A Search for Wide Low-Mass Companions to Spectroscopic Binaries

Transcription

1 A Search for Wide Low-Mass Companions to Spectroscopic Binaries Brian Devour Department of Physics and Astronomy Astro 490 Departmental Honors Thesis Graduating May 15, 2010, Submitted May 13, 2010

2 Abstract We report on a near-infrared survey of binary stellar systems for possible distant tertiary components. Theory suggests that for multiple stars forming inside a single cloud of gas, the most probable outcome is for a single binary to dominate and accrete the majority of the gas, while lower mass stars may remain in loose orbits about the binary. Thus, it is predicted that a large proportion of binary systems may possess distant tertiary members of lower mass than the primary components. To test this theory, we search the space around known binary systems for undiscovered dim companions. Also, since these companions will be primarily low mass objects, this offers a good chance to search for brown dwarfs. These extremely low mass 'failed stars' are dim and difficult to detect in many cases, especially for the low temperature objects that we would like to discover. Thus, this presents a good opportunity to expand the population of known brown dwarfs. As part of a continuing survey, we have taken narrow-field images of 21 close spectroscopic binaries with the 8m Gemini North telescope and wide-field images of a further 20 systems with the Kitt Peak 4m telescope. From our analysis of these images, we have identified a number of candidate tertiary members for further investigation. 1 Introduction 1.1 Binaries and Tertiary Components It is generally accepted that the vast majority of stars form in clusters, from the hierarchical fragmentation of a single large cloud of gas into smaller and smaller clumps as it collapses. In this environment it would not be unusual for several stars to begin to form inside a single clump of gas. These protostars will compete for the available supply of gas (Bonnell et al, 2001), and the complex interactions between them produce the variety of stellar masses and system types we see in the heavens today. Through numerical simulations (Delgado-Donate et al, 2003), the possible outcomes of these interactions can be investigated. In general, the star or stars forming nearest the center of the cloud accrete gas the fastest and come to dominate the system. In some cases, two stars form close enough together to become a binary early on, while in 1

3 others, the central more massive star acquires a close companion from among the more distant protostars. In any case, the system soon becomes dominated by a binary system of comparatively massive protostars near the center of the cloud (or, occasionally, a single star if no companion is acquired). Subsequent interactions between the central binary and other growing protostars tend to drive the smaller interlopers onto larger orbits, while tightening the orbits of the two central stars about each other. In the outer reaches of the cloud, the smaller protostars have a harder time accreting gas, widening the mass disparities. By the time the gas is exhausted or driven off, in some cases the lower mass stars have been ejected from the system entirely, while in others one or more may remain in orbit about the central binary at relatively large distances. This theory makes two major predictions. First, it predicts that many binary star systems, especially tight binaries, will have distant companions. Second, it predicts that these companions (along with the stars ejected from the system) will tend to have relatively smaller masses, while binaries will preferentially consist of heavier stars. By searching for distant companions to binary systems and comparing their mass to that of their primaries, we can test this theory. Our search focuses on what are called spectroscopic binaries binary systems where the two stars are so close together that they cannot be resolved visually as separate objects. These systems can only be identified by the periodic Doppler shift in their spectra as they orbit each other, hence the name. Since these are very tight binaries, they are the most likely to have tertiary companions. 1.2 Brown Dwarfs These tertiary companions are predicted to be lower mass than their primaries. Thus, at least for intermediate mass primary stars, a significant fraction of them are likely to be brown dwarfs. Generally speaking, a brown dwarf is an object similar to a very small star or large gas giant planet, whose mass is low enough that its central pressure is insufficient for hydrogen fusion to occur in its core. They are sometimes referred to as 'failed stars', as they failed to acquire enough mass to ignite hydrogen fusion and take their place on the main sequence. 2

