Radial Velocity Detection of Planets:II. Results
1. Period Searching: How do you find planets in your data?
2. Exoplanet discoveries with the radial velocity method
Finding a Planet in your Radial Velocity Data
1. Determine if there is a periodic signal in your data.
2. Determine if this is a real signal and not due to noise.
3. Determine the nature of the signal, it might not be a planet!
4. Derive all orbital elements
1. Period Analysis
How do you know if you have a periodic signal in your data? Here are RV data from a pulsating star
What is the period?
Try 16.3 minutes:
Lomb-Scargle Periodogram of the data:
1. Period Analysis: How do you find a signal in your data
1. Least squares sine fitting:
Fit a sine wave of the form:
V(t) = A·sin(t + ) + Constant
Where = 2/P, = phase shift
Best fit minimizes the 2:
2 = di –gi)2/N
di = data, gi = fit
Note: Orbits are not always sine waves, a better approach would be to use Keplerian Orbits, but these have too many parameters
1. Period Analysis
2. Discrete Fourier Transform:
Any function can be fit as a sum of sine and cosines
FT() = Xj (T) e–itN0
j=1
A DFT gives you as a function of frequency the amplitude (power = amplitude2) of each sine wave that is in the data
Power: Px() = | FTX()|2
1N0
Px() =
1N0
N0 = number of points
[( Xj cos tj + Xj sin tj ) ( ) ]2 2
Recall eit = cos t + i sint
X(t) is the time series
A pure sine wave is a delta function in Fourier space
t
PAo
FT
Ao
1/P
1. Period Analysis
3. Lomb-Scargle Periodogram:
Power is a measure of the statistical significance of that frequency (period):
12
Px() =[ Xj sin tj–]
2
j
Xj sin2 tj–
[ Xj cos tj–]2
j
Xj cos2 tj–j
+12
False alarm probability ≈ 1 – (1–e–P)N = probability that noise can create the signal
N = number of indepedent frequencies ≈ number of data points
tan(2) = sin 2tj)/cos 2tj)j j
The first Tautenburg Planet: HD 13189
Least squares sine fitting: The best fit period (frequency) has the lowest 2
Discrete Fourier Transform: Gives the power of each frequency that is present in the data. Power is in (m/s)2 or (m/s) for amplitude
Lomb-Scargle Periodogram: Gives the power of each frequency that is present in the data. Power is a measure of statistical signficance
Am
plitu
de (m
/s)
Noise level
Alias Peak
False alarm probability ≈ 10–14
Alias periods:
Undersampled periods appearing as another period
Raw data
After removal of dominant period
To summarize the period search techniques:
1. Sine fitting gives you the 2 as a function of period. 2 is minimized for the correct period.
2. Fourier transform gives you the amplitude (m/s in our case) for a periodic signal in the data.
3. Lomb-Scargle gives an amplitude related to the statistical signal of the data.
Most algorithms (fortran and c language) can be found in Numerical Recipes
Period04: multi-sine fitting with Fourier analysis. Tutorials available plus versions in Mac OS, Windows, and Linux
http://www.univie.ac.at/tops/Period04/
2. Results from Doppler SurveysButler et al. 2006, Astrophysical Journal, Vol 646, pg 505
Telescope Instrument Wavelength Reference1-m MJUO Hercules Th-Ar1.2-m Euler Telescope CORALIE Th-Ar1.8-m BOAO BOES Iodine Cell1.88-m Okayama Obs, HIDES Iodine Cell1.88-m OHP SOPHIE Th-Ar2-m TLS Coude Echelle Iodine Cell2.2m ESO/MPI La Silla FEROS Th-Ar2.7m McDonald Obs. 2dcoude Iodine cell 3-m Lick Observatory Hamilton Echelle Iodine cell3.8-m TNG SARG Iodine Cell3.9-m AAT UCLES Iodine cell3.6-m ESO La Silla HARPS Th-Ar8.2-m Subaru Telescope HDS Iodine Cell8.2-m VLT UVES Iodine cell9-m Hobby-Eberly HRS Iodine cell10-m Keck HiRes Iodine cell
Campbell & Walker: The Pioneers of RV Planet Searches
1980-1992 searched for planets around 26 solar-type stars. Even though they found evidence for planets, they were not 100% convinced. If they had looked at 100 stars they certainly would have found convincing evidence for exoplanets.
1988:
„Probable third body variation of 25 m s–1, 2.7 year period, superposed on a large velocity gradient“
Campbell, Walker, & Yang 1988
Eri was a „probable variable“
Filled circles are data taken at McDonald Observatory using the telluric lines at 6300 Ang.
