Skip to main content

Full text of "Modeling Multi-Wavelength Stellar Astrometry. III. Determination of the Absolute Masses of Exoplanets and Their Host Stars"

See other formats


Accepted for Publication in the Astrophysical Journal 

Preprint typeset using I^'T^]X style emulatcapj v. 5/2/11 



(N 

o 



(N 

6 



p 

rn 

O 



X 



MODELING MULTI- WAVELENGTH STELLAR ASTROMETRY. IIL DETERMINATION OF THE ABSOLUTE 

MASSES OF EXOPLANETS AND THEIR HOST STARS 

Jeffrey L. Coughlin^"' and Mercedes Lopez-Morales^'^ 

Draft version February 24, 2013 

ABSTRACT 

Astrometric measurements of stellar systems are becoming significantly more precise and common, 
with many ground and space-based instruments and missions approaching 1 /xas precision. We examine 
the multi-wavelength astrometric orbits of exoplanetary systems via both analytical formulae and 
numerical modeling. Exoplanets have a combination of reflected and thermally emitted light that cause 
the photocenter of the system to shift increasingly farther away from the host star with increasing 
wavelength. We find that, if observed at long enough wavelengths, the planet can dominate the 
astrometric motion of the system, and thus it is possible to directly measure the orbits of both the 
planet and star, and thus directly determine the physical masses of the star and planet, using multi- 
wavelength astrometry. In general, this technique works best for, though is certainly not limited to, 
systems that have large, high-mass stars and large, low-mass planets, which is a unique parameter 
space not covered by other exoplanet characterization techniques. Exoplanets that happen to transit 
their host star present unique cases where the physical radii of the planet and star can be directly 
determined via astrometry alone. Planetary albedos and day-night contrast ratios may also be probed 
via this technique due to the unique signature they impart on the observed astrometric orbits. We 
develop a tool to examine the prospects for near-term detection of this effect, and give examples of 
some exoplanets that appear to be good targets for detection in the K to N infrared observing bands, 
if the required precision can be achieved. 



Subject headings: astrometry — planetary systems 



1. INTRODUCTION 

As part of a Space Interferometry Mission (SIM) Sci- 
ence Study, in lCoughlin et all (|2010bl ). hereafter referred 
to as Paper I, we examined the implications that multi- 
wavelength microarcsecond astrometry has for the detec- 
tion and characterization of interacting binary systems. 
In Paper I we found that the astrometric orbits of binary 
systems can vary greatly with wavelength, as astrometric 
observations of a point source only measure the motion of 
the photocenter, or center of light, of the system. For sys- 
tems that contain stellar components with different spec- 
tral energy distributions, the motion of the photocenter 
can be dominated by the motion of either component, 
depending on the wavelength of observation. Thus, with 
multi-wavelength astrometric observations it is possible 
to measure the individual orbit of each component, and 
thus de rive absolut e masses for both objects in the sys- 
tem. In lCoughlin. H arrison. & Gelino (2010a), hereafter 
referred to as Paper II, we showed that multi-wavelength 
astrometry can also be used to directly measure the incli- 
nation and gravity darkening coefhcient of single stars, as 
well as the temperature, size, and position of star spots. 

Astrometry has long been used to measure fundamen- 



tal quantities of binary stars, and more recently has 
been used to study extrasolar planets. Although no 
independently confirmed planet has yet been initially 
discovered via astrometry, many planets discovered via 
radial- velocity, (which only yields the planetary mass as 
a function of the system's inclination and host star's 
mass), have had follow-up astrometric measurements 
taken in order to determine their inclinations, and thus 
true planeta ry mass as a function of only the assumed 
stellar mass (iMcArthur et al. 2001 [Benedict et al.ll2006l: 
Bean et al.l 120071: iMartioU et all l2010t iMcArthur et al.l 



20101 : IRbUerail I2OIOI: IReffert fc Ouirrenbachl I201H ) 



Mlcough@ninsu.edul 

^ Uepartment of Astronomy, New Mexico State University, 
P.O. Box 30001, MSC 4500, Las Cruces, New Mexico 88003- 
8001 

2 Institut de Ciencies de I'Espai (CSIC-IEEC), Campus UAB, 
Facultat de Ciencies, Torre C5, parell, 2a pi, E-08193 Bellaterra, 
Barcelona, Spain 

^ Carnegie Institution of Washington, Department of Terres- 
trial Magnetism, 5241 Broad Branch Road NW, Washington, 
DC 20015-1305, USA; Visiting Investigator 

'^ NSF Graduate Research Fellow 



There are many ground and space-based microarcsec- 
ond precision astrometric projects which are either cur- 
rently operating or on the horizon. The proposed SIM 
Lite Astrometric Observatory, a redesign of the earlier 
proposed SIM PlanetQuest Mission, was to be a space- 
based 6-meter baseline Michelson interferometer capable 
of 1 //as precision measur ements in ^80 sp ectral channels 
spanning 450 to 900 nm (jDavidson et al.i r2009). thus al- 
lowing niulti- wavelength microarcsecond astrometry. Al- 
though the SIM Lite mission has been indefinitely post- 
poned at the time of this writing, it has already achieved 
all of its technological milestones, and it, or another 
similar mission, could be launched in the future. The 
PHASES project obtained as good as 34 /ias astr omet- 
ric p recision of close stellar pairs (Muterspaugh "et" al.l 
l2010f l. The CHARA array has multi- wavelength capa- 
bilities, and can provide ang ular resolution to ~200 /ias 
(jten Brummelaar et alll2005h . PRIMA/ VLTI is working 
towards achieving ^3 0-40 /zas precision in the K-band 
()van Belle et al.l l2q08D. with GRAVITY /VLTI expected 
to obtain 10 uas (jKudrvavtseva et~aIll2010D . The AS- 



TRA/KECK project will be able to simultaneously ob- 
serve and measure the distance between two objects to 
better than 100 ^as precision. The GAIA mission will 
provide astrometry for ~10^ objects with 4 - 160 //as ac- 
curacy, for stars with V = 10-20 mag respecti vely, and 
does posses some multi-wavelength capabilities (jCacciarll 
120091 ). The MIC ADO instrument on the proposed E- 
ELT 40-meter class telescope will be able t o obtain bet- 
ter t han 50 /^as accuracy at 0.8-2.5 /xm (jTrippe et al.l 
120101 ). Finally, the NEAT mission proposes to obtain 
as low as 0.05 / xas astrometric me asurements at visi- 
ble wavelengths (|Malbet et al.l 120111 ). Thus, astrometric 
measurements of extrasolar planets are going to become 
significantly more common in the future. 

