ( o ) Measured: RV ratio: K P /K S Mass ratio: M P /M S! Inferred: Orbital inclination i Planet mass M P = ƒ(m S )! Uncertainties in M P dominated by uncertainties in M S For τ Boo b: i = (45.5±1.5)º, M P = (5.95±.28) M Jup 17

24 Molecular detections to date Snellen+ 21, 214; Brogi+ 212, 213, 214, 216; Brikby+ 213, de Kok+ 213; Schwarz+ 214 Rodler+ 212, 213; Lockwood planets (2 transiting, 3 non transiting, 1 directly-imaged) CO (6), H2O (4) detected Detections on 9/1 datasets Masses and inclinations for non-transiting planets! CH4 and CO2 not (yet) detected Low abundances expected for CO 2? (e.g. Heng & Lyons 215) Uncertainty: Incorrect / incomplete line lists VLT/CRIRES R=1, 2.3µm, 3.2µm, 3.5µm dayside, transit, night-side 18

25 Modeling the planet atmosphere log(p) C = (TC,pC) C = (TC,pC ) non-inverted! inverted Parametrized T/p profile H 2 -dominated (hot-jupiters) Trace gases: CO, H 2 O, CH 4, CO 2 B = (TB,pB) A = (TA,pA) VMR: for CH 4, CO 2 VMR: for H 2 O, CO Clear atmosphere (inclusion of an optically-thick cloud deck is possible) ΔT T What do we measure? 1. Degree of match between data and models (strength of CCF) 2. Average line/continuum depth! Transit spectra Weak dependence on T or lapse rate Possible influence of hazes/clouds! Emission spectra Line depth T degeneracy between T A, lapse rate, abundances 19

26 Modeling the planet atmosphere log(p) C = (TC,pC) C = (TC,pC ) non-inverted! inverted Parametrized T/p profile H 2 -dominated (hot-jupiters) Trace gases: CO, H 2 O, CH 4, CO 2 B = (TB,pB) ΔT A = (TA,pA) T VMR: for CH 4, CO 2 VMR: for H 2 O, CO Clear atmosphere (inclusion of an optically-thick cloud deck is possible) What do we measure? 1. Degree of match between data and models (strength of CCF) 2. Average line/continuum depth!! Transit spectra Removing degeneracies Weak dependence on T or lapse rate Possible influence! of hazes/clouds Larger spectral range (ideally! the whole NIR) Absolute fluxes from broad-band / low-res measurement Emission spectra Line depth T degeneracy between T A, lapse rate, abundances 19

27 Dayside spectra and thermal inversions log(p) non-inverted! inverted Line cores at low pressures Continuum at high pressures Line depth traces T difference (continuum vs. core) dt/dlog(p) > Absorption lines dt/dlog(p) < Emission lines ΔT T The cross-correlation naturally detect inversion layers (Models with the wrong T-p profile produce anti-correlation) All high-resolution observations have detected absorption lines (But stay tuned for Brogi+ in prep.) No evidence of inversion layers to date in the literature (Knutson+ 28, 21; Diamond-Love+ 214; Schwarz+ 214) 2

28 Testing the synchronous rotation of hot Jupiters HJs become tidally locked on short timescales: P orb = P rot HJs have 2 main regimes of atmospheric circulation Equatorial super-rotation Day- to night-side winds Rotation and winds broaden and distort the planet line profiles (Showman+ 212; Miller-Ricci Kempton+ 212, 214; Rauscher & Kempton 214) 21

Testing the synchronous rotation of hot Jupiters HJs become tidally locked on short timescales: P orb = P rot HJs have 2 main regimes of atmospheric circulation Equatorial super-rotation Day- to