The first extrasolar planet around a normal star: HD 114762 with Msini = 11 MJ discovered by Latham et al. (1989)
The Brown Dwarf Desert
Mass Distribution Global Properties of Exoplanets:
Planet: M < 13 MJup → no nuclear burning
Brown Dwarf: 13 MJup < M < ~80 MJup → deuterium burning
Star: M > ~80 MJup → Hydrogen burning
Up-to-date Histograms with all planets:
One argument: Because of unknown vsini these are just low mass stars seen with i near 0
i decreasing
probability decreasing
P(i < ) = 1– cos Probability an orbit has an inclination less than
e.g. for m sin i = 0.5 MJup for this to have a true mass of 0.5 Msun sin i would have to be 0.01. This implies = 0.6 deg or P =0.00005: highly unlikely!
Argument against stars #1
Argument against stars #2
Some planetary systems have multiple planets, for example m1 x sini = 5 MJup, and m2 x sini = 0.03 MJup. To make the first planet a star requires sini =0.01. Other planet would still be mtrue=3 MJup
Brown Dwarf Desert: Although there are ~100-200 Brown dwarfs as isolated objects, and several in long period orbits, there is a paucity of brown dwarfs (M= 13–50 MJup) in short (P < few years) as companion to stars
An Oasis in the Brown Dwarf Desert: HD 137510 = HR 5740
A note on the naming convention:
Name of the star: 16 Cyg
If it is a binary star add capital letter B, C, D
If it is a planet add small letter: b, c, d
55 CnC b : first planet to 55 CnC
55 CnC c: second planet to 55 CnC
16 Cyg B: fainter component to 16 Cyg binary system
16 Cyg Bb: Planet to 16 Cyg B
The IAU has yet to agree on a rule for the naming of extrasolar planets
Semi-Major Axis Distribution
Semi-major Axis (AU) Semi-major Axis (AU)N
umbe
r
Num
ber
The lack of long period planets is a selection effect since these take a long time to detect
The short period planets are also a selection effect: they are the easiest to find and now transiting surveys are geared to finding these.
Updated:
Eccentricity distribution
Fall off at high eccentricity may be partially due to an observing bias…
e=0.4 e=0.6 e=0.8
=0
=90
=180
…high eccentricity orbits are hard to detect!
For very eccentric orbits the value of the eccentricity is is often defined by one data point. If you miss the peak you can get the wrong mass!
2 ´´
Eri
Comparison of some eccentric orbit planets to our solar system
At opposition with Earth would be 1/5 diameter of full moon, 12x brighter than Venus
EccentricitiesMass versus Orbital Distance
There is a relative lack of massive close-in planets
Classes of planets: 51 Peg Planets: Jupiter mass planets in short period orbits
Discovered by Mayor & Queloz 1995
• ~35% of known extrasolar planets are 51 Peg planets (selection effect)
• 0.5–1% of solar type stars have giant planets in short period orbits
• 5–10% of solar type stars have a giant planet (longer periods)
Classes of planets: 51 Peg Planets
Another short period giant planet
Butler et al. 2004
McArthur et al. 2004Santos et al. 2004
Msini = 14-20 MEarth
Classes of planets: Hot Neptunes
If there are „hot Jupiters“ and „hot Neptunes“ it makes sense that there are „hot Superearths“
Mass = 7.4 ME P = 0.85 d
CoRoT-7b
Classes: The Massive Eccentrics
• Masses between 7–20 MJupiter
• Eccentricities, e > 0.3
• Prototype: HD 114762 discovered in 1989!
m sini = 11 MJup
There are no massive planets in circular orbits
Classes: The Massive Eccentrics
Planet-Planet Interactions
Initially you have two giant planets in circular orbits
These interact gravitationally. One is ejected and the remaining planet is in an eccentric orbit
Lin & Ida, 1997, Astrophysical Journal, 477, 781L
Red: Planets with masses < 4 MJup
Blue: Planets with masses > 4 MJup
• Most stars are found in binary systems• Does binary star formation prevent planet formation?
• Do planets in binaries have different characteristics?
• For what range of binary periods are planets found?• What conditions make it conducive to form planets?
(Nurture versus Nature?)• Are there circumbinary planets?
Why should we care about binary stars?
Planets in Binary Systems
Star a (AU)16 Cyg B 80055 CnC 540
HD 46375 300Boo 155 And 1540
HD 222582 4740HD 195019 3300
Some Planets in known Binary Systems:
There are very few planets in close binaries. The exception is Cep.
