Red Giants in Eclipsing Binaries as a Benchmark for AsteroseismologyMeredith L. Rawls Department of Astronomy New Mexico State University
April 8, 2016 • PhD Thesis Defense
@merrdiff
Image: G Perez, IAC, SMM
Meredith L. Rawls • @merrdiff
✦ Eclipsing binaries with light curves and radial velocities let us directly measure M, R
✦ Asteroseismology is a powerful tool to characterize lots of stars quickly, using light curves alone
✦ We can study oscillating stars in eclipsing binaries from two independent perspectives
✦ A set of well-characterized binaries lets us explore why some don’t oscillate when we think they should
How to measure stellar properties?
An asteroseismic revolution
Meredith L. Rawls • @merrdiff Figure: Huber et al. 2014
An asteroseismic revolution
Meredith L. Rawls • @merrdiff Figure: Huber et al. 2014
Meredith L. Rawls • @merrdiff
✦ Larger stars oscillate more slowly✦ Evolved giants: hours–days✦ Sun-like stars: minutes
Stochastic “ringing” oscillations in stars’ convective zones
100 μHz 1 mHz 500 Hz
Figure concept: P. Gaulme
Solar-like oscillations tell us about a star’s mean gravity and density
Meredith L. Rawls • @merrdiff
Δν
νmax
Power
Frequency ν
✦ Acoustic pressure modes✦ Buoyant gravity modes✦ Scaling relations
Meredith L. Rawls • @merrdiff
Chaplin & Miglio 2013
left side x-axis range
MS
MS
Subgiant
Subgiant
RGBbase
RGB
RGB
RGB
Redclump
Meredith L. Rawls • @merrdiff
✦ Kepler red giant catalog: ~14,000✦ Kepler eclipsing binary catalog: 2,500+✦ After cross-correlating: 19✦ Most systems are SB2, but 4 are SB1✦ Most are oscillators, but 4 are not
Searching for red giants in eclipsing binary systems
Gaulme et al. 2013, 2014
Meredith L. Rawls • @merrdiff
✦ Process light curves to “fill” eclipses and remove gaps
✦ Single set of oscillations✦ Broad and damped modes✦ Evolved onto the red clump
Characterizing a star: asteroseismology
70 80 90 100 110 120 130 140Frequency (µHz)
0
100
200
300
400
500
600
700
800
Smoothed
Pow
erDensity
(ppm
2µHz−
1)
0
100
200
300
400
500
600
700
800
KIC 9246715
Reference star
Rawls et al. 2016
KIC 9246715
Meredith L. Rawls • @merrdiff
Characterizing a star: binary modeling✦ Process Kepler light curve to
preserve eclipses✦ Extract radial velocities from
spectra via the broadening function✦ Get all velocities on one zeropoint✦ Suite of broadening function
programs in python:
github.com/mrawls/BF-rvplotter
Rawls et al. 2016
0.0
0.2
0.4 0.773 φ
2012-03-01
TRES 0.831 φ
2012-03-11
TRES 0.960 φ
2012-04-02
TRES 0.170 φ
2012-05-08
TRES
0.0
0.2
0.4 0.275 φ
2012-05-26
TRES 0.316 φ
2012-06-02
TRES 0.374 φ
2012-06-12
ARCES 0.460 φ
2012-06-27
ARCES
0.0
0.2
0.4 0.479 φ
2012-06-30
TRES 0.618 φ
2012-07-24
TRES 0.811 φ
2012-08-26
ARCES 0.812 φ
2012-08-26
ARCES
0.0
0.2
0.4 0.817 φ
2012-08-27
ARCES 0.864 φ
2012-09-04
ARCES 0.869 φ
2012-09-05
TRES 0.015 φ
2012-09-30
TRES
0.0
0.2
0.4 0.155 φ
2012-10-24
TRES 0.318 φ
2012-11-21
TRES 0.091 φ
2013-04-02
TRES
−60 −40 −20 0 20 40 60
0.196 φ
2013-04-20
ARCES
−60 −40 −20 0 20 40 60
0.0
0.2
0.4 0.511 φ
2013-06-13
ARCES
−60 −40 −20 0 20 40 60