In this paper, we examine the multi-wavelength astro- 
metric signature of exoplanets. A star-planet system is 
a specialized case of a binary system with extreme mass 
and temperature ratios, and thus the findings of Paper I 
apply to exoplanets. Specifically, an extrasolar planet 
has a combination of reflected and thermally emitted 
light that cause the photocenter to be displaced from 
the center of mass of the star. Since the planet's tem- 
perature is very different from that of the host star, the 
amount of photocenter displacement due to the planet 
will greatly vary with wavelength. Although the luminos- 
ity ratio between a star and planet is extreme, the planet 
also lies a much farther distance from the barycenter of 
the system compared to the star, and thus it has a large 
"moment-arm" with which to influence the photocenter. 
While conventional single-wavelength astrometric mea- 
surements can yield the inclination and spatial orienta- 
tion of a system's orbital axis, with multi- wavelength as- 
trometry it should be possible to measure the individual 
orbits of both the star and planet, and thus determine 
the absolute masses of both. 

In Sj2]we derive analytical formulae for estimating the 
astrometric motion of a star-planet system at a given 
wavelength. In iJ3] we perform numerical simulations of 
the multi- wavelength astrometric orbits of a few systems 
of interest using the reflux code, and examine a few 
features speciflc to transiting planets. In both sections 
we present the most promising systems for future ob- 
servation and detection of this effect. Finally, in [2] we 
discuss our results and what future work is needed to 
achieve these observations. 



2. ANALYTICAL FORMULAE FOR COMPUTING 
THE REFLEX MOTION 

Our objective is to derive an analytical expression for 
the amplitude of the sky-projected angular astrometric 
reflex motion of a star-planet system with respect to the 
wavelength of observation, a. In all of the following equa- 
tions, we are dealing with sky-projected distances mea- 
sured along the semi-major axis of the system, and thus 
they are independent of the inclination of the system. We 
consider the case of a star and single planet in a circular 
orbit, with masses M^, and M^ respectively, separated 
by an orbital distance, a, as illustrated in Figure [TJ The 
system's barycenter, marked via a "-I-" symbol, lies in- 
between the star and planet, at a distance of r^, from the 
star, and r^ from the planet. 

Defining the mass ratio, g, as 




Figure 1. An illustration of a system containing a star, shown on 
the left, and a planet, shown on the right, separated by a distance 
a, not to scale. The star and planet lie at distances of r* and rp, 
respectively, from the barycenter of the system, which is marked 
via a "-I-" symbol. Similarly, the star and planet lie at distances 
of s* and Sp, respectively, from the photocenter of the system, 
which is marked via a "x" symbol. All distances are sky-projected 
distances along the semi-major axis of the system, and thus are 
independent of the system's inclination. Note that although in 
this illustration the photocenter is to the left of the barycenter, it 
can lie anywhere between the star and planet. 



the values for r^ and r^ are then 



Ti, 



a ■ q 



q+l 



(1) 



(2) 



(3) 



where by definition r^, + rp — a. 

In the case where all the light from the system is as- 
sumed to come from the star, i.e., the system's photo- 
center is the star's center, the wavelength-independent 
amplitude of the angular astrometric reflex motion of 
the system, ao, is 



ao = arctan 



m 



arctan 



a ■ q 



D-{q + l) 



(4) 



where D is the distance to the system from Earth, and 
a, via Kepler's third law, is 



(G(M, -f Afp))- 



(5) 



where G is the gravitational constant, and P is the or- 
bital period of the system. 

When the planet's luminosity is not negligible, in order 
to determine the wavelength-dependent value of a, the 
location of the system's photocenter, which varies with 
wavelength, must be determined. We define s,t and Sp 
to be the distance to the system's photocenter from the 
star and planet respectively, as shown in Figure [TJ where 
the photocenter is marked with a " x " symbol. We define 
the luminosity ratio at a given wavelength, Lr, as 






(6) 



where Lp is the luminosity of the planet, and L^, is the 
luminosity of the star. Thus, similar to the previously 
presented derivations, the values for Si, and Sp are 



a ■ Lr 



s* 



Lr 



1 



(7) 



(8) 



where by definition s^, + Sp — a. The observed astro- 
metric motion results from the movement of the system's 
photocenter around the system's barycenter. Thus, tak- 
ing into account light from both the star and planet. 



a = arctan I * „ * I = arctan ' — 



D 



D 



and thus 



a — arctan 



{q- Lr) 



D-iq + l)-iLr + l) 



(9) 



(10) 



where we have defined a so that a > signifies that 
the star dominates the observed reflex motion, i.e., Lr < 
q, and a < signifies that the planet dominates the 
observed reflex motion, i.e., Lr > q. Note that when the 
barycenter and photocenter are at the same point, i.e., 
Lr — q, and thus a = 0, no reflex motion is observable. 

We now estimate the value of Lr based upon the values 
of readily measurable system parameters. Light emitted 
from the planet consists of both thermally emitted light, 
as well as incident stellar light reflected off the planet. 
Thus, 



J_j^ - — - 



Lf 



La 



L. 



