Orbital motion of the young brown dwarf companion TWA 5 B
aa r X i v : . [ a s t r o - ph . S R ] M a y Astronomy&Astrophysicsmanuscript no. 13917 c (cid:13)
ESO 2018November 15, 2018
Orbital motion of the young brown dwarf companion TWA 5 B ⋆ Ralph Neuh¨auser , Tobias O.B. Schmidt , Valeri V. Hambaryan , and Nikolaus Vogt , Astrophysikalisches Institut, Universit¨at Jena, Schillerg¨asschen 2-3, 07745 Jena, Germany Departamento de F´ısica y Astronom´ıa, Universidad de Valpara´ıso, Avenida Gran Breta˜na 1111, Valpara´ıso, Chile Instituto de Astronomia, Universidad Catolica del Norte, Avda. Angamos 0610, Antofagasta, ChileReceived 21 Dec 2009; accepted 4 March 2010
ABSTRACT
Context.
It is di ffi cult to determine masses and test formation models for brown dwarfs, because they are always above the main sequence, sothat there is a degeneracy between mass and age. However, for brown dwarf companions to normal stars, such determinations may be possible,because one can know the distance and age of the primary star. As a result, brown dwarf companions are well-suited to testing formationtheories and atmosphere models. Aims.
With more adaptive optics images available, we aim at detecting orbital motion for the first time in the system TWA 5 A + B. Methods.
We measured separation and position angle between TWA 5 A and B in each high-resolution image available and followed theirchange in time, because B should orbit around A. The astrometric measurement precision is about one milli arc sec.
Results.
With ten year di ff erence in epoch, we can clearly detect orbital motion of B around A, a decrease in separation by ∼ . ′′ per yearand a decrease in position angle by ∼ . ◦ per year. Conclusions.
TWA 5 B is a brown dwarf with ∼
25 Jupiter masses (Neuh¨auser et al. 2000), but having large error bars (4 to 145 Jupitermasses, Neuh¨auser et al. 2009). Given its large projected separation from the primary star, ∼
86 AU, and its young age ( ∼
10 Myrs), it hasprobably formed star-like, and would then be a brown dwarf companion. Given the relatively large changes in separation and position anglebetween TWA 5 A and B, we can conclude that they orbit around each other on an eccentric orbit. Some evidence is found for a curvature inthe orbital motion of B around A - most consistent with an elliptic (e = ∼
18 mas) periodic sinusoid with ∼ . + b.Measuring these residuals caused by the photocenter wobble - even in unresolved images - can yield the total mass of the inner pair, so can testtheoretical pre-main sequence models. Key words.
Astrometry – Stars: binaries: visual – Stars: brown dwarfs – Stars: formation – Stars: individual: TWA 5 – Stars: pre-main sequence
1. Introduction: The brown dwarf TWA 5 B
The star TWA 5 is one of the five original members of the TWHya association (TWA), a group of 5 to 12 Myr young stars(Kastner et al. 1997), where no gas clouds are left from thestar formation process (Tachihara et al. 2009); see Torres et al.(2008) for a recent review on TWA. TWA 5 is an M1.5 weak-line T Tauri star (Webb et al. 1999) with variable H α emission,hence still ongoing accretion (Mohanty et al. 2003). The centralstar itself is either a very close ( ≤
66 milli arc sec or mas)binary (Konopacky et al. 2007, henceforth K07) or even triple(Torres et al. 2003). The close inner pair TWA 5 Aa + b has atotal dynamical mass of 0 . ± .
