Imaging search for the unseen companion to Eps Ind A -- Improving the detection limits with 4 micron observations
M. Janson, D. Apai, M. Zechmeister, W. Brandner, M. Kuerster, M. Kasper, S. Reffert, M. Endl, D. Lafreniere, K. Geissler, S. Hippler, Th. Henning
aa r X i v : . [ a s t r o - ph . E P ] J un Mon. Not. R. Astron. Soc. , 1–8 (2009) Printed 17 September 2018 (MN L A TEX style file v2.2)
Imaging search for the unseen companion to ǫ Ind A –Improving the detection limits with 4 µ m observations ⋆ M. Janson † , D. Apai , M. Zechmeister , W. Brandner , M. K¨urster ,M. Kasper , S. Reffert , M. Endl , D. Lafreni`ere , K. Geißler ,S. Hippler , Th. Henning Department of Astronomy, University of Toronto, 50 St George St, Toronto, M5S 3H4, Canada Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA Max Planck Institute for Astronomy, K¨onigstuhl 17, Heidelberg, D-69117, Germany ESO, Karl-Schwarzschild-Strasse 2, Garching bei M¨unchen, D-85748, Germany Landessternwarte, K¨onigstuhl 12, Heidelberg, D-69117, Germany McDonald Observatory, 1 University Station, Austin, TX 78712, USA
N/A
ABSTRACT ǫ Ind A is one of the nearest sun-like stars, located only 3.6 pc away. It is known tohost a binary brown dwarf companion, ǫ Ind Ba/Bb, at a large projected separationof 6.7 ′ , but radial velocity measurements imply that an additional, yet unseen com-ponent is present in the system, much closer to ǫ Ind A. Previous direct imaging hasexcluded the presence of any stellar or high-mass brown dwarf companion at smallseparations, indicating that the unseen companion may be a low-mass brown dwarf orhigh-mass planet. We present the results of a deep high-contrast imaging search for thecompanion, using active angular differential imaging (aADI) at 4 µ m, a particularlypowerful technique for planet searches around nearby and relatively old stars. We alsodevelop an additional PSF reference subtraction scheme based on locally optimizedcombination of images (LOCI) to further enhance the detection limits. No companionis seen in the images, although we are sensitive to significantly lower masses than pre-viously achieved. Combining the imaging data with the known radial velocity trend,we constrain the properties of the companion to within approximately 5-20 M jup , 10-20 AU, and i > o , unless it is an exotic stellar remnant. The results also imply thatthe system is probably older than the frequently assumed age of ∼ Key words: stars: low-mass, brown dwarfs – planetary systems
Direct imaging of exoplanets is a field of research in rapiddevelopment. The past year has seen a number of interest-ing planet candidates imaged directly around stars. Mostnotable is arguably HR 8799, showing three planetary com-panions (see Marois et al. 2008) so far. These planets havebeen shown to exhibit Keplerian motion around the star,and have estimated masses in the range of 7-10 M jup fromtheoretical models (Baraffe et a. 2003). The system is knownto also host a debris disk (e.g. Moor et al. 2006; Rhee et al.2007). Along with the fact that there are three almost equal- ⋆ Based on observations collected at the European Southern Ob-servatory, Chile (ESO No. 082.D-0251 and 081.C-0430). † Reinhardt fellow. E-mail: [email protected] mass companions, all orbiting the star in what appears tobe a co-planar manner, this implies that the objects mostlikely formed in a circumstellar disk, by a process distinctfrom the star or brown dwarf formation process. This canbe contrasted with the case of the 2M1207 system (Chauvinet al. 2005), where the low primary-to-secondary mass ratiois more reminiscent of a brown dwarf binary system than astar-planet system. Two other intriguing planet candidatesin systems with debris disks were reported around the sametime as HR 8799 b/c/d: Fomalhaut b (Kalas et al. 2008)and Beta Pic b (Lagrange et al. 2009), though additionalfollow-up observations would be desirable to provide moreinformation on these systems. ǫ Ind A is a K4V-type southern sky star located at adistance of 3.6 pc, with a very high proper motion of 4.7 ′′ yr − (Perryman et al. 1997), see Table 1. Comoving on the c (cid:13) M. Janson et al.
Table 1.
Summary of ǫ Ind A properties.Property ValueRight Ascension 22 03 21.66Declination -56 47 09.5Spectral Type K4.5VDistance 3 . ± .
