Harvard-Smithsonian Center for Astrophysics Laser Frequency Combs for Precision Radial Velocity Measurements in Astrophysics David F. Phillips Harvard-Smithsonian Center for Astrophysics A collaboration between astronomers (CfA), physicists (CfA) and electrical engineers (MIT) ~15 scientists

“Astro-Comb” • Precision spectroscopy essential for advances in astrophysics • Currently limited by spectrograph wavelength calibration • Laser frequency combs can solve calibration problem • Rapid progress in last 3 years: concept => lab demo => operation at observatories • Coming soon: discovery & characterization of Earth-like planets • 10-20 years: direct measurement of cosmological dynamics Diddams et al., Nature 445, 627 (2007). Murphy et al., MNRAS 380, 839 (2007).

Unique Opportunity Origins of Life in the Universe Over the next decade, we will: • Discover the first Earth-like planets around other stars • Search these planets for chemical signatures of life • Image planets directly Is the Earth special, or just another planet? => Complete Copernican revolution? http://origins.harvard.edu/

Extrasolar Planets 400 exoplanets found to date Hot Jupiters Primarily hot Jupiters “Super Earths” now being found Soon, earth-like planets Hot Jupiters Jupiter mass planets close to their stars Primarily Radial Velocity Method to date Sensitive to massive close-in planets Automated transit surveys underway Will find earth-like planets Super-Earths Radial velocity sensitivity Habitable zone (Sasselov 2008)

Super Earths Image: S.Cundiff Planets could have similar size, but be very different in mass and composition (Sasselov 2008)

Exoplanet Detection Methods Direct detection (107-1010 brightness difference) Radial velocity (determine mass – m sin(i) – from reflex motion) Transit method (determine radius from eclipses) Microlensing (no follow-up possible) Pulsar timing (exotic environment for planets)

Direct Detection HR 8799 d=129 lt-yr Rplanets = 20-70 AU m = 10 mJ Challenging Stars are bright, planets are dim and close to stars Two techniques -- adaptive optics and a coronagraph -- to minimize the glare from the star and reveal the dim glow of the much fainter planets. http://www.jpl.nasa.gov/news/news.cfm?release=2010-128

Spitzer space telescope at 8 µm Transit Method HD 189733B Hot jupiter 63 light-years away Mass = 1.13 MJ Radius = 1.14 RJ T=1000 K P = 2.2 days secondary eclipse Spitzer space telescope at 8 µm primary eclipse Heather Knutson & Dave Charbonneau (2007)

Transits: Kepler spacecraft launched 3/6/09 Transit method, free of Earth’s atmosphere http://kepler.nasa.gov/ Search 100,000 nearby stars over next 3 years Expect to find many Earth-size planets in “habitable zone” 43 days of science data on more than 156,000 stars recently released Only gives planet size & orbit period, not mass

Radial Velocity Method Unseen exoplanet Reflex motion of star due to planet • Heavier planet causes larger Doppler shift in star light • Gives planet mass and orbital period, not size • Doppler-shift from Earth-mass planet is VERY SMALL

Stellar Spectrum 100,000 absorption lines

Spectrograph Small frequency shifts => Use echelle spectrograph => broad spectral coverage & high-resolution corrector collimator slit 2D CCD array From telescope cross disperser (low-dispersion) echelle grating (high dispersion) Slit is often a multimode fiber, allowing the spectrograph to reside in an environmentally controlled room, away from the telescope State-of-the-art astrophysical spectroscopy is broadband, photon-starved => spectrograph resolution R = λ/Δλ ~ 100,000 => minimum resolution element Δν > 5 GHz

Resolution Linewidth of typical stellar lines ~1 GHz Doppler shift: Δλ/λ = ΔRV/c Δλ ΔRV 10–7 Å 1 cm/s 10–6 Å 10 cm/s 10–5 Å 1 m/s 0.01 Å 1 km/s Earth-mass planet around Sun-like star Pre-comb sensitivity “Hot Jupiter” Resolution of spectrograph Δλ/λ ~ 105 at 500 nm: 5 GHz line splitting: 1000-10,000 Need high signal-to-noise ratio and many lines

Octave-Spanning Femtosecond Ti:Sapphire Laser pump laser 532 nm BaF2 plate & wedge lens Ti:Sa crystal red comb • octave-spanning • ~1 GHz rep rate • 105 comb lines PZT Double-chirped mirror pairs => compensate broadband dispersion • Stabilize ωr => feedback to external cavity length • Stabilize ωCE => feedback to pump laser power < 4 fs Broadband gain, Kerr effect => octave spanning Kerr-lens mode-locking => stabilize pulsing Cavity sets Trep ~1 GHz

Octave-Spanning Femtosecond Ti:Sapphire Laser pump laser 532 nm BaF2 plate & wedge lens Ti:Sa crystal red comb • octave-spanning • ~1 GHz rep rate • 105 comb lines PZT Double-chirped mirror pairs => compensate broadband dispersion • Stabilize ωr => feedback to external cavity length • Stabilize ωCE => feedback to pump laser power < 4 fs => spectrograph resolution R = λ/Δλ ≤ 100,000 => minimum resolution element Δν > 5 GHz => optimal calibrator line-spacing > 10 GHz But such high-repetition-rate, octave-spanning combs do not yet exist…

