IDENTIFICATION OF GALACTIC BULGE SURVEY X-RAY SOURCES WITH TYCHO-2 STARS

Our starting point is the X-ray source lists from the GBS. We merge the published list, which we will refer to as the northern sample (Jonker et al. 2011) with additional sources detected in the final southern strip of the survey, the southern sample (P. G. Jonker et al. in preparation). For ease of recognition, we will use abbreviated source names of CXnnnn for sources from the northern sample, and CXBnnnn for sources in the southern sample. Besides being more compact, this notation has the advantage that each list is sorted in order of X-ray brightness, so the CX number conveys information that is absent from the longer positional names such as CXOGBS J175024.4–290216.

In performing our optical matches, we identified two cases where two of the X-ray sources appear to coincide with the same Tycho star. In both cases, the two X-ray sources were detected in different images, and one has a much larger off-axis angle than the other, and consequently less precise coordinates. In these cases, we believe that they are duplicate detections of the same source, with position differences just too large for Jonker et al. (2011) to have identified them as duplicates. The redundant sources are CX191, which appears to have been an off-axis detection of CX360, and CX230, which appears to be an off-axis detection of CX33. Another similar case has been found with the optically fainter catalysmic variable candidate CX93, for which CX153 appears to be an off-axis detection (Ratti et al. 2012). On examination of the GBS source catalog, we identified a further 15 likely duplications which share the characteristics of being from different Chandra observation IDs and having one or both detection at sizeable off-axis angle. These are: CX16 = CX515, CX38 = CX225, CX61 = CX136, CX175 = CX464, CX177 = CX384, CX345 = CX233, CX270 = CX571, CX597 = CX273, CX280 = CX382, CX394 = CX285, CX292 = CX323, CX304 = CX303, CX1155 = CX483, CX636 = CX635, and CX646 = CX851. In each case, we have listed the detection with smallest off-axis angle first, so the second source of each pair can be considered the duplicate. In most cases, the detection closer to the axis is also the stronger one. A few more cases were found where close objects appear to be detected in the same observation ID, and we believe these are distinct X-ray sources. CX916 and CX917 are one such pair that match an optical double, HD316666A and B. We removed all 18 of the apparent duplicate objects from the catalog, leaving 1216 unique sources in the northern sample.

Following this experience, we were careful to check the southern sample for repeat detections of objects from the northern sample, and for duplicates. As a result, the southern sample includes 424 new and unique X-ray sources with no credible duplications. After duplicate removal, our combined catalogs total 1640 unique sources. All statistics reported here on numbers of matches expected and found are based on this combined unique subset of the full source lists.

We primarily consider matches with the Tycho-2 Astrometric Catalog (Høg et al. 2000a). This provides a reasonably uniform catalog of the 2.5 million brightest stars in the sky, with excellent positions and lower quality, although still useful, two-color photometry. We also check for matches with the original Henry Draper Catalog (Cannon & Pickering 1918–1924), the Henry Draper Catalog Extension (Cannon 1925–1936), and the Henry Draper Extension Charts (see Nesterov et al. 1995, and references therein) as these provide spectroscopic classifications for many of the matches found in the Tycho-2 Catalog.

We consider a candidate match to be an X-ray source within 10 arcsec of a Tycho-2 star to allow for the large uncertainties in Chandra positions for large off-axis angles (see Hong et al. 2005). Proper motions are applied to the Tycho-2 stars pair-by-pair to obtain the correct optical coordinates corresponding to the epoch of each Chandra observation. The proper motions between the J2000 reference epoch of the Tycho-2 catalog, and the 2006–2012 Chandra observations are mostly small compared to the Chandra positional uncertainty, but not always, with the largest accumulated proper motion (for CXB93, a nearby M dwarf) being 3.3 arcsec.

We identify 70 Chandra sources that lie within 10 arcsec of 69 Tycho-2 stars (both CX916 and CX917 lie close to the same star; see below). Information about these matches is summarized in Table 1 for the objects from Jonker et al. (2011) and Table 2 for the southern objects from P. G. Jonker et al. (in preparation). Most of these actually match to within 1 arcsec, with a few at substantially larger deviations: the median X-ray/optical offset is 0.60 arcsec, while the mean offset is 1.19 arcsec. After removing the 10 objects identified as either chance coincidences or alignments with resolved companions (see Section 2.3), the median X-ray/optical offset drops to 0.49 arcsec, and the mean to 0.74 arcsec. These numbers are consistent with the absolute astrometric calibration of Chandra, for which the 90% uncertainty is 0.6 arcsec.⁷

Of these sources, 28 also clearly coincide with a source in one of the Henry Draper catalogs. In addition in one case, CX77, the positional offset between the Tycho-2 position and the nearest HD star, HD159571, is 45 arcsec, but the photometry of the two stars agree. The discrepancy cannot be due to proper motion as the Tycho-2 star only has a proper motion of 2.9 mas yr⁻¹. Examining Digital Sky Survey images of the region, there is no bright star at the quoted position of HD159571, so we believe that the coordinates quoted are inaccurate and count this as a match, in agreement with Fabricius et al. (2002).

Finally, we identify two additional GBS sources with resolved companions to X-ray detected Tycho-2 stars. CX917 appears to coincide with HD 316666B, the companion to HD 316666A (CX916). CXB36 coincides with HD$\dot {3}$18327B, the companion to HD 318327A (CXB93).

With the precise localization possible with Chandra, we expect few of our matches to be coincidences with bright stars. We can assess the probability of chance coincidences by comparing the number of matches between GBS X-ray sources and Tycho-2 stars within a specified matching radius to the number of matches found within an annulus around the X-ray source. This preserves information about the spatial distributions of X-ray and optical sources to provide a robust estimate, for this region of the sky, of the probability of a chance coincidence.

We plot the number of matches found as a function of offset in Figure 1. The central peak contains 45 candidate matches with offsets below 1 arcsec, 11 with offsets of 1–2 arcsec, and 14 with offsets of 2–7 arcsec. There are no candidate matches between 7–10 arcsec, so we have a total of 70 candidate matches with Tycho-2 stars. Comparing this central peak in the distribution with the number found at larger radii it is clear that few can be expected to be coincidences. Statistically we would expect ∼5 chance coincidences within 10 arcsec, a 20% chance of finding one match within 2 arcsec, and a 5% chance of finding one within 1 arcsec. Given these statistics, it seems likely that a few of the matches with offsets of 2–7 arcsec are due to chance coincidence, but most are probably not. In some cases the match may be with a resolved binary companion rather than to the Tycho-2 star itself.

