THE SOLAR NEIGHBORHOOD. XXIII. CCD PHOTOMETRIC DISTANCE ESTIMATES OF SCR TARGETS—77 M DWARF SYSTEMS WITHIN 25 pc

For more than a century, there has been serious effort invested in compiling a complete survey of stars near the Sun. Recent work of note includes Lépine's SUPERBLINK program that has detected dwarfs, giants, and subgiants in the northern sky with spectroscopic distances ⩽ 25.0 pc (Lépine 2005b) and additional objects in the south, but without spectral types or distance estimates (Lépine 2005a, 2008). Reid and collaborators have also made progress in this area, focusing on objects within 20 pc using spectrophotometry, as reported in the Meeting the Cool Neighbors papers (Reid et al. 2008, and references therein). These efforts include both previously known and new members of the solar neighborhood. The survey most similar to our survey discussed here is that of Deacon who used SuperCOSMOS I plates and Two Micron All Sky Survey (2MASS) to reveal new nearby red objects in the southern sky (Deacon et al. 2005; Deacon & Hambly 2007). This paper focuses specifically on new discoveries from our SuperCOSMOS RECONS (SCR) effort (Hambly et al. 2004; Henry et al. 2004; Subasavage et al. 2005a, 2005b; Finch et al. 2007).

The most widespread method of revealing new members in the Sun's vicinity is through proper motion searches, as nearby objects generally have larger proper motions than those further away. Much of the nearby star work in the latter half of the 20th century was based upon the Palomar and UK Schmidt sky surveys performed by Luyten (1979a, 1979b) and the Lowell proper motion surveys by Giclas et al. (1971, 1978a, 1978b). Long-term efforts to obtain optical photometry by Weis (1996) in the northern hemisphere and Eggen (1987) in the southern hemisphere were carried out by observing one star at a time. This photometry is often used to estimate distances to stars using various color–absolute magnitude calibrations. Much of these data were then incorporated into the Catalog of Nearby Stars (CNS; Gliese & Jahreiß 1991), which provided a snapshot view of the solar neighborhood population, albeit one that was incomplete. NStars continued in this vein with a comprehensive database of objects with trigonometric parallaxes (see Henry et al. 2002, 2003 for more complete discussions). As we entered the 21st century, large sky surveys such as 2MASS (Skrutskie et al. 2006), DENIS (The Denis Consortium 2005), and SDSS (Adelman-McCarthy et al. 2009) gathered magnitudes in optical and infrared filters of huge numbers of sources that could be examined for nearby star candidates.

Once a new nearby star candidate is found, distances can be determined using a variety of astrometric, photometric, and spectrophotometric techniques. In this paper, we focus on distance estimates made using photometry from a combination of optical CCDs and infrared arrays, which is an effective intermediate method for refining membership of the solar neighborhood. This method provides distances that are more accurate than those determined using photographic plates, but that are not as precise (or as time and labor intensive) as those calculated via trigonometric parallax.

A primary goal of the RECONS group (the Research Consortium on Nearby Stars⁶) is to identify hidden members of the solar neighborhood within 25 pc to match the horizons of the CNS and NStars efforts. We have focused our searches in the southern hemisphere because that portion of the sky has been historically underrepresented and is expected to yield rich treasures in the form of new stellar neighbors. In this paper, we first discuss recent photometric distance efforts by others (Section 2), then provide descriptions of our two methods used to estimate distances photometrically: via scanned photographic plates (Section 3) and improvements based on CCD photometry (Section 4), with complementary spectroscopy (Section 5). We sum up the results in Section 6.

Table 1 lists the main photometric distance estimation efforts in the southern sky since the NStars Database was compiled. The efforts considered are purely photometric and are independent of any spectral types or relations based on spectroscopy. The criteria for inclusion in the counts are that the systems (1) be new discoveries, (2) have colors of red dwarfs (effectively spectral type M, (V − K) ≳ 3.0), (3) be found in the southern hemisphere (decl. ⩽ 0°), and (4) are estimated to be within 25.0 pc via a photometric distance estimate.

We used the NStars Database as a benchmark to evaluate solar neighborhood completeness, as NStars incorporated only those objects for which a parallax had been measured; therefore, it was the most definitive collection of systems found within 25 pc. Great care was taken when extracting the sample of 329 southern M dwarf systems from NStars. All northern hemisphere objects, as well as those objects brighter than M_V = 9.00, the adopted cutoff for the brightest M dwarfs, were eliminated. For the remaining objects with no designated spectral type, various colors such as (V − I), (V − J), and (J − K), depending on the availability of photometry, were used to eliminate white dwarfs that remained in the sample. If the primary component of a multiple system was discovered to be anything other than an M dwarf, or if the system contained a white dwarf, the system was omitted. Finally, the individual components of multiple systems were combined to give a count of southern M dwarf systems.