009 pcProper Motion 4705 mas yr − Mass 0.7 M sun Age 1-5 Gyr sky at the same rate, at a separation of 6.7 ′ , is ǫ Ind B,which was first detected by Scholz et al. (2003), and shortlythereafter resolved into the brown dwarf binary ǫ Ind Ba/Bb(McCaughrean et al. 2004) as it is known today. Being themost nearby binary brown dwarf, and with a physical sepa-ration small enough to determine its dynamical mass withina reasonable timeframe, ǫ Ind Ba/Bb will be a benchmarkobject for the physical understanding of brown dwarfs. How-ever, in addition to the Ba/Bb components, the ǫ Ind sys-tem may provide an additional possibility to study an evenlower-mass and cooler object. ǫ Ind A displays a linear ra-dial velocity trend (Endl et al. 2002; Zechmeister et al., inprep.) which, unless due to some exotic stellar remnant, isindicative of a giant planet or very low-mass brown dwarfcompanion. If this object could be directly imaged, it wouldconstitute yet another important benchmark object, givenits probable low mass, low temperature, and the possibilityto estimate both its luminosity as well as its dynamical masswithin a reasonable timeframe. It would also be the closestplanet or very low-mass brown dwarf companion directlydetected outside of our Solar System.Here, we will present our deep imaging campaign of ǫ Ind, using narrow-band 4 µ m imaging with two differenthigh-contrast techniques. One technique is pure active angu-lar differential imaging (aADI) as already implemented anddemonstrated as a powerful technique for detecting closecompanions to bright nearby stars (Janson et al. 2008). InaADI (also known as roll subtraction), images are takenat two different instrument rotator angles, and one is sub-tracted from the other, thereby removing the bulk of thestellar PSF including static instrumental speckles. The othertechnique is a combination of PSF reference subtraction andaADI (PSFR+aADI), using the LOCI (locally optimizedcombination of images) algorithm developed by Lafreniereet al. (2007). We also compare the performance of aADI at4 µ m with the same technique in the L’-band. The conceptof using aADI with L’ was proven by Kasper et al. (2007;the usefulness of L’ for high-contrast imaging purposes wasalso independently demonstrated by Hinz et al. 2006). Theconcept of using NB4.05 to enhance the physical contrastwas introduced in Janson et al. (2008).The outline is as follows: In Sect. 2, we summarize theobservational parameters and ambient conditions of the ob-serving runs. The two paths of data reduction employed aredescribed in Sect. 3. This is followed by a presentation ofthe results and the associated analysis in Sect. 4, includinga comparison between filters (Sect. 4.1), a comparison be-tween techniques (Sect. 4.2), a discussion of what we can Table 2.
Observational conditions for the four runs.A1 A2 B1 B2Date (2008) 31 Oct 2 Nov 3 Jul 3 JulFilter NB4.05 NB4.05 NB4.05 L’Seeing a ′′ ′′ ′′ ′′ Strehl b
83% 79% 85% 83%Humidity 8% 13% 3% 3%Coh. time a c
30 33(per angle)DIT 1.0 s 1.0 s 0.2 s 0.2 sNDIT 61 61 150 150Tot. time 1159 s 1159 s c
900 s 990 s(per angle) a Values given by the atmospheric seeing monitor at a wavelengthof 500 nm. b Strehl ratio given by the AO system, rescaled to the observingwavelengths. c Two frames were de-selected for 0 o , hence the effective time is1037 s for that case. learn from the dynamical input in Sect. 4.3, and the finaldetection limits and their interpretation in Sect. 4.4. Finally,we conclude in Sect. 5. The data presented here are based on two different sets ofVLT/NACO observations of ǫ Ind. One set of observationsconsisted of deep imaging with aADI in the NB4.05 filter,executed in service mode and split into two equal observ-ing blocks (henceforth observations A1 and A2), on 31 Oct2008 and 2 Nov 2008. The other set of observations weretaken in visitor mode on 3 Jul 2008, as part of a larger sur-vey searching for planets around a volume-limited sample ofnearby stars (Apai et al., in prep.). Those observations wereless deep, but consisted of aADI imaging in both the NB4.05and broad-band L’ filters (henceforth observations B1 andB2, respectively). For observations A1 and A2, the samestrategy was used as in previous observations (see Janson etal. 2008): Jittering was applied to enable a good subtractionof the thermal background, and the aADI was performed attwo different instrument rotator angles, using a differentialangle of 33 degrees. For B1 and B2, a four-point large-throwdithering scheme was applied for the background subtrac-tion purposes, and the differential angle used for aADI was20 degrees. All observations were taken with the L27 objec-tive, providing a pixel scale of 27 mas/pixel, and a field ofview of 28 ′′ by 28”. The weather conditions and observa-tional parameters of each run are listed in Table 2. The data reduction for A1 and A2 was done differently withrespect to the two different techniques applied, hence sep-arate descriptions are provided below. The reduction of B1and B2 was only done with aADI. c (cid:13)000