For more examples see Mugrauer & Neuhäuser 2009, Astronomy & Astrophysics, vol 494, 373 and references therein
The first extra-solar Planet may have been found by
Walker et al. in 1992 in a
binary system:
Ca II is a measure of stellar activity (spots)
2,13 AUa0,2e
26,2 m/sK
1,76 MJupiterMsini2,47 YearsPeriod
Planet
18.5 AUa 0,42 ± 0,04e
1,98 ± 0,08 km/sK
~ 0,4 ± 0,1 MSunMsini56.8 ± 5 YearsPeriod
Binary Cephei
Walker et al. Excluded the planet hypothesis largely because the Ca II line strength showed variations with the same period as the velocity data. However, if you divide the Ca II in half (two time series) a signal is seen in the first half but not the last half. The signal in the last half is not the same period as the planet signal.
CepheiPrimary star (A)
Secondary Star (B) Planet (b)
Neuhäuser et al. Derive an orbital inclination of AB of 119 degrees. If the binary and planet orbit are in the same plane then the true mass of the planet is 1.8 MJup.
The planet around Cep is difficult to form and on the borderline of being impossible.
Standard planet formation theory: Giant planets form beyond the snowline where the solid core can form. Once the core is formed the protoplanet accretes gas. It then migrates inwards.
In binary systems the companion truncates the disk. In the case of Cep this disk is truncated just at the ice line. No ice line, no solid core, no giant planet to migrate inward. Cep can just be formed, a giant planet in a shorter period orbit would be problems for planet formation theory.
The interesting Case of 16 Cyg B
Effective Temperature: A=5760 K, B=5760 KSurface gravity (log g): 4.28, 4.35
Log [Fe/H]: A= 0.06 ± 0.05, B=0.02 ± 0.0416 Cyg B has 6 times less Lithium
These stars are identical and are „solar twins“. 16 Cyg B has a giant planet with 1.7 MJup in a 800 d period
Kozai Mechanism: One Explanation for the high eccentricty of 16 Cyg B
Two stars are in long period orbits around each other.
A planet is in a shorter period orbit around one star.
If the orbit of the planet is inclined, the outer planet can „pump up“ the eccentricity of the planet. Planets can go from circular to eccentric orbits.
This was first investigated by Kozai who showed that satellites in orbit around the Earth can have their orbital eccentricity changed by the gravitational influence of the Moon
Kozai Mechanism: changes the inclination and eccentricity
Planetary Systems: 49 Multiple Systems
49 Extrasolar Planetary Systems (18 shown) Star P (d) MJsini a (AU) eHD 82943 221 0.9 0.7 0.54 444 1.6 1.2 0.41GL 876 30 0.6 0.1 0.27 61 2.0 0.2 0.1047 UMa 1095 2.4 2.1 0.06 2594 0.8 3.7 0.00
HD 37124 153 0.9 0.5 0.20 550 1.0 2.5 0.4055 CnC 2.8 0.04 0.04 0.17 14.6 0.8 0.1 0.0 44.3 0.2 0.2 0.34 260 0.14 0.78 0.2 5300 4.3 6.0 0.16Ups And 4.6 0.7 0.06 0.01 241.2 2.1 0.8 0.28 1266 4.6 2.5 0.27HD 108874 395.4 1.36 1.05 0.07
1605.8 1.02 2.68 0.25HD 128311 448.6 2.18 1.1 0.25 919 3.21 1.76 0.17HD 217107 7.1 1.37 0.07 0.13 3150 2.1 4.3 0.55
Star P (d) MJsini a (AU) eHD 74156 51.6 1.5 0.3 0.65 2300 7.5 3.5 0.40 HD 169830 229 2.9 0.8 0.31 2102 4.0 3.6 0.33HD 160691 9.5 0.04 0.09 0 637 1.7 1.5 0.31
2986 3.1 0.09 0.80
HD 12661 263 2.3 0.8 0.35
1444 1.6 2.6 0.20HD 168443 58 7.6 0.3 0.53 1770 17.0 2.9 0.20HD 38529 14.31 0.8 0.1 0.28 2207 12.8 3.7 0.33HD 190360 17.1 0.06 0.13 0.01 2891 1.5 3.92 0.36HD 202206 255.9 17.4 0.83 0.44 1383.4 2.4 2.55 0.27HD 11964 37.8 0.11 0.23 0.15
1940 0.7 3.17 0.3
The 5-planet System around 55 CnC
5.77 MJ
Red lines: solar system plane orbits
•0.11 MJ ••
0.17MJ
0.03MJ
0.82MJ
The Planetary System around GJ 581
7.2 ME
5.5 ME
16 ME
Inner planet 1.9 ME
Can we find 4 planets in the RV data for GL 581?