Figure 1 shows results from these photometric distance efforts in concert with the 329 systems from NStars (small dots). Declination is plotted against right ascension in a polar plot with R.A. progressing counterclockwise from 0 to 24 hr and decl. starting at 0° at the outer boundary and moving inward to −90° at the center. Phan-Bao's 37 objects⁷ are denoted by open triangles, Reylé's 34 objects⁸ are denoted by open circles, Finch's 15 objects⁹ are noted by upside-down open triangles, and Costa & Méndez's three objects¹⁰ are indicated with open squares. The 77 systems found to date in SCR surveys (discussed in the next section) are represented by solid stars. In all cases, the counts and points in Figure 1 represent newly discovered southern systems.

Zoom In Zoom Out Reset image size

SCR objects are previously undiscovered systems that have been revealed through searches utilizing the digitally scanned SuperCOSMOS photographic plates. A sample of 72 star systems was harvested from five earlier papers in this series (Hambly et al. 2004; Henry et al. 2004; Subasavage et al. 2005a, 2005b; Finch et al. 2007). To date, an additional 28 systems have been found during continuing searches and are reported here for the first time. Table 2 lists the names, coordinates, and SuperCOSMOS B_J, R_59F, and I_IVN plate magnitudes. Additional data for these objects are found in Table 3.

The 100 SCR systems (104 objects) listed in Table 3 comprise all SCR red dwarfs with proper motions ⩾018 yr⁻¹ estimated to be within 25.0 pc based upon "plate distance estimates." These estimates use SuperCOSMOS B_J, R_59F, and I_IVN plate magnitudes (hereafter BRI) and JHK_s¹¹ magnitudes from 2MASS. The 100 systems all have M dwarf primaries, two of which are binaries with separations large enough to permit separate plate distance estimates for the secondaries, SCR2241-6119B and SCR2335-6433B. Two additional objects, SCR1107-3420B and SCR1746-8211, have CCD distance estimates that place them within 25.0 pc, but they are not included in the counts of M dwarf discoveries because the primary of the system is not an M dwarf.

In Table 3, Columns 1–3 are the names, R.A., and decl. (epoch and equinox 2000.0). Columns 4 and 5 are the proper motions and position angles as measured by SuperCOSMOS. Columns 6–9 list the VRI magnitudes and the number of nights each system was observed. The 2MASS JHK magnitudes are listed in Columns 10–12. Columns 13–18 provide the plate distance estimates, related errors, and number of color–absolute magnitude relations utilized (see next paragraph), followed by the CCD distance estimates (see Section 4.3), the corresponding errors, and number of relations utilized. The tangential velocities, based on the CCD distance estimates and the SuperCOSMOS proper motions, are given in Column 19. Column 20 lists the spectral types, where available, and Column 21 includes the references for the publications that originally reported each object, where the 29 new objects in this paper are noted with reference 6. A final Column 22 is provided for notes.

A system's plate distance estimate determines its membership in this sample and is based on up to 11 useful M_K versus color relations, as described in Hambly et al. (2004). Relevant errors in the SuperCOSMOS plate photometry for this sample, based on the ranges in BRI magnitudes for the sources, are 0.2–0.6 mag for B = 11–21, 0.1–0.3 mag for R = 9–18, and 0.2–0.4 mag for I = 8–16 (Hambly et al. 2001). The errors on the individual passbands are largely systematic in origin and are correlated between the passbands because of the calibration technique used; however, the errors on colors such as (B − R) and (R − I) are smaller, which is more important for their use in distance–color relations (see also Section 4.4).

A plate distance estimate is considered reliable if all 11 relations are applicable, i.e., if a star's color falls within the range covered by the calibrations detailed in Hambly et al. (2004) for single, main-sequence stars. Between six and ten relations indicates that the distance estimate is indicative, but less reliable. Six relations are adopted because if one plate magnitude is erroneous, up to five relations may drop out. Thus, if at least six relations are valid, then at least two of the three BRI magnitudes provide optical/infrared colors consistent with normal main-sequence stars. In the present sample, only two stars have fewer than six valid relations, so their distance estimates are highly suspect. In fact, the improved distance estimates discussed next indicate that both are beyond 50 pc.

