Free Access
Issue
A&A
Volume 560, December 2013
Article Number A21
Number of page(s) 11
Section Galactic structure, stellar clusters and populations
DOI https://doi.org/10.1051/0004-6361/201321045
Published online 29 November 2013

© ESO, 2013

1. Introduction

The direction toward the inner Milky Way presents a formidable challenge for proper motion (PM) studies because of the crowding and confusion (for previous attempts see Lépine et al. 2002b; Lépine 2008. VISTA Variables in the Via Lactea (VVV) is a new ESO Public survey (Minniti et al. 2010, 2012a; Saito et al. 2012a) that may help to alleviate these problems. The VVV is carried out with the Visible and Infrared Survey Telescope for Astronomy (VISTA; Dalton et al. 2006; Emerson et al. 2006) at Paranal Observatory, and will obtain ZYJHKS coverage and multi-epoch (up to 100 for some pointings) KS observations of ~562 deg2 in the Milky Way bulge and inner disk at sub-arcsec seeing. After two years of operation, we have already demonstrated, that VVV is producing new, interesting results: the discovery of new star clusters (Minniti et al. 2011a, Borissova et al. 2011; Moni Bidin et al. 2011), investigation of the structure and stellar populations content of the Milky Way (Minniti et al. 2011b; Gonzalez et al. 2011a,b, 2012; Saito et al. 2012c), the study of variable stars and transients (Catelan et al. 2011; Saito et al. 2012b), and others. One of the main goals is to obtain a three-dimensional tomographic map of the Milky Way bulge based on red clump giants, RR Lyr, and Cepheid variables. However, some corollary science objectives are also considered, including a PM study, that take advantage of the projected ≥5 yr survey duration. Early PM science with VVV is possible if it is used as a second epoch to a previous infrared survey, the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006). The 2MASS observations provide ≥10 yr baseline.

The VVV footprint on the sky is relatively small, ~20 deg × 15 deg centered on the bulge, and ~55 deg × 4.5 deg along the adjacent southern disk, or just above 1% of the total sky, but it encompasses the regions with the highest stellar surface density in the Galaxy. This study is based on the available multi-filter imaging that was taken during the first VVV observing season, covering ~500 deg2. The typical image quality is 0.8–1.0 arcsec, and the pixel scale is ~0.34 arcsec pix-1, which compare favorably to other Galactic surveys. The final VVV data products will be a ZYJHKS atlas and a catalog of ~109 sources, ~106 of which are variables1.

thumbnail Fig. 1

Examples for the co-moving companion search. North is up and east is left. The red images are 2MASS KS, and the blue images are VVV KS. The objects of interest are circled. From left to right: LP 922-16 A and B (~2×2 arcmin), LHS 3188 A and B (~1.5×1.5 arcmin), and C* 1936 (~1 × 1 arcmin).

Open with DEXTER

Table 1

High proper motion stars in the VVV area.

We embarked on a project to improve the solar neighborhood census by searching for common PM companions to known nearby high proper motion (HPM) stars. Our effort has the potential to improve the local stellar multiplicity fraction estimate, a key constraint to star formation theories, with implications for the stellar population modeling of unresolved stellar systems. We were driven by the argument that relatively bright new solar neighborhood stars could be found only in a survey covering the densest regions of the Milky Way, like VVV, because these stars far away from the Galactic plane were easy to discover with the previous generation of surveys. We build upon the success of the RECONS PM and parallax measurements project (Finch et al. 2007; Henry et al. 2004; Jao et al. 2009; Subasavage et al. 2009), the work of Léipne and collaborators (Lépine et al. 2002a; Lépine & Bongiorno 2007), Raghavan et al. (2010), Faherty et al. (2010), Allen et al. (2012), and others, with the advantage that VVV has better spatial resolution and higher sensitivity to low mass red objects than the previous surveys.

The paper is organized as follows: Sect. 2 describes the sample, Sect. 3 summarizes the search method, the follow up spectroscopy is reported in Sect. 4, and Sect. 5 gives the results.

2. Sample selection

The sample was selected with SIMBAD, and it includes all stars with PM ≥ 200 mas yr-1, implying a total movement of ≥2 arcsec (≥6 VIRCAM pixels) over the ~10-year interval separating 2MASS and VVV. Our previous experience shows that movements of this magnitude are easily detected. Some of the stars are saturated on the VVV images, but this still allows us to search for fainter co-moving companions because the motions of the “doughnuts” with burned out cores are still discernible.

The VVV survey plan for the first year envisioned separate visits of each point of the survey area for ZY and JHKS observations, and up to six visits in KS, on separate nights, for variability studies. At the time we carried out this study, 144 out of 152 disk tiles and 188 out of 196 bulge tiles were completed, covering ~216 and ~282 deg2, respectively.

