M A S T E R’S T H E S I S
2006:353 CIV
NATHALIE ÖSTBERG
Exoplanet Transit Search with the Westerlund Telescope
MASTER OF SCIENCE PROGRAMME Space Engineering
Luleå University of Technology
Department of Applied Physics and Mechanical Engineering
Division of Physics
E XOPLANET T RANSIT S EARCH
WITH THE
W ESTERLUND T ELESCOPE
- Nathalie Östberg -
Luleå University of Technology
2006
A BSTRACT
More than 200 exoplanets have been found to this day, and the quest continues.
The work behind this master thesis in Space Engineering at Luleå Univer- sity of Technology was conducted at the Uppsala Astronomical Observatory.
The purpose was to inspect if the Westerlund Telescope could be used for extra-solar planetary transit searches by photometry. Observations were made during two nights in the fall of 2005 with the 90-cm cassegrain telescope.
The centre point of the observed field had coordinates RA 20h 51min 20s,
Dec 36º 52’ 32’’. No exoplanets were found, but detection of a 1% dimming
in the light intensity was confirmed. Six variable stars were mapped and pe-
riods were calculated for four of these. A total of 32 interesting light-curves
were kept for further investigation, tentatively by another Master of Science
student.
A CKNOWLEDGEMENTS
I would like to thank my co-supervisor Johan Warell for all the help with the Westerlund telescope and for always taking time to answer my questions.
Thanks to my supervisor Claes-Ingvar Lagerkvist for his support, to Bertil Petterson for all the help with computer difficulties and Göran Henriksson for taking time to explain his computer programme used for period determina- tion. Further acknowledgement goes to my examiner Sverker Fredriksson for general advice and for carefully and constructively reading my manuscript.
Many thanks as well to my coworkers Carolina Bergfors and Jill Håkansson for good cooperation and many interesting discussions around related topics.
I would like to thank Helene Holmgren for proofreading my report and Karin Ågren for the good company during all very much needed coffee breaks.
Last but not least, loving thanks to my husband Hugo for his help with the
layout, for the illustrations of the spectrum of a cool and hot star and for his
patience during stressful times through this thesis work.
C ONTENTS
C HAPTER 1 - I NTRODUCTION 1
C HAPTER 2 - D ISCOVERY METHODS 3
2.1. D
YNAMICALEFFECTS...3
2.1.1. Radial velocity measurements ... 3
2.1.2. Astrometry ... 4
2.2. P
HOTOMETRY...4
2.2.1. Search for microlensing events ... 4
2.2.2. Planetary transits ... 5
2.3. D
IRECTIMAGING...6
2.4. H
ELPFULTOOLS...6
2.4.1. Adaptive optics ... 7
2.4.2. Coronagraphs ... 7
2.4.3. Nulling interferometry ... 7
2.5. T
ELESCOPES–
TODAYANDINTHEFUTURE...7
2.5.1. CoRoT (Convection, Rotation and planetary Transits) ... 7
2.5.2. Kepler ... 8
2.5.3. Eddington ... 8
C HAPTER 3 - D IVERSITY OF EXOPLANETS 9 3.1. O
CCURRENCE...9
3.2. E
CCENTRICITY...10
3.3. M
ASSDISTRIBUTION...10
3.4. S
EMIMAJORAXISANDMIGRATION...10
3.5. M
ETALLICITYANDMULTI-
PLANETSYSTEMS...11
3.6. P
ULSARPLANETSANDFREE-
FLOATINGPLANETS...11
C HAPTER 4 - V ARIABLE STAR TYPES 13 4.1. C
ATACLYSMICVARIABLES...13
4.1.1. Supernovæ (SN) ... 13
4.1.2. Novæ ...14
4.1.4. Dwarf novæ ...14
4.2. E
RUPTIVEVARIABLES...15
4.2.1. Luminous Blue variables (LBVs)/S Dor stars ... 15
4.2.2. Wolf-Rayet stars (WR) ... 15
4.2.3. Pre-main-sequence stars (PMS) ... 15
4.2.4. Flare stars or UV Ceti stars ... 15
4.2.5. R Coronae Borealis (RCBs) ... 15
4.3. R
OTATINGSTARS...15
4.4. E
CLIPSINGBINARYSTARS...16
4.4.1. Algol ... 16
4.4.2. β Lyrae ... 16
4.4.3. W UMa ... 16
4.5. P
ULSATINGVARIABLES...16
4.5.1. β Cephei variables ... 17
4.5.2. Be stars ... 17
4.5.3. 53 Per/slowly pulsating B variables (SPBs) ... 17