Astro-Comb Phase lock box Synthesizer Photo-detector Fabry-Perot cavity Astro-comb beam Octave-spanning 1-GHz laser frequency comb PZT λ/2 EOM Photo-detector Diode laser PID lock box Synthesizer Remove most comb lines with a stabilized Fabry-Perot Cavity to yield line-spacing up to ~30 GHz

Astro-Comb intensity wavelength Fabry-Perot cavity Astro-comb beam PID lock box λ/2 Photo-detector Fabry-Perot cavity PZT Octave-spanning 1-GHz laser frequency comb Diode laser Astro-comb beam EOM Synthesizer Phase lock box intensity wavelength

Astro-Comb intensity wavelength Fabry-Perot cavity Astro-comb beam PID lock box λ/2 Photo-detector Fabry-Perot cavity PZT Octave-spanning 1-GHz laser frequency comb Diode laser Astro-comb beam EOM Synthesizer Phase lock box intensity wavelength

Filter Cavity Dispersion-compensating mirrors Bragg-Stack Mirror Pair Effective path wavelength Layers of alternating refractive index Complementary-Chirped Mirror Pair wavelength Effective path Layers of alternating refractive index

Astro-comb Astro-comb tested using a 1.5 m telescope with high-resolution echelle spectrograph (TRES) TRES spectrograph R~40,000

Calibration Spectrum Th:Ar calibrator Astro-comb spectrum 2D CCD echelle grating (high dispersion) 100 µm fiber lens cross disperser (low-dispersion) Th:Ar calibrator Astro-comb spectrum Red astro-comb and Th:Ar spectrum TRES spectrograph on 1.5 m Tillinghast reflector at Mt. Hopkins 2 dimensional spectrum on spectrograph CCD

Red Astro-Comb 100 nm bandwidth 32 GHz spacing 10 GHz FWHM on spectrograph 100 nW per line 25000-40000 counts in 5 minutes through integrating sphere.

Red Astro-Comb SNR A ≈ 0.7 FWHM ≈ 10 GHz (S/N)300 s ≈ 200 (S/N)max ≈ 300 n ≈ 6 pixels ∆ν ≈ 15 MHz or 15 m/s for one peak 250 peaks/order: ∆ν ≈ 1 MHz in one order Murphy (2007)

One order of comb lines superimposed Fit Model Comb ∆ν = 1 GHz S • δI/I S = side mode suppression ∆ν = 3 MHz • δI/I Negligible at the 1 m/s level Fiber 100 µm fiber Modeled as flat top in 2D Half circle in 1D Add skew Optics Modeled as Hermite- Gaussians Up to 16 terms CCD Model as flat Tails from charge diffusion don't appear to help One order of comb lines superimposed Residuals of fit near shot noise floor

Calibration Calculate wavelength calibration as above order by order. Compare calibration to reference frame. Calculate deviation of all orders from mean drift.

Calibration ≤ 1 m/s uncertainties Calculate wavelength calibration as above order by order. Compare calibration to reference frame. Calculate deviation of all orders from mean drift.

Blue Astro-Comb 1 GHz Ti:Sapphire source laser BBO doubling crystal 50 nm bandwidth Fabry-Perot cavity: Bragg stack mirrors Bandwidth limited by air dispersion to ~15 nm Improved flux to spectrograph Phase scrambling

One order of comb lines superimposed Blue Astro-Comb One order of comb lines superimposed faster data rate Residuals of fit 50 cm/s resolution Much closer to shot noise! 420 nm with less fringing. Fiber shaking active.

Green Astro-Comb Photonic Crystal Fiber (PCF) PID lock box λ/2 Photo-detector Fabry-Perot cavity PZT Octave-spanning 1-GHz laser frequency comb Diode laser Astro-comb beam EOM Synthesizer Phase lock box Photonic Crystal Fiber (PCF) High nonlinear coefficient Zero-dispersion wavelength ~700 nm Four-wave mixing Near Gaussian mode profile FP Cavity Complementary-Chirped Mirror Pair

Green Astro-Comb 45 GHz green astro-comb Various Ti:Sapphire settings Astro-comb lines not resolved on OSA Lab demo. Next, test at telescope 240 GHz green astro-comb demo for low-resolution OSA Side suppression minimal

What’s Next? • Observe Earth candidates found by Kepler HARPS clone for northern hemisphere Astro-comb calibrator => ΔRV < 10 cm/s William Herschel telescope, Canary Islands, 4.2 meter mirror • Observe Earth candidates found by Kepler • Use Doppler-shift method to distinguish new Earths from exotics Harvard, Smithsonian, Geneva Observatory

Calibration Wavelength calibration sufficient for RV <10 cm/s => only combs can do the job Murphy et al., Mon. Not. R. Astron. Soc. 380, 839 (2007)

RF beat frequencies => lock ωr & ωCE Octave-Spanning Comb Octave-spanning comb => atomic clock accuracy & stability for all comb lines τpulse < 4 fs • 105 comb lines • narrow lines (<kHz) • coherent • equally-spaced • intense (10 μW/line) ωr ~ 1 GHz ~300 nm nH = 2nL RF beat frequencies => lock ωr & ωCE to atomic clock