We arrived at a target list of 167 objects: 67 in the disk and 100 in the bulge, for which the VVV can provide a new epoch of observations (Table 1). The stars come from the catalogs of LHS (Luyten 1979a), LTT (Luyten 1957, 1961, 1962), MACHO (Alcock et al. 2001), NLTT (Luyten 1979b,c, 1980), OGLE (Sumi et al. 2004), and several other works: Finch et al. (2007), Lépine (2005), Rattenbury & Mao (2008), Subasavage et al. (2005), Terzan et al. (1980).

The selection is dominated by nearby dwarfs (SIMBAD lists: 8 F, 19 G, 24 K, and 22 M-types). The distances to the few stars with parallaxes range from ~1.3 to ~136 pc, with a median of ~44 pc.

The sample is heterogeneous, subjected to different biases, and the PMs from different catalogs have different error budgets, so our work cannot present a basis for strict statistical studies of the solar neighborhood. This question will be addressed after the completion of VVV, when it has generated a long baseline coverage with self-consistent data.

3. Analysis

We visually searched around known HPM stars on false three-color images, generated combining the reddest available POSSII band, and J bands from 2MASS and VVV (Fig. 1). Three field sizes were used to provide different levels of zoom into the vicinity of program stars: 0.9, 1.8, and 3.6 arcmin, centered on each candidate. Inspecting larger images was found to be impractical.

For each object in our sample we performed the following steps. First, we identified the known HPM star from its coordinates and the apparent change of position between the epochs of the surveys used to create the false three-color images. The large PM made these identifications unambiguous. Second, we selected candidate companions looking by eye for other stars in the field with similar apparent motion to the known HPM star. Some candidates were discarded later, after we calculated their PMs (see below) and found them inconsistent with the PMs of the known HPM star. In a few cases the candidate companions were not fully resolved on the older images. Then, we used the non-circular PSF of older surveys as an argument supporting the companionship (Fig. 1, middle). Finally, we inspected the selected candidates on three-color images built from VVV JHKS data to minimize missidentification, for example from extreme colors or artifacts. The subjective nature of this procedure, together with the varying point spread function (PSF) are difficult to quantify, making our results unsuitable for a rigorous statistical constraint on the completeness of HPM stars.

The astrometric calibration of VVV data is based on hundreds of 2MASS stars that fall onto each tile (Irwin et al. 2004; Minniti et al. 2010). This procedure removes the systematic bulk motion of the unmoving background stars between the VVV and the 2MASS epochs. Therefore, we directly compared 2MASS and VVV coordinates, to measure PMs. This makes our PMs relative in nature, because the 2MASS reference stars that were used to derive the astrometric solution for the VVV have some average common motion that remains unaccounted for.

We measured PMs only for the new co-moving HPM candidates, for their hosts from the known HPM star list, and for the stars from the HPM list that appeared to move with much slower PM that given in the literature (Tables 2, and 3). We calculated the stellar positions as unweighted centroids. The cores of the bright stars (KS ≤12 mag) are saturated, and to investigate the effect of the saturation, we set to zero the central pixels that are above 60% of the saturation limit for 50 stars below the saturation limit. The result was much stronger than the typical saturation effect for the stars in our sample. The differences of the coordinates with and without saturation was 0.03 ± 0.03 arcsec; in other words, the wings of the images are sufficient to measure the stellar positions accurately.

The final PMs are simple arithmetic averages of the PMs determined between 2MASS and various VVV observations, and the error is the rms of the measurements, if more than three are available (Tables 4 and 5). Adding older photographic epochs usually worsens the fit because of crowding and contamination. The 2MASS sets the faint magnitude limit of our new HPM candidates, and the minimum primary–companions separation: J ≤ 16 mag and d ≥ 1.5–1.7 arcsec, respectively Skrutskie et al. 2006. The maximum separation was determined by the size of the cut outs.

