The four telescopes used for the imaging campaign are one-meter class telescopesa and have
CCD cameras that provide similar pixel scales. As listed in Table 2.1, the CTIO/SMARTS 0.9m has pixels 0′′.401 in size, the CTIO/SMARTS 1.0m has 0′′.289 pixels, Lowell’s 42in has 0′′.327 pixels, and the USNO 40in has 0′′.67 pixels. While the 0.9m has a slightly larger pixel scale than the 1.0m and the 42in, the seeing was typically better at that site, allowing for better resolution, as shown in Figure 7.1. Only 22 (fewer than 2% of the survey) were observed at the USNO 40in, so we do not consider the coarser pixel scale to affect the overall survey in any significant way. Therefore, it was felt that overall, the data from the four telescopes used was of similar quality and that the results could be combined without
a
132
modification. We note that all data were acquired without binning pixels.
Figure 7.1 The seeing conditions experienced for the observations of the multiplicity sample in the I−band at each of the telescopes used. The four different telescopes are symbolized by the different colors: royal blue for the CTIO/SMARTS 0.9m, dark green for the Lowell 42in, cyan for the CTIO/SMARTS 1.0m, and bright green for the USNO 40in.
7.2.2 Photometry
The long history of published absolute photometry from the CTIO/SMARTS 0.9m as a by-product of the astrometry program (Winters et al. 2011, 2015, and citations in §3.2) is a testament to the quality of the data obtained with that telescope and instrument set-up. But it is important to ensure that the photometric data from the 42in and the 0.9m agree. Four targets were observed for photometry at both locations. Results match to 0.06 mag, except
for the R−band magnitude of GJ1167, as shown in Table 7.1, which can be attributed to a possible flare event observed at the time of observation at the 42in, as theV andImagnitudes are consistent. This object is in fact included in a flare star catalog of UV Cet-type variables (Gershberg et al. 1999). An additional six targets had published high quality photometry from the literature to which results from the Lowell 42in were comparedb. The photometry
matches to within 0.08 mags for all six objects. As described in Winters et al. (2015), our typical 1σ errors are 0.03 mag, so the comparison of data presented here have differences of 2σ or less in 28 of the 30 cases shown in Table 7.1. No absolute photometry data were obtained from either the SMARTS/CTIO 1.0m or the USNO 40in telescopes.
Table 7.1: Overlapping Photometry Data
Name (V −K) VJ RKC IKC # obs tel/ref
2MA0738+2400 4.86 12.98 11.81 10.35 1 42in 12.98 11.83 10.35 2 0.9m G043-002 4.76 13.23 12.08 10.67 1 42in 13.24 12.07 10.66 2 0.9m 2MA1113+1025 5.34 14.55 13.27 11.63 1 42in 14.50 13.21 11.59 2 0.9m GJ1167 5.59 14.16 12.67 11.10 1 42in 14.20 12.82 11.11 1 0.9m LTT17095A 4.22 11.12 10.12 9.00 1 42in 11.11 10.11 8.94 ... Weis (1993) GJ0015B 5.12 11.07 9.82 8.34 2 42in 11.06 9.83 8.26 ... Weis (1996) GJ0507AC 3.96 9.52 8.56 7.55 1 42in 9.52 8.58 7.55 ... Weis (1996) GJ0507B 4.64 12.15 11.06 9.66 1 42in 12.12 11.03 9.65 ... Weis (1996) GJ0617A 3.64 8.59 7.68 6.85 1 42in 8.60 7.72 6.86 ... Weis (1996) GJ0617B 4.67 10.74 9.67 8.29 1 42in 10.71 9.63 8.25 ... Weis (1994) b
We note again that all photometry from the literature has been converted to the JKC system, where applicable.
134
7.3 Detection Limits
It is a difficult but important task to the quantify the subjectivity contributed by human analysis. Outlined here are the detection limits for both the blinking and imaging campaigns, where an effort is made to account for the human element of the searches.
7.3.1 Blinking Search
The common proper motion (CPM) search had two elements that needed to be evaluated in order to confidently identify objects moving with the primary star in question: companion brightness and magnitude of each system’s proper motion.
As archival SuperCOSMOS plates were used for the blinking campaign, it is important to determine which companions would have been bright enough to be detected on the plates during this part of the search. A companion would have to have been detectable on at least two of the three plates in order to notice its proper motion, so any companion would need to be brighter than the magnitude limits given in Table 5.3. As the search is for stellar
companions, it is only necessary to be able to detect a companion as faint as the faintest object in the sample, effectively an M9.5 V at 25 pc.
