Duncan Wright
Transcription
Duncan Wright
The Hunt for Habitable-Zone Planets with CYCLOPS on the AAT D.J. Wright and C. G. Tinney Department of Astrophysics, University of New South Wales, Sydney, NSW, 2052 focused purely on the characterization of known planets with well-constrained orbits. That Abstract is, a future TPF -style instrument could likely be realized sooner if the census of low-mass Background planets in the habitable zones around nearby stars was known. The most viable candidate stars for finding Habitable zone planets using the radial velocity Filling takingstars. the census of Ear th-size planets around nearby method are the therole M of Dwarf The radial velocity method works by solar-type detecting stars the reflex goingvariation to be theofprime goalasofthe theplanet Space orbits Interferometry (SIM). was radial was velocity the star around Mission it. For the lowSIM mass MaDwarf concept for a space telescope that would indirectly detect exoplanets through the stars the flux from the star is much lower than for higher mass stars and hence the habitable astrometric perturbation of their host stars. The requisite astrometric precision for detecting zone (crudely defined radii range over which liquidThe water be expected to exist small planets wasas tothe be obtained using interferometry. 2010can Decadal Report did not on a planetsrecommend surface) isthat at this much smaller radii due andtohence thecost planets are lead-time. at short periods. mission proceed its large and long With the I will outlineend myof new project to start nextofsemester the of AAT to detect the SIM program, the task taking the on census planets around habitable-zone nearby stars thatsupercan be directly falls solely to ground-based radial velocities. earth planets aroundimaged M Dwarf stars using CYCLOPS. Finding Earth-like planets orbiting in the ‘habitable zone’ of other stars is one of the major goals of modern astronomy. Detecting Earth-like planets, in Earth-like orbits around Sunlike stars is currently beyond reach of any facility except Kepler (and is problematic even then). What is possible now, is the detection of Earth-like planets in habitable-zone orbits around low-mass stars using the radial velocity technique. Early indications from both Doppler planet searches and Kepler are that rocky planets around low-mass stars (<1 M¤) are common (20-30%). We are initiating a search, sensitive to low-mass planets orbiting in the habitable zones of M4-M6 dwarf stars. This search will utilize a new high-precision arclamp wavelength calibration method developed for the CYCLOPS fibre-feed to the UCLES spectrograph. Table 1 – Orbital Properties of a 5MEarth Habitable Zone Planet Orbiting an M Dwarf Host M Dwarf Semi-major Axis Period Doppler Amplitude Mass (M¤) Sp. Type Inner Outer Inner Outer Inner (m s-1) Outer (m s-1) 0.1 M6 0.021 AU 0.061 AU 3.5 d 18 d 9.8 5.7 0.3 M3 0.09 AU 0.21 AU 4.5 d 70 d 4.4 1.8 Expected RV Amplitudes Searching for low mass (rocky) planets around the lowest mass stars (late M Dwarfs) has two very significant advantages over larger stars: 1 – The Doppler signature of any resident planets scales inversely with the mass of the star. M4-M6 dwarfs are 0.3-0.1 M¤ hence Doppler amplitudes of planets are 3-10 times larger than for the same planet around a solar-type star. 2 – M Dwarfs have low luminosity, so the lower flux they deliver to a planetary surface means that the orbital radius corresponding to the location of the habitable zone is much smaller. This makes the Doppler amplitudes larger again (e.g. Fig. 1). Figure 2.1 The theoretical continuous (> 5 Gyr) habitable zone as a function of stellar mass (shaded region). different shading areas correspond to various assumptions on cloudiness and the efficiency of the FigureThe 1: The theoretical habitable zone (shaded region) as a function of stellar mass (from runaway greenhouse. The dotted lines delineate the extreme outer edges of the habitable zone. Planets Selsis inward et al. of2007). Venus, Earth, andlocked Mars arethan indicated for Earth, the Sun. Theareorbital the dashed line become tidally in less 1 Gyr. Venus, and Mars indicatedpositions for the Sun. sizes The orbital positions relative sizes ofdiscovered the first three planets discovered orbiting M dwarf GJ and relative of the firstand three planets orbiting the 0.3 M¤ the M-dwarf GJ 581 581 (Udry et al. 2007) are also shown. The corresponding orbital periods of habitable-zone planets around (Udry stars et al. shown for inward dashed with2007) masses are between 0.1 and 0.2comparison. M! are between Planets approximately 4.5 andof 70 the d. These planetsline havebecome a probability transiting for atmospheric with upcoming tidallyhigh locked in of less thanand 1 would Gyr be – excellent thoughtargets planets with an studies atmosphere are facilities nonetheless like JWST. Figure from Selsis et al. 2007. Table 1 demonstrates why M-dwarfs make such compelling targets for habitable zone searches – such a planet will orbit a 0.3 M¤ star in periods of just 4.5 to 70d. For an even smaller 0.1M¤ star they orbit in just 3.5-18d. These short orbital