4.1. VRI CCD Photometry

Our goal was to obtain at least two sets of V_J, R_KC and I_KC¹² (hereafter VRI) photometry for all 104 objects to improve our distance estimates. These new distance estimates are dubbed "CCD distance estimates" to reflect the combination of CCD photometry at VRI and 2MASS photometry at JHK, and all R and I magnitudes refer to CCD values for the remainder of this paper.

Care is taken during a night of photometry to ensure that previously observed objects are sandwiched between targets that are new to the program. Multiple results for a single object are then compared to ensure consistency and can also be used to indicate possible variability. The photometric observations used in making the distance estimates reported here span a seven year interval from 2003 January through 2010 May.¹³ In total, over 800 frames were acquired on the 0.9 m telescope at the Cerro Tololo Inter-American Observatory (CTIO) for this study. A small subset of 16 frames was acquired on the CTIO 1.0 m in 2005 August; results are consistent to 0.03 mag with the 0.9 m data.

All data were reduced using IRAF. Calibration frames taken at the beginning of each night were used for typical bias subtraction and dome flat fielding. Standard star fields from Graham (1982), Bessel (1990), and/or Landolt (1992, 2007) were observed multiple times each night in order to derive transformation equations and extinction curves. In order to match those used by Landolt, apertures 14'' in diameter were used to determine the stellar fluxes, except in cases where close contaminating sources needed to be deconvolved. In these cases, smaller apertures were used, and aperture corrections were applied. Further details about the data reduction procedures, transformation equations, etc., can be found in Jao et al. (2005).

Figures 2–4 illustrate the total errors versus magnitude in each of the three filters for the 268 stars with three or more nights of photometry from our program, including some of the SCR stars reported here. These errors incorporate signal-to-noise errors, nightly errors determined by fits to standard stars, and night-to-night measurement differences to provide a comprehensive assessment of our photometric errors. Signal-to-noise errors are 0.05 mag or less for objects brighter than 16th mag at V, brighter than 18th mag at R, and brighter than 17th mag at I. Standard star fits on most nights have errors of 0.01–0.02 mag. Typical variations in photometric measurements from night to night are 0.02–0.03 mag. As can be seen in Figures 2–4, the convolution of all three errors typically results in total errors of 0.03 mag or less for 66% of objects at V and 84% of objects at both R and I, as evident from the inset histograms. Note that the lack of a general increase in errors for fainter stars is caused by observers increasing integration times to provide adequate signal to noise on even the faintest targets. Nonetheless, these red dwarfs are faintest at V, so the resulting errors are somewhat larger than at R and I. At V, 94% of the objects presented have total errors of 0.06 mag or less; for R and I, this statistic is 99%. The SCR systems reported here follow the general trend reported above: the fraction of objects with errors less than 0.03 mag is 73%, 94%, and 92%, respectively, at VRI. At V, 94% of the objects have total errors less than 0.06 mag, while this is true of 100% of the objects at R and I.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.2. JHK Photometry from 2MASS

4.3. CCD Distance Estimates

The CCD distance estimates,¹⁴ as described in Henry et al. (2004), are found using a method similar to that used for the photographic plate distance estimates. The difference is that we use the accurate VRI magnitudes obtained at CTIO instead of BRI plate magnitudes from SuperCOSMOS. The maximum number of relations possible from the combination of VRIJHK magnitudes is 15 (the same number of combinations as for the plate relations), but only 12 yield useful results. The color spread is not large enough in three of the relations (M_K versus J − H, J − K, and H − K) to give reliable results and so are omitted.

Of the 100 systems with photographic plate distance estimates that place them closer than 25.0 pc, we find 77 to have CCD distance estimates within the 25.0 pc horizon. Thus, we retain 77% of the SuperCOSMOS discoveries as likely nearby stars. All systems with CCD distance estimates within 15.0 pc and additional compelling systems are targeted for trigonometric parallax measurements as part of our Cerro Tololo Inter-American Observatory Parallax Investigation (CTIOPI) (Jao et al. 2005; Henry et al. 2006; Subasavage et al. 2009; Riedel et al. 2010). At present, 31 of the systems discussed here are on the CTIOPI program.