4 Despite their status as 'failures', brown dwarfs are still quite interesting objects. As we have seen in the previous section, they figure strongly into various scenarios of star formation, and they are likely one of the most common types of 'stars' in the universe. Since their only internal energy source is the gravitational energy released during their formation, their surface temperatures are relatively low and they glow mostly in the infrared. These low temperatures allow many different molecules to exist in their upper layers, which give them very complex spectra with many absorption features. In terms of spectral types, there are three main categories of brown dwarfs. First, there are M type brown dwarfs, whose spectra contain characteristic lines of metal oxides such as TiO and VO. These have surface temperatures of perhaps K and infrared J band absolute magnitudes of perhaps 9 to 11. Next, there are L type brown dwarfs, whose spectra are characterized by metal hydrides like FeH and CrH. Their temperatures are in the K range, and their J band magnitudes lie roughly in the 11 to 15 range. Beyond this are T type brown dwarfs at less than 1500 K, which contain methane features in their spectra and have J band magnitudes of about 15 to 17 (Kirkpatrick et al, 1999, Gelino et al, 2009). It is important to note that not all M and L dwarfs are brown dwarfs these two spectral classes also include the smallest true stars. The brown dwarfs in these classes are mostly young and still have high enough temperatures to be confused with low mass stars. Care must be taken to distinguish these from actual stars. An important question is, what do brown dwarfs look like at extremely low temperatures? It has been speculated that below roughly 700 K, ammonia may play a significant role in their spectra. This would dictate the creation of a new spectral type, tentatively named Y (Delorme et al, 2008). To date, several objects that may possibly be members of this new class have been discovered (Lucas et al, 2010), but an unambiguous detection has proved elusive. As such, the second objective of our survey is to attempt to discover new brown dwarfs, especially extremely low temperature ones. Although it is perhaps not likely, detection of a candidate cool Y brown dwarf would be a very exciting prospect. 3

5 2 Methods and Analysis 2.1 Identifying Tertiary Components If we are looking for dim tertiary companions to existing systems, we must find a way to differentiate these objects from the background of intrinsically brighter but farther away stars. Trying to determine the distance to every single object in the field would be far too time consuming, as well as possibly unreliable. Thus, we exploit the proper motion of the stars involved that is, their apparent movement across the sky due to the relative motion between them and the Solar System. All of the binary systems studied are relatively nearby they have to be if we are to have any chance of actually detecting any brown dwarf companions they may have. Since brown dwarfs are so dim we cannot detect them at great distances, and so we must limit our search to nearby binary systems. Therefore, these systems will be close enough that their proper motion will be large enough to measure. On the other hand, the dim background stars are all far enough away that their proper motions are extremely small (if they are measurable at all). Thus, any dim foreground objects in the field could distinguish themselves by having a large proper motion. If such a foreground object were to have a proper motion with the same direction and magnitude as the primary system, it would be a very strong indicator that that object was associated with the primary such an agreement would be very unlikely to come about by chance. Therefore, to detect tertiary components of the system, we search for dim objects with a proper motion equal to that of the primary. Of course, we cannot instantaneously measure proper motion from a single image. To obtain the proper motion of each object, we compare multiple images taken at different times. With a sufficiently long interval (a few years), any nearby objects will change positions noticeably in our images, allowing their proper motion to be measured. 2.2 Image Types and Analysis This survey involves two different types of images narrow field images intended to uncover any companions relatively nearby the central binary, and wide field images to 4

6 detect companions farther out. The details and methods of processing these images are somewhat different for each type. The narrow field images were taken with the 8 meter Gemini North telescope on Mauna Kea, using the Near Infrared Imager (NIRI) instrument (Hodapp et al, 2003) at the f/14 setting. This telescope is equipped with adaptive optics, allowing it to greatly compensate for the blurring effects of our atmosphere. This enables us to get images showing detail extremely close to the central stars, where with a normal telescope all we would see is a large, spread out smear of light. Thus, we can resolve faint objects (and hence potential candidates) nearer to our primaries than would otherwise be possible. These observations were taken in the CH4s narrowband filter in the near infrared H band, as brown dwarfs radiate more strongly in infrared than they do in visible light. These images were taken at five different dither positions to allow for subtraction of the background sky brightness and transients. The images were then flat fielded and mosaiced together to form the final images using the Altair IRAF data reduction script prepared by our collaborator Laird Close of the University of Arizona. Due to their small field of view (only 50" by 50"), these images do not usually contain large numbers of background objects. Thus, once the images are properly processed, a simple visual comparison is sufficient to detect potential companions. If one takes two images separated by an appropriate interval, both centered on the primary star, and compares them, the other objects in the field will appear to move due to the primary's motion. If, however, there is another object in the field that shares the primary's exact proper motion, it, unlike everything else in the field, will remain as motionless as the primary. By simply overlaying the two images and blinking back and forth between them, it is easy to see which objects move and which do not. Those that do not are potential companions. The wide field images were obtained with the 4 meter Mayall telescope at the Kitt Peak National Observatory, using the FLAMINGOS imager and spectrograph (Elston et al, 2003). These observations were taken in the near infrared J, H, and K bands. While this telescope is not equipped with adaptive optics, the large field of view (10' by 10') of 5