14 M ⊙ (assuming 44 pc asdistance) and an orbital period of 5 . ± .
09 years (K07). Thewide companion TWA 5 B was originally discovered by Webb
Send o ff print requests to : Ralph Neuh¨auser, e-mail: [email protected] ⋆ Based on observations collected at the European SouthernObservatory, Chile, in runs 79.C-0103(A) and 81.C-0393(A) as wellas on data obtained from the public ESO science archive. et al. (1999) and Lowrance et al. (1999) and confirmed as co-moving with TWA 5 A by Neuh¨auser et al. (2000).The spectral type of TWA 5 B is M8-9 (Webb et al. 1999,Lowrance et al. 1999, Neuh¨auser et al. 2000, Mohanty et al.2003). The mass of the companion is between 15 and 40Jupiter masses just from temperature, luminosity, and theoret-ical hot-start model tracks (Neuh¨auser et al. 2000). The masslies anywhere between 4 and 145 Jupiter masses, if calculatedfrom temperature (2800 ±
100 K), luminosity (log( L bol / L ⊙ ) = − . ± .
30 at 44 ± g = . ± . + B was observed by several teamswith ground-based adaptive optics (AO) and / or the HubbleSpace Telescope (HST). We obtained two more recent images,so that we can now investigate possible orbital motion of Baround A with a 10 year di ff erence in epoch (first preliminaryresults in Schmidt et al. 2008). We can then also try to detectthe orbital motion of Ab and Aa around each other as residu- Ralph Neuh¨auser et al.: Orbital motion of TWA 5 B als of the much longer orbit of B around A due to a periodicwobble of the photocenter of the close Aa + Ab pair.
2. Astrometry with VLT/NACO
We observed TWA 5 with the adaptive optics imager NACO(for NAOS CONICA for the Nasmyth Adaptive Optics System,NAOS, with the COude NearInfrared Camera and Array,CONICA, Rousset et al. 2003) at the ESO VLT in 2007 and2008 with the S13 camera, i.e., a 14 ′′ × ′′ field of view. InJuly 2007, we obtained AO images in the following jitter set-up. Each individual image had a detector integration time (DIT)of 30 sec; the number of DITs (NDIT) co-added together im-mediately after exposure, i.e. without shifting, but added up andsaved in one single file, was 3, resulting in 90 sec total expo-sure per fits file; and the number of such integrations (NINT)was 14, so that we had 30 × ×
14 sec =
21 min total inte-gration time. We always used the neutral density filter, becauseof the brightness of TWA 5 and the good seeing (0 . ′′ ). On 12June 2008, we obtained four (NINT) times 174 (NDIT) times0.3454 sec (DIT) images under less good conditions, repeatedin the next night under better conditions (with NINT 5), i.e., 9min total integration time.All science and flat-field frames taken were subtracted bya dark frame, then the science frames were divided by a nor-malized flat field. A shift + add procedure was applied to sub-tract the background and to add up all frames for each run.We used ESO eclipse and MIDAS. The same procedure wasperformed for the astrometric standard star binary HIP 73357.Astrometric data on separations and position angles (PA) in-clude Gaussian centering errors in the science targets and theastrometric standards, as well as possible motion in the stan-dards (see e.g. Neuh¨auser et al. 2008 for details on typical as-trometric calibration procedure and maximum orbital motionin HIP 73357).The NACO S13 pixel scale for 2007 July 8 was determinedto be 13 . ± .
079 mas / pixel with the detector orientationshifted by 0 . ± . ◦ ; the pixel scale for 2008 June 13 wasdetermined to be 13 . ± .
086 mas / pixel with the detectororientation shifted by 0 . ± . ◦ . These orientation valueshave to be added to a value measured on a raw frame. For the2003 NACO data by Masciadri et al. (2005), which we reducedagain, no astrometric calibration targets were observed, so thatwe obtained the (rough) calibration from the fits file headers(from the position keywords compared to the star position inthe image), namely 13 . ± .
13 mas / pixel (close to the nominalvalue) and a detector orientation 0 . ± . ◦ . In Table 1, we listall imaging observations used here with separations and anglesmeasured.TWA 5 A is itself a close visual pair, where Aa is slightlybrighter than Ab (Macintosh et al. 2001, Brandeker et al. 2003,K07). The close pair Aa + b (separation ≤
66 mas, see K07) isnot resolved in most of the observations listed in Table 1 (ex-cept the two obtained with Keck in 2000.1). We thus have tocorrect the separation of B wrt A for the photocenter motionof Aa + b. K07 have solved the orbit of Ab around Aa, which Fig. 1.