29 Testing the synchronous rotation of hot Jupiters HJs become tidally locked on short timescales: P orb = P rot HJs have 2 main regimes of atmospheric circulation Equatorial super-rotation Day- to night-side winds Rotation and winds broaden and distort the planet line profiles (Showman+ 212; Miller-Ricci Kempton+ 212, 214; Rauscher & Kempton 214) Testing predictions on HD b 1.1 M Jupiter, 1.2 R Jupiter, K1-2V star! 2 hrs of VLT/CRIRES = 1 2.3µm 21

30 15 KP (km/s) CO CO CHO 4 CO+H 2 2 analysis analysis CO 5.5σ Water vapor 5 CH H H2O4 HH2O 4 CH 2O 24O CH CH H 4 H 2O 2O 5.1σ Carbon monoxide 5 CO2 CO CO2 CO CO2 CO CO KP (km/s) Change in planet transit depth x CH H 4 H 2O 2O KP (km/s) KP (km/s) CH H2O KP (km/s) 2 2 CO Brogi+ S/N analysis S/N analysis S/N analysis 216 analysis CO CO CO KP (km/s) CO Model Model spectra spectra KP (km/s) Model -2.5spectra.. 1 KP (km/s) Planet orbital KPvelocity (km/s) (km s-1) -2. KP (km/ KP (km/ -1.5 Depth lines Change / Fstar ( 1) Change in planet transit depthofx molecular 1 in planet transit depth x 1 15 The transmission spectrum of HD b -1. CHO 4 CH4 CO+H CO2 CO 7.6σ Combined signal 5 15 VMR(CO) = VMR(H2O) = (km/s) (km/s) (km/s) CO+H CO+H CH CH CH2O CO+H CO+H 2O 4 CH 4 CH O 2O CH Wavelength (µm) Vrest (km/s) Vrest (km/s) Wavelength (µm) Vrest (km/s) 2 2 Vrest (km/s) No detection of CH4 or CO

31 Modeling the broadened planet line profile Rigid rotation (V rot ) Equatorial band super-rotating (V eq ) No day-to-night side winds (directly from offset of CCF) Line intensity (normalized) Radial velocity (km s -1 ) 23

32 Measuring line broadening from the CCF Planet RV amplitude (km/s) Co-added CCF / σ CCF Best-fit value Synchronous rotation Challenge: CCF is not the true planet profile P rot = Planet rotational velocity (km/s) 1 2 χ 2 = P rot =1.7 6 Equatorial excess vel. (km/s) χ 2 = Planet rest-frame velocity (km/s) Full χ 2 maps of fitted parameters P rot =1. χ 2 =65.5 Observed CCF CCF of models Residuals Planet rotational velocity (km/s) σ 6σ 5σ 4σ 3σ 2σ 1σ 24

33 Measuring line broadening from the CCF Planet RV amplitude (km/s) Co-added CCF / σ CCF Best-fit value Synchronous rotation Challenge: CCF is not the true planet profile P rot = Planet rotational velocity (km/s) 1 2 χ 2 = P rot = Planet rest-frame velocity (km/s) Full χ 2 maps of fitted parameters 6 Equatorial excess vel. (km/s) 1 Data is consistent with 6 synchronous rotation V rot = ( ) km/s 4 P rot = ( ) days χ 2 = P rot =1. χ 2 =65.5 Observed CCF CCF of models Residuals Planet rotational velocity (km/s) σ 6σ 5σ 4σ 3σ 2σ 1σ 24

34 Measuring line broadening from the CCF Planet RV amplitude (km/s) Co-added CCF / σ CCF Best-fit value Synchronous rotation Challenge: CCF is not the true planet profile P rot = Planet rotational velocity (km/s) 1 2 χ 2 = P rot =1.7 6 Equatorial excess vel. (km/s) χ 2 = Planet rest-frame velocity (km/s) Full χ 2 maps of fitted parameters 1 8 Fast rotation 6 excluded P rot < 1.3 days (1σ) P rot < 1.4 days (3σ) 2 1 P rot =1. χ 2 =65.5 Observed CCF CCF of models Residuals Planet rotational velocity (km/s) σ 6σ 5σ 4σ 3σ 2σ 1σ 25