The faintest target in the sample in both the V− and R−bands, RG0050.5-2722, has
V = 21.54, R = 19.19, and I = 16.70. The B magnitude for this object will therefore be much fainter than the∼20.5 faintness limit of theB plate, and thus RG0050.5-2722 was not found on the B plate; however, it was identified on both the R and I plates, as its R and
faint to be seen on theB plate, but as is the case with RG0050.5-2722, all are bright enough for detection on the R and I plates. The faintest target in the I−band is DEN0909-0658, with I = 17.18, well within the magnitude limits of the SuperCOSMOS I plate. (This star is slightly bluer than RG0050.5-2722, and thus its V magnitude is slightly brighter.)
As the sample has no lower limit to the proper motion, the epoch spread between the plates needed to be large enough to detect the primary star moving in order to then notice a companion moving in tandem with it. Fifty-eight of the objects studied have µ less than Luyten’s historical cut-off of 0′′.18 yr−1; the slowest hasµ= 0′′.03 yr−1. In order to detect this
slowest object moving by 1′′.5, the epoch spread would need to be 50 years. The three main SuperCOSMOS plates used provide coverage of only up to 28 years, so in eighteen cases, the POSS-I plate was used for primaries with Declination −20 < δ < +5◦. This extended the
epoch spread by eight to twenty-four years, enabling the movement of these eighteen objects to be perceived. As noted in the histogram ofµshown in Figure 7.2, these 58 slower-moving objects with µ < 0′′.18 yr−1 make up only∼5% of the sample.
The motions of an additional 150 primaries were not able to confidently be detected due to the epoch spread of the plates being less than 5 years. These will be compared to the
I−band images taken during the imaging portion of the survey in the near future. Because this is the CPM part of the survey, these companions would have angular separations greater than 10′′, the separation subset that has a multiplicity rate of 4.6% (as discussed in§6.2 and
shown in Figure 6.2). Currently, it is possible that seven CPM companions were missed due to this small epoch spread.
136
Figure 7.2 Histogram of the proper motion of the primary (or single) component in each system is plotted, with the vertical line indicatingµ= 0′′.18 yr−1, the canonical lower proper
motion limit of Luyten.
7.3.2 Imaging Search
A study of the detection limits of the imaging campaign was done for objects with a range of I magnitudes at ρ = 1— 5′′ and at ∆mags = 0 up to 10 in one-magnitude increments
for different seeing conditions at the two main telescopes where the bulk of the data were acquired: the CTIO/SMARTS 0.9m and the Lowell 42in.
Because the apparent I−band magnitudes for the stars in the multiplicity sample range from 5.32 — 17.18 (see Figure 5.2), objects with I−band magnitudes of 8, 12, and 16 were selected for investigation. Only 88 primaries (7.8% of the sample) have I < 8, so it was not
felt necessary to create a separate set of simulations around these brighter stars. These stars are listed in Table 7.2 with theirI magnitudes, the FWHM at which they were observed and at which telescope, and any relevant notes.
As noted above, the faintest primary in theI−band found in the sample is DEN0909-0658 (I = 17.18), withπ = 42.5 mas (corresponding to 23.5 pc), (V −K) = 8.9, andMV = 19.58, all parameters that place it at the boundary of the sample in both distance and spectral type. Therefore, ∆mags of up to 10 magnitudes were probed in order to determine if a companion with I = 18 could be detected within 1 — 5′′ of a primary withI = 8.
Table 7.2: Stars Used for Detection Limit Study
Star I FWHM Telescope Note
(mag) (arcsec) GJ0285 8.24 0.8 0.9m LP848-050AB 12.47J 0.8 0.9m ρAB <2′′ SIP1632-0631 15.56 0.8 0.9m L032-009A 8.04 1.0 0.9m ρAB = 22′′.40 SCR0754-3809 11.98 1.0 0.9m BRI1222-1221 15.59 1.0 0.9m GJ0709 8.41 1.0 42in GJ1231 12.08 1.0 42in
Reference Star 16 (scaled) 1.0 42in
GJ2060AB 7.83J 1.5 0.9m ρAB = 0′′.485
2MA2053-0133 12.46 1.5 0.9m Reference Star 16 (scaled) 1.5 0.9m
GJ0109 8.10 1.5 42in
LHS1378 12.09 1.5 42in
2MA0352+0210 16.12 1.5 42in Reference Star 8 (scaled) 1.8 0.9m SCR2307-8452 12.00 1.8 0.9m Reference Star 16 (scaled) 1.8 0.9m
GJ0134 8.21 1.8 42in LHS1375 12.01 1.8 42in SIP0320-0446AB 16.37 1.8 42in ρAB <0′′.33 GJ0720A 8.02 2.0 42in ρAB = 112′′.10 LHS3005 11.99 2.0 42in 2MA1731+2721 15.50 2.0 42in
138
Each of the stars was analyzed in seeing conditions of 1′′.0, 1′′.5, and 1′′.8, but because the seeing at CTIO is typically better than that at Anderson Mesa, where the Lowell 42in is located, we were able to push to 0′′.8 for the CTIO 0.9m, but had to extend to 2′′.0 for the Lowell 42in. These test stars were verified to have no known detectable companions within the 1 — 5′′ radii explored in this part of the project. We note that LP848-050AB has an
astrometric perturbation due to an unresolved companion, but that at FWHM of 0.8′′, the
two stars were still not resolved. As the imaging search probes separations 1 — 5′′, we argue
that using this star does not affect the detection limit analysis. The other binaries used all either had larger or smaller separations than the 1 — 5′′ regions explored.