Position angle (in degrees) versus observing epoch (JD- 2450000 in days) for corrected data listed in Table 1. Thedotted lines (starting from the 2008 data point opening to thepast) indicate maximum PA change due to orbital motion for acircular pole-on orbit. The full lines with strong positive slopein the lower right corner are for the background hypothesis, ifthe bright central star (TWA 5 A) had according to its knownproper motion, while the fainter northern object (now known asB) would be a non-moving object; the data points are inconsis-tent with the background hypothesis by many σ . All data pointsare fully consistent with common proper motion, but not ex-actly identical proper motion (plotted as dashed line). Instead,the data are fully consistent with orbital motion (inclination be-tween pole-on and face-on), because the PA appears to decreaseby ∼ . ◦ per year (plotted as a best fit, full line).we can use for the correction. That Aa + b was not resolved inthe HST / Nicmos images in 1998.3 (aquisition without corono-graph) and 1998.5 is consistent with Aa + b being close to theirclosest approach, as suggested by Macintosh et al. (2001) andconsistent with the orbit of K07. The closest approach was in1998.4 with a few mas separation.We used the orbit in K07 to correct the separations mea-sured between A and B for the photocenter shift in Aa + b, as-suming that Aa and Ab are equally bright. The K-band mag-nitude di ff erence is only 0 . ± .
09 mag (K07). TWA 5 Aborbits Aa with 5 . ± .
09 yr period and a semi-major axis of0 . ± . ′′ (K07). The correction for the epochs listed inTable 1 amounts to ∼
28 mas, so is significant. The correctedvalues are also listed in Table 1. There are two typos in Table 1 in K07: The position angle ofthe Aa + b pair should be 25 . ± . ◦ for the 2000 Feb 20 data fromMacintosh et al. (2001), who gave 25 . ± . ◦ ; and it should be24 . ± . ◦ for the 22 Feb 2000 data from Brandeker et al. (2003),who gave 24 . ± . ◦ ; the Feb 2000 positions of Ab relative to Aa arecorrectly plotted in Fig. 3 in K07; however, the signs of the RA valuesgiven in Fig. 3 in K07 are wrong. Since the slightly fainter Ab is lo-cated towards the NE in Feb 2000 (Macintosh et al. 2001, Brandekeret al. 2003) and Dec 2005 (K07), the RA axis in Fig. 3 in K07 shouldgo from west (left, negative RA changes) to east (right, positive RAchanges).alph Neuh¨auser et al.: Orbital motion of TWA 5 B 3T able 1. Astrometry of TWA 5 A and BEpoch JD - Telesope and Band separation [arc sec] (a) PA (b) Ref.year 2450000 instrument in RA in Dec total [ ◦ ] (c)1998.1 863.5 IRTF Speckle K − . ± . . ± . . ± . ± − . ± . . ± . . ± . ± / Nicmos H − . ± .
01 1 . ± .
01 1 . ± .
01 358 . ± . − . ± .
01 1 . ± .
01 1 . ± .
01 358 . ± . / Nicmos HK(d) − . ± .
001 1 . ± .
006 1 . ± .
006 358 . ± .
03 Wei00corrected: − . ± .
001 1 . ± .
006 1 . ± .
006 358 . ± . / KCam H n / a (g) n / a (g) 1 . ± .
012 n / a (g) Mac01corrected: n / a (g) n / a(g) 1 . ± .
012 n / a (g)2000.1 1596.6 VLT / FORS2 I − . ± .
037 1 . ± .
034 1 . ± .
05 357 . ± . − . ± .
037 1 . ± .
034 1 . ± .
04 357 . ± . / KCam JHK − . ± . . ± . . ± .
008 359 . ± .
08 Bra03corrected: − . ± . . ± . . ± .
057 358 . ± . / ISAAC (d) − . ± .
097 1 . ± .
096 1 . ± .
091 356 . ± . − . ± .
097 1 . ± .
096 1 . ± .