0.982 φ
2013-09-02
ARCES
−60 −40 −20 0 20 40 60
0.023 φ
2013-09-09
ARCES
Smoothed BF
Two-Gaussian fit
Uncorrected Radial Velocity (km s−1)
BroadeningFu
nction
KIC 9246715
Meredith L. Rawls • @merrdiff
Characterizing a star: binary modeling
9.30
9.35
9.40
9.45
9.50
Magnitude
ELC Model
−60
−40
−20
0
20
40
RadialVelocity(km
s−1)
0.0 0.2 0.4 0.6 0.8 1.0
−0.0040.0000.004
−202
9.259.309.359.409.459.50M
agnitude
0.20 0.21 0.22 0.23
−0.0040.0000.004
0.49 0.50 0.51 0.52
Orbital Phase (conjunction at φ = 0.5)
Secondary Primary
∆
∆
∆
✦ Double red giant KIC 9246715✦ Stars are nearly twins; eclipse
depths vary due to geometry✦ Simultaneously fit radial
velocities and light curve✦ Minimize χ2 with 16 free
parameters using DE-MCMC optimizers
Rawls et al. 2016
Radialvelocities
Keplerlight curve
Eclipsezoom
KIC 9246715
Meredith L. Rawls • @merrdiff
Putting the pieces together KIC 9246715
Meredith L. Rawls • @merrdiff
Putting the pieces together KIC 9246715
2.15 M⊙, 8.30 R⊙ 2.17 M⊙, 8.37 R⊙
Meredith L. Rawls • @merrdiff
Putting the pieces together
200 400 600 800 1000 1200 140098.5
99
99.5
100
100.5
Time (BJD - 2454833)
∆I/I
(%)
?
KIC 9246715
Rawls et al. 2016
2.15 M⊙, 8.30 R⊙ 2.17 M⊙, 8.37 R⊙
Meredith L. Rawls • @merrdiff
✦ KIC 9246715 is in good company✦ Oscillation modes are damped or
missing in many red giant binaries✦ All modes affected equally✦ Compare properties of binaries with
strong, damped, and no oscillations✦ Determine what physical conditions
prevent resonant convection zones
Some other systems also show damped or absent oscillations
— no oscillations
νmax ~ 4 μHzνmax ~ 80 μHz
Gaulme et al. 2013
Meredith L. Rawls • @merrdiff
✦ Fourier decomposition turns dozens of composite spectra into one spectrum per star
✦ Stellar atmosphere modeling✦ Lines sensitive to magnetic fields✦ Suite of python programs to assist
with disentangling:
github.com/mrawls/FDBinary-tools
Disentangled spectra hold clues to magnetic activity
One of 23 high-res spectra (light from both stars)Star 1, Teff = 4990 ± 90 KStar 2, Teff = 5030 ± 80 K
KIC 9246715
Meredith L. Rawls • @merrdiff
✦ Magnetically sensitive Fe I lines don’t differ from non-sensitive ones
✦ Ca H & K lines too blue, noisy✦ Ca II λλ8498,8542 show no emission✦ Excess Hα absorption?✦ Orbital solution + light ratio both
needed for successful disentangling
Disentangled spectra hold clues to magnetic activity
5554 5557 5560−0.4−0.20.00.20.40.60.81.01.2 Fe i, mag
5572 5575 5578 5581
Fe i, non-mag
5686 5689 5692 5695
Fe i, non-mag
6298 6301 6304 6307
Fe i, mag
6559 6562 6565
Hα
6840 6844−0.4−0.20.00.20.40.60.81.01.2 Fe i, mag
7088 7092
Fe i, non-mag
Observed
Model
Difference
8490 8500 8510 8520 8530 8540 8550
Caii λλ8498,8542
Wavelength (A)
Scaled
Flux
5554 5557 5560−0.4−0.20.00.20.40.60.81.01.2 Fe i, mag
5572 5575 5578 5581
Fe i, non-mag
5686 5689 5692 5695
Fe i, non-mag
6298 6301 6304 6307
Fe i, mag
6559 6562 6565
Hα
6840 6844−0.4−0.20.00.20.40.60.81.01.2 Fe i, mag
7088 7092
Fe i, non-mag
Observed
Model
Difference
8500 8510 8520 8530 8540
Caii λλ8498,8542