The IDL shifttask was used to shift and add the science star as a proxy for an embed- ded companion, scaled by a factor of 2.512 for each magnitude difference. In cases where the science star was saturated in the frame, a reference star was selected from the shorter exposure taken in similar seeing in which the science star was not saturated. Its relative magnitude difference was calculated so that it could be scaled to the desired brightness in the longer exposure, and then it was embedded for the analysis. In all cases, the background sky counts were subtracted before any scaling was done.
Eighth magnitude targets were searched using the 3-second exposure to probe ∆I = 0, 1, 2, 3, using the 30-second exposure for ∆I = 4, 5, and using the 300-second exposure for ∆I = 6, 7, 8, 9, 10. Similarly, twelfth magnitude objects were probed at ∆I = 0, 1, 2, 3 using the 30-second exposure and at ∆I = 4, 5 with the 300-second frame. Finally, the
Figure 7.3 Detection Limits for the CTIO/SMARTS 0.9m: Contour plots for L032-009A, with I = 8.04 in 1′′.0 seeing conditions for an embedded companion at ρ = 1— 5′′ with
∆mags = 0 — 10. The Y, N, and M labels indicateyes,no, ormaybefor whether or not the embedded companion is detectable. The 3-second exposure was used for ∆mags = 0 — 3, the 30-second exposure was used for ∆mags = 4 — 5, and the 300-second exposure was used for ∆mags = 6 — 10. Thirty-five simulated companions were detected, five were possibly detected, while fifteen were not detected.
140
Figure 7.4 Detection Limits for the CTIO/SMARTS 0.9m: Contour plots for SCR0754-3809, withI = 11.98 at 1′′.0 seeing conditions for an embedded companion atρ= 1−5′′with ∆mags
= 0−5. The Y, N, and M labels indicateyes,no, ormaybefor whether or not the embedded companion is detectable. The 30-second exposure was used for ∆mags = 0, 1, 2, and the 300-second exposure was used for ∆mags = 3, 4, 5. Twenty-two simulated companions were detected, two were possibly detected, while six were undetectable.
Figure 7.5 Detection Limits for the CTIO/SMARTS 0.9m: Contour plots for BRI1222-1221, with I = 15.59 at 1′′.0 seeing conditions for an embedded companion at ρ = 1−5′′ with
∆mags = 0 − 2. The Y, N, and M labels indicate yes, no, or maybe for whether or not the embedded companion is detectable. The 300-second exposure was used. Fifteen out of fifteen simulated companions were detected.
300-second exposure was used to explore the regions around the sixteenth magnitude objects for evidence of a stellar companion at ∆I = 0, 1, 2. The images were examined in IRAF
using both the contour and radial plot features; however, only contour plots are presented here. We note that the radial plots are often more sensitive.
Example contour plots for seeing conditions of 1′′.0 at the CTIO/SMARTS 0.9m telescope are shown in Figures 7.3, 7.4, and 7.5. The ‘Y’, ‘N’, or ‘M’ labels in each plot indicate yes,
no, or maybe for whether or not the companion was detectable by eye at the separation,
magnitude, and seeing conditions explored. In most cases, a companion would have been detected. As can be seen, the science star is saturated in the frames used for ∆I greater than 4. The companion can be detected in 72 of 100 simulated situations, possibly detected in seven more cases, and not detected in twenty-one cases. The conditions where the companion remains undetected are at small ρ, at ∆I > than ∼4, and at all ρ for ∆I = 10. We note that use of the radial plots will sometimes permit detection of an object that was not visible in the contour plot. Also, comparison of any elongated stars’ contour plots to contour plots of other stars in the frame allow judgement calls of whether or not the star in question is multiple.