091 357 . ± . / Hokupaa H (e) n / a (g) n / a (g) 1 . ± . / a (g) Neu01corrected: n / a (g) n / a (g) 1 . ± . / a (g)2003.1 2687.8 VLT / NACO K s − . ± .
013 1 . ± .
020 1 . ± .
020 357 . ± .
38 Mas05 (h)corrected: − . ± .
013 1 . ± .
020 1 . ± .
020 358 . ± . / NACO K s − . ± .
013 1 . ± .
011 1 . ± .
011 356 . ± .
40 this workcorrected: − . ± .
013 1 . ± .
011 1 . ± .
011 356 . ± . / NACO K s − . ± .
013 1 . ± .
012 1 . ± .
012 356 . ± .
40 this workcorrected: − . ± .
013 1 . ± .
012 1 . ± .
012 356 . ± . Corrected values are separations and PA between A and B after correction for photocenter shift of Aa + b due to orbit of Ab aroundAa (K07). Separation in α is negative for B west of A, and separation in δ is positive for B north of A. (b) PA is position angle measured fromnorth over east to south. (c) References are Lowrance et al. 1999 (Low99), Macintosh et al. 2001 (Mac01), Weintraub et al. 2000 (Wei00),Neuh¨auser et al. 2000 (Neu00), Webb et al. 1999 (Webb99), Neuh¨auser et al. 2001 (Neu01), Brandeker et al. 2003 (Bra03), and Masciadri etal. 2005 (Mas05). (d) Narrow band filter(s). (e) With Wollaston polarimeter. (f) Values (listed in the first row for Low99) are those given byWei00 according to priv. comm. with P. Lowrance, corrected compared to those published in Low99. (g) Values not available (not published).(h) Mas05 list neither separation nor PA, data reduced by us. Fig. 2.
Separation (in arc sec) versus observing epoch (JD -2450000 in days) for corrected data listed in Table 1. The dot-ted lines (starting from the 2008 data point opening to the past)indicate maximum possible separation change due to orbitalmotion for a circular edge-on orbit. The expectation for thebackground hypothesis is not shown for clarity and becauseit was already rejected in the previous figure. All data pointsare fully consistent with common proper motion, but not ex-actly identical proper motion (plotted as dashed line). Instead,the data are fully consistent with orbital motion (inclination be-tween pole-on and edge-on), because the separation decreasesby ∼ . corrected values for separations andPA. The data are inconsistent with B being a non-moving back-ground object by many σ , but fully consistent with TWA 5A + B being a common-proper motion pair, as known before(Neuh¨auser et al. 2000, Brandeker et al. 2003). We show inFigs. 1 and 2 the maximum possible orbital motion for a cir-cular orbit of TWA 5 B around A. We use 0 . ± .
14 M ⊙ astotal mass towards TWA 5 A (K07). The observed change inPA (Fig. 1) is close to the maximum expected for a circular or-bit, which would indicate that either the orbital plane is seennearly pole-on and / or that the orbit is eccentric. In the formercase, there should be almost no change in separation; however,the separation also changes significantly, so that we can con-clude that the orbit of B around A is eccentric (and maybe alsoinclined).The distance towards TWA 5 is not measured as parallax,but Hipparcos has measured the parallax of three to five mem-bers of the TW Hya association (TWA), where TWA 5 is acertain member, namely for TWA 1, 4, 9, 11, and 19. The useof the values for TWA 9 and 19 is dubious due to binarityand uncertain membership, respectively (Mamajek 2005). Theweighted mean of the distances towards TWA 1, 4, and 11 is61 . ± . ± . . ± . . ± . Ralph Neuh¨auser et al.: Orbital motion of TWA 5 B
Fig. 3.