4.5.4. δ Scuti variables ... 17
4.5.5. Cepheids ... 17
4.5.6. RR-Lyrae stars ... 17
4.5.7. RV Tau stars ... 18
4.5.8. Mira-stars ... 18
4.5.9. ZZ Ceti variables ... 18
C HAPTER 5 - A STRONOMICAL O BSERVATIONS 19 5.1. T
HEW
ESTERLUNDT
ELESCOPE...19
5.2. P
ROCEDURE...19
5.3. C
ALIBRATIONIMAGES...20
5.3.1. Bias and dark images ... 20
5.3.2 Flat-field frames ... 20
C HAPTER 6 – P ICTURE REDUCTION 23 6.1. C
ALIBRATIONOFTHEPICTURES...23
6.1.1. Master bias ... 23
6.1.2. Master dark ... 24
6.1.3. Master flat ... 24
6.1.4. Calibrated images ... 24
6.2. M
AKINGLIGHT-
CURVES...25
6.3. M
AGNITUDECORRECTIONBETWEENTHETWONIGHTS...26
C HAPTER 7 - R ESULTS 27 7.1. E
XOPLANETS...27
7.2. V
ARIABLESTARS...28
7.2.1. Temperature ... 28
7.2.2. Period ... 29
7.2.3. Star #1 ... 30
7.2.4. Star #2 ... 30
7.2.5. Star #3 ... 31
7.2.6. Star #4 ... 31
7.2.7. Star #5 ... 32
7.2.8. Star #6 ... 32
C HAPTER 8 - C ONCLUSIONS 35
B IBLIOGRAPHY 37
A PPENDIX 1 41
A PPENDIX 2 53
C HAPTER 1 - I NTRODUCTION
Mankind has always been adventurous and curious about unknown places.
Questions as to the uniqueness of our existence have long been discussed and have driven us to explore more and more of the Universe. One way to shed light on the matter has been to find planets in other solar systems. Many different search methods are used today to fulfill the task. Since the first exo- planet was discovered in 1995 (Mayor & Queloz 1995), vigorous diversity in the extra-solar planet community has been revealed. Since this thesis work was based on the transit method, and the procedure revealed a fruitful by- product – variable stars – an introduction on their different kinds is given.
In this survey, the Westerlund telescope at the Uppsala Astronomical Obser-
vatory was used. The observation procedure is explained just as well as the
mechanism behind the picture reductions. Finally the results are depicted and
improvements on the search manner are discussed.
C HAPTER 2 - D ISCOVERY METHODS
A planet exerts a gravitational tug at its mother star, which in turn makes the star wobble with the same period as the planet. This dynamical effect can either be directly observed (astrometry) or detected by the means of the Doppler effect in the star’s spectral lines (radial velocity). Other techniques that are commonly used today are based on photometry, such as searches for microlensing effects and planetary transits. Last but not least, the dream of every astronomer in quest of an exoplanet is of course to get a snapshot of a planet by direct imaging. To optimize these techniques there are a few tools available today, such as the use of coronagraphs, adaptive optics and nulling interferometry. Let us look a little further into the techniques for exoplanet discovery and the telescopes used, both today and in the future.
2.1. D YNAMICAL EFFECTS
2.1.1. Radial velocity measurements
This indirect search method has been the most fruitful among today’s tech- niques. A spectrograph is used to monitor the successive red- and blueshifts of the star’s spectrum, caused by the radial movement of the star. The star’s radial velocity is then given by
, (1)
where M
pis the planet’s mass in Jupiter masses, M* the star’s mass in solar masses, a the semi-major axis of the planet’s orbit in AU and i the inclina-
M
pM
*a
psin i
.
tion from the sky-plane (Clark 1998). Since the amplitude of the oscillation is small, detections require stabile and precise measurements; the radial velo- city ought to be four times higher than the sensitivity of the spectrograph. In spite of good instruments, the radial velocity method remains dependent on the inclination of the observed star-planet system and thus only a minimum mass can be estimated.
