Nicholas Law UNC-Chapel Hill

Christoph Baranec, Dani Atkin-son, Larissa Nofi (IfA Hawaii)

Carl Ziegler (UNC-Chapel Hill)Tim Morton (Princeton)Reed Riddle (Caltech)

How to observe 20,000 targets with laser-guide-star AO:Robo-AO, the First Robotic LGS-AO System

Robo-AO (Baranec et al. 2014 ApJ 790L), the world's first robotic laser guide star adaptive optics system, routinely images 20+ targets per hour completely autonomously, with overheads of only around 60 seconds per target. The system operates in the visible, producing 0.1" FWHM images in the 600-900nm range, covering a large fraction of the Kepler passband. Robo-AO operates autono-mously, with FAA clearance for laser operations without spotters and a Space Command laser-clearance model that allows pointing almost anywhere in the sky without advance notice.

▲ Robo-AO observations up to summer 2015; the colors correspond to different programs. Robo-AO has observed over 19,000 targets in 24 programs ranging from eoxplanet follow-up to young-star binarity surveys.

Data pipeline for the first 20,000-target AO surveyAll Robo-AO data is reduced by our automated data-reduction pipeline, which produces calibrated, stacked, PSF-subtracted images. The hundreds of targets observed each night provide us with detailed PSF characterization for each target.

▲ PSF subtraction for a typical KOI target with a close companion.

High-resolution imaging of every Kepler Planet Candidate System

The Robo-AO Comprehensive KOI SurveyThe Robo-AO Kepler survey has observe every Kepler Object of Interest (KOI) with LGS-AO imaging to search for blended stars, which may be physically associ-ated and/or responsible for transit false positives. Up to now, it has been extremely difficult to obtain adaptive optics images of the thousands of candidates generated by large surveys like Kepler because of the faintness of the targets and the exces-sive observing time required. Robo-AO’s sub-minute overheads and laser-guide-star AO system allow us to observe hundreds of KOIs per night at 0.15” resolution.

Robotic Laser Adaptive Optics Imaging of 715 Kepler Exoplanet Candidates! ! ! ! ! ! Law, Morton, Baranec, Riddle, et al. 2014 ApJ 791 35L

Companion rates vary with planetary population

Robo-AOKepler

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

300 400 500 600 700 800 900 1000

Rob

o-AO

QE

Kepl

er e

ffici

ency

Wavelength / nm

1.5 arcsec

Our first observing season (Law et al. 2014) searched for stars close to 715 Kepler planet candidate hosts. We found 53 companions, 43 of which are new discoveries. Our survey is sensitive to objects from 0.15” to 4” separation, with magnitude differ-ences up to ~6 magnitudes. We measured an overall nearby-star-probability for Kep-ler planet candidates of 7.4 +/- 1.0%, and calculated the effects of each detected nearby star on the Kepler-measured planetary radius. In the paper we discussed several KOIs of particular interest, including KOI-191 and KOI-1151, which are both multi-planet systems with detected stellar companions whose unusual planetary sys-tem architecture might be best explained if they are “coincident multiple” systems, with several transiting planets shared between the two stars. Finally, we found 98%-confidence evidence that short-period giant planets are 2-3X more likely than longer period planets to be found in wide stellar binaries.

▲ Binarity broken down by planetary population; there is tentative evidence that close-in giant planets are more likely to have nearby stars.

▲ Companion discoveries from our first observing sea-son, and the high-performance-target contrast curve for the Robo-AO observations.

▲ 1/5 of the KOIs we have observed (2” cutouts). Bright binaries are visible in this grid; fainter compan-ions pop out after PSF subtraction.

▲ The Robo-AO KOI passband, covering a large fraction of the Kepler passband itself, allowing easy and accurate assessment of the companion’s ef-fects on the Kepler light curves.

▲ Observed first-season targets as a fraction of the Q1-Q6 KOI sample (Batalha et al. 2012). We pri-oritized the brightest and the coolest KOIs in our first sample; our subsequent surveys have observed almost all of the remaining KOIs.