Le 
L, 



La 
L. 



(11) 



where Le is the luminosity of the planet from thermal 
emission, L^, is the luminosity of the star, and La is the 
luminosity of light reflected off the planet. To estimate 
the thermal component, we assume that both the star 
and planet radiate as blackbodies, and thus 



i?2 exp{ 



Le 



Li, Rl exp{ 



he 



1 



he 
XkT„ 



(12) 



where Rp is the radius of the planet, A is a given wave- 
length, h is Planck's constant, c is the speed of light, k is 
Boltzmann's constant, T^, is the effective temperature of 
the star, and Tp is the effective temperature of the planet. 
To derive Tp we first assume that the planet is in radia- 
tive equilibrium, and has perfect heat re-distribution, i.e. 
a uniform planetary temperature, and thus 



Tp^T, 



{l~As)-Rl 
4a2 



(13) 



where T^, is the temperature of the star. As is the plan- 
etary Bond albedo, and i?* is the radius of the star. 



To estimate the contribution due to reflected light, we 
first note that the flux received at the planet's surface 
is Li, divided by the surface area of a sphere at a dis- 
tance a, i.e., Atto^. The planet intercepts and reflects 
this light on only one of its hemispheres, which has effec- 
tive cross sectional area of TrRp, with an efficiency equal 
to the albedo. Combining these terms and re-arranging 
to obtain the luminosity ratio due to reflected light yields 



La 
L. 



AxRl 
4a2 



(14) 



where A\ is the planet's albedo at a given wavelength. 

Combining the above equations, and assuming values 
of Ab and A\, we can estimate a at a given A, using 
only Mi,, Ri,, T^, Mp, Rp, P, and D. We note that 
this assumes that the planet is in radiative equilibrium, 
but does not account for any additional internal heat 
sources from the planet, such as gravitational contrac- 
tion or radioactive decay. While internal heat sources 
are likely to be negligible for close-in planets, it could 
significantly contribute to the total luminosity of further 
out gaseous planets, thus making them even more eas- 
ily detectable. Our approximation for a also assumes 
that the planet's luminosity is constant over its orbit as 
observed from Earth. However, some planets have signif- 
icant flux differences between their day and night sides 
due to low day-to-night re-radiation efficiency and/or sig- 
nificant planetary albedos. In these cases, if the inclina- 
tion of the system is ^ 0°, then the planet's luminos- 
ity will vary with orbital phase as seen by the observer, 
and the projected astrometric orbit of the photocenter 
at wavelengths where the planet's luminosity dominates 
will deviate from an ellipse, with increasing deviation 
as the inclination approaches 90°. (This effect is fur- 
ther discussed and illustrated in Section [H) As well, 
we assumed a circular orbit, and thus eccentric planets 
with varying levels of stellar insolation and temperature 
would have unique orbital signatures resulting from time- 
variant planetary flux. Finally, we assumed in this ana- 
lytical derivation that the star and planet are effectively 
point sources, but of course in reality they have a physical 
size. If the star and/or planet have non-symmetric sur- 
face features, such as star spots or planetary hot spots, 
then the star and planet could each influence the loca- 
tion of the photocenter as these features rotated across 
their surface. The effect on the photocenter would only 
be a fraction of their physical radii, and would only cause 
signiflcant deviations to the observed astrometric orbit if 
the radii of either object was a signiflcant fraction of the 
object's distance from the system's barycenter. While 
this would likely be negligible for the planet, it could be 
significant for the star, e.g., the case of microarcsecond, 
wavelength-dependent, astrometric perturbations result- 
ing from star spots presented in Paper II. 

In Figure [2] we present plots of a versus A for a Jupiter- 
like planet, {Mp = 1.0 Mj, Rp = 1.0 Rj), around FOV, 
G2V, and MOV stars at 10 parsecs, with periods of 1, 
10, 100, and 1000 days. We also show various planetary 
albedos, assuming Ab = Ax, of 0.0, 0.25, 0.5, and 0.75. 
In Figure [3] we do the same for an Earth-like planet, 
{Mp = 1.0 Mq, Rp — 1.0 i?©). In general, systems that 
have large, high-mass stars and large, low-mass planets 
present the best opportunity to observe negative val- 



ues of a, and thus be able to directly determine their 
masses. (This is a unique parameter space not covered 
by other exoplanet characterization techniques such as 
radial- velocity or the transit method.) Short-period, and 
thus hot, planets around more massive stars transition 
to negative values of a at shorter wavelengths, but have 
lower overall amplitudes compared to long-period, and 
thus cool, planets around low-mass stars. Reflected light 
is a fairly minor contribution, only having some signifi- 
cant relevance for planets with very short orbital periods, 
i.e., ~1 day. For both hot Jupiters and hot Earths, neg- 
ative values of a can be observed with wavelengths as 
short as ~2 /xm, i.e., the K band. Considering A < 100 
/zm, a < could only be observed for P < 100 days for 
a Jupiter-like planet, and for P < 500 days for an Earth- 
like planet. Earth itself, (P = 365 days around a G2V 
star), would have a value of a « 0.3 /ias for A < 10 /im, 
and a ~ -0.05 /ias at 100 /im, and thus, theoretically, 
the absolute mass of an Earth-analogue and its host star 
could be determined via this technique. 