2.1.2. Astrometry
The same physical phenomenon as in radial velocity measurements is used in astrometry – the wobble of stars caused by planets in orbit. The differen- ce is that instead of measuring the Doppler effect, real observations of the oscillations are made. The movement, measured in microarcseconds, can be approximated by the following formula
, (2)
where the input data are the same as in equation (1) and D is the distance to the star in parsec. The downside of this method is the requirement for the star to be relatively near. As equation (2) shows, the magnitude of the observed movement will half when the distance to the star is doubled. Furthermore, only large planets in large orbits can be expected to be detected in earth-based astrometric searches, because a planet in larger orbit will make the star’s orbit larger. From a distance of 10 pc, Jupiter seems to cause the Sun to swing with a maximum movement of 500 microarcseconds. The Earth, on the other hand, contributes to the Sun’s dance with a meagre 0.3 microarcseconds at the same distance. Since the signals are faint, space-based interferometers are needed to detect them (Clark 1998).
2.2. P HOTOMETRY
A lot of information can be found when monitoring the intensity of light reaching us from stars. Regarding the quest for extra- solar planets, one can proceed in the following two manners. Either one looks for intensifica- tion of the light intensity due to microlensing events, or drops in the light-curves, which in turn are caused by planetary transits.
2.2.1. Search for microlensing events When a foreground object passes either through or nearby the line of sight of an observed background star, the light-curve of the latter shows a well-known pattern: the in- tensity of light increases symmetrically. This Figure 1. Plot of the gravi-
tational microlensing effect created from theoretical cal- culations by Benett & Rhie (1996). The axes show the number of times the inten- sity is magnified versus time (retrieved at http://www.as- tro.livjm.ac.uk/courses/one/
NOTES/smlens.htm).
M
pM
*D .
space and behaves like a lens (Bond et al. 2004). If that star has a planetary companion in orbit, a small disturbance can be detected in the light-curve, as shown in figure 1. To date, four planets have been found with the microlensing technique, of which the latest is one of the smallest exoplanets ever found, a planet of 5.5 Earth masses orbiting at 2.6 AU from its parent star, an M-dwarf (Beaulieu et al. 2006).
2.2.2. Planetary transits
When a planet passes in front of its parent star it causes a dip in the monitored light intensity curve, as illustrated by figure 2. The amplitude of the dimming is proportional to the ratio of the radii, as shown in equation 3:
, (3)
where F is the flux, Rp and R* the radii of the planet and star (Bordé et al.
2001). The duration of the intensity drop depends on the distance between the two objects, a. After timing the transit duration t and orbital period P, it is possible to calculate the impact parameter k – the height over the equator at which the planet is passing:
, (4)
Figure 2. Superposed light- curves of star HD 209458 showing transits occurring during two nights in Septem- ber 1999, detected by STARE project astronomers. The axes show the relative inten- sity flux versus time (Char- bonneau et al. 2000).
R
pR
*F =
2
cos k = P . R
*Finally, the inclination can be calculated as follows:
, (5)
revealing the true planet mass missed in the radial velocity searches. Even though thousands of stars can be monitored at the same time, only a few planetary transits can be detected. The Earth must lie near the plane of the planet’s orbit in order for it to produce a transit. The geometrical probability of transit at an inclination close to 90° making the transit visible is given by the following ratio (Bordé et al. 2001):
. (6)
Obviously, the probability increases for close orbiting planets. Furthermore, a planet the size of Jupiter would cause a 1% drop in the light intensity from a Sun-like star, whereas an earth-like planet would cause a drop of only 0.01%, which is only detectable from space (BEST homepage). The transit method is therefore best suited for giant planets in close orbits, at least for ground-based observations. The biggest challenge is to identify the planetary transits from anomalies. Light-curves emanating from eclipsing binary stars, triple systems or even stellar companions of planetary size but not planetary masses can easily be mistaken for planetary transits (Brown 2003). The transit method has though been useful as a complement to radial velocity searches as to determine exact mass and radius (Bouchy et al. 2005).
2.3. D IRECT IMAGING
There are many advantages with direct imaging. There are no restrictions to star types, masses or system inclination. Long-period planets do not require long observation times like those needed for radial velocity surveys (Lagrange et al. 2004). To date, only four possible candidates have been directly detected. The main discussion has been around the type of the com- panions because of their large masses: how does one tell if the companion is indeed a planet or a brown dwarf? To differentiate extra-solar planets from the failed stars, the International Astronomical Union has come forward with the following two criteria: first, brown dwarfs have masses above 13.6 M Jup , which is the minimum mass of deuterium burning. Second, the formation of the object plays an essential role; either the object has started its life by gravitational collapse, which is the case of a star, or it has formed due to core accretion and finally become a planet (Chauvin et al. 2005).