Zoom In Zoom Out Reset image size

Those sources observed by Chandra at small off-axis angles with good detections are most suspect. We can assess the likelihood of a chance match on a case-by-case basis using the results of Hong et al. (2005) who present 95% confidence radii as a function of off-axis angle and number of counts. Of the offsets greater than 2 arcsec, CX917 can immediately be excluded as this corresponds to the resolved companion HD316666B, and the large offset comes from matching it with HD316666A. We find that the offsets for CX728, CX925, CXB211, and CXB225 are not significant at the 95% confidence level so these may be true counterparts to the X-ray source observed at large off-axis angles. For the remainder of the objects with offsets greater than 2 arcsec, in every case the offset is at least twice the 95% confidence ranges, so these nine objects (CX53, 59, 77, 156, 360, 388, 622, and CXB17, and 296) are either matches with resolved binary companions or true coincidences.

We also checked the 10 objects with offsets between 1 and 2 arcsec by the same criteria. All of these objects were observed at greater than 5 arcmin off-axis angle, and none of the offsets seen are significant at the 95% confidence level, so we believe these are all true matches, as was expected statistically.

The first step in obtaining a credible classification for the objects found is a spectral type for the optical counterpart. The Henry Draper Catalog, Extension, and Charts provide a baseline classification but this is often crude and lacks luminosity classes. For objects in the original Henry Draper Catalog, improved classifications with luminosity classes are available in the Michigan Catalogue of two-dimensional spectral types for the HD stars (Houk 1982). We also checked each object against the online Catalog of Stellar Classifications (Skiff 2010) and for a number of objects, especially the handful of early-type stars, one or more modern classifications were available. Where we list no spectral types in Table 1 or 2, we could find no published spectral classification in the literature.

We also collate infrared photometry of our objects from the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006). While in general the GBS is working with the superior IR data from the Vista Variables in the Via Lactea project (VVV; Saito et al. 2012), for the objects considered here, VVV photometry is expected to be saturated and so 2MASS is generally more reliable.

Finally we checked each matched Tycho-2 star against the All-Sky Automated Survey (ASAS) Catalog of variable stars (Pojmanski 2002). Our findings are summarized in Table 3. We found seven matches with identified ASAS variables: CX4, CX6, CX9, CX33, CX352, CXB287, and CXB422. Of these, CX6 and CX352 are known variables (V3892 Sgr and V779 Sgr, respectively), and CX33 was a previously suspected variable (NSV 23882). CX9 is attributed a 5.72 day period in the ASAS catalog, but we could not reproduce this. Otero et al. (2006) instead find a period of 2.87 days (almost, but not quite, half the ASAS period) which we can reproduce. We were not able to confirm the putative variability in either CXB287 or CXB422 using the ASAS data.

To place limits on the variability present in each of the other objects, we also examined the ASAS photometry of each one. We show in Figure 2 a measure of the scatter of the light curves as a function of magnitude. All of our Tycho-2 stars were included in the ASAS photometry database, although the brightest are saturated and unusable. CX205 is by far the worst case at magnitude 6.7. CX156, CX183, CX256, CX275, and CX863 all suffer from mild saturation, but we can still exclude large amplitude variability in these objects. We initially used the standard deviation of the light curves to quantify the variability (considering only A grade photometry), but found that there were several objects showing large standard deviations caused by a small number of points with large excursions several magnitudes below the typical values. Many stars show these excursions, even well-understood objects such as the W UMa system V3892 Sgr (CX352), so we believe they are spurious. We opt to instead quantify the variance by measuring the deviation from the median enclosing 68.2% of the points. For a Gaussian distribution, this is equal to the standard deviation, but it is more robust to the presence of small numbers of outliers with extremely large deviations. Using this statistic, the plot became much cleaner than using the standard deviation.

Zoom In Zoom Out Reset image size

Several variables showed up in this way that were not included in the ASAS variability catalog, most notably V846 Oph (CX115), a known eclipsing Algol. We also performed a visual examination of all light curves as a function of time and a period search. For each object, we filter the data by the grade of photometry, typically keeping just grades A and B. Period searches were carried out separately for 1–100 days, and 2–24 hr using Lomb–Scargle periodograms, and verified by inspection of both unbinned and binned folded light curves. In this process, we picked up another four periodic variables that had not passed the ASAS acceptance criteria: CX27, CX82, CX333, and CX514, are all low-amplitude, periodic systems. CX7, CX10, CX12, CX25, CX32, CX72, CX360, CX524, CX1113, and CXB5 all show irregular variability of some form, either slow trends, irregular flickering-like behavior, or in the case of CX72 a single discrete outburst. Our final test was to examine Lomb–Scargle periodograms of each year of data separately. This is useful where a period may not be coherent on long timescales, for example if the starspot geometry on a star is changing. Only one periodic variable was added to the list, CX7, which showed quite complex behavior. We will discuss this in Section 6.4.

Several objects remain for which the variance is larger than typical for stars of their magnitude, but for which we could not confirm real variability in the light curve. The most pronounced cases are CX1087, CXB17, CXB200, and CXB296. These objects may show unresolved short timescale variability. We note that both CXB17 and CXB296 appear to be chance coincidences with Tycho-2 stars.

In all, we identify nine periodic variables, five definite irregular ones, and a further seven suspected irregular variables. Two X-ray bright objects, CX4 and CX7, show both periodic and aperiodic behavior. In total, a third (20/60) of the non-coincidental matches appear to be variable in ASAS data. This is probably an underestimate given the presence of other objects which appear to show an excess variance compared to other stars at a similar magnitude, but neither a detectable period, nor clear, resolved aperiodic variability.

In a few cases Hipparcos parallaxes exist for our optical counterparts (van Leeuwen 2007). These provide good estimates of distance, and hence X-ray luminosity (Table 4). In the absence of a parallax distance, if we have a spectral and luminosity classification then we can estimate a spectroscopic parallax distance (Table 5). To do this we use intrinsic (B − V) colors from Fitzgerald (1970) to estimate E(B − V), and then absolute magnitudes from Straižys & Kuriliene (1981) to derive a distance. In the absence of a luminosity class, we can obtain a lower limit on the distance by assuming a main-sequence star.

For the non-Hipparcos objects for which we estimate E(B − V) based on spectral classifications, we also list the Bulge reddening estimated by Gonzalez et al. (2011, 2012) based on red clump stars in VVV data. In every case, our inferred reddening is much less than the Bulge reddening, consistent with the relatively short distances deduced for these objects. Note that this is also true for the early-type stars at inferred distances of several kiloparsecs.