Utilizing exoplanets.org, we have collected the values 
for all the previously mentioned system parameters for 
all currently known exoplanets. Selecting those that have 
well-determined values of all the needed parameters, in 
Table [1] we list the top five exoplanets with the largest 
negative values of a for each of the K (2.19 /im), L (3.45 
/im), M (4.75 /im), and N (10.0 /im) infrared bandpasses, 
with a total of 11 unique exoplanets. We choose these 
wavelengths as they are the major ground-based infrared 
observing windows, and no systems examined had nega- 
tive a values at wavelengths shorter than ~2 /im. All of 
the candidate systems ended up being transiting planets 
both because they have well-determined values for the 
planetary radii, and transit surveys are most sensitive to 
close-in planets. As can be seen, the top candidates for 
detecting a < 0, and thus measuring the absolute mass 
of the planet, are WASP- 12 b in the K band with a = 
-0.05 /ias, HD 209458 b in the L and M bands with a 
= -0.23 and -0.66 /ias respectively, and HD 189733 b in 
the N band with a — -3.04 /ias. It is interesting that 
three low-mass Neptune and sub-Ncptune mass planets, 
55 Cnc e, Gliese 436 b, and GJ 1214 b, also make the 
list, illustrating that this technique can 'favor' the char- 
acterization of low-mass planets. 

3. NUMERICAL MODELING VIA REFLUX 

In order to provide a check on our analytical formulae, 
better illustrate the multi-wavelength astrometric orbits 
of exoplanet systems, and prob e some more subtle ef - 
fects, we use the reflux0 code (|Coughlin et al.ll2010bl ). 
which computes the flux- weighted astrometric reflex mo- 
tion of binary systems at multiple wavelengths, to model 
a couple known exoplanet systems. We discussed the 
code in detail in Papers I and II, but in short, it utilizes 
the Eclipsing Light Curve (ELC) code, which was writ- 
ten to compute light curv es of eclipsing binary systems 
(jOrosz fc: Hauschild^l2000[) . ELC includes the dominant 
physical effects that shape a binary system's light curve, 
such as non-spherical geometry due to rotation and tidal 
forces, gravity darkening, limb darkening, mutual heat- 
ing, reflection effects, and the inclusion of hot or cool 
spots on the stellar surface. The ELC code represents 
the surfaces of two stars, or a star-planet system, as a 
grid of individual luminosity points, and calculates the 



Table 1 

Currently Known Exoplanets with the Most Negative a Values 



Name 


D 


M« 


R* 


Tt 


Mp 


Rv 


P 


Q 




(pc) 


(Mo) 


(R©) 


(K) 


(Mj) 


(R.t) 


(Days) 


(/las) 








K Band 


(2.19 urn) 








WASP- 12 b 


427 


1.28 


1.63 


6300 


1.35 


1.79 


1.091 


-0.05 


WASP- 19 b 


250 


0.93 


0.99 


5500 


1.11 


1.39 


0.789 


-0.05 


WASP-33 b 


115 


1.50 


1.44 


7430 


2.05 


1.50 


1.220 


-0.04 


55 Cnc e 


12 


0.96 


0.96 


5234 


0.03 


0.19 


0.737 


-0.01 


CoRoT-1 b 


480 


0.95 


1.11 


5950 


1.03 


1.49 


1.509 


-0.01 






L Band 


(3.45 p 


m) 








HD 209458 b 


49 


1.13 


1.16 


6065 


0.69 


1.36 


3.525 


-0.23 


WASP-33 b 


115 


1.50 


1.44 


7430 


2.05 


1.50 


1.220 


-0.20 


WASP-19 b 


250 


0.93 


0.99 


5500 


1.11 


1.39 


0.789 


-0.15 


WASP- 17 b 


300 


1.19 


1.20 


6550 


0.49 


1.51 


3.735 


-0.11 


WASP- 12 b 


427 


1.28 


1.63 


6300 


1.35 


1.79 


1.091 


-0.10 






M Band 


(4.75 Atm) 








HD 209458 b 


49 


1.13 


1.16 


6065 


0.69 


1.36 


3.525 


-0.66 


HD 189733 b 


19 


0.81 


0.76 


5040 


1.14 


1.14 


2.219 


-0.47 


WASP-33 b 


115 


1.50 


1.44 


7430 


2.05 


1.50 


1.220 


-0.29 


WASP-19 b 


250 


0.93 


0.99 


5500 


1.11 


1.39 


0.789 


-0.21 


WASP- 17 b 


300 


1.19 


1.20 


6550 


0.49 


1.51 


3.735 


-0.19 






N Band 


(10.0 Aim) 








HD 189733 b 


19 


0.81 


0.76 


5040 


1.14 


1.14 


2.219 


-3.04 


HD 209458 b 


49 


1.13 


1.16 


6065 


0.69 


1.36 


3.525 


-1.53 


Gliese 436 b 


10 


0.45 


0.46 


3684 


0.07 


0.38 


2.644 


-0.95 


WASP-34 b 


120 


1.01 


0.93 


5700 


0.58 


1.22 


4.318 


-0.64 


GJ 1214 b 


12 


0.16 


0.21 


3026 


0.02 


0.24 


1.580 


-0.59 



resulting light curve given the provided systemic param- 
eters. REFLUX takes the grid of luminosity points at each 
phase and calculates the flux-weighted astrometric pho- 
tocenter location at each phase, taking into account the 
system's distance from Earth. Although ELC is capa- 
ble of using model atmospheres, for this paper we set the 
code to calculate luminosities assuming both the star and 
planet radiate as blackbodies. 

We choose to model Wasp-12, HD 209458, and HD 
189733, as they are all well-studied systems, and have 
the most negative a values for the K, L, M, and N band- 
passes presented in Table [T] For each system we set 
the values for M^,, i?^,, T^, Mp, Rp, P, D, and rota- 
tion period of the star to those in the Exoplanets.org 
database, and set the rotation period of the planet to 
the orbital period of the system, i.e., assume the planet 
is tidally locked, and assume a circular orbit. We as- 
sume that the spin axes of both the star and planet are 
perfectly aligned with the orbital axis. We employ the 
use of spots in the ELC code to simulate a day/night 
side temperature difference, by assuming a uniform day- 
side temperature for the planetary hemisphere facing the 
star, and a uniform night-side temperature for the plan- 
etary hemisphere facing away from the star. We employ 
the valu es for the day a nd night side temperatures de- 
rived by ICowan fc Agoll (|2011li which were 2939 K for 
the day-side of Wasp-12 b, 1486 and 1476 K for the day 
and night sides respectively of HD 209458 b, and 1605 
and 1107 K for the day and night sides respectively of 
HD 189733 b. We adopted a temperature of 1470 K for 

^ REFLUX can be run via a web in terface from 
http : //astronomy . nmsu . edu/j lcough7ref lux .html Additional 
details as to how to set-up a model arc presented there. 