7 these images allow us to examine many more potential companions, and to detect companions more distant from their primaries. These images are also dithered to allow for sky subtraction, with a 16 point semirandom dither pattern. Using IRAF and the computer scripts WCSTOOLs and SWarp (Mink 2002, Bertin et al 2002), the images are bad pixel masked, dark subtracted, linearity corrected, sky subtracted, and flat fielded. To correct for any optical distortions, we fit the positions of the stars to the 2 Micron All Sky Survey (2MASS) catalog. With the large number of background stars in each image, a more systematic method of proper motion analysis is needed for these than for the narrow field images. This is the reason we use the 2MASS catalog. This survey, carried out around the turn of the century, mapped the entire sky at infrared wavelengths in fact, in the J, H, and K bands, the exact bands we used. Since we are looking now for companions farther away from their primaries, it would have seen and catalogued most of the objects we seek. And although by itself it would not show any associations between the companions and their primaries, in combination with our observations it can. Since 2MASS recorded the precise position of all objects catalogued, we simply must match up the objects we see with objects in the 2MASS database and then calculate how far and in which direction they have moved effectively using the 2MASS images as the temporally separated second set of images to compare with our current ones. This is a somewhat tedious process given the large number of sources in each field, but with computer aid we can use it to obtain precise quantitative data on the motion of all the stars we observed. This data is then plotted for identification of potential companions. An important element in the detection of dim companions is the magnitude limit of the image. This, in combination with the distance to the system, determines the faintest objects that can be observed. Our systems are all relatively nearby, but not at a constant distance, so the faintest visible object in each band varies. Table 1 gives the parallax, distance, faintest detectable object, and the linear size of the field of view for each image. Table 2 gives the magnitude limit and angular resolution of each set of images we use. In this and all future sections, parallax and proper motion for all primary systems are obtained from the Hipparcos catalogue (Perryman, et al, 1997). 6

8 Table 1: Distances and absolute magnitude limits Object Parallax Distance Field size Faintest visible object (abs. mag) (mas) (pc) (au) J band H band K band Narrow-field images: 1 Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Wide-field images: 1 Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip Hip

9 Table 2: Image magnitude limits and resolutions Telescope / Instrument Band Magnitude Resolution Limit (arcsec/pixel) 4m Mayall / FLAMINGOS J H K MASS survey J H K m Gemini North / NIRI H Results and Conclusions 3.1 Example Images In the next few pages, we present several sample images and graphs that are typical of our work. Figure 1 is a narrow field image of the binary known as Hipparcos Several potential companion objects are apparent, along with a number of artifacts introduced by image processing. Figure 2 is a more heavily processed version of Figure 1, where the image has been smoothed and an idealized point spread function subtracted for each bright star in the image. This reveals another potential companion that had been lost inside the glare of the primary. Figures 3 and 4 are sample wide field images. As is apparent, these fields range from relatively sparse to extremely crowded. The crowded fields are a good example of why the computer aided analyses are so important to check every object in those fields by hand would be a difficult task indeed! Figures 5 and 6 show the proper motion plots for two of our wide field images. These plots give the proper motion for each object in right ascension and declination in milliarcseconds per year. The primary is marked with a star, while the other objects in the image are marked with Xs. As mentioned before, most of the objects in these fields are distant background stars, and thus their proper motions are clustered around zero. Figure 5 is an example of a non-detection. While there are five additional objects in this field with large proper motions, none of them has a proper motion in any way similar to that of the primary, and so none of them could be a potential companion. Figure 6, on the other hand, does shows a possible companion, as there is an object with 8

10 similar proper motion to that of the primary. Table 3 summarizes the proper motions for these two objects. While the differences are significantly larger than the 2σ uncertainties, this could still be a possible companion. Unfortunately, on further investigation, this turned out to be a false alarm due to an omitted cleanup step our script ended up counting the central binary twice, and so this is actually a distorted measurement of the proper motion of the primary. This particular misleading result occurred in two other fields, along with another false positive where a dim object was identified with the wrong object in the 2MASS catalog resulting in an incorrect proper motion calculation. It is obviously important to double check any potentially significant discoveries. Table 3: Proper motion for Hipparcos 3362 and candidate Object Proper motion Proper motion 2σ uncertainty 2σ uncertainty (RA, mas/yr) (Dec, mas/yr) (RA, mas/yr) (Dec, mas/yr) Hip Candidate Figure 1: Narrow field image of Hipparcos