Figure 5 shows a comparison of the two photometric distance estimates with error bars omitted for clarity. Primary stars and a few secondaries for which separate photometry is available are plotted. The single point in the upper right of the plot and one point with a plate distance estimate of 25–30 pc are secondary components (SCR2241-6119B and SCR2335-6433B). Twenty-three systems are estimated to be beyond 25 pc using CCD distance estimates. The two most likely causes of the distance offsets are poor plate photometry compared to CCD photometry and close multiples that are unresolved on the plates but resolved in CCD images. Errors in the plate photometry certainly cause most of the discrepancies. A distance of up to ∼35 pc is possible if the object is an unresolved binary in which the two components are identical and contribute an equal amount of light to the measured magnitude. All SCR targets have been inspected for duplicity in the CCD images. Seven were found to be binaries with separations between 1'' (typical resolution of our CCD images) and 5''. The typical resolution of the plate images for binaries with magnitude differences of ≲ 2 is 4''. Details on multiple systems can be found in Section 5.1.

Zoom In Zoom Out Reset image size

Eight objects are found to have CCD distance estimates of 35–65 pc. One object (SCR2335-6433B in the upper right of Figures 5 and 6) is a common proper motion companion that was discovered by eye when blinking plates. The remaining seven objects had 3–8 plate relations contributing to their distance estimates rather than the full suite of 11, indicative of inaccurate estimates. The corresponding CCD distances for these objects often rely on 6–10 relations rather than the full suite of 12, typically because they are bluer than some of the plate and CCD distance relations. Three of the eight objects have already been verified to be K dwarfs via spectroscopy.

Zoom In Zoom Out Reset image size

4.4. Photometric Distance Estimate Errors

When using both the plate and CCD distance estimating suites, there is inherent spread around the fit line for each relation that is a result of two causes: (1) the deviations due to errors in parallax and photometry for the stars used to derive the fits, and (2) cosmic scatter due to differences in stellar ages, compositions, and perhaps magnetic properties. Cosmic scatter dominates the offsets between distance estimates and true distances. Unresolved, unknown multiples may also contribute, although great care was taken to remove such systems before the fits were made, so this is not a significant issue. In order to estimate the reliability of the suites of relations in both the plate and CCD methods, we have run single, main-sequence, red dwarfs with known trigonometric distances back through the relations to derive representative errors. The median offsets between the trigonometric and photometric distance estimates are 26% for the plate suite and 15% for the CCD suite.

For all errors on distance estimates listed in Table 3, these percentage errors are combined in quadrature with each star's individual error, which comes from the up to 11 or 12 distances derived from the plate and CCD suites, respectively. Figure 6 is the same as Figure 5, but with the total errors shown for both estimates. Note that in no case can the plate distance estimate have an error less than 26% nor less than 15% for the CCD distance estimate.

Figure 7 illustrates a comparison of the SuperCOSMOS plate (R − I) color and the CCD (R − I) color for 100 SCR stars reported here. The fit is a polynomial described by (R − I)_ccd = 0.640 + 0.192*(R − I)_plate + 0.198*(R − I)²_plate − 0.035*(R − I)³_plate. This polynomial permits users of SuperCOSMOS data to predict the (R − I)_ccd color that would be measured for red objects, i.e., those with −0.2 = (R − I)_plate = 3.8.

Zoom In Zoom Out Reset image size

Two outlying points represented by open squares are SCR1448-5735 and SCR1637-4703, both of which were omitted from the fit because they are blended with other sources in R images. Two other objects for which I plate magnitudes were not available (SCR1931-0306 and SCR2230-5244) were also excluded from the fit. The mean absolute deviation is 0.27 mag, indicating that the SuperCOSMOS (R − I) color is roughly nine times less accurate than the CCD (R − I) color, which typically has errors less than 0.03 mag.

For 37 stars, we have spectra that were acquired on the CTIO 1.5 m between 2003 July and 2006 December. Observations were made using a 20 slit or wider in the RC Spectrograph with grating 32 and order blocking filter OG570 to provide wavelength coverage from 6000 to 9500 Å and resolution of 8.6 Å on the Loral 1200 × 800 CCD. Reductions were performed using standard procedures via IRAF. Spectral types have been assigned on the RECONS system (Jao et al. 2008). Thirty-four of the stars are confirmed to be M dwarfs, while three are K dwarfs. The three K dwarfs have slipped into the sample because of erroneous plate magnitudes, as evidenced by having fewer than 11 relations for their plate distance estimates.