Jupiter with P = 1 .0 Days around a FOV Star at 1 







Si!iaim™-i=illll!»s.. 










%\ 




\A\ 




\s\ 




A„,.aoo \\\ \ 




- A°t-0.25 \.,\\ 

Ab1.0.50 \..\\ 






Ab,a-o.75-.- V.;-,, V, 








\>-.""""- 




^c::::;;;^ 





Jupiter witti P = 1 D.D Days around a FOV Star at 1 D Parsecs 




Jupiter with P = 1 00.0 Days around a FOV Star at 1 Parsecs 





28 






- 






"^.-y 
\''\ 




26 




^Vv 


- 




24 




Y^* 


\ 


3- 


22 

20 
18 
16 


Ap'j =0.25 

- A^'^ 50 


v' 


\\"-- ""■"■" 


a 


- V^O-75 




14 










12 






^^^^ 



Jupiter with P = 1 000.0 Days around a FOV Star at 1 



130 
125 - 



ABJ..O00 

Am = 025 — - 






Jupiter with P = 1 .0 Days around a G2V Star at 1 Parsecs 




^^^\ 





%\ 


-s- -1 

"b -2 

-3 


Apj =0.00 \'.\ \ 

Apj-0.50 V.\ \ 

- Ab;, = 0.75-.-.-.- W\X, 


-4 


\s^- 

. .^~T^ T- 1 . 



Jupiter with P = 1 .0 Days around a MOV Star at 1 Parsecs 



?.(nm) 
Jupiter with P = 1 0.0 Days around a G2V Star at 1 D Parsecs 



Ag;j =0.00 . 

AbV = 0.25 ■ 

AbV = 0.50 . 

Ab',' = 0.75 . 




Jupiter with P = 1 00. D Days around a G2V Star at 1 



35 

_ 30 

g 25 

20 

15 




Jupiter with P = 1000.0 Days around a G2V Star at 10 



= 0.00 . 

= 0.25 - 

= 0.50 ■ 

= 0.75 . 





Jupiter with P = 1 0.0 Days around a MOV Star at 1 Parsecs 




?.(^m) 
Jupiter with P = 1 00. D Days around a MOV Star at 1 



50 



40 - "b,^ - 



= 0.00 

Ab\ = 0.25 

^ Ap'j = 0.50 

a 30 - A=*. 0.75 -.-.-.- 



?.{tim) 
Jupiter with P = 1000.0 Days around a MOV Star at 1 




Figure 2. Plots of the reflex motion amplitude, a, versus the wavelength of observations, A, for a Jupiter-like planet, {Mp = 1.0 Mj, Rp 
= 1.0 Rj) around FOV, G2V, and MOV stars at 10 parsecs, (left, middle, and right columns respectively), at periods of 1, 10, 100, and 
1000 days, (top to bottom rows, respectively). The solid, dashed, dotted, and dash-dotted lines represent planetary albedos of 0.0, 0.25, 
0.5, and 0.75 respectively. 



the night side of Wasp-12 b, i.e., half that of the day 
side, assuming very httle planetary heat redistribution. 
For all the systems, we set the st ar's gravity d arkening 
coefficients to those determined bv lClaretl ()200ClD . though 
do not enable gravity darkening for the planet. For both 
the planet and star, we assume zero albedo, since we are 
dealing principally with infrared wavelengths where the 
effect is negligible, and we have already shown that even 
in the optical reflected light is a minor contribution to the 
astrometric motions under investigation. Furthermore, 
the chosen planets are expected to have v ery low albe- 
dos JAfi < 0.3) fro m m odel atmosphere s (Marlcv et al.l 
[T999t iSeageret al.][2000l : iSudarsky et"all[2000 ') , and have 
even had their albedos co nstrained to very low val - 
ues from obse rvations, e.g ., iLopez-Morales et al.l ()201Clf ) 
for Wasp-12b IRowe et al.l (12008D for HD 209458b, and 
lWiktorowic3 (J200^~fo7HD 189733b. We also do not as- 



sume any limb-darkening since we are dealing principally 
with infrared wavelengths. 

In Figures SI [5l and [5] we present plots of the X and 
Y components of the photocenter versus phase, as well 
as the sky-projected X-Y orbit of the photocenter, in the 
V, J, H, K, L, M, and N passbands, for Wasp-12, HD 
209458, and HD 189733 respectively. The point (X,Y) = 
(0,0) corresponds to the barycenter of the system, and 
the projected orbital rotation axis is parallel to the Y- 
axis. Phase 0.0 corresponds to the primary transit, when 
the planet passes in front of the star and is closest to the 
observer, and phase 0.5 corresponds to the secondary 
eclipse, when the planet passes behind the star and is 
farthest away from the observer. 

Examining the modeling results, the values for a deter- 
mined via the analytical formulae appear to match the 
numerical modeling results fairly well. For example, via 



Earth with P = 1 .0 Days around a FOV Star at 1 Parsecs 



Earth with P = 1 .0 Days around a G2V Star at 1 F 




?.(nm) 
Earth with P = 1 0.0 Days around a FOV Star at 1 Parsecs 




Earth with P = 100.0 Days around a FOV Star at 10 Parsecs 




X (|.im) 
Earth with P = 1000.0 Days around a FOV Star at 10 Parsecs 



0.45 p- 

0.4 - 
0.35 - 



„„ Arj =0.00 - 

°-^ " Ab1 = 0.25. 