Exoplanet research program grant NNX15AC91G

12 N.M. Law et al.

�������������������������������

� �� ���

�� ����������

��������������

��� ���� ���� ������ ���

Fig. 10.— 1� uncertainty regions for binarity fraction as a func-tion of KOI period for two di↵erent planetary populations (we split“small” from “giant” at Neptune’s radius (3.9 R�), but the exactvalue of the split does not significantly a↵ect the uncertainty re-gion shape). The gas giants cut o↵ for shorter periods because ofinsu�cient targets for acceptable statistics.

It is expected that multiple-planet systems detectedby Kepler are less likely to be false positives thansingle-planet systems because there are far fewer false-positive scenarios which can lead to multiple-period false-positives. In Figure 9 we show the stellar multiplicityrates for single and multiple planet detections. There isa di↵erence in stellar multiplicity between the single andmultiple planet detections, but a Fisher exact test showsa 13% probability of this being a chance di↵erence due tosmall-number statistics. At least in the current datasetwe cannot distinguish stellar multiplicity between singleand multiple planet systems.

5.3.3. Stellar multiplicity and close-in planets

Stellar binarity has been hypothesized to be impor-tant in shaping the architectures of planetary systems,both by regulating planet formation and by dynami-cally sculpting planets final orbits, such as forcing Kozaioscillations that cause planet migration (Fabrycky &Tremaine 2007; Katz et al. 2011; Naoz et al. 2012) or bytilting the circumstellar disk (Batygin 2012). If planetarymigration is induced by a third body, one would expectto find a correlation between the presence of a detectedthird body and the presence of short-period planets.Figure 10 shows the fraction of Kepler planet candi-

dates with nearby stars as a function of the period of theclosest-in planet, grouping the planets into two di↵erentsize ranges. From these raw binarity fractions, wherewe have not accounted for the probability of physicalassociation, it appears that while small planets do notshow a significant change in third-body probability withthe orbital period of the Kepler candidate, giant planetsshow a significant increase at periods less than ⇠15 days.Binning all our targets into only four population groupsallows us to search for smaller changes in the binaritystatistics (Figure 11). We arbitrarily split “small” plan-ets from “giant” planets at Neptune’s radius (3.9 R�),but the exact value of the split does not significantly af-fect the results; only two of the detected systems haveplanetary radii within 20% of the cuto↵ value. We seethat small planets at short periods share the same bi-narity fraction as all sizes of planets with >15d periods(within statistical errors). However, the short period gi-ant planets again show a significantly increased binarity

Fig. 11.— Fraction of KOIs with nearby stars for four di↵er-ent planetary populations. Giant here is shorthand for a radiusequal to or larger than that of Neptune. We assign KOIs to thesepopulations if any planet in the system meets the requirements; asmall number of multiple-planet systems are therefore assigned tomultiple populations.

fraction. A Fisher exact test rejects the hypothesis thatthe two planetary populations have the same binarityfraction, at the 95% level.We can attempt to remove the background asterisms

by selecting on the basis of magnitude ratio, as faintbackground stars are more likely to be chance alignmentsthan roughly-equal-brightness companions. Our surveydisplayed an excess of close-separation bright compan-ions: there are 13 companions with �m < 2 with sep-arations <1.500, and only one at larger radii (Figure 6),while the numbers of fainter companions do not showsuch a bias. We suggest that this excess reveals a bright-companion population which is more likely to be physi-cally associated than an average companion in the survey.Selecting the companions with �m < 2 and separa-

tion <1.005 leads an increased di↵erence in stellar multi-plicity between the planetary populations (Figure 12),increasing the significance to 98%. This approach doesnot fully account for the probability of each companionbeing physically associated, and so its results should beinterpreted with caution. For example, close-in compan-ions are less likely to be rejected by the Kepler centroid-based false-positive tests, but it is not obvious why thisrejection would be di↵erent for planetary systems withshort-period (<15d) and longer-period KOIs (with a me-dian period of 54d for the KOIs we surveyed). In fact,the shorter-period systems have more eclipse events inthe Kepler dataset and it should therefore be easier todetect a small centroid shift from close-in companions.On the basis of our current analysis, we suggest that

the di↵erence of multiplicity rates between the planetarypopulations may be tentative evidence for third bodies instellar systems producing an excess of close-in giant plan-ets. We expect the full Robo-AO surveys to be able toevaluate this possibility at more than the 3� confidencelevel.