Wavelength (A)
Scaled
Flux
Rawls et al. 2016
KIC 9246715
Meredith L. Rawls • @merrdiff
✦ Robotically-controlled 1m telescope at APO
✦ Kepler cannot measure color✦ Goal: compare eclipse fluxes
with out-of-eclipse flux✦ Especially useful for red
giants with faint companions
Multi-band photometry potentially constrains light ratio and temperatures
Meredith L. Rawls • @merrdiff
✦ Two non-oscillators✦ Mix of orbital periods (19 – 235 days)✦ Three with variability > 1%✦ Two with phase effects✦ Mix of eccentricities (< 0.001 – 0.38)✦ Observed the most in 1m campaign
Choosing a representative subset of red giant binaries
Meredith L. Rawls • @merrdiff
Modeling red giants with faint companions
279.764 279.766 279.768
10000
30000
50000
T0 conj
−0.15652 −0.15648 −0.15644
5000
15000
25000
e cosω
0.1650 0.1660 0.1670
10000
30000
50000
e sinω
89.0 89.4
10000
30000
50000
i
1.12 1.16 1.20
5000
15000
25000
M1
5800 6200 6600
20000
60000
100000 Teff, 1
0.73 0.75 0.77
5000
15000
25000
T2/T1
0.0086 0.0088 0.0090
5000
15000
25000
R1/a
0.0688 0.0692 0.0696
5000
15000
25000
R2/a
25.5 26.5
5000
15000
25000
K1
0.05 0.15 0.25
5000
15000
25000 LD1 q1
0.56 0.60 0.64
10000
30000
50000 LD1 q2
0.4 0.6 0.8
10000
30000
50000
LD2 q1
0.35 0.45 0.55
10000
30000
50000
LD2 q2
0.01 0.03 0.05
20000
60000
100000
Kepler contam
207.1076 207.1080 207.1084
5000
15000
25000
Period
KIC 3955867Broadeningfunction fits
KIC 10001167Disentangledgiant spectrum
KIC 7037405ELC fit parameterdistributions
Meredith L. Rawls • @merrdiff
KIC 8702921
P = 19.4 dayse = 0.10variability ~ 1.4%
No Ca II emissionDeep Hα line
The only SB1in the sample
0.27 M⊙
0.29 R⊙
1.67 M⊙
5.32 R⊙1/7
Meredith L. Rawls • @merrdiff
KIC 92916291.11 M⊙
1.83 R⊙
1.13 M⊙
7.93 R⊙
P = 20.7 dayse = 0.0006variability ~ 16%
Ca II net emissionNormal Hα line
Fast rotatorBVRI used in fit
2/7
Meredith L. Rawls • @merrdiff
KIC 39558670.92 M⊙
0.97 R⊙
1.10 M⊙
8.24 R⊙
P = 33.7 dayse = 0.01variability ~ 23%
Ca II net emissionHα not too deep
Fast rotator
3/7
Meredith L. Rawls • @merrdiff
KIC 100011670.81 M⊙
0.97 R⊙
0.86 M⊙
12.73 R⊙
P = 120.4 dayse = 0.16variability ~ 0.4%
No Ca II emissionDeep Hα line
Phase effects
4/7
Meredith L. Rawls • @merrdiff
KIC 57861541.02 M⊙
1.56 R⊙
1.06 M⊙
11.01 R⊙
P = 197.9 dayse = 0.38variability ~ 0.2%
No Ca II emissionDeep Hα line
Most eccentric
5/7
Meredith L. Rawls • @merrdiff
KIC 70374051.15 M⊙
1.74 R⊙
1.27 M⊙
13.72 R⊙
P = 207.1 dayse = 0.23variability ~ 0.2%
No Ca II emissionDeep Hα line
25 APOGEE RVsMcKeever+ in prep
6/7
Meredith L. Rawls • @merrdiff
KIC 99703961.00 M⊙
1.08 R⊙
1.17 M⊙
7.70 R⊙
P = 235.3 dayse = 0.18variability ~ 0.2%
No Ca II emissionDeep Hα line
Kepler LC gapsBVRI used in fit
7/7
Meredith L. Rawls • @merrdiff
✦ Seismic M, R are both too large✦ Densities and gravities nearly agree✦ Stars least like the Sun differ the most✦ Spectroscopic gravities are too high
The scaling relations are not perfect
Coming soon: Gaulme et al. 2016
Mass Radius
Density(Δν)
Gravity(νmax)
Meredith L. Rawls • @merrdiff
✦ The most active giants lack oscillations, have short periods, and are rotationally synchronous