900 s 990 s(per angle) a Values given by the atmospheric seeing monitor at a wavelengthof 500 nm. b Strehl ratio given by the AO system, rescaled to the observingwavelengths. c Two frames were de-selected for 0 o , hence the effective time is1037 s for that case. learn from the dynamical input in Sect. 4.3, and the finaldetection limits and their interpretation in Sect. 4.4. Finally,we conclude in Sect. 5. The data presented here are based on two different sets ofVLT/NACO observations of ǫ Ind. One set of observationsconsisted of deep imaging with aADI in the NB4.05 filter,executed in service mode and split into two equal observ-ing blocks (henceforth observations A1 and A2), on 31 Oct2008 and 2 Nov 2008. The other set of observations weretaken in visitor mode on 3 Jul 2008, as part of a larger sur-vey searching for planets around a volume-limited sample ofnearby stars (Apai et al., in prep.). Those observations wereless deep, but consisted of aADI imaging in both the NB4.05and broad-band L’ filters (henceforth observations B1 andB2, respectively). For observations A1 and A2, the samestrategy was used as in previous observations (see Janson etal. 2008): Jittering was applied to enable a good subtractionof the thermal background, and the aADI was performed attwo different instrument rotator angles, using a differentialangle of 33 degrees. For B1 and B2, a four-point large-throwdithering scheme was applied for the background subtrac-tion purposes, and the differential angle used for aADI was20 degrees. All observations were taken with the L27 objec-tive, providing a pixel scale of 27 mas/pixel, and a field ofview of 28 ′′ by 28”. The weather conditions and observa-tional parameters of each run are listed in Table 2. The data reduction for A1 and A2 was done differently withrespect to the two different techniques applied, hence sep-arate descriptions are provided below. The reduction of B1and B2 was only done with aADI. c (cid:13)000 , 1–8 maging search for the companion to ǫ Ind A Since observations A1 and A2 were taken in the same wayas the observations of Janson et al. (2008), largely the samedata reduction could be applied as in the case for aADIpurposes: The most basic reduction steps (e.g. flat field-ing, bad pixel removal, background subtraction) were pro-vided by the ESO automatic pipeline. The subsequent stepswere performed with our dedicated IDL pipeline: The im-ages were shifted using bilinear interpolation to a commoncenter determined through cross-correlation, and to an ab-solute center using center of gravity. Low-frequency filteringwas applied by subtracting a smoothed counterpart of eachimage produced by convolution with a Gaussian kernel with0.5 ′′ FWHM (the FWHM of the stellar PSF was about 120mas). All the images corresponding to different rotation an-gles were subtracted from each other, and the results fromthe two nights were coadded. As an alternative analysis, allimages were also de-rotated back to a common angle andcoadded. This procedure yields a single co-added signatureof any companion, which provides a useful alternative wayto look at the data in the background-limited regime, withrespect to the aADI-subtracted data which instead producestwo independent signatures of half the amplitude each.For observations B1 and B2, the data reduction wasperformed with the IDL routines as described above, withthe exception that the centering was determined based on asub-frame of 200x200 pixels around the star instead of thewhole frame. This was done in order to avoid influence fromresidual features from the sky subtraction and the differentdithering scheme.The residuals as function of separation from the starwere calculated from the standard deviation of all pixelsin an annulus corresponding to each separation step. Thephysical brightness contrast was derived from the 3 σ resid-uals by division of the peak value of the primary. Since theprimary was saturated in the science images (the saturationradius was about 5-6 pixels), this had to be calibrated. Forruns A1 and A2, this was done by introducing a neutraldensity filter into the optical path during acquisition, thusgetting non-saturated images of the primary, allowing to de-termine a renormalized peak value. For run B1, the samenon-saturated images as for A1 and A2 were used, whichcould be done since the Strehl ratio was stable and almostequal between the epochs. For B2, an image was taken of afainter photometric standard star during the night. In applications involving LOCI, it is preferable to maximizethe number of PSF representations of a target or referencestar, hence for this case, reduction was done on individualframes, with the combination of frames only performed atthe very end. The PSF star used was ǫ Eri, which is prac-tically ideal for the purpose given the similar spectral type,brightness, and the fact that observations exist taken underalmost identical circumstances. All individual target and ref-erence star frames were manually subjected to flat fielding,dark subtraction and bad pixel correction using calibrationframes provided by ESO. Low-frequency filtering was ap-plied as described above. A master sky frame was then pro-duced by taking the median of the frames, where the stellar image is randomly placed in each frame, thereby removingthe star altogether. The individual frames revealed ring-likestructures in the background that could be reproduced inthe master sky frame. By subtracting the master sky framefrom each individual frame, the pattern could be removed.The pattern was found to be constant during the extent ofan observation, but variable between observations (e.g., theframes corresponding to ǫ Ind and ǫ Eri were different fromeach other), and is probably related to instrumental dustemitting at 4 µ m. The full background subtraction obtainedin this way was found to be equally good as that delivered bythe ESO ’jitter’ routine. It is interesting to note, that giventhe fact that the pattern is constant during an observation,it should be possible to calibrate it out of a generic obser-vation by making a master sky observation directly beforeor after the target observation. Hence, for any observationdedicated to the detection of point sources, it should be pos-sible to achieve the same degree of background