Separations in δ versus α (both in arc sec) for corrected data listed in Table 1. The brown dwarf B moves to the south-west (relative to A). We tried to fit a geometric orbit for dif-ferent models of motion of B wrt A: Solid (black) line for anelliptic orbit of B around A (best fit), dotted (red) line for acircular orbit of B around A, dashed (green) line for constantchange in separation in α and δ (as if B would be a backgroundobject, i.e. with negligible parallax for B, with proper motionof B being di ff erent from A, and taking into account parallac-tic motion of TWA 5 A at a distance of 44 ± ±
44 pc) andits uncertainty ( ± α and δ ) is not plotted. Welist the odds ratios from Bayesian statistics in the upper right:The ellipse is 2 to 3 times more likely than the constant change(background) and the hyperbola (ejected), respectively. Whilethis is only a geometric fit, we point out that the first two datapoints with small errors (1998 from HST) are in the upper left(NE), the data from the middle epochs (2000) are in the cen-ter of the plot (large errors), and the last two data points withsmall errors (2007 and 2008 from NACO) are in the lower right(SW), i.e. following the (geometric) orbit.through its proper motion to be 59 ± . ± . . ± .
94 pc.The ground-based trigonometric distance towards TWA 22 isonly 17 . ± . ± . ± ∼
50 pc,and also given the wide spread of TWA members on the sky,the individual distance towards TWA 5 of 44 ± ∼ . ± . ′′ (epoch 1998.5). With a distanceof 44 ± + b is 86 . ± . . ± .
14 M ⊙ (K07), the orbital period would then be ∼
950 yrs (fora circular orbit). For our best fit (geometric) orbit as an ellipse(with e = ∼
100 AU (at 44 pc) and theorbital period then ∼
3. Interpretation
The separation between TWA 5 A and B in both α and δ change slightly with time. This is typical of a common-proper motionpair, where orbital motion is seen. Because of orbital motion, α and / or δ do not stay exactly constant, but can change slightly.However, a constant change in the separation in both α and δ would also be consistent with B being a moving backgroundobject: If B moves slightly, but with a proper motion di ff er-ent than A, then we would see a slight change in separation. Ifboth objects are orbiting each other, we should see curvature inthe orbital motion after some time. Such a curvature could inprinciple also be consistent with hyperbolic motion. All thesealternatives are also possible in all the other substellar compan-ions detected directly so far. Curvature in orbital motion hasbeen shown for none of them, yet. Even if youth indicators aredetected in the spectrum of an apparently co-moving substel-lar companion (like low gravity or accretion), the object couldstill be an independent member of the same young associationas the primary object. In such a case, the expected di ff erencein proper motion could reach a few mas / yr, the typical veloc-ity dispersion in young clusters (see e.g. Herbig & Jones 1979,Mugrauer & Neuh¨auser 2005). A final proof of companionshipcould be curvature in orbital motion, consistent with a circularor elliptic orbit.Given the few data points with large error bars, we can-not yet fit a full physical orbit for the motion of B around A.However, we can try to fit a geometric orbit by testing di ff erenthypotheses by estimating the probabilities for di ff erent modelsof motion of B relative to A with Bayesian statistics: a constantchange in separation in α and δ as if B were a background ob-ject with proper motion di ff erent from A, a circular orbit of Baround A, an elliptic orbit of B around A, hyperbolic motionof B wrt A (i.e. B being ejected), and A and B as an exactlyco-moving pair; i.e., both objects having the same motion in α and δ , so that the separation in both α and δ remain constant.Assuming that a priori those five models are equally probableand provide a complete set of hypotheses, we calculated globallikelihoods (Bayesian model comparison, see Gregory 2005).According to these calculations, an elliptic orbit is mostlikely, with a probability for this hypothesis being 0.63, whichis not yet very significant, but 2 to 3 times larger than the prob- alph Neuh¨auser et al.: Orbital motion of TWA 5 B 5 Fig. 4.
Phase-folded residuals (in arc sec) from the (best) ge-ometric fit (elliptic orbit of TWA 5 B around the photocenterof Aa + b, similar to Fig. 3, but for un corrected data) with datapoints with error bars ( un corrected data from Table 1 after sub-traction from best-fit orbit), best-fitting sinusoid for residuals(full black line), and its ± σ limits (as dashed-dotted red lines).Period (5 . ± .