All 800 contour plots made with IDL are presented in Appendix IV. As noted in §7.1.4, the ∆I for the M dwarf sequence is slightly less than nine magnitudes, so ∆Is of ten would be detections of brown dwarf companions. There were no companions detected with ∆I = 10 around the brighter stars in the simulations. These simulations emphasize that this survey was not sensitive to brown dwarf companions at separations 1 — 5′′around the very bright
142
M dwarfs in the sample. Because this search was for stellar companions, we can then ignore the ∆I = 10 results for our summary of detection sensitivity, reducing the total contour plot number to 760. We will not consider the non-detections in the ∆I = 10 simulations in our summary or in 7.3.
Table 7.3 presents a summary of the results of the detections of the embedded compan- ions, with the fake companions being detected 68% of the time, possibly being detected in 9% of the simulations, and not detected in 23% of the simulations. For the simulations of bright stars (I = 8), 57% of the embedded companions were detected. For the stars withI = 12, the simulated companions were detected in 66% of the cases tested. For the faint stars (I = 16), the embedded companions were detected in 89% of the cases tested. At ρ = 1′′, the embedded companion was detected in 27% of cases, possibly detected in 20% of cases, and not detected 53% of cases. However, we do not claim high sensitivity at separations this small. The detections are better in simulations at larger ρ. For ρ = 2′′, the percentages are
68%, 1%, and 22% for the ‘yes’, ‘maybe’, and ‘no’ cases, respectively. They are 80%, 8%, and 13% for the test cases for ρ = 3′′, 83%, 5%, 13% for the simulations for ρ = 4′′, and
83%, 3%, 14% for the cases where ρ = 5′′.
Table 7.3: Detection Limit Summary
Seeing Yes Maybe No
Conditions CTIO 0.9m FWHM = 0′′.8 75 5 15 I = 8 mag 36 5 9 I = 12 mag 24 ... 6 I = 16 mag 15 ... ... FWHM = 1′′.0 72 7 16 I = 8 mag 35 5 10
I = 12 mag 22 2 6 I = 16 mag 15 ... ... FWHM = 1′′.5 65 11 19 I = 8 mag 33 5 12 I = 12 mag 20 3 7 I = 16 mag 12 3 ... FWHM = 1′′.8 58 10 27 I = 8 mag 28 4 18 I = 12 mag 18 3 9 I = 16 mag 12 3 ... TOTAL - CTIO 0.9m 270 33 77 Lowell 42in FWHM = 1′′.0 70 4 21 I = 8 mag 34 4 12 I = 12 mag 21 ... 9 I = 16 mag 15 ... ... FWHM = 1′′.5 64 10 21 I = 8 mag 33 6 11 I = 12 mag 17 3 10 I = 16 mag 14 1 ... FWHM = 1′′.8 60 11 24 I = 8 mag 29 6 15 I = 12 mag 19 2 9 I = 16 mag 12 3 ... FWHM = 2′′.0 54 12 29 I = 8 mag 24 9 17 I = 12 mag 18 2 10 I = 16 mag 12 1 2
TOTAL - Lowell 42in 248 37 95
TOTAL 518 70 172
Figure 7.6 illustrates the detection limits of the survey. Four companions have angular separations outside the range of the plot, but well within the 600′′sensitivity of the searches.
We note that the largest ∆I detected was ∼ 6.4 (GJ0184B), while the largest angular separation detected was 295′′(GJ0049B).
144
Figure 7.6 Rho vs. ∆I is plotted to show the observational limits of the survey. The solid points indicate companions that were detected during the two main campaigns (imaging and blinking). The open circles with arrows indicate companions that have lie outside the range of the plot: L977-016B with ρ = 200′′and ∆I = 1.38, G192-011B with ρ = 161′′and ∆I =
2.44, GJ0049B with ρ = 295′′and ∆I = 3.10, and GJ0644E with ρ = 221′′and ∆I = 5.71.
ubiquitous Malmquist bias, missing low mass primaries, user-imposed parallax criteria limits, as well as faint companions at small separations. An attempt has also been made to mitigate the human error factor that contributes to searches by eye. As we have shown, in at least two-thirds of the cases tested, a companion within the separation and ∆mag limits probed would have been discernible from images alone — additional close companions have been recovered via high resolution imaging techniques, astrometric perturbations, and literature searches.
CONCLUSIONS