For all of the stars, we use a color–color diagram to verify that the sample objects are red dwarfs. Figure 8 is a color–color diagram that plots (V − K) versus (J − K). Labels denoting spectral types are shown inside the left axis. SCR stars are open circles, with the combined photometry of seven SCR close binaries marked by crosses inside the circles. The dots are RECONS 10 pc sample members (Henry et al. 2006), and a few known giants are plotted as solid triangles for reference. As is evident from this diagram, none of the SCR stars presented in this paper fall in the region where known giants are found, so we are relatively confident that none of these objects are giants. We anticipate that these objects are not subdwarfs because their tangential velocities are ⩽100 km s⁻¹. The point that is slightly elevated (1.02, 5.90) toward the three giants is the binary system SCR1630-3633AB.

Zoom In Zoom Out Reset image size

5.1. Multiples

5.2. Objects Worthy of Note

The majority of the objects (71 of 104, or 68% of the sample) have V> 14 because these new discoveries are fainter than those targeted by Luyten in most of the southern sky. Only the faint companion object discovered by eye during the blinking process, SCR2241-6119B (R= 17.23), is fainter than CCD R= 16, a consequence of the plate R= 16.5 cutoff chosen in our searches to date. The color–color diagram in Figure 8 shows that most of the 104 SCR objects are M3−M6 type dwarfs, which corresponds to the "sweet spot" created by our search magnitudes through most of the 25 pc volume.

As can be seen in the polar plot of Figure 1, many of these SCR discoveries are south of −60°. Luyten's Palomar survey for proper motion stars extended to −45° for a quarter of the southern sky and to −33° for the entire southern sky (Luyten 1974), and his Bruce Survey covered the remainder of the southern sky. Giclas searched declinations north of −45° for stars of magnitudes 8–17 with proper motions ⩾0.20 and at even smaller μ for red objects (Giclas et al. 1978a). Still, we have found 11 systems between decl. = 0° and −30°, an area of the sky covered previously by both Luyten and Giclas. Four of these 11 are brighter than V= 14 and so were missed by those surveys.

In total, the 77 systems we determine to be within 25 pc, when compared to the 329 southern systems found in NStars, result in a 23% increase in the number of red dwarf systems nearer than 25.0 pc in the southern sky. In Table 5, we summarize the distance distribution for the 100 systems discussed in this paper, as well as the number of binaries found within each distance bin. We anticipate that seven will prove to be within 10 pc and 70 will be between 10 and 25 pc, while the remaining 23 systems are likely beyond 25 pc. If all of the new red dwarf systems from our SCR search to date, plus the 89 systems found by Phan-Bao et al., Reylé et al., Finch et al., and Costa & Méndez are found to have trigonometric parallaxes placing them within 25 pc, the number of red dwarf systems in the southern sky within that horizon would increase by 50%. Forthcoming in this series is a paper presenting trigonometric parallaxes for roughly one-third of the SCR systems in this paper, three of which (SCR0630-7643AB, SCR1138-7721, SCR1845-6357AB) have had results published previously (Henry et al. 2006).

We have several continuing efforts to reveal the Sun's red dwarf neighbors. We have found another ∼100 systems with 040 yr⁻¹> μ ⩾ 018 yr⁻¹ between decl. = 0° and −47° during our continuing sweep of the southern sky. This search is a complement to our Finch et al. (2007) effort and will complete the SCR search of the southern sky for stars that would have been members of Luyten's New Luyten Two-Tenths (NLTT) Catalogue (Luyten 1979b). Already underway is a project to gather photometric data for ∼150 additional SCR systems that have photographic plate distance estimates placing them between 25 and 30 pc. Some of these systems will, in effect, replace the 23 systems with CCD distances beyond the 25 pc original limit of the plate distances. Already, of the 38 systems for which we have obtained initial VRI photometry, nine are anticipated to be nearer than the 25 pc horizon. Still remaining are additional searches that probe beyond the R = 16.5 cutoff of our first sweep of the southern sky.

The support of additional RECONS team members was vital for this research, in particular Adric Riedel. We also thank Brian Mason for insight on preliminary optical speckle survey results, as well as Ed Nelan for early results from HST-FGS. This work has been supported under National Science Foundation grants AST 05-07711 and AST 09-08402, as well as by NASA's Space Interferometry Mission, and Georgia State University. We also thank the members of the SMARTS Consortium, who enable the operations of the small telescopes at CTIO, as well as the observer support at CTIO, especially Edgardo Cosgrove, Arturo Gomez, Alberto Miranda, and Joselino Vasquez. This research makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research also makes use of the SIMBAD database and the Aladin and Vizier interfaces, operated at CDS, Strasbourg, France. Thank you to the referees for their helpful comments, as well.