Where appropriate, we will discuss the location of the more distant (non-local) sources. We assume that the distance of the Sun from the Galactic center is 8.33 ± 0.35 kpc (Gillessen et al. 2009). We adopt a distance of the Sun above the plane of 26 ± 3 pc (Majaess et al. 2009). These numbers are of some importance as we are conducting a Bulge survey, not a plane survey, and so our lines of sight diverge from the plane at larger distances. Consequently, we would only expect to see relatively nearby early-type stars in our sample, at least among the higher latitude sources (e.g., CX863, CXB9, CXB306). The spiral structure of the Milky Way along our sight lines can be seen in the maps of Churchwell et al. (2009). Beyond the local stars of the Orion Spur, we encounter the Sagittarius Arm at around 1.2 kpc, the Scutum-Centaurus Arm around 2.8 kpc, the Norma Arm around 4.4 kpc, and the Near 3 kpc Arm around 5.8 kpc. Among the luminous stars considered here, the Be stars CX6 and CX33 (and the coincidental counterpart to CXB296), together with CX863, appear coincident with the Sagittarius Arm, while CX31, CXB9, and CXB306 appear to lie in the Scutum-Centaurus Arm. As many of the objects are variable, we compile them in Figures 8–10. In Figure 8, we show light curves of periodic variables, in Figure 9 those of clear aperiodic variables, and in Figure 10 light curves showing marginal aperiodic variability. We will discuss these in the context of individual objects in the following sections.

This X-ray source was previously detected by ROSAT/PSPC (Sidoli et al. 2001) who noted that the source coincides with the supernova remnant G359.1+0.9 (Gray 1994). With a better position, we find that the Chandra source is inside the remnant, but off-center, but is a very close match to HD 316072. It was also serendipitously observed by XMM-Newton, and Farrell et al. (2010) note that the X-ray spectrum is consistent with thermal plasma. This, together with the close association with HD 316072, supports interpretation of the X-ray source as stellar coronal emission from this star, rather than being associated with the supernova remnant. The X-ray to bolometric flux ratio is unusually high for a late-type giant (log (F_x/F_bol) = −3.8), as is the inferred X-ray luminosity, L_X ≃ 3 × 10³¹ erg s⁻¹, but it is near the top of the range observed in other late-G giants, rather than outside it (Gondoin 2005).

HD 316072 is identified as a variable in ASAS data (Pojmanski 2002), strengthening the likelihood of it being the true counterpart further. A Lomb–Scargle periodogram yields a single clear period of 16.0811 ± 0.0026 days, with the uncertainty estimated by the bootstrap method. The periodicity has an amplitude of about 0.1 mag, is apparent in individual data points as well as in phase-binned data, and is present in multiple independent subsets of the light curve. We show the folded light curve in Figure 8. The rotation rate is quite rapid for a giant, so there is no need to look for X-rays from a companion star and we can comfortably attribute the X-rays to coronal activity in the giant.

Zoom In Zoom Out Reset image size

HD 161103 is one of the prototypical Be stars identified as a class by Merrill et al. (1925). Early classifications include B2 iii–v (Hiltner 1956) and B1 iii (Garrison et al. 1977). Recent works favor a B0.5 iii–ve star (Steele et al. 1999; Lopes de Oliveira et al. 2006). As expected for a Be system, it shows a pronounced infrared excess over the colors expected for an isolated star due to the presence of circumstellar material and is also a strong outlier in (J − H) versus (H − K).

This object is a long-known variable star, V3892 Sgr, and confirmed as a variable by ASAS (Pojmanski 2002). We show its light curve in Figure 9. It exhibits relatively low amplitude irregular variations, at least by comparison with the other Be star in our sample, CX33.

Zoom In Zoom Out Reset image size

HD 161103 emerged as a possibly more interesting object when Motch et al. (1997) matched it with a ROSAT source and inferred an unusually high luminosity for its spectral type (unabsorbed L_x ≃ 10³² erg s⁻¹), and a quite hard color. Both characteristics stand out in our Chandra data as well. Based on these characteristics, Motch et al. (1997) identified this as a candidate X-ray binary. Since the X-ray luminosity is lower than known Be + neutron star systems, they suggested a white dwarf companion. Lopes de Oliveira et al. (2006) studied this object further, identifying it as one of a number of objects analogous to γ Cas. They reported a 3200 s pulsation, although it was unclear if this was truly a coherent pulsation from a rotating compact object, or a quasi-periodic oscillation as seen in γ Cas. The nature of this object, and of the γ Cas analogs in general, remains unknown (Motch et al. 2007).

TYC 6839-513-1 was spectroscopically classified as a K0 v star by Torres et al. (2006). These authors measured a Li equivalent width of 80 mÅ, comparable to the lower end of the distribution seen in Pleiades stars of the same color, suggesting that this is a young object.

This star is not identified as a variable in the ASAS variability catalog, but on examination it does show quite large amplitude, very slow variability (Figure 9). It also does appear to show periodic variability with a period around 3.5 days, although this is only apparent after examining yearly periodograms (Figure 11). A primary period around 3.5 ± 0.1 days is apparent on most years, although the frequency is not stable, and sometimes two periods are seen. The quoted uncertainty primarily reflects the variations seen in the dominant period. In addition on a few years there is significant power at the first harmonic. Together, these frequencies and their aliases account for all the features seen in the periodograms. We interpret this behavior to be variations caused by starspots that form at different latitudes subject to different rotation rates. Sometimes multiple spots may result in significant power at harmonics.

The X-ray to bolometric flux ratio of CX7 is very high, log (F_X/F_bol) = −3.33 (Figure 4), close to the limit of coronal saturation, and together with CX9, it also stands out in X-ray-to-optical flux ratio above all of the other 67 stars considered (Figure 3). We can use the spectral type to derive a convective turnover time (Wright et al. 2011) and combine this with the rotation period to obtain the Rossby number, R₀ = P_rot/τ = 0.23. Comparing this with Figure 2 of Wright et al. (2011) this is close to the saturated regime, consistent with the very high value of log (F_X/F_bol). The strong aperiodic variability, quite rapid rotation, and the high level of X-ray activity all support the classification of this object as a young K star.

Superficially, this object presents a challenge for interpretation as an A star with the highest X-ray-to-bolometric flux ratio in our sample. The spectrum is among the harder objects in the sample, so if it is a coronal spectrum it must be rather hot, above 10 MK, atypical of later A stars, which can show low-luminosity cool coronal emission. We then expect this system to be a binary of some kind, with the X-rays attributed to a companion star. This is indeed the case. This is an ASAS variable, showing shallow eclipses with period 2.8723 days (Otero et al. 2006); see Figure 8. No secondary eclipse is seen, and the IR colors are clearly redder than expected for a single A star (Figure 5), so the companion is a late-type star, and the X-rays can probably be attributed to coronal activity. With F_X/F_Bol so close to the saturation line at log (F_X/F_Bol) ∼ −3, however, the companion must itself be very close to saturation. For the companion to have an acceptable F_X/F_Bol ratio, it must have a bolometric luminosity close to the A star, and so must be larger, and hence an evolved star. This may then be an Algol system.