After inspecting the 167 objects in our sample (67 in the disk and 100 in the bulge),

  • (1)

    seven new co-moving companions to bright(J ≤ 16 mag) HPM stars with PM ≤ 200 mas yr-1 were found: L 149-77, LHS 2881, L 200-41, LHS 3188, LP 487-4, LHS 5333, and LP 922-16; particularly notable is the discovery of a low-mass M5V companion to LHS 3188, at ~21 pc from the Sun;

  • (2)

    six known co-moving binaries were recovered: LTT 5140 A + LTT 5140 B, L 412−3 + L 412−4, LTT 6990 A + LTT 6990 B, GJ 2136 A + GJ 2136 B, LP 920−25 + LP 920−26, and MACHO 124.22158.2900 + MACHO 124.22158.2910;

  • (3)

    LTT 7318 and LTT 7319, that were considered co-moving stars, appeared not to be;

  • (4)

    the PMs of all co-moving pairs of HPM stars in our sample (Table 4), and of the stars with previously overestimated PMs (Table 5), were measured;

  • (5)

    spectral types ranging from G8V to M5V of seventeen members of the co-moving pairs were determined from new near-infrared spectroscopy (Table 6);

  • (6)

    HPMs of eight stars (C* 1925, C* 1930, C* 1936, CD−60 4613, LP 866−17, OGLE BUL−SC20 625107, OGLE BUL−SC21 298351, and OGLE BUL−SC32 388121) reported in at least some previous works appear to have been grossly overestimated.

4. Follow-up spectroscopic observations

Near-infrared spectra of co-moving pairs were obtained to determine their spectral types at the ESO NTT with SofI (Son of ISAAC; Moorwood et al. 1998) in two low-resolution modes, with blue (λ = 0.95−1.64  μm) and red (λ = 1.53−2.52  μm) grisms to cover the entire near-infrared spectral range. The slit was 1 arcsec wide during the Apr 2011 run, and 0.6 arcsec during the May 2012 run, delivering an average resolution of R ~ 600 and ~1000, respectively. It was aligned along the axis connecting the two candidate companions, except if their apparent magnitudes were too different–in which case they were observed separately. Typically, four (six in the case of the relatively faint MACHO 124.22158.2900–MACHO 124.22158.2910 binary candidate) images were obtained, into a two-nodding ABBA or ABBAAB sequence, with nodding of 30–60 arcsec. Each image constituted 48–1050 s of integration, averaged over 3–12 individual detector integrations to ensure the peak values are well below the non-linearity limit of the detector (Table 6). The atmospheric conditions varied during the observations, but most often they were mediocre, with a seeing above 1.5 arcsec, thin to thick cirrus, because these targets were poor weather fillers which accounts for somewhat longer than usual integration times.

The data reduction steps were: (i) sky/dark/bias removal by subtracting from each other the two complementary images in a nodding pair; (ii) flat fielding with dome flats; (iii) extraction of one-dimensional (1D) spectra from each star, on each individual image, by tracing the stellar continuum with 6–8 pixel (1 pixel~0.29 arcsec) wide apertures, with the IRAF2 task apall; (iv) wavelength calibration of each 1D stellar spectrum with 1D Xenon lamp spectrum, extracted from Xenon lamp images with the same trace as each target spectrum; (v) combination of the four or six 1D spectra of each star in wavelength space with the IRAF task scombine, with appropriate masking or rejection of remaining detector artifacts and cosmic ray affected regions; (vi) telluric correction with spectra of near-solar analogs (G1V-G3V), observed just before or after the science target, at similar airmass, and reduced the same way; and (vii) recovery of the original spectral shape and removal of the artificial emission lines Maiolino et al. (1996) by multiplying with spectra of corresponding spectral type star from the flux-calibrated IRTF library (Cushing et al. 2005; Rayner et al. 2009).

The signal-to-noise of the final spectra varies significantly with the target’s brightness and with wavelength, but the areas clear from telluric absorption have S/N ~ 10–30. The final spectra are plotted in Fig. 2.

The spectral typing was performed comparing the overall shape of the SofI spectra with spectra from the IRTF library (Fig. 3) and the results are listed in Table 6. The typical uncertainty, estimated from a comparison with template stars of neighboring subtypes (Fig. 3), is one subtype. It was determined by comparing our targets with IRTF spectra of stars with close subtypes, and comparing stars with multiple IRTF observations. Finally, we corrected for telluric absorption the telluric standard HIP 084636 with HIP 098813, and re-determined its spectral type obtaining a best match with G2V star, to be compared with G3V reported by Gray et al. (2006).

thumbnail Fig. 2

Near-infrared spectra of our targets. The spectral areas with poor atmospheric transmission are omitted. The spectra were normalized to 0.5 in the overlapping region (bracketed with dotted lines) and shifted vertically by 0.5 for clarity. M124A and M124B indicate MACHO 124.22158.2900 and MACHO 124.22158.2910, respectively; L 412−3 B is an alternative notation of HD 322416 and LP 920−25 B is the same for LP 920−26. The spectrum of the telluric HIP 084636 shown here was corrected for the atmospheric absorption with HIP 098813.