6. CONCLUSIONS

We observed 715 Kepler planetary system candidateswith the Robo-AO robotic laser adaptive optics system.Our detection of 53 planetary candidates with nearbystars from 715 targets implies an overall nearby-starprobability of 7.4%±1.0% at separations between 0.001and 2.005 and �m

⇠

< 6. We have detailed the e↵ects of the

▲ 1σ uncertainty regions for binarity fraction as a function of KOI period for two different planetary populations (we split “small” from “giant” at Neptune’s radius (3.9 R⊕), but the exact value of the split does not significantly affect the uncertainty region shape). The gas giants cut off for shorter periods because of insufficient targets for acceptable statistics in our first results; later papers will push to smaller radii.

Robo-AO Imaging of 715 Kepler Exoplanet Candidates 13

����

����

����

����

����

������� �

����������

������� �

����������

�� ����������

Fig. 12.— Fraction of KOIs with nearby stars for four di↵erentplanetary populations – as figure 11 with only companions with�m < 2 and separations <1.005, removing faint nearby stars whichare less likely to be physically associated (we did not detect anybright companions around the 84 longer-period giant planet KOIsin our survey, so we only show an upper limit). There is a 98%-confidence detection of a di↵erence in stellar multiplicity rates forclose-in giant planets compared to further-out giants.

detected nearby stars on the interpretation of the Keplerplanetary candidates, including the detection of prob-able ”co-incident” multiples (KOI-191 and KOI-1151),multiple-planet systems likely containing false positives(KOI-1845), and the confirmation of five KOIs as roughlyEarth-radius planets in multiple stellar systems (KOI-1613, KOI-1619, KOI-2059, KOI-2463, and KOI 2657).We have also found tentative, 98%-confidence, evidencefor stellar third bodies leading to a 2-3⇥ increased rateof close-in giant planets.We expect the ongoing Robo-AO surveys to complete

observations of every Kepler planet candidate by theend of 2014. The increased survey numbers will allowus to search for stellar multiplicity correlations only inmultiple-detected-planet systems, which are expected tohave a much lower false-positive probability, and thus willimprove our ability to disentangle false-positives from as-trophysical e↵ects. The number of multiple systems in

our current sample is not large enough to verify our ten-tative conclusions on the e↵ects of stellar multiplicity onshort-period giant planets (in particular, we have onlycovered one multiple-planet system with a short-periodgiant planet), but we plan to investigate these possi-bilities in future data releases. We are also continuingobservations of our detected companions to search forcommon-proper-motion pairs. The completed Robo-AOsurvey will also allow us to confirm many more Keplerplanet candidates and likely find more exotic planetarysystems.

ACKNOWLEDGEMENTS

We thank the anonymous referee for careful analysisand useful comments on the manuscript. The Robo-AO system is supported by collaborating partner insti-tutions, the California Institute of Technology and theInter-University Centre for Astronomy and Astrophysics,and by the National Science Foundation under GrantNos. AST-0906060 and AST-0960343, by the MountCuba Astronomical Foundation, by a gift from SamuelOschin. We are grateful to the Palomar Observatorysta↵ for their ongoing support of Robo-AO on the 60-inch telescope, particularly S. Kunsman, M. Doyle, J.Henning, R. Walters, G. Van Idsinga, B. Baker, K. Dun-scombe and D. Roderick. We recognize and acknowledgethe very significant cultural role and reverence that thesummit of Mauna Kea has always had within the indige-nous Hawaiian community. We are most fortunate tohave the opportunity to conduct observations from thismountain. C.B and J.A.J. acknowledge support from theAlfred P. Sloan Foundation. J.A.J acknowledges supportfrom the David and Lucile Packard Foundation.Facilities: PO:1.5m (Robo-AO), Keck:II (NIRC2-

REFERENCES

Adams, E. R., Ciardi, D. R., Dupree, A. K., Gautier, III, T. N.,Kulesa, C., & McCarthy, D. 2012, AJ, 144, 42