We attempt to fit the infrared excess more quantitatively with a simple model. We assume that the primary is an A5 main-sequence star (we will justify this assumption later), with appropriate absolute magnitude and colors, and then consider a range of late-type giant companions, with the radius and spectral type of the companion allowed to vary freely, along with the reddening. We find acceptable fits to the photometry for a wide range of companions of spectral type G6 or later. All solutions involve a companion underluminous for a giant, so we classify the companion as a G6 or later sub-giant. Formally, the best fits are found for a K1 IV companion with radius around 4.5 R_☉, and with E(B − V) ≃ 0.14 (close to that found above neglecting the companion). This is not a unique solution, but we will examine it further to show that it is consistent with other characteristics of the system. The K1 IV solution corresponds to a companion which is about 1 bolometric magnitude fainter, and the X-ray to bolometric ratio of the companion star alone then increases to log₁₀(F_X/F_bol) = −2.7, placing the companion on or a little above the saturation line, but not implausibly so. With an A5/K1 binary, the ratio of V-band surface brightnesses is about 30:1, so the companion star is too dim for a secondary eclipse to be seen, consistent with the light curve (Figure 8). Finally, we can estimate the binary size given a 2.87 day period and plausible masses. For a ∼2 M_☉ A5 V primary, the expected binary separation is then a = 10.6(1 + q)^1/3 R_☉. This is quite consistent with the inferred companion radius; for example if we assume M₂ = 1.0 M_☉, then we expect its Roche lobe to have radius 3.9 R_☉, certainly consistent within the uncertainties on our crude estimates above. We note that if the A5 star were larger than a main-sequence star, then the companion would also have to be larger, and would not fit within the binary; the assumption of a main-sequence primary made above is thus justified.

We conclude that the properties of CX9 can be explained well by an eclipsing binary with an A5 V primary and a late-type sub-giant companion (with K1 IV adopted as a representative example). The companion must be either filling its Roche lobe (and actively transferring mass) or very close to it. It will therefore be tidally locked, explaining the very high level of coronal activity in this system, which is then either an Algol, possibly in a mass-transferring phase, or an Algol-like system.

HD 315992 was initially classified as an F8 star of undetermined luminosity class (Nesterov et al. 1995). More recently, HD 315992 was included in the RAVE 3rd Data Release (Siebert et al. 2011) from which we obtain T_eff = 5369 K and log g = 4.18, pointing instead to a mildly evolved G7 star. The location in the (J − H) versus (H − K) diagram (J − H = 0.50 ± 0.04, H − K = 0.20 ± 0.040) also favors a later spectral type.

This appears to be a rather active star, with a high X-ray-to-optical flux ratio and quite hard X-ray spectrum. It appears to show irregular variability in ASAS data, although this is one of the more marginal variables in our sample (Figure 10). There is no evidence for an IR excess, and it lies on the single-star line in Figure 5. While we cannot draw a definite conclusion, CX10 appears consistent with an active single star with no evidence for binarity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

HD 318207 is one of the stars that appears to have an infrared excess, and correcting for extinction will only make this more dramatic, so this is likely a binary with a later-type companion. The location in the (J − H) versus (H − K) also suggests the presence of a companion cooler than G5. Like CX10, this object shows a high X-ray-to-optical flux ratio, and quite a hard X-ray spectrum. It appears to show slow long-term variability in ASAS data (Figure 10). We tentatively classify this as a binary containing one or two active stars, although the activity may in this case be unrelated to the binarity as the luminosity is not outside the range expected for single stars.

TYC 7381-792-1 shows no detectable period, but does appear to exhibit irregular variability in ASAS data. The colors are consistent with a single star, with the JHK colors suggesting an early K main-sequence or late G giant type. It has a high X-ray-to-optical flux ratio and quite hard spectrum (like CX10 and 12). It is likely an active single star or binary.

TYC 6839-636-1 has the third highest X-ray-to-optical flux ratio in our sample. The JHK colors suggest a late spectral type, and are quite similar to CX4, possibly pointing to a giant. We identify this as a variable in ASAS data with a periodic modulation about 0.15 mag and period 31.13 days. The light curve is shown in Figure 8. Combined with the colors, this period suggests this is an active late-type giant.

CX31 is identified with LS 4306s, the southern object of a pair of O stars about 25 arcsec apart. It has an O9 v spectral type (Vijapurkar & Drilling 1993). We estimate E(B − V) = 0.78 ± 0.05. Savage et al. (1985) included this star in a catalog of UV extinction determinations and estimate a slightly higher but roughly consistent E(B − V) = 0.99. Our distance estimate for E(B − V) = 0.78 is 2.6 kpc, consistent with Savage et al. (1985), with a height above the plane of about 80 pc.

Both the high X-ray luminosity and the hard X-ray spectrum flag this as an unusual massive star. The inferred interstellar reddening is higher than for the other OB stars in the sample, but not outside the range considered in Section 5, so if there is no additional absorption, the X-ray colors favor temperatures above 40 MK. Local absorption is not generally seen in O stars, where the X-rays are believed to originate sufficiently far from the stellar surface to not be seen through a large column. If there is substantial local absorption, then the unabsorbed X-ray luminosity, and hence the F_X/F_bol ratio, becomes even larger.

This object does appear in the Washington Double Star Catalog (Mason et al. 2001). The companion star is about 2 arcsec away, corresponding to about 5000 AU at 2.6 kpc, making this an extremely wide binary if it is indeed a physical association. The companion is 3.5 mag fainter than the O star, so if it is a main-sequence star, would be have a spectral type around B5. The X-rays clearly originate from the O star, however, and not from the companion or a point in between, and it is unlikely that the presence of the companion is pertinent to the origin of the X-rays.

It remains possible that the O star itself is a much closer inner binary. In the optical/IR color–color plot it lies on the single star track, but that could only indicate two stars of similar spectral type. In this case, the unusual X-ray emission could indicate that this is a colliding wind system. A compact companion is also a possibility, which would make this a quite low-luminosity X-ray binary.

We find TYC 6839-218-1 to be variable in ASAS data with slow, and probably irregular variations of about 0.1 mag. Its X-ray-to-optical flux ratio and X-ray color both are unremarkable for stellar X-ray emission. The JHK colors are consistent with a G or K star.

HD 316341 is a known Be star, although rather less well studied than HD 161103 (CX6) and has never been proposed to be a Be X-ray binary. Like CX6, and as expected for a Be system, it shows a pronounced infrared excess over the colors expected for an isolated star due to the presence of circumstellar material, and is an outlier in (J − H) versus (H − K). It is also a known variable, NSV 23882, and was clearly identified as variable by ASAS. Its light curve is shown in Figure 9. There is substantial variability, both an overall decline and several outbursts, with an overall range of around 0.5 mag. The outbursts appear to have a characteristic recurrence time of about 200 days, but are not strictly periodic and outburst separations range from 180 to 230 days. CX230 appears to be an off-axis detection of this object, indicating that substantial X-ray variability is also present, since this corresponds to a four-times-fainter X-ray flux than its detection as CX33.