1 = 0.317 cycles/d
2 = 0.186
3 = 0.077
4 = 0.015
Note: for Fourier analysis we deal with frequencies (1/P) and not periods
The Period04 solution:P1 = 5.38 d, K = 12.7 m/s
P2 = 12.99 d, K = 3.2 m/s
P3 = 83.3 d, K = 2.7 m/s
P4 = 3.15, K = 1.05 m/s
P1 = 5.37 d, K = 12.5 m/s
P2 = 12.93 d, K = 2.63 m/s
P3 = 66.8 d, K = 2.7 m/s
P4 = 3.15, K = 1.85 m/s
=1.53 m/s=1.17 m/s
Almost:
Conclusions: 5.4 d and 12.9 d probably real, 66.8 d period is suspect, 3.15 d may be due to noise and needs confirmation.
A better solution is obtained with 1.4 d instead of 3.15 d, but this is above the Nyquist sampling frequency
Published solution:
Resonant Systems Systems Star P (d) MJsini a (AU) eHD 82943 221 0.9 0.7 0.54 444 1.6 1.2 0.41
GL 876 30 0.6 0.1 0.27 61 2.0 0.2 0.10
55 CnC 14.6 0.8 0.1 0.0 44.3 0.2 0.2 0.34
HD 108874 395.4 1.36 1.05 0.07 1605.8 1.02 2.68 0.25
HD 128311 448.6 2.18 1.1 0.25 919 3.21 1.76 0.17
2:1 → Inner planet makes two orbits for every one of the outer planet
→
→
2:1
2:1
→ 3:1
→ 4:1
→ 2:1
•
Eccentricities
Period (days)Red points: SystemsBlue points: single planets
EccentricitiesMass versus Orbital Distance
Red points: SystemsBlue points: single planets
On average, giant planets in planetary sytems tend to be lighter than single planets. Either 1) Forming several planets in a protoplanetary disks „divides“ the mass so you have smaller planets, or 2) if you form several massive planets they are more likely to interact and most get ejected.
The Dependence of Planet Formation on Stellar Mass
Exoplanets around low mass stars
Ongoing programs:
• ESO UVES program (Kürster et al.): 40 stars
• HET Program (Endl & Cochran) : 100 stars
• Keck Program (Marcy et al.): 200 stars
• HARPS Program (Mayor et al.):~200 starsResults:
• Giant planets (2) around GJ 876. Giant planets around low mass M dwarfs seem rare
• Hot neptunes around several.
Currently too few planets around M dwarfs to make any real conclusions
0123456789
101112
0.025 0.225 0.425 0.625 0.825 1.025 1.225 1.425
2 planets with masses 2.1, 2.3 MJup
1 Planet with mass 4.9 MJup
GL 876 System
1.9 MJ
0.6 MJ
Inner planet 0.02 MJ
Exoplanets around massive stars
Difficult with the Doppler method because more massive stars have higher effective temperatures and thus few spectral lines. Plus they have high rotation rates. A way around this is to look for planets around giant stars. This will be covered in „Planets off the Main Sequence“
Result: Only a few planets around early-type, more massive stars, and these are mostly around F-type stars (~ 1.4 solar masses)
Galland et al. 2005
HD 33564
M* = 1.25
msini = 9.1 MJupiter
P = 388 days
e = 0.34
F6 V star
HD 8673
A Planet around an F star from the Tautenburg Program
Frequency (c/d)
Scar
gle
Pow
erP = 328 days
Msini = 8.5 Mjupiter
e = 0.24
An F4 V star from the Tautenburg Program
M* = 1.4 Mּס
Mstar ~ 1.4 Msun Mstar ~ 1 Msun
Mstar ~ 0.2 Msun
Preliminary conclusions: more massive stars have more massive planets with higher frequency. Less massive stars have less massive planets → planet formation is a sensitive function of the planet mass.
Astronomer‘s
Metals
More Metals ! Even more Metals !!
Planets and the Properties of the Host Stars: The Star-Metallicity Connection
The „Bracket“ [Fe/H]
Take the abundance of heavy elements (Fe for instance)
Ratio it to the solar value
Take the logarithm
e.g. [Fe/H] = –1 → 1/10 the iron abundance of the sun
These are stars with metallicity [Fe/H] ~ +0.3 – +0.5
There is believed to be a connection between metallicity and planet formation. Stars with higher metalicity tend to have a higher frequency of planets. This is often used as evidence in favor of the core accretion theory
Valenti & Fischer
The Planet-Metallicity Connection?