35 Measuring line broadening from the CCF Planet RV amplitude (km/s) Co-added CCF / σ CCF Best-fit value Synchronous rotation Challenge: CCF is not the true planet profile P rot = Planet rotational velocity (km/s) 1 2 χ 2 = No rotation marginally excluded (1.5σ) 4 7 P rot =1.7 6 Equatorial excess vel. (km/s) χ 2 = Planet rest-frame velocity (km/s) Full χ 2 maps of fitted parameters P rot =1. χ 2 =65.5 Observed CCF CCF of models Residuals Planet rotational velocity (km/s) σ 6σ 5σ 4σ 3σ 2σ 1σ 26

36 Measuring line broadening from the CCF Planet RV amplitude (km/s) Co-added CCF / σ CCF Best-fit value Synchronous rotation Challenge: CCF is not the true planet profile P rot = Planet rotational velocity (km/s) 1 2 χ 2 = P rot =1.7 6 Equatorial excess vel. (km/s) χ 2 = Planet rest-frame velocity (km/s) Full χ 2 maps of fitted parameters 1 8 Equatorial 6 super-rotation unconstrained P rot =1. χ 2 =65.5 Observed CCF CCF of models Residuals Planet rotational velocity (km/s) σ 6σ 5σ 4σ 3σ 2σ 1σ 27

37 Day-to-night side winds 5 Co-added CCF / σ CCF Planet rest-frame velocity (km/s) V rest = ( ) km/s Compare with optical high-resolution transmission spectra (Na doublet) Wyttenbach+ (215) V rest = ( 8±2) km s -1 Louden+ (215) V rest = ( ) km s -1 28

38 Day-to-night side winds 5 Co-added CCF / σ CCF Planet rest-frame velocity (km/s) V rest = ( ) km/s Compare with optical high-resolution transmission spectra (Na doublet) Wyttenbach+ (215) V rest = ( 8±2) km s -1 Louden+ (215) V rest = ( ) km s -1 NB: Our data constrains independently orbital, rotational, and wind velocity! 28

39 Targeting smaller and fainter planets Brogi+ in prep. 1. Sample size: 124 Observable: 1 Detectable S/N>5: 6 Current sample (H < 11 mag) Planet mass (M Jup ) Planet radius (R Jup ) S/N < Equilibrium temperature (K) Sample: H < 11 mag, M P >.5 M Jup, R P >.35 R Jup Simulations: APO 2.5m + Apogee (R=22,5, full H band) Spectrum: µm, H2O spectrum for HD b, scaled by: scale height, planet/star radius, host-star H-band magnitude

40 Trading mirror size for better spectrographs ~7% S/N of VLT/CRIRES for equal observing time VLT Apogee SNR (Apogee/VLT) Mirror size 8.2m 2.5m.35 Spectral resolution 1, 22,5.47 Spectral range 5 nm 19 nm 1.9 Throughput % ~15% 2.6 Total.71 Detection estimated by observing a target every time there is a transit Work in progress CH 4 and CH 4 +H 2 O models: relative VMRs Apply the calculations to the expected yield of TESS (Sullivan+ 215)

41 Work in progress: HDS in the TESS era TESS yield from Sullivan , target stars 1,7 detections 67 R P > 4 R 1,1 with 2 R < R P < 4 R 556 with R P < 2 R : 419 around M-dwarfs 137 around FGK stars 13 brighter than K = 9 mag Adapting the APO simulations to range of spectra for super-earths Exploring a range of current and next-generation NIR spectrographs

Spectral and spatial resolution combined: β Pic b 1996 - ESO 3.

42 Spectral and spatial resolution combined: β Pic b ESO 3.6m + ADONIS (ﬁrst imaged in 1984) VLT 8m + NACO Image: courtesy of D. Ehrenreich

Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) 2.