subtractionwith and without jittering. This is an important realizationwith respect to high-contrast imaging at these wavelengthsusing techniques such as passive ADI (pADI) or coronagra-phy, where it is desirable to maintain the stellar primary ata fixed position on the detector, and simultaneously achievethe best possible background subtraction. In summary, thereappears to be no conflict between these two requirements, aslong as the master sky calibration step is performed duringobservations.PSFR and aADI were performed separately, in se-quence. For PSFR, every target and reference image wasde-rotated such that the spider patterns were aligned. Foreach target frame, an optimized PSF reference frame wasthen produced from the full set of reference frames usingthe LOCI (Lafreniere et al. 2007) algorithm and subtractedfrom the target frame. The optimization was performed in10 regions, five for the image range contaminated by thefour spiders, covering different radial sections of the PSF,and five for the image range not contaminated by spiders,also covering different radial sections. The spider optimiza-tion areas were rectangular with a fixed width of 25 pixels,inner radii of 10, 20, 40, 70, and 120 pixels, and outer radiiof 60, 70, 90, 150, and 200 pixels. The remaining areas wereannuli excluding the spider regions, between inner radii of20, 30 40, 50, and 60 pixels and outer radii of 50, 60, 70, 80,and 100 pixels. The subtractions were performed sequen-tially outwards with the subtraction zone defined from theinner radius of the optimization zone and outwards. Follow-ing this procedure, each of the target frames were re-rotatedto their true parallactic angle. The aADI step was then per-formed through a second LOCI PSF construction, using all33 o frames as PSF library for each 0 o frame, and vice versa.For this case, the optimization regions were simply five an-nuli between inner radii of 10, 20, 30, 50, and 70 pixels andouter radii of 40, 50, 60, 70 and 100 pixels. The optimiza-tion regions were chosen to provide a good balance betweenthe two main criteria of LOCI: to maximize the efficiency ofstellar PSF structure subtraction, and minimize subtractionof actual companions. The latter was tested by generating aseries of runs where false companions had been introducedin the target frame – in total 3600 companions distributedbetween 10 and 100 pixels separation from the center of thestar, and over all azimuthal angles. The partial subtractionsin each case were used to construct a radial profile of con- c (cid:13) , 1–8 M. Janson et al. served companion flux fractions. As expected, a significantflux loss occurs at 10 pixels, but decreases rapidly outwards.At 100 pixels, the fraction of restored companion flux ap-proaches unity, as indeed expected, given that the LOCIoptimization is not applied beyond 100 pixels for the vastmajority of the image space.Finally, all frames corresponding to each rotator anglewere combined using 3 σ -clipping. The radial profile of resid-uals was created in the same way as for aADI, but with theadditional step that it was normalized by the radial profileof conserved companion flux fraction to provide an accuratemeasure of the actual achieved contrast. The output images from runs A1+A2 from each of the tworeduction paths are shown in Fig. 1. No companion candi-dates were detected in the images. In the following, we dis-cuss the implications of this result, and compare the meth-ods used.
Although the B1 and B2 images are less deep than A1+A2,the fact that they were obtained for the same target at aboutthe same time, and with an almost identical observationalsetup, makes them ideal for comparing L’ and NB4.05 imag-ing for planet detection purposes around bright stars. Acomparison was already made in Janson et al. (2008) be-tween L’ aADI, NB4.05 aADI, and SDI+aADI (from Jan-son et al. 2007). While a fully relevant comparison could bemade between SDI+aADI and NB4.05 aADI, where NB4.05aADI was found to perform better under all circumstances,the comparison with L’ was preliminary, since no compa-rable data was available. Instead, the comparison betweenL’ and NB4.05 was based entirely on physical contrast givenby the theoretical models, and the instrumental contrast wasassumed to be the same. While this is relevant for a largepart of the parameter space, there will in reality be differ-ences in instrumental contrast due to differences in Strehlratio, PSF diffraction, and thermal background between thefilters. Using the B1 and B2 observations, we can now pro-vide a comparison that takes all these issues into account.The comparison was done by translating the bright-ness contrasts into mass detection limits using the spectraland photometric evolutionary models of Baraffe et al. (2003)and Burrows et al. (2003) for various ages. The method isdescribed in detail in Janson et al. (2008). Note that thecomparison is done for almost identical observing time, andwith virtually no difference in overheads, i.e. the telescopetime investment is also the same in both cases. As expected,the instrumental contrast is almost identical in the contrast-linited range, confirming the assumptions of the previousanalysis, and thus the difference in the inner range is almostentirely set by the expected flux distribution of the compan-ion. We show an example that demonstrates the favourablespectral range of NB4.05 in Fig. 2, for 10 M jup and 15 M jup objects, at an age of 1 Gyr. The flux density is higher inNB4.05 than in both L’-band and M-band. For cooler ob-jects, the bulk of the flux moves redward, hence M-bandbecomes better in terms of flux density, but the thermal Figure 2.