Adams, E. R., Dupree, A. K., Kulesa, C., & McCarthy, D. 2013,AJ, 146, 9

Baranec, C., Riddle, R., Law, N. M., Ramaprakash, A.,Tendulkar, S. P., Bui, K., Burse, M. P., Chordia, P., Das,H. K., Davis, J. T., Dekany, R. G., Kasliwal, M. M., Kulkarni,S. R., Morton, T. D., Ofek, E. O., & Punnadi, S. 2013, Journalof Visualized Experiments, 72, e50021

Baranec, C., Riddle, R., Ramaprakash, A. N., Law, N., Tendulkar,S., Kulkarni, S., Dekany, R., Bui, K., Davis, J., Burse, M., Das,H., Hildebrandt, S., Punnadi, S., & Smith, R. 2012, Proc. SPIE8447, Adaptive Optics Systems III, 8447, 844704

Barclay, T., Rowe, J. F., Lissauer, J. J., Huber, D., Fressin, F.,Howell, S. B., Bryson, S. T., Chaplin, W. J., Desert, J.-M.,Lopez, E. D., Marcy, G. W., Mullally, F., Ragozzine, D.,Torres, G., Adams, E. R., Agol, E., Barrado, D., Basu, S.,Bedding, T. R., Buchhave, L. A., Charbonneau, D.,Christiansen, J. L., Christensen-Dalsgaard, J., Ciardi, D.,Cochran, W. D., Dupree, A. K., Elsworth, Y., Everett, M.,Fischer, D. A., Ford, E. B., Fortney, J. J., Geary, J. C., Haas,M. R., Handberg, R., Hekker, S., Henze, C. E., Horch, E.,Howard, A. W., Hunter, R. C., Isaacson, H., Jenkins, J. M.,Karo↵, C., Kawaler, S. D., Kjeldsen, H., Klaus, T. C., Latham,D. W., Li, J., Lillo-Box, J., Lund, M. N., Lundkvist, M.,Metcalfe, T. S., Miglio, A., Morris, R. L., Quintana, E. V.,Stello, D., Smith, J. C., Still, M., & Thompson, S. E. 2013,Nature, 494, 452

Barrado, D., Lillo-Box, J., Bouy, H., Aceituno, J., & Sanchez, S.2013, in European Physical Journal Web of Conferences,Vol. 47, European Physical Journal Web of Conferences, 5008

8 N.M. Law et al.

���

���

���

���

��� ��� ��� ��� ��� ���

��� ���� ������ ��

��� � ����� �����Fig. 6.— The separations and magnitude di↵erences of the

detected companions compared to the survey’s typical high-performance 5� contrast curve (one very faint companion was de-tected around a bright KOI in exceptional conditions). The distri-bution of companion properties has no evidence for unaccountedincompleteness e↵ects, although there is an excess of bright com-panions at close separations, suggesting that those companions aremore likely to be physically associated.

Fig. 7.— Keck-AO NIRC2 J-band images confirming two Robo-AO companion detections.

for which they discovered companions within a 2.005 ra-dius are also in our survey. Both surveys detect KOI-401at a separation of 2.000 and at a contrast of 2.6 magnitudes(L12 i-band) or 2.9 magnitudes (Robo-AO LP600). Thecompanions to KOI-628 were visible in our survey butat contrasts that placed them in the “likely detections”group. L12 detected a companion to KOI-658 at 1.009radius and a contrast of 4.6 magnitudes in i-band. Atthat radius, for the performance achieved on KOI-658,the Robo-AO snapshot-survey limiting magnitude ratiois ⇠4.0 magnitudes and so we do not re-detect that com-panion. For the same reason we also do not re-detect thecompanions to KOI-703 (6.4 magnitudes contrast), KOI-704 (5.0 magnitudes contrast) and KOI-721 (3.9 mag-nitudes contrast). The 0.0013-radius companion to KOI-1537 detected in Adams et al. (2013) is at too close aseparation to be detectable in our survey. The L12 com-panion to KOI-1375 is visible in our dataset, but has acontrast ratio of 4.0 magnitudes, under our formal de-tection limit and well below the 2.75 magnitude i-bandcontrast measured by L12. The target is not stronglycoloured according to L12 and it is not obvious why thecompanion is so much fainter in our survey.