Open with DEXTER

thumbnail Fig. 3

Spectral classification example. Solid lines show our spectra, and dotted lines-the template spectra from the IRTF library (Cushing et al. 2005; Rayner et al. 2009).

Open with DEXTER

5. Discussion and summary

Why have the new HPM stars not been detected before? Some of them appear on old photographic surveys but the contamination from nearby stars, worsened by the poor spatial resolution of those surveys, makes the identification of the stars as HPM objects difficult. The extreme differences between optical and infrared brightness of stars that is often found in the Galactic plane often led to misidentifications and some spurious HPM detections while true HPM stars were missed. Even with the high-quality of the VVV data we cannot consider the position of the stars reliable because of the uneven background. Multiple measurements are needed, separated by some years, to let the stars move by at least 2–3 arcsec, so they lay on a completely different background, averaging out the contamination effects.

The PM errors in Table 4 are the rms for three or four measurements (Table 2), and they only include statistical uncertainties. A comparison with the measurements in the literature suggest that the real uncertainties are larger. Excluding the obvious errors which yield differences exceeding 100 mas yr-1, due to missidentification, for example we find an average difference of 2 mas yr-1, with an rms of 17 mas yr-1, and we suggest that the reader use the second number as the real error of our PMs that includes both internal and external uncertainties. The scatter gives an upper limit to the unaccounted bulk PM of the filed stars used for the astrometric calibration of the VVV data, and these are indeed small. More accurate measurements will become available in the future as the VVV survey progresses. The planned survey duration of five years is likely to be extended to seven years.

Some of the objects with overestimated PMs are very red. Interestingly, three of them were considered HPM objects despite being classified as Carbon stars, suggesting that they were giants. Unaccounted astrometric color terms, combined with the extreme colors, have most likely led to the erroneous classification.

Table 5

List of stars with overestimated PMs.

Table 6

Details of the IR spectroscopic observations and derived spectral types.

Notes on some individual objects:

  • LHS 2881 B is ~8.1 arcsec away from a HPM object listed in Monet et al. (2003) with μ(RA) = 198 ± 38 and μ(Dec) = 848 ± 319 arcsec yr-1 which is absent in our data and it is likely a result of a missidentification or a spurious entry in the USNOB1.0. Interestingly, the LHS 2881 pair has a similar PM to that of LHS 2871: μ(RA) = −461.01 ± 1.67 and μ(Dec) =  −645.32 ± 1.31 arcsec yr-1, as reported by van Leeuwen (2007). The wide separation of ~44 arcmin makes it unlikely that they are bound, but may indicate a common origin.

  • LP 487-4 is projected on the sky close to the open cluster NGC 6475 (M7), but it is not a physical member because the cluster has μ(RA) = 2.58 and μ(Dec) = −4.54 arcsec yr-1 (Loktin & Beshenov 2003). Furthermore, the optical spectroscopy of James et al. (2000) yields a radial velocity Vrad = 78.6 ± 0.2 km s-1, inconsistent with Vrad =  −14.21 ± 1.39 km s-1 of NGC 6475 (Kharchenko et al. 2005).

  • LTT 5140 A parameters were derived from optical spectroscopy and Strömgren photometry by Nordström et al. (2004): log   Teff = 3.785, [Fe/H] = 0.04, MV = 3.67 mag, Age = 3.3 Gyr Vrad = 15.9 ± 0.2 km s-1. Later, Holmberg et al. (2009) updated them to log   Teff = 3.774, [Fe/H] = −0.06, and MV = 3.63 mag to reflect the revised Hipparcos parallaxes. Desidera et al. (2006) estimated from chromospheric activity log   age = 9.82 and 9.58 for the primary and the secondary, respectively.

  • LTT 7318 and LTT 7319 were considered a binary by Dommanget (1983), but later measurements by Salim & Gould (2003) and van Leeuwen (2007) indicate that the two stars are not physically connected. Our data support this conclusion.

  • Some objects were included in our sample just because one source, namely Monet et al. (2003) reported HPM for them, even though other works have estimated low PM. For example, C* 1925, C* 1930, and C* 1936, which are known carbon stars, i.e., distant giants, as reported by Alksnis et al. (2001).

  • CD−60 4613 was considered a HPM star by Turon et al. (1992), but it was probably misidentified with the nearby LTT 5126 because of the large error in the NLTT coordinates of that star reported by Salim & Gould (2003). Indeed, van Leeuwen (2007) reported correct position and low PM for this star in the revised Hipparcos catalog under HIP 65056.

  • the HPMs for OGLE BUL−SC20 625107, OGLE BUL−SC21 298351, and OGLE BUL−SC32 388121 are subject to various sources of systematics, blending, contamination from variable sources, and seeing variations, worsened by the crowded OGLE fields (see Sect. 7 in Sumi et al. 2004).