periods make such planets far more detectable – a 5MEarth planet orbiting a M-dwarf with mass between 0.1-0.3M¤ delivers a Doppler amplitudes of 1.8-9.8 m s-1. expected to host regions capable of supporting liquid water (e.g. Wordsworth et al. 2011). The second track for studying habitable planets endorsed by the 2010 Decadal Report is focused on the opportunity offered by low-mass M dwarfs. In contrast to solar-type stars, UCLES + CYCLOPS on the AAT CYCLOPS is a single object Integral field unit made up of 15 close-packed 0.6” hexagonal fibres (only 12 functioning) that are arranged to have a ~2.5” diameter on the sky (see Figure 2). The fibres are reformatted to make a pseudo slit that injects light into UCLES at resolution ~70000. Each of the 12 fibres produces it’s own spectrum that is extracted. To be capable of detecting the <10 m s-1 Doppler amplitudes expected from habitable zone planets around M Dwarfs we need to be able to calibrate the UCLES spectrograph to < 2 m s-1. This is possible with CYCLOPS due to the tremendous amount of position information available when we take a calibration Thorium-Argon (ThAr) image because of the 12 independent fibre spectra present. By extracting all the ThAr spectra from a reference image and those from another ThAr image (Figure 3-a) we can measure the changes in position of the emission lines precisely by fitting a Gaussian to the crosscorrelation of a section of the two spectra (Figure 3-b and c). By doing this for different sections of ThAr spectra over the image wecan build up information on the spatial dependence of the positions of the ThAr lines (Figure 3-d). Finally the spatial dependence of the ThAr line shifts can be fitted with a polynomial surface of order 3-by-3, an examination of the residuals (e.g. Figure 3-e) presents a generalised error distribution (an exponential power function), for the example in Figure 3 this indicates an overall precision of 0.86 m s-1. A similar process is used to find the RV change of a stellar observation, though there is insufficient room on this poster for a detailed explanation. 250 2.5 0.4 500 Figure 2: The CYCLOPS fibre bundle on-sky layout. 3.45 200 ref. ThAr 0.35 3.4 Stellar Calib. ThAr 400 2 Pixel 400 300 500 250 600 è 200 700 0.25 è 0.2 0.15 150 Crossïcorrelation value 350 Normalised Intensity 0.3 300 1.5 è 1 è 3.25 3.2 3.15 0.05 1000 500 500 1000 1500 2000 2500 3000 3500 4000 0 Pixel (a) A gray-scale ThAr image. Each fibre produces it’s own separately extracted ThAr spectrum. 0 800 100 3.05 1500 50 1000 150 50 0.5 100 900 3.3 3.1 0.1 800 200 3.35 # of pieces of ThAr spectrum 450 Extracted ThAr shift (pixel) 100 850 900 950 1000 1050 1100 1150 0 ï15 1200 0 ï10 ï5 0 5 10 15 (b) Section of 400 pixels from one fibre spectrum from a reference ThAr (red) and a stellar calibration ThAr (blue). 1000 Pixel crossïcorrelation shift (pixel) Position (extracted pixel) 0 (c) Cross-correlation of ThAr spectrum pieces (blue +’s) and Gaussian fit (red). 2000 3000 5000 4000 0 ï80 ï60 ï40 ï20 0 20 40 60 80 Residual ThAr position (km sï1) Pixel (d) 3D plot of shifts from crosscorrelations of pieces all over the image (blue +’s) and a polynomial surface fit (red +’s). (e) Histogram of the residuals to the 3D fit in m s-1. Figure 3: Figures a – e demonstrate how to calibrate the CYCLOPS + UCLES spectrum to high precision using a Thorium-Argon emission arc lamp. In this example we achieve an overall precision of <1ms-1. Working in the Infra-red 100 100 180 200 200 200 200 160 300 300 300 300 140 400 400 400 400 120 500 100 500 500 Pixel 100 100 Pixel The M Dwarf stars that will be observed are nearby, but are intrinsically faint. Most M dwarfs are V > 9 mag. They emit most of their flux in the Infra-red and tend to be ~2 magnitudes brighter in the I band. The M Dwarfs are cool, and so have a molecular absorption ‘forest’ in the infra red which can provide a large amount of RV information. To try to maximise our RV precision we can observe with UCLES+CYCLOPS in the Infra-red, however the ThAr lamp used for calibration has several bright Argon lines in that wavelength region that make calibration difficult (see Figure 4 left). To try to overcome this obstacle we have tried other calibration lamps including Thorium-Krypton and Thorium-Xenon. Example images are shown alongside the ThAr images in Figure 4. It is clear the Thorium-Xenon lamp will offer superior infra-red calibration and so shall be used throughout our M Dwarf planet search. Pixel Pixel 200 500 600 600 600 600 700 700 700 700 800 800 800 800 900 900 900 900 1000 1000 1000 1000 500 500 1000 1000 1500 1500 2000 2000 Pixel Pixel 2500 2500 3000 3000 3500 3500 4000 4000 500 1000 1500 2000 Pixel 2500 3000 3500 4000 80 60 40 20 500 1000 1500 2000 2500 3000 3500 4000 0 Pixel Figure 4: Left: A 120s Thorium-Argon arc lamp exposure. Middle: The same for a Thorium-Krypton lamp. Right: The same for a Thorium-Xenon lamp. All images are on the sam color-scale shown on the right. Note the saturating lines in the top of the left and middle images, the top is the infra-red part of the spectrum and these saturating lines make calibration difficult.