5. DISCUSSION

5.1. Implications for Kepler Planet Candidates

The detection of a previously unknown star within thephotometric aperture of a KOI host star will a↵ect the

derived radius of any planet candidate around that hoststar, because the Kepler observed transit depth is shal-lower than the true depth due to dilution. The degreeof this e↵ect depends upon the relative brightness of thetarget and secondary star, and which star is actually be-ing transited. In particular, if there is more than one starin the photometric aperture and the transiting object isaround a star that contributes a fraction Fi to the totallight in the aperture, then

�true = �obs

✓1

Fi

◆, (1)

where �true is the true intrinsic fractional transit depthand �obs is the observed, diluted depth. Since � /

(Rp/R?)2, the true planet radius in the case where thetransit is around star i is

Rp,i = R?,i

✓Rp

R?

◆

0

r1

Fi, (2)

where R?,i is the radius of star i, and the 0 subscriptrepresents the radius ratio implied by the diluted transit,or what would be inferred by ignoring the presence of anyblending flux.Thus, for each planet candidate in KOI systems ob-

served to have close stellar companions, the derivedplanet radius must be corrected—and there are two po-tential scenarios for each candidate: the eclipsed star iseither star A (the brighter target star) or star B (thefainter companion).In case A, the corrected planet radius is

Rp,A = Rp,0

r1

FA, (3)

and in case B,

Rp,B = Rp,0RB

RA

r1

FB. (4)

Case A is straightforward, with nothing needed exceptthe observed contrast ratio (in order to calculate FA).It should be noted, however, that this assumes that theestimated host stellar radius RA is unchanged by thedetection of the companion star. As the radii for mostKepler stars are inferred photometrically, this may notbe strictly true, as light from the companion might causethe primary stellar type to be misidentified. We do notattempt to quantify the extent of this e↵ect in this paper.We do, however, note that it is likely to be negligible forlarger contrast ratios where the colors of the blendedsystem are dominated by light from the primary.Case B, in addition to needing FB , needs also the ratio

RB/RA. If the observed companion is an unassociatedbackground star, then the single-band Robo-AO obser-vation does not constrain RB . However, under the as-sumption that the companion is physically bound, thenwe can estimate its size and spectral type, given assumedknowledge about the primary star A.In order to accomplish this, we use the Dartmouth stel-

lar models (Dotter et al. 2008) and the measured pri-mary KOI star properties listed in the NASA ExoplanetArchive. For the mass and age of the primary, we use theDartmouth isochrones to find an absolute magnitude inthe observed band (approximating the LP600 bandpass

�������������� �� �� ����

�������������� ����� ��

���������� ���

������������������� ��

��� !��� ������"#

$ ������%������ �����&������'��� �����

� �������� � ���� �������

�������(���������

����� �))��

$ ���* ����+�������,�������!��������-�$��� ���)

������.�-�$� � ��� !+������/�������

(��������

������������� ����������� ������(��)�� �0���������

1�����)����!���������-�$�����!���,����0������

� !� ����!���,��-�$

� ������ ����-�$�� !� ���������2&���3 �������������-�$

4 �����5������ ���6��6������

$��������,�������������5������������!����������

�� � ��0����� +4� �����,����� ��� +

� �����3���,���� ���� ��� +�����0����� +

1 �� ��� ������ ����2&�� ���

0�� �

�0���������� �����������2&�,�����0� ��

7 ��+����������������!+�+

8 ���0� �

�������������

��� �� ��������

�4���1�����

��� ��� �0� ������

������������� ���� �

������������� �����

�����2�����(���� �� � ����

����%��

�6 ������9.�

� ������ ���� � ����� 2&���3 �������������-�$

$ ������������� ��

- �� ����� ��(�:��

����! ����

������������� �������� ����

������ ���������������������� �����

��!�� ��������������"��# � ������ ����

;�&�.<���&=�&�.<���&

$��% �� ���#��&�

�����'��� ����(��� ���

����)�����9��������������9.��������4���1��� �����������!����� ��������

*� ���������5� ��������>��������������������?������������ � �����2��0���������������@�����&�����������������

��������� �������.��� ���� �����

!�� ��������������@����� ���� �����