Example of two model spectra from Burrows et al.(2003), and the corresponding flux densities in filters L’, NB4.05,and M. Upper lines: A 15 M jup object. Lower lines: A 10 M jup object. The age is 1 Gyr in both cases. background is also much worse in M-band. The improve-ment of NB4.05 over L’ increases further for cooler objects.We show the results of the detection limit comparisonfor 1 Gyr, 3 Gyr, and 5 Gyr in Fig. 3. It is seen that for allthese ages, NB4.05 performs better in the contrast-limitedinner part, and L’ performs better in the outer background-limited part, as expected. The crossover point for ǫ Ind Ain our dataset is at about 4 ′′ . The position of the crossoverpoint will vary as a function of stellar brightness and inte-gration time. The brighter the star and the longer the inte-gration time, the larger the parameter range where NB4.05will be favourable, and vice versa. We conclude that NB4.05is likely to be an excellent choice for very deep planet searchimaging close to bright stars, although it should be notedthat this depends on the validity of the theoretical models.A first test of the models could be provided by the HR 8799system. As can be seen in the images (Fig. 1), the main differencebetween aADI and PSFR+aADI is that spiders are moreefficiently removed in the latter case. However, the impactof this is largely cosmetic, as a comparable amount of fluxis lost from the companion in the spider regions. This canbe clearly seen in a comparison of the respective contrastcurves for the two methods (see Fig. 4), which have beennormalized with respect to flux losses. PSFR+aADI slightlyimproves the performance at large separations, but providesno improvement at all for small separations. It should benoted that the results are based on a single PSF referencestar (though with multiple representations) – it would bepreferable to use multiple reference stars, and doing so mightsubstantially improve the performance. In any case, we donot reach as promising results as those achieved with PSFRusing LOCI on space-based HST data (see Lafreniere et al.2009), where a significant improvement over aADI is readilyseen. As we have demonstrated, 4 µ m imaging provides avery high Strehl ratio, so if this was the limiting PSF stabil-ity factor at this level of contrast, we should have expected c (cid:13) , 1–8 maging search for the companion to ǫ Ind A Figure 1.
Final output images for aADI (top) and PSFR+aADI (bottom) to the same flux scale, for the images A1+A2. The mostobvious difference between the methods is in the treatment of the spiders.c (cid:13) , 1–8
M. Janson et al. an improvement in the inner image range. Hence, the re-sults imply that other PSF effects become dominant oncethe Strehl ratio is high enough, such as low-order aberra-tions arising in the telescope, and differences in PSF repre-sentation resulting from dithering. This in turn implies thata stable telescope configuration is the best way forward forimproving the contrast in 4 µ m imaging even further. Thereis an obvious and well-tested technique for achieving this,called passive ADI, in which the pupil is stabilized duringobservations, while the field is allowed to rotate (see Maroiset al. 2006). Indeed, the LOCI algorithm was originally de-signed for this purpose (Lafreniere et al. 2007). In fact, wehave a passive ADI sequence at 4 µ m showing exquisite per-formance at small separations, but those data are taken witha different telescope and of a different target, so a rigorouscomparison can not be made. The passive ADI data will bepart of a separate publication. There exist extensive radial velocity measurements of ǫ IndA, as well as some limited astrometric information, whichcan be used to constrain the properties of any sufficientlymassive companion, as discussed in the following.
The linear radial velocity trend of ǫ Ind A was first re-ported by Endl et al. (2002). The original dataset coveredan observational baseline of about 5.2 years, taken with theESO CES instrument in the period 1992-1998. Since then,HARPS data have been taken from 2003 to 2008 (Zechmeis-ter et al., in prep.). The linear trend of the HARPS data isconsistent with that of the aforementioned CES data, with aslope of 4.4 m s − yr − . Hence, we adopt this slope over the16 year total baseline. One contributor to the linear trend issecular acceleration. This is the apparent acceleration thatan observer measures over time in projected motion (in thiscase along the line of site) of an object with constant ve-locity, due to the actual motion in 3D space. Using all themeasured spatial coordinates and velocity components of ǫ Ind, the secular acceleration can be calculated to 1.8 m s − yr − . This is quite large, due to the fast motion of ǫ Ind inthe plane of the sky, but still leaves a 2.6 m s − yr − trendthat must be due to actual acceleration.Since ǫ Ind Ba/Bb is known to be physically bound to ǫ Ind A, it needs to be tested whether it could be responsiblefor the observed trend. We do this with the following order-of-magnitude estimate: The projected separation between Aand Ba/Bb is about 1500 AU, hence for masses of 0.7 M sun ,0.047 M sun , and 0.028 M sun respectively (see McCaughreanet al. 2004), the orbital period of the A/B system is at least66000 years for a circular Keplerian orbit. Such an orbitwould lead to a radial velocity semi-amplitude for ǫ Ind Aof 62 m s − , which in turn gives an average peak-to-peakacceleration of 4 ∗ − m s − yr − . Thus, the gravitationalinfluence of ǫ Ind Ba/Bb is several orders of magnitude toosmall