1

The Validation of Kepler planetsF. Fressin, G. Torres & the Kepler team

Planet or Blend?An observed periodic transit signal could be due to:

Transiting Planet Eclipsing Binary Physically bound or Chance alignment

We use Blender, a light-curve fitting software• It attempts to explain Kepler candidates assuming they are the

result of a pair of eclipsing objects in the photometric aperture.

Planet or Blend?

We use Blender, a light-curve fitting software• It attempts to explain Kepler candidates assuming they are the

result of a pair of eclipsing objects in the photometric aperture.

Transiting Planet Eclipsing Binary Physically bound or Chance alignment

Primary Star

Secondary Star (MS or not)

Tertiary Star or planet

An observed periodic transit signal could be due to:

Blender results : Example of Kepler-10cEclipsing Binary Stars can mimic a transiting planet signal

Period = 45.3 daysRadius = 2.23 R⊕

Blender results : Example of Kepler-10cEclipsing Binary Stars can mimic a transiting planet signal

Period = 45.3 daysRadius = 2.23 R⊕

Blender results : Example of Kepler-10c

Blender results show that only a very small fraction can actually reproduce the exact transit shape

Eclipsing Binary Stars can mimic a transiting planet signal

Period = 45.3 daysRadius = 2.23 R⊕

•Speckle interferometry (WIYN at Kitt Peak)

•Adaptive optics imaging (PHARO at Palomar)

•Centroid shift analysis (Kepler data)

Blender Inputs

Separation vs. Magnitude

Color vs. Magnitude

Combining Follow-Up Observations

• Kepler photometry

• Precise host star characterization (Kepler asteroseismology)

• Spectroscopy (Hires at Keck)• Multi-color photometry (KIC, 2MASS)

• Infrared transit observation (WarmSpitzer)

Background star + star Background star + planet

Physically bound star + planet

Quantifying the blend probability

• Blend frequency = (0.41 + 1.21) x 10-8

= 1.62 x 10-8

• Planet prior = 157 / 156,453 = 1.0 x 10-3

The planet hypothesis is 60,000 times more likely than a blend

9

Kepler-10bKepler-9d

Kepler-10c

Kepler-11g

Kepler-19b

CoRoT-7b

Gj-1214b

55 Cnc e

Gj-436b

Kepler planet

Blender Validation

Other transiting planet

Supporting the challenging Radial Velocity confirmations

Kepler-10b shows a clear RV signal, but has inconclusive Bisector variations.

Blender validates the transiting planet scenario.

CoRoT-7b• Leger et al. 2009 strong support for validation• Queloz et al. 2009 confirmation M = 4.8 ± 0.8M⊕• Hatzes et al. 2010 confirmation M = 6.9 ± 1.4M⊕• Ferraz-Mello et al. 2011 confirmation M = 8.0 ± 1.2M⊕• Pont et al. 2011 marginal detection M = 2.3 ± 1.5M⊕• Boisse et al. 2011 marginal detection M = 5.7 ± 2.5M⊕• Hatzes et al. 2011 confirmation M = 7.4 ± 1.2 M⊕

Batalha et al. 2011M = 4.56 ± 1.29M⊕

Blender + Spitzer validation of CoRoT-7b

In visible light

In Infrared

Background star + star Background star + planet

Physically bound star + planet

Validation of CoRoT-7b

• Blend frequency = 4.2 x 10-7

• Planet prior = 231 / 156,453= 1.5 x 10-3

The planet hypothesis is 3,500 times more likely than a blend

Kepler-10bKepler-9d

Kepler-10c

Kepler-11g

Kepler-19b

CoRoT-7b

Gj-1214b

55 Cnc e

Gj-436b

Kepler planet

Blender Validation

Other transiting planet

• Improving Centroid shift resultsS. Bryson

• Target color using SpitzerJ.M. Desert* oral04.03

• Dynamical Constrains from multiple systemsD. Fabricky, M.Holman

• Kepler detection efficiencyC. Dressing* poster24.01

• Constrains from multiple systemsD. Ragozzine (Co-planarity boost)J. Lissauer* oral04.05(Multiplicity boost)

The Blender Connections

Further constraining the Blend Frequency Refining the planet prior

Blender results strongly scale with the amount of data.

Allowed Eclipsing binaries using Q1 – Q5

... are divided by 3 using Q1 – Q8

Sub-Earth-size Kepler candidate

Gathering more data won’t only provide more critical KOIs, but also strongly help validating them (improved Centroid and Blender)

By ESS3,our goal will be to prove that Earth-size planets, as hard to find as they may be, are not Extreme Solar Systems