Beyond being a Be star, it is unclear how to classify this. The X-ray luminosity is not remarkable for single Be stars, especially not B0 stars, and is lower than the typical range for γ Cas stars (Motch et al. 2007). The spectrum is harder than is typical, however, with a hardness ratio comparable to the γ Cas analog, CX6. As a B0 star, it also lies very close to the narrow range of spectral types around B0.5 seen in the other γ Cas analogs. It may be a lower luminosity member of the class or one temporarily caught in a low-luminosity state.

CX53 was matched against HD 161117, although with a quite large offset of 3.33 arcsec. Initially we flagged this object as a possible chance alignment. HD 161117 is a resolved double star, however, with a 12th magnitude companion listed in the Tycho Double Star Catalog (Fabricius et al. 2002). Using the coordinates and proper motions of this star we find an offset of 0.52 arcsec from the Chandra position at the epoch of the Chandra observations, so we believe this is likely to be the true counterpart to the X-ray source. If it is a true binary, then the reddening, distance, and X-ray luminosity will be the same as calculated in Table 4. This double is also listed in the Washington Double Star Catalog (Mason et al. 2001). The two stars have very similar proper motions, with a common proper motion of 46 mas yr⁻¹, and a difference between them of just 7 mas yr⁻¹. The latter is approximately consistent with the orbital proper motion expected for a binary star separated by about 3 arcsec at 100 pc, so we believe this is indeed a true binary.

The companion is 3.4 mag fainter than HD 161117, but otherwise no information is known about it. If it is a main-sequence star, then we expect a spectral type around K4 V. Associating this star with the X-ray source implies a true log (F_X/F_opt) = −2.52, placing it among the X-ray brighter objects in our sample. The X-ray hardness, however, is unremarkable for coronal emission from late-type stars.

CX59 was matched with TYC 6832-663-1 with a quite large offset of 3.53 arcsec. This source was detected only 3 arcmin off-axis by Chandra with 27 photons, so there is no reason to expect such a large offset. This star has not been reported as a double, and the X-ray source is also the hardest of the sources considered here for which we can calculate a hardness ratio. We conclude that this is likely to be a chance alignment.

The X-ray spectrum of CX72 is quite soft, consistent with coronal emission. The object lies above the line in the IR/optical color–color diagram (Figure 5), with a significant IR excess relative to its Tycho-2 color. In general this indicates a cooler companion, but in this case, the JHK colors are consistent with its F8 spectral type, with moderate reddening, so the Tycho-2 color may be at fault, leaving no convincing evidence for a cool companion.

We identified an outburst in the ASAS light curve of HD 316356 (Figure 10). This is quite significant with amplitude 0.15 mag and lasting for three months. The outburst behavior is quite different to the flares of coronally active stars and requires a different explanation further arguing against associating the X-rays with a cool, active companion. The outburst itself is more reminiscent of dwarf nova outbursts in morphology, although the duration is substantially longer than typical dwarf nova outbursts. Aside from the outburst, HD 316356 shows no significant variability in the ASAS data, and in particular no evidence for ellipsoidal variations as would be expected if it were the donor star in a cataclysmic variable. A chance alignment with an unrelated dwarf nova or a hierarchical triple cannot be ruled out, but it would then be a quite improbable coincidence with a quite atypical dwarf nova, so this seems unlikely. At this point, the nature of the outburst remains unclear.

CX77 lies close to TYC 6839-487-1, although with a 6.32 arcsec offset this is likely to be a chance alignment. This star is also HD 159571 (Fabricius et al. 2002), although we note that the SIMBAD coordinates listed for HD 159571 lie rather far from TYC 6839-487-1 and do not appear to coincide with a bright star at all.

As for CX53, this star is listed as a double (Fabricius et al. 2002). Both stars have astrometry included in the UCAC-3 catalog (Zacharias et al. 2010), although the proper motions for the companion star are flagged as suspect. The offset of the companion star from the X-ray source is 1.17 arcsec. The companion star has been detected by both 2MASS and the Spitzer GLIMPSE Point Source Catalog.¹⁰ The colors from J to 8 μm are flat, and difficult to reconcile with a cool star. The colors and apparent magnitude are as would be expected for an A star at around the distance of HD 159571. The X-ray source itself is strongly variable, being not detected in an observation closer to on-axis, and is rather hard, and neither these characteristics, nor the inferred luminosity, would be typical for an A star. It seems quite likely it is indeed an unrelated object and that this object should be classified as a chance coincidence.

CX82 is matched with TYC 6849-1294-1. There is very little information about this star and no published spectral type. The Tycho-2 colors are consistent with a G-type star or earlier, but have large uncertainties so a cooler star cannot be ruled out. The colors lie above the single-star line in Figure 5, but the (B − V) color is too uncertain to say this with confidence. The JHK colors favor a K3–5 classification, with a V−K color also consistent with this.

We find this object to be variable in ASAS data with a 6.3 day period and 0.1 mag amplitude (Figure 8). The X-ray-to-optical flux ratio is among the highest objects in the sample. Both characteristics would point to a rapidly rotating active star or binary. Combining the 6.3 day period with a possible spectral type around K4, we would expect a Rossby number of R₀ ∼ 0.29 (see Section 6.4), consistent with a star near coronal saturation.

HD 314883 shows no peculiarities in our data, with no variability and an insignificant deviation from the single-star line in Figure 5. It is however a noted (close) resolved binary (see, e.g., Fabricius et al. 2002). The two components are separated by 0.87 arcsec with both consistent with the X-ray position to within an arcsecond. They appear to share common proper motion, and have very similar V_T and B_T magnitudes. This appears to be a real binary with two very similar stars. The X-rays could originate from either component or from a combination of the two. The X-ray-to-optical flux ratio is unremarkable for one, or two, late F stars, and so most probably reflects normal coronal emission.

CX115 is clearly the X-ray counterpart to V846 Oph (HD 316070), a known eclipsing Algol system. Budding et al. (2004) list this object as having an A2 primary and G7 iv secondary star. Their fits to the light curve give mass ratio q = 0.15, i = 78°, and a reasonable likelihood that this is a semi-detached system.

CX115 is one of the objects showing a pronounced infrared excess (Figure 5) relative to single stars. If we assuming an A2 primary and G7 IV secondary as above, we estimate that the companion contributes about 15%–20% of the light in the V band and has a negligible effect on the reddening estimate, so the distance estimate above should be increased by only about 10%. For comparison, Singh et al. (1996) cite a distance of 580 pc, and a ROSAT/PSPC X-ray luminosity of 6 × 10³⁰ erg s⁻¹.

The eclipses are very clearly seen in the ASAS data, although it is surprisingly omitted from the catalog of variables. We show the folded light curve in Figure 8. The period we derive from this photometry is 3.12678 ± 0.00014 days, in perfect agreement with the period of 3.1268 days cited by Budding et al. (2004). The secondary eclipse is small, but detectable, and the light curve shows additional variability out of eclipse.