2.4. H ELPFUL TOOLS
The two major problems to overcome when looking for the weak signal of an exoplanet are the turbulence of the atmosphere in earth-based observations and the weakness of the planet’s signal compared to the one originating from the parent star. To compensate, a few helpful tools are at hand, like adaptive cos i = R
*. sin ka
R
*P
g= a
2.4.1. Adaptive optics
An incoming wave front is constantly broken into different rays in the at- mosphere, causing a smeary disc instead of a bright spot on the image plane.
This distorted beam can be contorted back to its original uniform pattern by adapting the optics to the turbulence of the atmosphere. The distortions in the wavefront of a guide star in the same field of view than an observed star can be measured and used to calculate the corrections needed to the deformable mirrors. However, this procedure must be extremely fast because the varia- tions in the atmosphere are of a fiftieth of a second (Clark 1998).
2.4.2. Coronagraphs
In order to detect an exoplanet by direct imaging, the intensive light from the parent star must be blocked since the flux ratio between the two objects is huge. An earth-like exoplanet would typically be 6x10 9 times fainter than its stellar companion in the visible and 7x10 6 times fainter in the thermal infra- red (Mawet & Riaud 2005). Thus, coronagraphs are used to produce artificial stellar eclipses.
2.4.3. Nulling interferometry
An alternative way to extinct the intensive stellar light is to use destructive interference between two coherent beams. Before combining the beams of light recovered in two separate telescopes, one ray is delayed by half a wave- length. Peaks from one beam are then added to the slopes of the other, leaving no starlight at all. This canceling procedure works only on objects that are directly pointed at. Thus, the light coming from an orbiting planet will be slightly off-axis and even though one of the beams is delayed as well, the light will rather be reinforced than cancelled out.
2.5. T ELESCOPES – TODAY AND IN THE FUTURE
Only experiments based on planetary transits will be mentioned here, because it is the procedure used in this thesis work and because of the promising pros- pects of the transit method in space-based telescopes. More than 20 surveys focused on planetary transits are either ongoing or planned in the near future (Horne 2001), of which the following three are space-based.
2.5.1. CoRoT (Convection, Rotation and planetary Transits)
A 27-cm telescope equipped with four CCD wide-field cameras will be
launched in December 2006 for a 2.5-year long mission. The satellite will be
placed in polar circular orbit at an altitude of 896 km. The project is led by
french CNES and is focusing on terrestrial planets, just as well as asteroseis-
mology (CoRoT homepage).
2.5.2. Kepler
Kepler is a 1.4-metre telescope with a large field-of-view and four CCD cameras. The launch of the satellite is planned for 2008 by NASA. The instrument will monitor more than 100,000 stars for planetary transits during a period of four years from an earth-trailing heliocentric orbit. The mission’s focus will be on finding earth-like planets in the habitable zone around stars (distances from the star where liquid water can be found) and examine the diversity and structure of planetary systems (Kepler homepage).
2.5.3. Eddington
‘It would indeed be rash to assume that nowhere in the Universe has Nature repeated the strange experiment which she performed on the Earth’. Arthur Eddington’s words from 1933 will follow the European Space Agency’s satel- lite Eddington on its journey to the second Lagrangian point (L2), 1.5 million km in the anti-Sun direction, in 2008. The advantages of L2 are stable ther- mal and radiation environment and the fact that the Sun, Earth and Moon are always behind the satellite’s viewing direction. This precision photometer will search for planetary transits during three years of its five-year long mission.
The other goal of its mission is stellar seismology, just as it is for CoRoT
(Eddington homepage).
C HAPTER 3 - D IVERSITY OF EXOPLANETS
To date, 179 planetary systems and a total of 209 exoplanets have been found.
As many as 21 of the stars harbour more than one planet (Schneider 2006).
The following properties are derived from the ongoing 18-year search of 1330 stars of spectral classes F, G, K or M at Lick, Keck and the Anglo-Australian Telescopes that provide uniform Doppler precision of 3 m/s. Different masses, semimajor-axes and eccentricities were reported among the discovered planets and a preliminary planet occurrence rate could be established. A correlation between the metallicity of the host star and the probability of orbiting planets was determined and a few multiplanet systems were found (Marcy et al. 2004).
Furthermore, planets tend to be spotted in the most extraordinary places, around pulsars just as well as floating on their own.