14 yr), amplitude (18 ± ◦ and an eccentricity of e ≃ . + B would be the first substellar companion outsidethe solar system, where such evidence is reported from directimaging observations.Then, we can also try to investigate whether we can detecta small periodic wobble in the separation of B from A, whichwould stem from the expected photocenter shift of the closepair Aa + b. This wobble should be seen in periodic residuals tothe best fit (ellipse of B around A) in the un corrected data fromTable 1. The period would then give the orbital period of Aa + b,and the amplitude would yield the total mass of Aa + b. For theuncorrected (i.e. directly observed) data in Table 1, we also ob-tain an ellipse as the best geometric fit to the data by Bayesianstatistics. As can be seen in Fig. 4, the residuals to that bestfit do show a small-amplitude (periodic) sinusoid. We searchedfor the periodicity only in a small window of 3-8 years (aroundthe known orbital period of 5.94 yr) and detected a best-fit pe-riod of ∼ . ± .
14 yr. The (half-)amplitude of the sinusoidalwobble is 18 ± + b in K07, where they give 5 . ± .
09 yrperiod. Given the semi-major axis, eccentricity, and inclination,the maximum photocenter shift on the sky (for equal brightnessof Aa and Ab) is ∼
28 mas (their Fig. 3 and Table 2). Our resid-uals are close to zero (minimum of residuals) at epoch ∼ + b pair is unresolved in our NACO data, we candetect and measure the photocenter wobble of TWA 5 Aa + b inthe separation changes between A and B. In principle, we couldalso measure the total mass of the TWA 5 Aa + b pair; however,we refrain from doing so, because our images do no resolve forthe close pair and, thus, have much lower precision comparedto K07. This method should now also be applicable to othercases, at least for detecting close pairs, possibly also for testingand calibrating pre-main sequence models. Acknowledgements.
We would like to thank the ESO Paranal Teamand ESO Users Support group. We are grateful to Markus Mugrauerfor useful discussion. TOBS would like to thank EvangelischesStudienwerk e.V. Villigst for financial support. NV acknowledgessupport by FONDECYT grant 1061199. RN wishes to acknowl-edge general support from the German National Science Foundation(Deutsche Forschungsgemeinschaft, DFG) grants NE 515 / / References
Biller, B. A. & Close, L. M. 2007, ApJ, 669, L41Brandeker, A., Jayawardhana, R., & Najita, J. 2003, AJ, 126,2009 (Bra03)Ducourant, C., Teixeira, R., Chauvin, G., et al. 2008, A&A,477, L1Gizis, J. E., Jao, W., Subasavage, J. P., & Henry, T. J. 2007,ApJ, 669, L45Gregory, P. C. 2005, Bayesian Logical Data Analysis forthe Physical Sciences: A Comparative Approach with‘Mathematica’ Support, ed. Gregory, P. C. (CambridgeUniversity Press)Jones, B. F. & Herbig, G. H. 1979, AJ, 84, 1872Kastner, J. H., Zuckerman, B., Weintraub, D. A., & Forveille,T. 1997, Science, 277, 67Konopacky, Q. M., Ghez, A. M., Duchˆene, G., McCabe, C., &Macintosh, B. A. 2007, AJ, 133, 2008 (K07)Lowrance, P. J., McCarthy, C., Becklin, E. E., et al. 1999, ApJ,512, L69 (Low99)Macintosh, B., Max, C., Zuckerman, B., et al. 2001, inAstronomical Society of the Pacific Conference Series, Vol.244, Young Stars Near Earth: Progress and Prospects, ed.R. Jayawardhana & T. Greene, 309 (Mac01)Mamajek, E. E. 2005, ApJ, 634, 1385Masciadri, E., Mundt, R., Henning, T., Alvarez, C., & Barradoy Navascu´es, D. 2005, ApJ, 625, 1004 (Mas05)Mohanty, S., Jayawardhana, R., & Barrado y Navascu´es, D.2003, ApJ, 593, L109Mugrauer, M. & Neuh¨auser, R. 2005, AstronomischeNachrichten, 326, 701Neuh¨auser, R., Guenther, E. W., Petr, M. G., et al. 2000, A&A,360, L39 (Neu00)