CX156 appears to be a chance alignment with HD 161907, with an offset of over 6 arcsec. The most likely match to the X-ray source is 2MASS J17492831–2918593, located just 0.09 arcsec from the X-ray position. The colors (J − K = 0.65 ± 0.11) would suggest an early-K spectral type. At K = 9.30, this star would then be about the right brightness for an early K main-sequence star at a distance of 100 pc. It is therefore possible there is a physical association of this star with CX156, although the only evidence in favor of this is that they plausibly lie at the same distance.

HD 162120 lies off the single-star track in Figure 5, although it is not one of the most significant IR excess cases. It shows no apparent variability or other peculiarities. It probably has a late-type companion star contributing the X-ray emission, although there is no evidence for or against identifying this as a close Algol-like system.

HD 161852 is the brightest star in the sample and saturated in ASAS data. This is classified as an F2 iv/v star (Houk 1982). The optical brightness would suggest a main sequence rather than sub-giant classification at the Hipparcos distance of 50 pc.

HD 161852 was included in the X-ray survey of Hipparcos F stars using ROSAT by Suchkov et al. (2003). The ROSAT estimate of the X-ray luminosity of 7.6 × 10²⁸ erg s⁻¹ is in a softer band than our Chandra observation. It was found to be a little below average for F stars of approximately Solar metallicity as this is ([Fe/H] = −0.1), and indeed this is among the lowest F_X/F_bol objects in our sample.

With log (F_X/F_bol) = −5.9, HD 162761 is quite typical of single K0 III stars (log (F_X/F_bol) = −5.6 ± 0.5; Hünsch et al. 1998). Richichi et al. (2008) observed a Lunar occultation of this object, and identified a companion 0.11 arcsec away, with a brightness ratio of approximately 1:120 relative to the giant. If this is a physical binary, the separation at 240 pc would be about 25 AU. A K0 main-sequence companion star would be consistent with the brightness ratio, but would then have an X-ray to bolometric flux ratio of around 10⁻⁴, possible, but among the most active late-type stars in our sample. Identifying the X-rays with the visible giant star is more plausible.

CX275 lies about 2 arcsec away from HD 160627. While the offset is quite large, this was observed at quite a large off-axis angle, and the offset is consistent with the X-ray positional uncertainty (Jonker et al. 2011). The X-ray luminosity is quite high for an F0 V star. HD 160627 is quite rapidly rotating, however (vsin i ≃ 30 km s⁻¹; Nordstrom et al. 1997), and the X-ray luminosity is consistent with other F dwarfs with similar rotation (Maggio et al. 1987).

HD 316432 is a very good positional match to CX296. It lies above the single-star line in Figure 5; however, the uncertainty in its Tycho-2 color is large, and the Tycho-2 color is bluer than an F0 star should be. If we instead assume a typical F0 color (consistent within uncertainties), then there is no significant IR excess. The only real peculiarity about this object is the high implied X-ray luminosity, and the high F_X/F_bol. This suggests an unusually active F star, possibly a binary, although there is no detected variability to support this classification.

HD 159509 is a resolved double (Fabricius et al. 2002), with the companion lying closer to the X-ray position, so it is possible that HD 159509 is not the true counterpart. This was observed far enough off-axis that neither star can be ruled out with confidence. The brighter star is a suspected variable (NSV 22873; Samus et al. 2009). The ASAS data confirm a clear modulation with a period of 14.6565 days, or more likely twice that with ellipsoidal variations (Figure 8). The optical/IR colors, too, indicate that this is not a single A-type giant as there is a clear infrared excess, corresponding to IR magnitudes about 1 mag brighter than expected for the A star. Together, these facts would suggest that HD 159509 has a late-type companion. Since an A3/5 III star should be about 10 times too small to fill its Roche lobe in a 30 day binary, it may be that the (low-amplitude) ellipsoidal variations actually arise from the small contribution to the visible light from a larger late-type giant companion. This would make this another Algol-like system with an evolved late-type companion to an A star, although in this case a giant. Fixing the primary to A4 III, we find a best fit to the BVJK photometry for a K1 IV/III companion. Given that it appears to be an Algol-like system, it seems natural to associate the X-rays with HD 159509 rather than the slightly closer alternate star. There are comparable Algol systems. For example, RZ Cas has a 32.3 day period, an A5 IIIe primary, and K1 III secondary (Malkov et al. 2006).

CX337 also matches an A star, HD 315995. Like many of the A stars in our sample, this shows colors inconsistent with a single star (Figure 5) indicating the presence of a cooler companion. If we assume a main-sequence primary, we can fit the BVJK photometry well for a range of late-G to early-K sub-giant companions, much as for the other A stars in the sample.

CX352 clearly coincides with the contact binary, V779 Sgr (HD 316675). Its spectral type in the Henry Draper Charts is listed as F8 (Nesterov et al. 1995), with the companion inferred to be F9 (Svechnikov & Kuznetsova 2004). Its Tycho-2 color is anomalous, presumably due to the large amplitude variability, so we can make no reddening estimate.

V779 Sgr is included in the ASAS Catalog of Variable Stars. A Lomb–Scargle periodogram gives a very sharp peak at 0.222515 days. This is the first harmonic as dominates in all ellipsoidal variables, and the orbital period is twice this, 0.44503040 ± 0.00000018 days, with a bootstrap uncertainty. The folded light curve is shown in Figure 8 and shows a modulation amplitude of about 0.6 mag.

Estimates of system parameters are presented by Brancewicz & Dworak (1980), who attribute 56% of the light to the brighter component, which has a radius 1.48 R_☉. Svechnikov & Kuznetsova (2004) alternatively attribute the brighter component 70% of the light and radius 1.44 R_☉. We then expect an absolute magnitude about 1 mag brighter than a single F8 star if the light from both stars is included. At maximum light, when we see both stars well, it is observed by ASAS to be around V ≃ 11.4, so allowing for the light from both stars, we estimate a distance around 500 pc, neglecting the unknown extinction. This would correspond to L_X ≃ 7 × 10²⁹ erg s⁻¹.

CX360 lies 4.07 arcsec from TYC 6840-525-1. CX191 also appears to be an off-axis detection of this source. CX360 was observed at only 2.8 arcmin off-axis, so there is no reason to expect such a large positional error and this star is almost certainly not the true counterpart. It is not noted as a double, so this is likely a chance alignment with an unrelated object. We note that TYC 6840-525-1 does appear to show irregular variability in the ASAS light curve (Figure 10), although it is one of the weaker detections in our sample.

CX388 lies 4.79 arcsec from TYC 68390615-1. This was observed at only 0.9 arcmin off-axis, so the positional offset is highly significant and this star is almost certainly not the true counterpart. It is not noted as a double so, again, this is likely a chance alignment with an unrelated object.