to make any significant contribution to the observedtrend.With ǫ Ind Ba/Bb out of the picture, we are left withcloser, as of yet unseen companions. A previous H- and K-band imaging campaign (Geißler et al. 2007) has excluded the presence of stellar and massive brown dwarf compan-ions, down to 53 M jup outside of a projected separation of1.5 AU and 21 M jup outside of 4.7 AU. This also excludeswhite dwarfs, since at ages up to several Gyrs, they are muchbrighter in H-and K-band than a 50 M jup object (see e.g.Holberg & Bergeron 2006 and Baraffe et al. 2003). Stellarobjects outside of the field of view can be excluded, as theywould be detectable with wide-field or all-sky surveys suchas 2MASS (Skrutskie et al. 2006). While more exotic formsof stellar remnants (e.g. neutron stars) can perhaps not becategorically excluded, for the remainder of this paper wewill assume that the observed acceleration is due to a low-mass brown dwarf or giant planet. The combined constraintsfrom the imaging and the radial velocity trend are given inSect. 4.4. As will be seen in the following, the companion is expectedto have a mass in the range of ∼ M jup , and an orbitalsemi-major axis in the range of ∼ ǫ Ind system, this corresponds to a strong astrometricamplitude signature imposed on the primary of about 15-60mas. However, with an orbital period of a few decades, itwould not be possible to detect orbital motion with, e.g.,
Hipparcos data alone. On the other hand, one might ex-pect a systematic difference between the proper motion asmeasured by
Hipparcos versus that measured in long-termground based monitoring, such as from the Fifth Funda-mental Catalog (FK5). This type of signature is referredto as ∆ µ binarity, see Wielen et al. (2001). For ǫ Ind, anapproximate conversion between the FK5 and HIP systemsimplies that there is a difference between the
Hipparcos andFK5 proper motions of ∆ µ α = − . ± .
68 mas yr − , and∆ µ δ = − . ± .
98 mas yr − . This corresponds in total to asignificance level of F = 2 .
54, where the F value is roughlyto the same level of confidence as the equivalent σ -numberfor Gaussian statistics (Wielen et al. 2001). In other words,there is an indication of a companion in the data, but notat a very high level of significance. We can make an order-of-magnitude estimation of whether these numbers are con-sistent with the RV companion by assuming that the orbitalmotion is completely averaged out in the FK5 data, thatthe orbit is circular, and that a sufficiently small fraction ofthe orbit is covered by Hipparcos such that local curvaturein the motion during that period is negligible. The limitingcases quoted above then yield astrometric motions of π ∗ π ∗
60 mas in 89 years respectively, i.e.1.5 mas yr − and 2.0 mas yr − , both of which are consistentwith the given ∆ µ within the errors. Hence, the astrometryis indeed consistent with the RV trend, though we reiteratethat the significance is rather limited for the astrometry. Since A1+A2 are the deepest images, they provide thestrongest detection limits, and therefore we concentrate onthem in this section. The instrumental contrast for A1+A2is determined at each separation as the maximum perfor-mance out of the aADI and PSFR+aADI contrasts at thatseparation. The corresponding mass limits for A1+A2, cal-culated in the same way as for B1 and B2 in section 4.1, are c (cid:13) , 1–8 maging search for the companion to ǫ Ind A shown in Fig. 5 for ages of 1, 3, and 5 Gyr. Several age de-terminations exist pointing to an age in the range of 1 Gyrfor ǫ Ind (e.g. Lachaume et al. 1999; Barnes 2007). How-ever, preliminary analysis of the astrometric masses of ǫ IndBa/Bb (Cardoso et al. 2008) implies that the componentsare probably under-luminous with respect to model predic-tions (Baraffe et al. 2003) at 1 Gyr, such that the ǫ Indsystem has to be significantly older, perhaps up to 5 Gyr,if the models are accurate (which may not be the case, seee.g. Dupuy et al. 2009). On the other hand, such an old agewould be incompatible with the observed spectra of ǫ IndBa and Bb according to the analysis of Kasper et al. (2009).It is with these uncertainties in mind that we consider thefull range of 1 to 5 Gyr in our analysis.Also plotted is the mass as function of semi-major axisderived from the slope of the linear trend, under the as-sumption that the inclination is 60 o (the mean inclinationof randomly oriented orbits). The minimum possible semi-major axis is set by the minimum possible period, which inturn is some multiple q of the observational baseline. Theexact value of q depends on the amount of curvature presentin the trend, the determination of which would be an over-interpretation of the data at hand. As discussed in Jansonet al. (2008), q = 1 would be the most conservative limitpossible to set, but it is unrealistic, since it would require adiscrete change in velocity state. Here, we set q = 2, whichis still conservative, and more realistic.The mass limits and RV trend shown in Fig. 5 provide agood illustration of the detectability of the dynamical com-panion under the assumption of a circular orbit. However,given the large eccentricity spread of the exoplanet popu-lation outside of 0.1 AU, it is necessary to perform moredetailed simulations in order to constrain the possible phys-ical and orbital parameters of the companion. The methodfor doing so is described in detail in Janson et al. (2008), andwe follow it here for q = 2. In brief, based on the empiricaldistribution of eccentricities for known exoplanets outsideof 0.1 AU, we