43 Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast imaging, Snellen+ 215) GPI first light - β Pic b" Macintosh et al. (214) +

44 Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast imaging, Snellen+ 215) MACAO@VLT" (Simulated) + +

45 Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast imaging, Snellen+ 215) MACAO@VLT" (Simulated) slit

46 Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast imaging, Snellen+ 215) MACAO@VLT" (Simulated) CRIRES@VLT µm - seeing spectra slit The stellar spectrum is everywhere the same (scaling factor) Spectral direction Spatial direction

6 1-4 contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast

47 Spectral and spatial resolution combined: β Pic b Snellen+, Nature, 214 Young exoplanet (12-21 Myr).44 separation (8.8 AU) contrast " No change in planet RV during the night (Combining high-dispersion spectroscopy and high-contrast imaging, Snellen+ 215) MACAO@VLT" (Simulated) CRIRES@VLT µm - seeing spectra slit Star position Planet position The stellar spectrum is everywhere the same (scaling factor) Spectral direction Spatial direction

48 β Pic b rotates in only 8 hours! Cross-correlation function broadened by 27 km/s Cross correlation with model CO signal (S/N=6.5) 34

β Pic b rotates in only 8 hours! Cross-correlation function broadened by 27 km/s Cross correlation with model CO signal (S/N=6.5) A test for the next decade!

49 β Pic b rotates in only 8 hours! Cross-correlation function broadened by 27 km/s Cross correlation with model CO signal (S/N=6.5) A test for the next decade! High Contrast Imaging: 1-4 planet/star contrast High-Dispersion Spectroscopy: 1-5 core/continuum contrast (photon limited) 1-9 planet/star contrast achievable 34

Simulating HDS+HCI: Metis @ E-ELT Earth-like planet (R=1.5 R, T=3K) orbiting α Cen B (Snellen+ 215).4 39m E-ELT 3 hours 4.85 µm R=1, (IFU) Sky y-distance ( ).2. -.

50 Simulating HDS+HCI: E-ELT Earth-like planet (R=1.5 R, T=3K) orbiting α Cen B (Snellen+ 215).4 39m E-ELT 3 hours 4.85 µm R=1, (IFU) Sky y-distance ( ) Traditional imaging 8σ (3 hrs) 5σ (1 hrs) Traditional imaging + Integral field, hi-res spectroscopy -.4 CCF sliced at RV=3 km/s Sky x-distance ( )

51 Terrestrial planets around dwarf stars The planet/star contrast increases for smaller stars Solar-type star M5-dwarf You are here! M-dwarfs are the most-common stars in the solar neighborhood!

4 mag (τ Boo) -5 Signal x 1 5-1 -15 O2 of Earth around an M5V star CO in τ Bootis b (F7V star) 76 762 764

52 Transiting terrestrial planets around M-dwarfs O 2 in transmission Earth-size planets orbiting M-dwarfs 1/3 of the dayside signal from τ Boo b Challenge: I = 1-11 mag (M-dwarf) vs. K = 3.4 mag (τ Boo) -5 Signal x O2 of Earth around an M5V star CO in τ Bootis b (F7V star) Wavelength (nm) Extremely Large Telescopes + Hi-res spectroscopy 39m E-ELT, 3 transits (3 years) 5σ detection!

53 Conclusions High-resolution transmission spectroscopy can characterize exoplanets Robust molecular detections Masses, inclinations of non-transiting planets Winds and planet rotation Inversion layers Relative molecular abundances and C/O ratio Can be combined with direct imaging Ideal to complement JWST in following-up TESS targets Potentially suitable for targeting rocky planets in HZ of M-dwarfs! 38

Prospects for ground-based characterization of Proxima Centauri b

Prospects for ground-based characterization of Proxima Centauri b Ma#eo Brogi Hubble Fellow, CU-Boulder Ignas Snellen, Remco de Kok, Henrie5e Schwarz (Leiden, NL) Jayne Birkby (CfA, USA), Simon Albrecht