3.1. O CCURRENCE
Of the 1330 monitored stars, 88 have been found to harbour planets of masses
M < 13 M Jup . In other words, 6.6% of the stars in the survey have planetary
companions within 5 AU. This should be seen as a lower limit since stars
between 3 and 5 AU are not effectively detected because of the restricted dura-
tion (6 - 8 years) of the Doppler measurements. In addition, low-mass planets
beyond 1 AU are certainly missed because of their only subtle wobble. About
5% of the stars in the survey seem to have long-term velocity disturbances,
which could be caused by a planetary companion at a distance between
5 and 20 AU. Bearing this in mind, an extrapolation of the statistics suggests
a 12% occurrence of extra-solar giant planets (EGPs) within 20 AU (Marcy
et al. 2005).
3.2. E CCENTRICITY
Eccentric orbits tend to dominate within the extra-solar planet community, on average e = 0.34 for planets within 0.2 - 3.5 AU (Marcy et al. 2004). Close orbiting planets have near-circular orbits though, probably due to tidal effects.
Furthermore, massive planets are connected to high eccentricities, posing the question about the nature of their initial orbits. High-mass planets have the vastest resistance to perturbations that are needed to force them out of their initial orbits.
3.3. M ASS DISTRIBUTION
Fewer high-mass than low-mass planets have been spotted. Figure 3 depicts the planet mass distribution, decreasing as M -1.05 with rising masses (Marcy et al.
2005). It is important to remember that, as mentioned before, only mini- mum masses can be calculated by radial velocity measurements because of the uncertainty in the inclination i, although it should affect the distribution only slightly.
3.4. S EMIMAJOR AXIS AND MIGRA -
TION
As shown in figure 4, the orbits of the 104 planets found in the survey have semi- major axes in the interval 0.02 - 6.0 AU.
Of them, 16 have been detected within 0.1 AU, suggesting that 1.2% of FGK main-sequence stars harbour hot Jupiters – gas giants in close orbits. The fact that EGPs have been detected as close as 0.022 AU (Heap et al. 2005) from their mother star, can be explained by migra- tion processes. Giant planets ought to form beyond 3 AU where large amounts of cool gas can be accreted. The planets may then lose energy and angular mo- mentum to the disk, which causes them to move inwards (type I). Another expla- nation would be that the gas disk itself would accrete onto the star and thus tug planets along with it (type II) (Marcy et al. 2005).
Figure 3. Planet mass distri- bution of the 104 exoplanets, following the best fit negative power law. Axes show the number of planets versus minimum masses (Marcy et al. 2005).
Figure 4. Distribution of
planets according to the
semimajor axes of their orbits
(Marcy et al. 2005).
3.5. M ETALLICITY AND MULTI - PLANET SYSTEMS
Stars with high metallicity tend to harbour more planets. In the above men- tioned survey, about 25.0% of the most metal-rich stars have a planetary companion, while less than 3.0% of the stars with low metallicity harbour planets (Marcy et al. 2005). The probability of forming planets can be featured in the following power law:
, (7)
which figure 5 illustrates (Fisher & Va- lenti 2005).
More and more stars with multiplanet systems are discovered each year. Al- though too few such stars have been found so far in order to establish statis- tics, the following trend has been noted.
The mean metallicity of stars with mul- tiple-planet systems is higher than for those with only one detected planet (Fisher & Valenti 2005).
3.6. P ULSAR PLANETS AND FREE - FLOATING PLANETS
The first exoplanets were discovered around a rapidly rotating neutron star through pulsar timing. In 1991, Wolszczan and Frail detected anomalies in the pulses emanating from the millisecond pulsar PSR1257+12, due to planets in orbit (Wolszczan & Frail 1992). A circumstellar disk resembling the proto- planetary disks found around ordinary young stars has recently been spotted around a pulsar, suggesting that planets are born around the dead remnants (Chakrabarty 2006).
Planets have just as well been found to be ‘alone’ in the interstellar medium and their origin is still a mystery. These free-floating planets could have been formed – like the favoured theory of planet formation today states – by accreting matter from the surrounding disk of a star and then by being ejected into the surrounding interstellar space due to gravitational perturbations. Or, they might have been born out of the gravitational collapse of molecular gas and dust (Boss 1997).
Figure 5. Percentage of stars harbouring planets with re- gard to the metallicity of the parent star (Marcy et al.
2005).
N
FeN
HN
FeN
H 2Sun