simulate all possible orbits and orbital phasesand test whether they are consistent with the observed lineartrend. The fraction of such orbits as function of semi-majoraxis is named φ . Out of these allowed orbits, we test whatfraction would lead to a detectable companion. This fractionas a function of semi-major axis is termed χ . One additionhas been made to this procedure with respect to what waspresented in Janson et al. (2008): In the case of ǫ Eri, theplane of the disk, the rotational plane of the star, and theorbital plane of the planet candidate ǫ Eri b all gave a con-sistent orbital inclination of about 30 o , hence this numberwas fixed in the simulations. In the case of ǫ Ind, we have noprior information of the inclination, hence it is treated as afree parameter in the simulations. This is done by perform-ing the simulations over several different inclination anglesand averaging the results. The input inclination angles areset to correspond to the actual probability of a given incli-nation occurring (i.e., accurately taking into account thatthe inclination is more likely to be edge-on than face-on).The results of the simulation are shown in Fig. 6. It canbe seen that if the age is 1 Gyr, the probability of detectingthe companion is always about 90% or higher for any semi-major axis, hence since no companion is detected, it is quiteunlikely that the system is that young. On the other hand,if the age is 3 Gyr, or even 5 Gyr as discussed above, there is still a substantial parameter range in which the companioncould hide. Given these results, in approximate numbers wecan constrain the planet or brown dwarf mass to about 5-20 M jup and its semi-major axis to about 10-20 AU. Also,the inclination must be larger than at least 20 o , otherwisethe projection effects could never bring the companion closeenough to the star to hide it, and the actual mass wouldbe sufficiently larger than the projected mass to make itbrighter than the background at any reasonable age. We have attempted to image the indirectly discovered com-panion to ǫ Ind A, using imaging in the 4 µ m filter as well asthe L’-band. As expected, 4 µ m imaging was found to be apreferable choice over L’-band in the inner, contrast-limitedregime, whereas the opposite was found to be true in theouter, background-limited range. This conclusion is basedon theoretical models that ultimately need to be confirmedthrough observations of known planets. The overlap occursat a radius of 4 ′′ in our images, a number that will dependon target brightness and integration time. Two PSF subtrac-tion techniques were employed: regular active ADI as usedpreviously, and a new combination of techniques, using PSFreference subtraction and aADI with the LOCI algorithm.While PSFR+aADI performs slightly better at large sep-arations, the techniques are virtually indistinguishable formost of the contrast-limited regime. Using more than onePSF reference star may change this picture. In addition, themethod of combining 4 µ m imaging and LOCI is also wellsuited for passive ADI, which has the potential to substan-tially enhance the performance even further.In spite of the high sensitivities achieved in our images,we did not detect any potential companion candidate. Unlessthe known radial velocity companion to ǫ Ind A is a neutronstar or even more exotic stellar remnant, the non-detectionin all images implies that the system is probably older than 1Gyr, possibly consistent with preliminary results presentedby Cardoso et al. (2008). Furthermore, we can constrain theplanet or brown dwarf mass to within approximately 5-20 M jup , the semi-major axis to ∼ > o . An analysis based on astrometry from FK5and Hipparcos is consistent with such a companion. Giventhe high significance of the RV trend, the fact that we canexclude all stellar, white dwarf and high-mass brown dwarfcompanions, and the fact that exotic stellar remnants arerare, it seems very plausible that ǫ Ind A is one of the near-est stars to host a massive giant planet or very low-mass ob-ject. Furthermore, it is likely that this companion would bedetectable through further imaging with either the presentlyavailable facilities, or facilities that come online in the rel-atively near future. Hence, ǫ Ind is a high-profile target forthe study of substellar objects, even aside from the fact thatit hosts the nearest binary brown dwarf.Finally, we note that no sophisticated coronagraphadapted for observations beyond 3 µ m presently exists on anyof the 8m-class or larger AO-assisted telescopes (althoughsimple coronagraphs do exist, e.g. a Lyot coronagraph forNACO). The potential coronagraphic performance is inti-mately connected to the adaptive optics performance, whichleads to an interest in coronagraphs in the context of ’ex- c (cid:13) , 1–8 M. Janson et al. treme AO’ facilities currently in development (e.g. Petit etal. 2008). However, given the fact that a demonstrated Strehlratio in the range of 85% can be reached even with NACO at4 µ m, an ’extreme AO’-type performance in this particularwavelength range is available already today. The develop-ment of a coronagraph for this wavelength range could there-fore be another promising avenue to further increase thenear-future capacity of detecting extrasolar planets throughdirect imaging. ACKNOWLEDGMENTS
The authors wish to thank Marten van Kerkwijk and Yan-qin Wu for useful discussion. The study made use of theCDS and SAO/NASA ADS online services. M.J. is sup-ported through the Reinhardt postdoctoral fellowship fromthe University of Toronto.