Ar'i =0.50 ■■ 

0.25 - a|;; = 0.75. 

0.2 - 




iMUKAUiEnAnaun 



-0.01 - 
-S- -0.02 
~a -0.03 
-0.04 
-0.05 
-0.06 



Ar 1 = 0.00 

AbV = 0.25 

Ar\" = 0.50 

Ar\: = 0.75 



Earth with P = 1 0.0 Days around a G2V Star at 1 Parsecs 



a -0.04 
-0.06 - 
-0.08 
-0.1 



■0-02 ^ J-:o25 — 

Ari =0.50 

ArV = 0.75 



Earth with P = 1 00.0 Days around a G2V Star at 1 Parsecs 



Ar , = O.DO 

Ab; = 0.25 .... 

Ar/ = 0.50 

ArV = 0.75 — . 



Earth with P = 1 000.0 Days around a G2V Star at 1 



Agi =0.00 - 
Ab\ = 0.25 - 
Ag;^ = 0.50 . 






Earth with P= 1.0 Days 


around 


1 MOV Star at 10 Parsecs 


0.02 








' ' 




' ' 





„ ^ 


%\ 


- 


0.02 
0.04 
0.06 


Arj =0.00 

a1' =0.25 

" Ar', -0.50 

Alt = 0.75-.- 




^ : 


0.08 









Earth with P = 1 0.0 Days around a MOV Star at 1 Parsecs 




Earth with P = 1 00.0 Days around a MOV Star at 1 Parsecs 



0.25 

0.2 

0.15 

0.1 

0.05 

^ 

t -0-05 

-0.1 

-0.15 

-0.2 

-0.25 



' ' 


' 


- 






- Ar'> -0.25 

. a|1 = 0.50 

Ab:i = 0.75 


w 


l^v-,; 












\*v """ 









Earth with P = 1 0OO.O Days around a MOV Star at 1 Parsecs 




Figure 3. Plots of the reflex motion amplitude, a, versus the wavelength of observations, A, for an Earth-like planet, {Mp = 1.0 M^g, Rp 
= 1.0 fi®) around FOV, G2V, and MOV stars at 10 parsecs, (left, middle, and right columns respectively), at periods of 1, 10, 100, and 
1000 days, (top to bottom rows, respectively). The solid, dashed, dotted, and dash-dotted lines represent planetary albedos of 0.0, 0.25, 
0.5, and 0.75 respectively. 



Table[Tl Wasp-12b was predicted to have a values of -0.05 
and -0.10 /zas in the K and L bands respectively, com- 
pared to the maximum, out-of-transit, numerical model 
resuhs of -0.05 and -0.08 fias. For HD 209458b, expected 
a values were -0.23, -0.66, and -1.53 /las for the L, M, and 
N bands, compared to -0.30, -0.74, and -1.63 /ias from the 
numerical models. For HD 189733b, expected a values 
were -0.47 and -3.04 ^as for the M and N bands, com- 
pared to -2.10 and -4.68 /zas from the numerical models. 
The differences are principally due to the use of obser- 
vationally determined day and night side temperatures 
in the numerical models, whereas the analytical formu- 
lae assumed perfect radiative equilibrium and a uniform 
planetary temperature. 

Although a transition from positive to negative a ap- 
pears to occur around the H, K, and L bands for Wasp-12 
b, HD 209458 b, and HD 189733 respectively, a deviation 



from the visible light signature is clearly visible at shorter 
wavelengths, and thus it may be possible to disentangle 
the astrometric motion due to the planet even at shorter 
wavelengths where it does not dominate the reflex motion 
of the photocenter. For Wasp-12 b and HD 189733 b the 
out of transit /eclipse signature deviates from a sinusoid 
due to the extreme day/night temperature differences on 
these planets, as discussed in Section [2l The different 
inclinations of the systems are immediately apparent in 
the X-Y orbit plots, and when actually measured on sky, 
would directly yield the three-dimensional orbit of the 
system. 

The presence of the primary transit and secondary 
eclipse is clearly visible in all three cases, with the pri- 
mary transit dominating the maximum amplitude of the 
astrometric shift for the visible wavelengths, particularly 
in the Y-direction. As no limb-darkening was assumed 




0.1 


CS^ V-band 

\X J-band 

\ \ H-band 

\ \ K-band 

\ \ L-band 

\ \ M-band 

\ \ N-band 




/ 


0.08 


\\ 


/ 


V 


0.06 


\\. 


j/y 


/ 




\ ^^^=s= 


^^^ / 




0.04 


\ ^^^^5^ 


/ 


- 


0.02 


( d^X 


^^ 


^ 





^ Vvj^. 


m^J ^ 


- 











Figure 4. Plots of the mu It i- wavelength astrometric orbit for the Wasp-12 system. Left: The X and Y components of motion versus 
phase. Right: The sky-projected, X-Y, orbit. The point (X,Y) = (0,0) corresponds to the system's barycenter, and the projected orbital 
rotation axis is parallel to the Y-axis. Phase 0.0 corresponds to the primary transit, when the planet passes in front of the star and is 
closest to the observer, and phase 0.5 corresponds to the secondary eclipse, when the planet passes behind the star and is farthest away 
from the observer. 





Figure 5. Plots of the mult i- wavelength astrometric orbit for the HD 209458 system. Left: The X and Y components of motion versus 
phase. Right: The sky-projected, X-Y, orbit. The point (X,Y) = (0,0) corresponds to the system's barycenter, and the projected orbital 
rotation axis is parallel to the Y-axis. Phase 0.0 corresponds to the primary transit, when the planet passes in front of the star and is 
closest to the observer, and phase 0.5 corresponds to the secondary eclipse, when the planet passes behind the star and is farthest away 
from the observer. 