11 Figure 2: Narrow field image of Hipparcos with additional processing 10

12 Figure 3: Wide field image of Hipparcos 3362 in the J band 11

13 Figure 4: Wide field image of Hipparcos in the H band 12

14 Figure 5: Proper motion plot for Hipparcos Figure 6: Proper motion plot for Hipparcos

15 3.2 Results For the Gemini narrow field images, we are currently working with the first epoch of images. This basically entails finding and cataloging every background object in each field for later followup. These objects are described in Table 4. Since we do not yet have the second epoch of data, we cannot yet carry out the proper motion comparison to determine if any of these objects are actual companions. The second epoch observations for this particular set of objects will be carried out in the next few months, and although I will not be here to perform the analysis, we should have some results soon after that. The analysis of the Kitt Peak wide field images is not yet complete we have analyzed 10 of 20. So far our most notable result is our observation of a definite tertiary companion to Hipparcos (also known as BY Draconis). This object, BY Dra C, had been previously observed and identified as a companion in 1997 (Zuckerman et al, 1997), but had not been revisited since. While it seems to be an ordinary M5 star rather than a brown dwarf, it is still an interesting find. Our observations, separated from theirs Table 4: Candidate objects in Gemini North NIRI fields Field Objects found # Description 1 Hip Dim stars at 13" W, 25" S, 22" SE 2 Hip Bright star at 13" WNW (known) 3 Hip Bright star at 17" SSW 4 Hip Bright star at 22" WSW, dim stars at 11" SW, 22" ESE, 24" SSW 5 Hip Many dim stars 6 Hip Dim stars at 18" SSW, 15" W, 20" NW, two at 24" SW 7 Hip Dim stars at 30" ESE, 27" SW 8 Hip Bright star 26" W, medium bright star 35" W 9 Hip Many dim stars 10 Hip Bright star 15" N, dim stars at 29" NW, 30" NNE 11 Hip Dim stars at 19" SE, 30" E, 23" ENE, 21" NNE 12 Hip Medium bright star at 17" N 13 Hip Dim star at 26" SW 14 Hip None 15 Hip None 16 Hip None 17 Hip dim stars very far NNE, 2 dim stars very far SSW 18 Hip None 19 Hip None 20 Hip Bright stars at 28" ENE, 3.5" NW, Dim stars at 28" NW, 5" SE 21 Hip None 14

16 by more than a decade, will allow us to precisely pin down its proper motion. Additionally, we will obtain a spectrum of this object to confirm its identity, something that the 1997 observations did not contain. BY Draconis and its companion are shown in Figure 7. The companion is the medium bright star below and to the left of the primary. The proper motion plot for this field is Figure 8 in this case, the proper motion match is so close that the X denoting BY Dra C overlaps the star denoting the primary. The other nearby X is another false positive due to detecting the central star twice. Some of the vital statistics of these two objects are given in Table 5. Another result is the disproving of a possible companion for Hipparcos 3362 proposed by Tokovinin et al in 2006 in a similar survey. They concluded that a nearby star was a companion through photometric analysis, but were unable to do a proper motion analysis. Our data show that they were mistaken, since this star has no detectable proper motion. Referring to Figure 6, the X representing this star is lost in the central blob the nearby X is the false alarm. Hipparcos 3362 and its proposed companion are shown in Figure 9. The disproved companion is the moderately bright star above the primary. A summary of our results for the Kitt Peak wide fields is given in Table 6. Table 5: Properties of Hipparcos and companion System distance (pc) Separation (pc, AU) J magnitude Absolute J magnitude Spectral type Primary spectral type ", M5 K6 Object Proper motion (RA, mas/yr) Proper motion (Dec, mas/yr) 2σ uncertainty (RA, mas/yr) 2σ uncertainty (Dec, mas/yr) Hip BY Dra C Table 6: Candidate objects in Kitt Peak 4m FLAMINGOS fields Field Objects found 1 Hip None 2 Hip None 3 Hip None 4 Hip None 5 Hip False positive: dim star mismatch 6 Hip False positive: central star recorded twice 7 Hip None 8 Hip 3362 False positive: central star recorded twice. Disproved other proposed tertiary component 9 Hip False positive, and known tertiary object reobserved: BY Dra C 10 Hip None 15