✦ They also show differential rotation✦ Ca II net emission agrees with
strongest photometric variability✦ Hα may trace chromospheric activity
Magnetic activity roundup
Robinson et al. 1990
Meredith L. Rawls • @merrdiff
10 100 30010
100
2005866138
9246715
7377422
4473933
10015516
8430105
8435232
5179609
4569590530877871332863955867
6307537
5193386
929162911235323
8702921
7943602
Porb [days]
Pvar[d
ays]
5 : 14 : 1 3 : 1 2 : 1 1 : 1 12: 1 Evidence for orbital &
rotational resonances
More stellar variability
Less stellar variability
Stronger oscillationsWeaker oscillations
Gaulme et al. 2014and Ravenel et al. in prep
✦ Binaries with no oscillations more likely to be synchronized
✦ Binaries with synchronization more likely to be active (variable)
Meredith L. Rawls • @merrdiff
✦ MESA: 1D stellar profiles over time✦ Initialize model with M & metallicity;
let star evolve to present R✦ KIC 9246715 contains the only red
clump stars in this sample✦ Adjusted mixing-length parameter
(increase efficiency of convection) to get sufficiently small RC stars
Stellar evolution models constrain tidal history and age KIC 9246715
Meredith L. Rawls • @merrdiff
✦ Depends on stellar evolution and influence of companion star
✦Δe is a function of M, M2/M, Porb, I(t)✦ I(t) is a function of Teff(t), Menv(t), R(t)
Tidal forces over time
Verbunt & Phinney 1995
KIC 9246715
Meredith L. Rawls • @merrdiff
✦ Depends on stellar evolution and influence of companion star
✦Δe is a function of M, M2/M, Porb, I(t)✦ I(t) is a function of Teff(t), Menv(t), R(t)
Tidal forces over time
Verbunt & Phinney 1995
M2/M = 0.989KIC 9291629
Meredith L. Rawls • @merrdiff
✦ Rotation synchronization should happen before orbit circularization
✦ Two most active giants with short periods are nearly circularized
✦ Eccentric systems have not had enough time for tides to circularize
✦ Two outliers/in-between cases
Predicted changes in eccentricity
Prediction: a range of oscillation populations
Meredith L. Rawls • @merrdiff
Tidally-induced“heartbeat” pulsations
No solar-likeoscillations
Damped solar-like oscillations
Oscillates likea single star
8702921
9291629
395586710001167
57861549246715
7037405
9970396
Meredith L. Rawls • @merrdiff
✦ Distances computed via theoretical bolometric corrections, updated KIC extinctions, & JHK magnitudes from 2MASS
✦ No clear galactic [Fe/H] gradient
Galactic context
Southworth et al. 2005
Meredith L. Rawls • @merrdiff
Pipeline comparisons
temperature
surfacegravity
metallicity
distance
Meredith L. Rawls • @merrdiff
✦ Solar-like oscillations are damped by magnetism & tides, which often occur together
✦ Most pronounced for short-period, spotty, synchronized and circularized systems
✦ The asteroseismic scaling relations are least accurate for stars least like the Sun
✦ Scaling relations overestimate giant M, R✦ These stars are references for survey pipelines✦ Asteroseismic surveys will tend to miss active
stars and close binaries
Red giant binaries are powerful benchmarks for asteroseismology