4.00 
3.50 
3.00 
2.50 
2.00 
1.50 
1.00 
0.50 
0.00 
-0.50 



V-band - 
J-band - 
H-band 
K-band - 
L-band - 
M-band 
N-band 






Figure 6. Plots of the mult i- wavelength astrometric orbit for the HD 189733 system. Left: The X and Y components of motion versus 
phase. Right: The sky-projected, X-Y, orbit. The point (X,Y) = (0,0) corresponds to the system's barycenter, and the projected orbital 
rotation axis is parallel to the Y-axis. Phase 0.0 corresponds to the primary transit, when the planet passes in front of the star and is 
closest to the observer, and phase 0.5 corresponds to the secondary eclipse, when the planet passes behind the star and is farthest away 
from the observer. 



in these models, the variation in the primary and sec- 
ondary echpse signatures with wavelength is due to the 
relative flux of the s t ar an d planet in those passbands. 
As noted by iGaudil ()2010t ). measuring the astrometric 
shift of the primary transit directly yields the angu- 
lar radius of the host star, and if the distance to the 
system is precisely known, one can directly derive the 
physical radius of the star. Additionally, if the den- 
sity of the star is directly determined froin the photo- 
metric light curve (jSeager fc Mallen-Ornelasl[2003| ). then 
one can also directly derive the mass of the star. We 
also note, for the first time, that measuring the astro- 
metric signature of the primary transit and, if observ- 
ing at longer wavelengths, the secondary eclipse, specif- 
ically the duration of ingress and egress, similarly di- 
rectly yields the angular radius of the planet. Since one 
may directly determine the surface gravity of the planet 
from t he photometric light and radial-velo city curves 
alone (jSouthworth. Wheatlev. fc Sams! [20071 ) . one may 
also directly determine the mass of the planet. Thus, 
for transiting planets, multi- wavelength astrometric mea- 
surements yield two independent methods of measuring 
the physical stellar and planetary masses. 

4. DISCUSSION AND SUMMARY 

We have shown that the multi-wavelength astromet- 
ric measurements of exoplanetary systems can be used 
to directly determine the masses of extrasolar planets 
and their host stars, in addition to the inclination and 
spatial orientation of their orbital axis. If the planet 
happens to transit the host star, then the angular radius 
of both the star and planet can be directly determined, 
and when combined with the trigonometric parallax of 
the system, the absolute radii of the planet and host star 
can directly determined via astrometry alone. We found 
that this technique is best suited, though is certainly not 
limited to, large, low-mass planets that orbit large, high- 
mass stars, and thus covers a unique parameter space 
not usually covered by other exoplanet characterization 
techniques. 

We have provided analytical formulae and numerical 
models to estimate the amplitude of the photocenter mo- 
tion at various wavelengths. We found that, for some sys- 
tems, the planet can dominate the motion of the system's 
photocenter at wavelengths as short as ~2 /xm, though 
the amplitude of the effect is only ^0.05 /las. If one is 
able to obtain astrometric measurements at wavelengths 
up to 10 /im, then the motion of the photocenter due to 
the planet could be as high as several microarcseconds, 
and can often be of a much larger magnitude than seen 
at optical wavelengths when the photocenter motion is 
due solely to stellar motion. 

We performed numerical modeling of several exoplanet 
systems via the reflux code, and found it to be consis- 
tent with the predictions of our analytical model. The 
numerical modeling revealed that, even at shorter wave- 
lengths where a > 0, the planet has a visible impact 
on the observed astrometric orbit of the system. As 
well, deviations from pure sinusoidal motions due to day- 
night flux differences are clearly visible, and thus multi- 
wavelength astrometry could probe planetary properties 
of albedo and heat redistribution eSiciency. 

One caveat when working to extract the planetary and 
stellar masses from actual observations is that one will 



likely need to either precisely know the luminosity ratio 
of the system, or make assumptions about the luminosity 
of the planet, e.g., it radiates as a blackbody and is in 
thermal equilibrium. It may be possible that other ob- 
servations could yield this information, such as the sec- 
ondary eclipse depth if the planet happens to transit. 
The remaining parameters of the system's distance and 
period should be well determined via other methods such 
as microarcsecond precision parallax and radial-velocity 
or photometric light curves. 

For the prospects of detection, it is clear that this ef- 
fect will probably not be detected in the very near-term. 
Although astrometric measurements are approaching 1 
^as accuracy, they have not yet been performed. Much 
of the ground-based work is being focused on the opti- 
cal and K bands, where in the latter the effect is just 
barely detectable. The development of microarcsecond 
precision astrometric systems in the mid-infrared, or sub- 
microarcsecond precision in the near-infrared, are clearly 
needed, and the methods presented here will serve to 
preselect the best planetary system candidates to be ob- 
served by those systems. 

The work presented in this paper assumed that 
both the star and planet radiate as blackbodies, how- 
ever it is known that both can significantly devi- 
ate from that assumption , esp e cially in the near 
infrared (e.g., 'Gil lon et all 120091: [R ogers et al.' 120091: 
Gibson ct al. 2010: iCroU et all 1201 1:' dc Moo ii et al.l 
2011; Coughlin & Loocz-Moralcs 201 21). At th e extreme 
end, Swain et al. (2010 ) and Waldmann et al.l ()2012l ) re- 
cently found evidence for a very large non-LTE emission 
feature around 3.25 /im in the atmosphere of HD 189733 
10. Although via blackbody approximations we calculate 
that the planet-to -star flux rati o should be 8 .3xl0~^, 
iSwain et al.l (|2010[ ) and Waldma nn et al.l (|20ia) measure 
the 3.25 /im emission feature to be ~8.5xl0~'^ times the 
stellar flux, or about ten times greater than expected. 
Assuming blackbody emission, the expected value for a 
for this system at visible wavelengths is 2.15 /las, and 
at 3.25 /im is 0.83 /ias. If the emission feature is real 
however, the expected value for a at 3.25 /im is a very 
large -11.3 /xas, dominated due to the planetary motion. 
Thus, the key in performing these types of observations 
may be to select particular wavelengths where the plan- 
ets are unusually bright. 