There are several problems with this hypothesis
Endl et al. 2007: HD 155358 two planets and..
…[Fe/H] = –0.68. This certainly muddles the metallicity-planet connection
The Hyades
• Hyades stars have [Fe/H] = 0.2
• According to V&F relationship 10% of the stars should have giant planets,
The Hyades
• Paulson, Cochran & Hatzes surveyed 100 stars in the Hyades
• According to V&H relationship we should have found 10 planets
•We found zero planets!Something is funny about the Hyades.
False Planets
or
How can you be sure that you have actually discovered a planet?
HD 166435
In 1996 Michel Mayor announced at a conference in Victoria, Canada, the discovery of a new „51 Peg“ planet in a 3.97 d. One problem…
HD 166435 shows the same period in in photometry, color, and activity indicators.
This is not a planet!
What can mimic a planet in Radial Velocity Variations?
1. Spots or stellar surface structure
2. Stellar Oscillations
3. Convection pattern on the surface of the star
Starspots can produce Radial Velocity Variations
Spectral Line distortions in an active star that is rotating rapidly
Tools for confirming planets: Photometry
Starspots are much cooler than the photosphere
Light Variations
Color Variations
Relatively easy to measure
Ca II H & K core emission is a measure of magnetic activity:
Active star
Inactive star
Tools for confirming planets: Ca II H&K
HD 166435
Ca II emission measurements
Bisectors can measure the line shapes and tell you about the nature of the RV variations:
What can change bisectors:• Spots• Pulsations • Convection pattern on star
Span
Curvature
Tools for confirming planets: Bisectors
Correlation of bisector span with radial velocity for HD 166435
Spots produce an „anti-correlation“ of Bisector Span versus RV variations:
Activity Effects: Convection
Hot rising cell
Cool sinking lane
•The integrated line profile is distorted.
•The ratio of dark lane to hot cell areas changes with the solar cycle
RV changes can be as large as 10 m/s with an 11 year period
This is a Jupiter!One has to worry even about the nature long period RV variations
The Great 51 Peg Controversy, or
My personal piece of professional rubbish
Variations of Bisectors with Pulsations
Gray & Hatzes
Gray reported bisector variations of 51 Peg with the same period as the planet. Gray & Hatzes modeled these with nonradial pulsations
A beautiful paper that was completely wrong.
The 51 Peg Controversy
Hatzes et al.
More and better bisector data for 51 Peg showed that the Gray measurements were probably wrong. 51 Peg has a planet!
How do you know you have a planet?
1. Is the period of the radial velocity reasonable? Is it the expected rotation period? Can it arise from pulsations?
• E.g. 51 Peg had an expected rotation period of ~30 days. Stellar pulsations at 4 d for a solar type star was never found
2. Do you have Ca II data? Look for correlations with RV period.
3. Get photometry of your object
4. Measure line bisectors
5. And to be double sure, measure the RV in the infrared!
Radial Velocity Planets30 90 1000Period in years →
Red line: Current detection limitsGreen line detection limit for a precision of 1 m/s
Summary Radial Velocity Method
Pros:• Most successful detection method• Gives you a dynamical mass• Distance independent
• Will provide the bulk (~1000) discoveries in the next 10+ years
Summary
Radial Velocity Method
Cons:• Only effective for late-type stars• Most effective for short (< 10 – 20 yrs)
periods
• Only high mass planets (no Earths!)
• Projected mass (msin i)
• Other phenomena (pulsations, spots) can mask as an RV signal. Must be careful in the interpretation
Summary of Exoplanet Properties from RV Studies
• ~5% of normal solar-type stars have giant planets
• ~10% or more of stars with masses ~1.5 Mּס have giant planets that tend to be more massive (more on this later in the course)
• < 1% of the M dwarfs stars (low mass) have giant planets, but may have a large population of neptune-mass planets
→ low mass stars have low mass planets, high mass stars have more planets of higher mass → planet formation may be a steep function of stellar mass
• 0.5–1% of solar type stars have short period giant plants
• Exoplanets have a wide range of orbital eccentricities (most are not in circular orbits)
• Massive planets tend to be in eccentric orbits
• Massive planets tend to have large orbital radii
• Stars with higher metallicity tend to have a higher frequency of planets, but this needs confirmation