17 Figure 7: Narrow field image of Hipparcos and companion Figure 8: Proper motion plot for Hipparcos

18 Figure 9: Narrow field image of Hipparcos 3362 and supposed companion 17

19 3.3 Conclusions and Further Work As for future work, obviously we will be completing the analysis of the remaining ten Kitt Peak images. There are still plenty of fields to examine that could contain brown dwarf companions. Similarly, the second epoch for the current batch of narrow field images will be collected this summer, and there could be brown dwarfs hiding there too. Beyond that, we hope to obtain a second set of wide field observations at Kitt Peak of each object starting next spring. As a focused search done with a large aperture telescope, our images actually go more than 3 magnitudes fainter than 2MASS and have approximately 6 times the angular resolution as well see Table 2. At the moment, this nets us little benefit as we are comparing our images to the 2MASS data and are thus operating under its limits. However, comparing our first epoch images to a set of second epoch images of the same sensitivity could allow us to detect fainter objects than are in the 2MASS database. Also, the increased resolution allows us to more easily pick out individual faint stars in crowded fields. Since the cool brown dwarfs we seek are so dim, these advantages could be very important in discovering any that may be present. So far, our detection rate is low. However, this is not yet cause for any great alarm on behalf of our star formation theories. There are two major factors that are likely responsible for our current slim pickings. First, while it is possible that companions may be found as far as perhaps 25,000 AU from the primary (Faherty, et al, 2010), so far our images mostly scan space out to less than 20,000 AU. Second, our current use of the 2MASS data, with its smaller magnitude limits, leaves our sensitivity to fainter dwarfs low. Considering that our images go almost 4 magnitudes deeper than 2MASS, an examination of Table 1 can show that, while our images are sufficient to detect almost all brown dwarfs, the 2MASS images may well miss many. This is the main reason why the second epoch of wide field images is so important. In conclusion, while we have only a couple of results so far, this project is still ongoing and the prospects look pretty good. We have confirmed the existence of a companion to BY Draconis, and disproved a candidate companion to Hip The bugs have been largely worked out of our analysis methods, and there are a variety of new data coming in soon which will open up new possibilities. 18

20 References: Bonnell, I. A., Bate, M. R., Clarke, C. J., Pringle, J. E. (2001), Competitive accretion in embedded stellar clusters, Monthly Notices of the Royal Astronomical Society, 323, Bertin et al (2002), "The TERAPIX Pipeline," Astronomical Data Analysis Software and Systems XI, ASP Conference Proceedings, 281, 228 Delgado-Donate, E. J., Clarke, C. J., Bate, M. R. (2003), Accretion and dynamical interactions in small-n star-forming clusters: N = 5, Monthly Notices of the Royal Astronomical Society, 342, Delorme, P. et al (12 authors) (2008), CFBDS J : reaching the T-Y brown dwarf transition?, Astronomy and Astrophysics, 482, Elston, R. et al (8 authors) (2003), Performance of the FLAMINGOS near-ir multi-object spectrometer and imager and plans for FLAMINGOS-2: a fully cryogenic near-ir MOS for Gemini South, Instrument Design and Performance for Optical/Infrared Ground-based Telescopes. Proceedings of the SPIE, 4841, < Faherty, J. K. et al (7 authors) (2010), The brown dwarf kinematics project. II. Details on nine wide common proper motion very low mass companions to nearby stars, The Astronomical Journal, 139, Gelino, C. R., Kirkpatrick, J. D., Burgasser, A. J., Dwarf Archives: A Compendium of M, L, and T Dwarf Data < Hodapp, K. W. et al (14 authors) (2003), The Gemini Near-Infrared Imager (NIRI), The Publications of the Astronomical Society of the Pacific, 115, Kirkpatrick, J. D. et al (10 authors) (1999), Dwarfs cooler than M : The definition of spectral type L using discoveries from the 2-Micron All-Sky Survey (2MASS), The Astrophysical Journal, 519, Lucas, P. W. et al (15 authors) (2010), Discovery of a very cool brown dwarf amongst the ten nearest stars to the Solar System, submitted to Nature on Apr 2, < Mink, D. J. (2002), "WCSTools 3.0: More Tools for Image Astrometry and Catalog Searching", Astronomical Data Analysis Software and Systems XI, ASP Conference Proceedings, 281, 169 Perryman, M. A. C. et al (19 authors) (1997), The HIPPARCOS Catalogue, Astronomy and Astrophysics, 343, L49-L52 Tokovinin, A., Thomas, S., Sterzik, M., Udry, S. (2006), Tertiary companions to close spectroscopic binaries, Astronomy and Astrophysics, 450, Zuckerman, B, Webb, R. A., Becklin, E. E., McLean, I. S., Malkan, M. A. (1997), BY Draconis is a triple star system, The Astronomical Journal, 114,