Finally, although we did not assume any limb- 
darkening in our models since we were examining near 
to mid-infrared wavelengths, limb- darkening will be sig- 
niflcant when observed at different optical bandpasses. 
The astrometric signature of transiting planets will vary 
greatly due to limb-darkening in the optical regime, and 
thus multi-wavelength astrometry of transiting planets 
may be used to explore the limb-darkening profiles of 
stars, or visa versa, stellar limb- darkening may need to 
be precisely understood in order to extract planetary and 
stellar parameters of interest. 

J.L.C acknowledges support from a NSF Graduate Re- 
search Fellowship. We thank Dawn Gelino and Thomas 
Harrison for heading up the SIM Science Study which 



■* We note that IMandell et al.l II2011I ) reported a n on-detection 
of a portion of this feature between the pubUcations of lSwain et al.l 



20101 ) and lWaldmann et al.l |[20T1 ). 



started these series of papers. This research has made 
use of the Exoplanet Orbit Database and the Exoplanet 
Data Explorer at exoplanets.org. This research has made 
use of NASA's Astrophysics Data System. 



REFERENCES 



Bean, J. L., McArthur, B. E., Benedict, G. F., Harrison, T. E., 

Bizyaev, D., Nelan, E., & Smith, V. V. 2007, AJ, 134, 749 
Benedict, G. F., et al. 2006, AJ, 132, 2206 
Cacciari, C. 2009, Mem. Soc. Astron. Italiana, 80, 97 
Claret, A. 2000, A&A, 359, 289 
Coughlin, J. L., Harrison, T. E., & Gclino, D. M. 2010a, ApJ, 

723 1351 
Coughlin, J. L., & Lopez-Morales, M. 2012, AJ, 143, 39 
Coughlin, J. L., et al. 2010b, ApJ, 717, 776 
Cowan, N. B., & Agol, E. 2011, ApJ, 729, 54 
CroU, B., Lafreniere, D., Albert, L., Jayawardhana, R., Fortney, 

J. J., & Murray, N. 2011, AJ, 141, 30 
Davidson, J., Edberg, S., Danner, R., Nemati, B., & Unwin, S., 

eds. 2009, SIM Lite: Astrometric Observatory (National 

Aeronautics and Space Administation) 
de Mooij, E. J. W., de Kok, R. J., Ncfs, S. V., & Snellen, I. A. G. 

2011, A&A, 528, A49 
Gaudi, B. S. 2010, in Bulletin of the American Astronomical 

Society, Vol. 42, American Astronomical Society Meeting 

Abstracts 215, 349.05-+ 
Gibson, N. P., et al. 2010, MNRAS, 404, L114 
Gillon, M., et al. 2009, A&A, 506, 359 
Kudryavtseva, N., et al. 2010, in In the Spirit of Lyot 2010 



Lopez-Morales, M., Coughlin, J. L., Sing, D. K., Burrows, A., 

Apai, D., Rogers, J. C, Spiegel, D. S., & Adams, E. R. 2010, 

ApJ, 716, L36 
Maltjet, F., et al. 2011, Experimental Astronomy, 87 
Mandell, A. M., Drake Deming, L., Blake, G. A., Knutson, H. A., 

Mumma, M. J., Villanueva, G. L., & Salyk, C. 2011, ApJ, 728, 

18 
Marley, M. S., Gelino, C, Stephens, D., Lunine, J. I., & 

Freedman, R. 1999, ApJ, 513, 879 
MartioU, E., McArthur, B. E., Benedict, G. F., Bean, J. L., 

Harrison, T. E., & Armstrong, A. 2010, ApJ, 708, 625 
McArthur, B. E., Benedict, G. F., Barnes, R., Martioli, E., 

Korzennik, S., Nelan, E., & Butler, R. P. 2010, ApJ, 715, 1203 
McArthur, B. E., et al. 2004, ApJ, 614, L81 
Muterspaugh, M. W., et al. 2010, AJ, 140, 1579 
Orosz, J. A., & Hauschildt, P. H. 2000, A&A, 364, 265 
Reffert, S., & Quirrenbach, A. 2011, A&A, 527, A140-I- 
Rogers, J. C, Apai, D., Lopez-Morales, M., Sing, D. K., & 

Burrows, A. 2009, ApJ, 707, 1707 
Roll, T., Seifahrt, A., Neuhauser, R., & Kohler, R. 2010, in 

Astronomical Society of the Pacific Conference Series, Vol. 435, 

Binaries - Key to Comprehension of the Universe, ed. A. Prsa 

& M. Zejda, 419—1- 
Rowe, J. F., et al. 2008, ApJ, 689, 1345 
Seager, S., & Mallen-Ornelas, G. 2003, ApJ, 585, 1038 
Seager, S., Whitney, B. A., & Sasselov, D. D. 2000, ApJ, 540, 504 
Southworth, J., Wheatley, P. J., & Sams, G. 2007, MNRAS, 379, 

Lll 
Sudarsky, D., Burrows, A., & Pinto, P. 2000, ApJ, 538, 885 
Swain, M. R., et al. 2010, Nature, 463, 637 
ten Brummelaar, T. A., et al. 2005, ApJ, 628, 453 
Trippe, S., Davies, R., Eisenhauer, F., Schreiber, N. M. F., Fritz, 

T. K., & Genzel, R. 2010, MNRAS, 402, 1126 
van Belle, G. T., et al. 2008, The Messenger, 134, 6 
Waldmann, I. P., Tinetti, G., Drossart, P., Swain, M. R., Dcroo, 

P., & Griffith, C. A. 2012, ApJ, 744, 35 
Wiktorowicz, S. J. 2009, ApJ, 696, 1116