REFERENCES
Baraffe I., Chabrier G., Barman T., Allard F., HauschildtP., 2003, A&A, 402, 701Barnes S.A., 2007, ApJ, 669, 1167Burrows A., Sudarsky, D., Lunine, J., 2003, ApJ, 596, 587Cardoso C., McCaughrean M., King R. et al., 2008, AIPConf. Proc., 1094, 509Chauvin G., Lagrange A.-M., Dumas C. et al., 2005, A&A,438, 25Dupuy T., Liu M., Ireland M., 2009, ApJ, 692, 729Endl M., K¨urster M., Els S., Hatzes A., Cochran W., Den-nerl K., D¨obereiner S., 2002, A&A, 392, 671Geißler K., Kellner S., Brandner W. et al., 2007, A&A, 461,665Hinz P., Heinze A., Sivanandam S. et al., 2006, ApJ, 653,1486Holberg J., Bergeron P., 2006, AJ, 132, 1221Janson M., Brandner W., Henning Th. et al., 2007, AJ,133, 2442Janson M., Reffert S., Brandner W., Henning Th., LenzenR., Hippler S., 2008, A&A, 488, 771Kalas P., Graham J., Chiang E. et al., 2008, Science, 322,1345Kasper M., Apai D., Janson M., Brandner W., 2007, A&A,471, 655Kasper M., Burrows A., Brandner W., 2009, ApJ, 695, 788Lachaume R., Dominik C., Lanz T., Habing H., 1999, A&A,348, 897Lafreni`ere D., Marois C., Doyon R., Nadeau D., ArtigauE., 2007, ApJ, 660, 770Lafreni`ere D., Marois C., Doyon R., Barman T., 2009, ApJ,694, 148Lagrange A.-M., Gratadour D., Chauvin G. et al., 2009,A&A, 495, 335Marois C., Lafreni`ere D., Doyon R., Macintosh B., NadeauD., 2006, ApJ, 641, 556Marois C., Macintosh B., Barman T. et al., 2008, Science,322, 1348McCaughrean M.J., Close L., Scholz R.-D. et al., 2004,A&A, 413, 1029 Moor A., Abraham P., Derekas A. et al., 2006, ApJ 644,525Perryman M.A.C., Lindegren L., Kovalevsky J. et al., 1997,A&A, 323, 49Petit C., Fusco T., Charton J., 2008, in Hubin N., Max C.,Wizinowich P., eds, Proc. of the SPIE Vol. 7015, Marseilles,pp. 70151URhee J., Song I., Zuckerman B., McElwain M., 2007, ApJ,660, 1556Scholz R.-D., McCaughrean M., Lodieu N., Kuhlbrodt B.,2003, A&A, 398, 29Skrutskie M.F., Cutri R.M., Stiening R. et al., 2006, AJ,131, 1163Wielen R., Lenhardt H., Schwan H., Dettbarn, C., 2001,A&A, 368, 298Zechmeister M. et al., in preparation c (cid:13) , 1–8 maging search for the companion to ǫ Ind A Figure 3.
Comparison between L’ (dashed line) and NB4.05(solid line) imaging for high-contrast purposes, for 1 Gyr (toppanel), 3 Gyr, (middle panel), and 5 Gyr (bottom panel). Asexpected, NB4.05 provides a better performance than L’ in theinner image range, and the opposite is true in the outer range.The comparison is based on sets B1 and B2, note that the A1+A2detection limits are better.
Figure 4.
Comparison of contrast for aADI (solid line) andPSRF+aADI (dashed line). The performance is generally verysimilar.
Figure 5.
Detection limits at 1, 3, and 5 Gyr for ǫ Ind A. Thesolid line that increases outwards is the mass as function of semi-major axis corresponding to the 2.6 m s − yr − slope of theobserved RV trend, and the dashed line is the reference for min-imum projected separation at a typical inclination of 60 o , bothunder the assumption of a circular orbit. The dotted vertical lineis the minimum semi-major axis from the RV baseline at q = 2.c (cid:13) , 1–8 M. Janson et al.
Figure 6.
Detection probability in our images as function of semi-major axis for 1, 3, and 5 Gyr. Also plotted is φ , the fraction oforbits at a given semi-major axis that are consistent with thelinear RV trend, denoted with the subscript ”All” to signify thatit is independent of age, in contrast to χ . c (cid:13)000