High proper motion stars are nearby objects, and finding seven of them implies an incompleteness of ~4% (over 167 HPM stars) in the solar neighborhood census. However, this is not a firm limit because (i) it refers only to the bright stars considered here; (ii) the starting list of 167 stars is probably incomplete; and (iii) it is contaminated by non-moving stars, as we showed. Therefore, we refrain from making statements on the completeness of the solar neighborhood census; we only demonstrated that the HPM census is lacking stars, and that high angular resolution surveys help to address this issue in the most crowded regions of the Galaxy.

The new generation of near-infrared surveys of the Milky Way will produce enormous amounts of data, allowing the possibility of many discoveries. This work allows us to refine the strategy for future surveys and HPM star searches in the densest regions of the Southern Milky Way disk and the bulge. We expect that many more HPM stars and companions to them–including brown dwarfs and even planetary mass objects-will be discovered by these surveys when the baseline of observations reaches a few years, helping to complete the census of faint nearby stars, and their multiplicity.


1

For further details see the VVV web page at: http://vvvsurvey.org

2

IRAF is distributed by the NOAO, which is operated by the AURA under cooperative agreement with NSF.

Acknowledgments

We acknowledge support by the FONDAP Center for Astrophysics 15010003; BASAL CATA Center for Astrophysics and Associated Technologies PFB-06; the Ministry for the Economy, Development, and Tourism’s Programa Iniciativa Científica Milenio through grant P07-021-F, awarded to The Milky Way Millennium Nucleus; FONDECYT grants No. 1090213 and 1110326 from CONICYT, and the European Southern Observatory. J.C.B. acknowledge support from a Ph.D. Fellowship from CONICYT. M.G. is financed by the GEMINI-CONICYT Fund, allocated to the project 32110014. R.K. acknowledges partial support from FONDECYT through grant 1130140. E.L.M. acknowledges support from grant AyA2011-30147-C03-03; J.B. acknowledge support from FONDECYT No. 1120601; A.N.C. acknowledges support from GEMINI-CONICYT No. 32110005 and from Comitee Mixto ESO-GOBIERNO DE CHILE. J.A.G. acknowledges support from Proyecto Fondecyt Postdoctoral 3130552, Fondecyt Regular 1110326, and Anillos ACT-86. We gratefully acknowledge use of data from the ESO VISTA telescope, and data products from the Cambridge Astronomical Survey Unit. We have also made extensive use of the SIMBAD Database at CDS Strasbourg, of the 2MASS, which is a joint project of the University of Massachusetts and IPAC/CALTECH, funded by NASA and NSF, and of the VizieR catalogue access tool, CDS, Strasbourg, France. Last but not least, we thank the anonymous referee for the thoughtful and helpful comments that greatly improved the scientific content of paper.

References

Online material

Table 2

Multi-epoch observations of target stars.

Table 3

Multi-epoch observations of stars with over-estimated PMs.

Table 4

Measured PMs for new, known, and rejected co-moving pairs of stars.

All Tables

Table 1

High proper motion stars in the VVV area.

Table 5

List of stars with overestimated PMs.

Table 6

Details of the IR spectroscopic observations and derived spectral types.

Table 2

Multi-epoch observations of target stars.

Table 3

Multi-epoch observations of stars with over-estimated PMs.

Table 4

Measured PMs for new, known, and rejected co-moving pairs of stars.

All Figures

thumbnail Fig. 1

Examples for the co-moving companion search. North is up and east is left. The red images are 2MASS KS, and the blue images are VVV KS. The objects of interest are circled. From left to right: LP 922-16 A and B (~2×2 arcmin), LHS 3188 A and B (~1.5×1.5 arcmin), and C* 1936 (~1 × 1 arcmin).

Open with DEXTER
In the text
thumbnail Fig. 2

Near-infrared spectra of our targets. The spectral areas with poor atmospheric transmission are omitted. The spectra were normalized to 0.5 in the overlapping region (bracketed with dotted lines) and shifted vertically by 0.5 for clarity. M124A and M124B indicate MACHO 124.22158.2900 and MACHO 124.22158.2910, respectively; L 412−3 B is an alternative notation of HD 322416 and LP 920−25 B is the same for LP 920−26. The spectrum of the telluric HIP 084636 shown here was corrected for the atmospheric absorption with HIP 098813.

Open with DEXTER
In the text
thumbnail Fig. 3

Spectral classification example. Solid lines show our spectra, and dotted lines-the template spectra from the IRTF library (Cushing et al. 2005; Rayner et al. 2009).

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.