The optical counterpart to CX402 appears to be the G5 star HD 316033. The activity level is not atypical for G dwarfs, but the optical/IR colors would suggest the presence of a cooler companion as well, although this is not one of the more significant IR excesses in our sample.

The optical counterpart to CX485 appears to be the F8 star HD 316059. This object shows pronouncedly redder IR than optical colors, indicating the presence of a cooler companion. The X-rays can most likely be attributed to coronal activity from one or both of the stars.

CX506 is well aligned with HD 160390. This star actually lies rather close to the single-star line in Figure 5, unlike most of the A stars in our sample. A small IR excess may be present, but we find by fitting the BVJK colors that this is only sufficient to accommodate a late-type main sequence or slightly evolved star, no more than a magnitude above the main sequence. Late-type giant and probably sub-giant companions can probably be ruled out. The allowed companions (later than F0) all have bolometric magnitudes at least 3.2 mag fainter than the A giant, and so if we attribute the X-rays to the (putative) companion, they would have F_X/F_Bol ≳ −3.5, so would be close to coronal saturation and very active. Such a case cannot be ruled out, but we should stress that there is no evidence (other than the presence of X-ray emission) for such a companion.

CX514 lies near the B9 star HD 314884. With an X-ray luminosity of at least 10³⁰ erg s⁻¹ and F_X/F_bol = −5.4, this falls well outside the norm for late B stars, so either this is a chance alignment or a companion is present.

We do find HD 314884 to be variable in ASAS data, showing a 0.89 day period (Figure 8). The periodicity is stable in amplitude and phase through the period covered by ASAS. Besides variability and X-ray emission, there is no other evidence for multiplicity in this system. The optical/IR colors are consistent with a single B9 star, ruling out late-type sub-giant or giant companions (if the primary is main sequence), but a main-sequence companion could still be present or an evolved companion to a B giant. If the putative companion is an early K or earlier type then the X-ray emission can be accounted for with log (F_X/F_bol) < −3. Such a companion would have to be quite active, but this is expected since it would also have to be quite young if in a binary with a B9 star.

TYC 6853-80-1 appears to be an irregular variable in ASAS data (Figure 10). Its JHK colors favor a K star, or possibly a G giant like CX4. This then appears to be a late-type, somewhat active star.

CX622 is offset 2.37 arcsec from TYC 6839-19-1. This was observed at 3 arcmin off-axis, so we would not expect such a large error in the X-ray position. There is very little information available about TYC 6839-19-1 but it shows no peculiarities that would support identifying it as the true counterpart to the X-ray source. This then is likely to be a chance alignment.

CX632 lies 1.28 arcsec from HD 315998. This was originally classified as a K5 star (Nesterov et al. 1995). It is one of our stars that is included in the RAVE 3rd Data Release (Siebert et al. 2011) from which we obtain T_eff = 4533 K and log g = 2.08. From this, it is clearly a giant and the temperature then indicates a spectral class around K1. Hence we classify this as K1 iii. With log (F_X/F_bol) = −5.6, the X-ray emission is quite typical of single K1 III stars for which Hünsch et al. (1998) find a distribution of ratios −5.5 ± 0.5. The optical/IR colors are consistent with a single giant star.

TYC 6853-1354-1 matches well with the position of CX698. The optical color would suggest a mid-F star or earlier, whereas the IR color is redder, possibly indicating the likely presence of a cooler companion, although the IR colors are flagged by 2MASS as suspect.

CX785 coincides with HD 160769. This object falls above the single-star line in Figure 5, but the 2MASS J and H photometry is flagged as suspect. The DENIS survey¹¹ is in reasonable agreement, with J − K_S = 0.55 ± 0.09, so the IR excess in this object appears to be real indicating a cooler companion. We can find a reasonable fit to the BVJK photometry with a late-K companion star a little above the main sequence. It is quite possible that this is the origin of the X-rays instead of the F3 star.

The optical counterpart to CX863 is clearly HD158902. This is a mid-B supergiant, originally classified as B8 by Cannon & Pickering (1918–1924). More recent two-dimensional classifications from objective prism surveys have given spectral types of B5 ia (MacConnell & Bidelman 1976), B5 ib (Garrison et al. 1977), and B3 ii (Houk 1982). The implied distance would then range from 1.3 kpc (B3 ii) to 4.5 kpc (B5 ia). Given that the GBS is observing out of the plane, and this source is at b = +194, less luminous stars seem most likely, since they imply shorter distances from the plane. For a distance of 1.3 kpc, the distance above the plane is about 70 pc, whereas for 4.5 kpc it is 160 pc. We therefore suggest this is a B3 ii star at a distance of 1.3 kpc, although a more distant and more luminous object remains possible. The properties of this star are quite normal for its spectral type. Its F_X/F_bol ratio is quite reasonable, it shows no variability, and its optical/IR colors are close to the single star line. The IR colors are a little redder than expected but the deviation is not highly significant.

CX904 is a good match to TYC 6849-1144-1. It does fall above the single-star line in Figure 5 suggesting it may be a binary, but the IR excess is not significant without either a spectral type or a more reliable optical color. The JHK colors alone would suggest a K spectral type.

CX916 matches TYC 6849-227-1 (HD 316666A) within 0.69 arcsec. CX917 matches HD 316666B within 1.5 arcsec. HD 316666A does show something of an IR excess in Figure 5, although it is not highly significant. It may itself be an unresolved binary. This was identified as a candidate variable by Piquard et al. (2001) with a period of 0.173946 days. We could not reproduce this, or any other periodicity in the ASAS data.

We have much less information on the companion. It is independently detected by 2MASS with K = 9.878 ± 0.033, 1.17 mag fainter than HD 316666A. The difference in magnitudes would be appropriate for a G0 V star, and the JHK colors are consistent with such a classification.

CX1113 matches well with TYC 6835-312-1. This star shows marginal evidence for irregular variability and its JHK colors would suggest a G star, but we have no other constraining information.

CXB5 coincides very closely to HD 315961, originally classified as a K5 star (Nesterov et al. 1995). Neese & Yoss (1988) instead classified this star (listed as PSD 157–562; Barbier-Brossat & Figon 2000) as K0 III, with E(B − V) = 0.3, and a distance of 504 pc, based on DDO photometry. Neese & Yoss (1988) compare their photometric classifications with those from objective prism spectroscopy and find scatters of up to around five spectral subclasses, and two luminosity classes, so this should be considered rather uncertain. The quite high X-ray luminosity, L_X ≃ 1.0 × 10³¹ erg s⁻¹, together with a high log (F_X/F_bol) = −4.3 (compared to 5.6 ± 0.5; Hünsch et al. 1998) points to a quite active giant.

We see no evidence for a periodicity in the ASAS data, but the star does seem to be an irregular variable (Figure 10). The X-ray color is rather hard, comparable to other stars at high log (F_X/F_bol) such as CX4, 10, and 12. It appears to be quite comparable to CX4 in some ways, and that also shows irregular variability in addition to its periodicity.

This is noted as a double with a fainter, yet bluer companion 1.6 arcsec away (Fabricius et al. 2002). The X-rays are clearly associated with the K star, however, and it is unclear if the nearby star is physically associated.

CXB9 is associated with HD 161853. This is clearly an OB star, but there is uncertainty about its classification. It was originally listed as B3 by Cannon & Pickering (1918–1924), and then revised to B0/1 ii in the Michigan HD classification (Houk 1982). Other estimates have included O7.5 (Crampton 1971), O8 v ((n)) (Walborn 1973), O7.5 v n (Garrison et al. 1977), O9 i (Hu et al. 1993), and finally an O8 iii post-asymptotic giant branch (AGB) star (Parthasarathy et al. 2000) based on its association with IRAS 17460–3114. Suárez et al. (2006) instead argue that it is not a post-AGB star, but instead a young object.

Based on UV data, Savage et al. (1985) estimated E(B − V) = 0.54, in fair agreement with our estimate of E(B − V) = 0.44 ± 0.02. The implied distance is then 2.7 kpc and the distance below the plane is about 70 pc (Straižys & Kuriliene 1981). This would be consistent with the Scutum-Centaurus Arm. Main-sequence types would situate it closer to 2.0 kpc, between Scutum-Centaurus and Sagittarius, while a supergiant classification would put it at 4–6 kpc, consistent with either the Norma Arm or the 3 kpc Arm. In the latter cases, the distance below the plane is, however, larger, 110–190 pc. Overall, an O8 iii star at around 2.7 kpc in the Scutum-Centaurus Arm appears to fit best. For this case, the X-ray luminosity is about 3 × 10³² erg s⁻¹. This is quite high, but the F_X/F_bol ratio is not excessive, and with quite soft colors this appears consistent with normal emission for O star winds. There is no variability seen in this object, and no evidence of binarity is reported. While the optical/near-IR colors we examine here are normal for a single star, it does have an mid-IR excess (Clarke et al. 2005). As noted above, Suárez et al. (2006) suggest that this is evidence that it is a young object.

CXB17 lies near the F8 star HD 316565 (Nesterov et al. 1995), but with an offset of 4.98 arcsec, this is likely to be a chance alignment. We note that a 2MASS J17543011–2923572 is coincident with the X-ray source and is a much more likely counterpart.

CXB93 coincides with a nearby M dwarf, LTT 7072 = GJ 2130A = HD 318327A. It is one component of a 20 arcsec separation binary, with the companion GJ 2130B = HD 318327B also detected by the GBS as CXB36. The spectral type of GJ 2130A has been variously estimated as M0 (Nesterov et al. 1995), M1.5 (Bidelman 1985), M2 (Stephenson & Sanduleak 1975), and M3 (Raharto et al. 1984), with the order reflecting that the determinations, not of the publication. The same works cite spectral types of M2–5 for GJ 2130B. GJ 2130A appears above the single-star line in Figure 5. In this case, it appears to reflect the large uncertainty in the Tycho-2 B−V color, or inaccuracy of the transformations from B_T − V_T for very red objects. An independent estimate gives B − V = 1.465 (Koen et al. 2010), in better agreement with its spectral type.

The X-ray luminosity of GJ 2130A is actually rather low among M dwarfs, for which Schmitt et al. (1995) find a range from 10²⁶ to 10²⁹ erg s⁻¹. While there is uncertainty in the Hipparcos parallax distance, spectroscopic parallax distances favor short distances, and low luminosities as well, ranging from 16 pc for M0 to 12 pc for M3. Partly this discrepancy may reflect the softer bandpass used by ROSAT, with our Chandra observations possibly missing much of the X-ray flux. The companion is X-ray brighter with L_X ≃ 1.5 × 10²⁷ erg s⁻¹.

Optically, GJ 2130A was studied by Rauscher & Marcy (2006) as part of a survey of chromospheric emission in late K and M dwarfs. It was found to show the strongest Ca ii emission among stars of comparable absolute magnitude.

CXB211 lies 2.61 arcsec from HD 160826, but this was observed at a large off-axis angle (8.6 arcmin), so this is still a plausible match. HD 160826 is classified as a B9 V star (Houk 1982), so should not be an X-ray source in its own right. This object shows no noticeable IR excess, but we do note this is a resolved double with a separation of 0.6 arcsec (Fabricius et al. 2002). The nearby star has V = 10.15 ± 0.04 and B − V = 0.43 ± 0.05, and lies at 2.90 arcsec from CXB211. If this is a physical binary with the same reddening and distance then this would be a late-F or early-G giant or sub-giant. It is likely that this is the true X-ray counterpart rather than HD 160826 itself as both stars are consistent with the poorly constrained X-ray position.

CXB287 lies 1.39 arcsec from HD 158982, but the offset is not significant as this was observed 5.9 arcmin off-axis. HD 158982 is classified as an A2 IV/V star by Houk (1982), so should not be an X-ray source. This is listed as a resolved double by Mason et al. (2001), with a separation 0.3 arcsec. The companion may be the true counterpart. It is only 0.4 mag fainter than the A2 star, so is also unlikely to be a significant X-ray source in its own right unless it is a cooler sub-giant or giant, rather than a main-sequence star of type A4 or so, or the companion is closer than the A star.

CXB296 lies 5.95 arcsec from HD 161839, making this star itself inconsistent with the X-ray source given a modest off-axis angle (2.1 arcmin). This star was originally classified as B5, but the Michigan Catalog updates this to B5/7 ii/iii (Houk 1982). An alternative classification of B2 v was proposed by Garrison et al. (1977). All of these classifications would correspond to E(B − V) between 0.25 and 0.35, and a range of distances of 1–5 kpc. HD 161839 has a pronounced infrared excess, one of the largest in our sample (Figure 5). It was previously identified by Clarke et al. (2005) as an IR excess object based on matching MSX and Tycho-2 detections. Combining 2MASS and GLIMPSE IR photometry, with colors of H − K = 0.23 and K − [8] = 1.40, it appears consistent with a Be star (see Figure 4 of Clarke et al. 2005). While a Be star might be expected to be an X-ray source, it is not consistent with the X-ray position, and is not noted as being a double, so this is probably a chance alignment.

TYC 6849-01627-1 appears to be the optical counterpart to CXB302. There is very little information about this star. The optical colors would be consistent with a range of spectral types earlier than early G, depending on reddening, and the JHK colors suggest early F. It appears to show marginal irregular variability in ASAS.

CXB306 coincides with HD 163613. This appears to be a normal early B supergiant. Its optical/IR colors are consistent with a single normal star, it shows no variability and there is no evidence for binarity. It was originally classified as B5 (Cannon & Pickering 1918–1924), and updated to B1 i–ii by Hoffleit (1956). Its F_X/F_bol ratio is quite reasonable. The expected distance ranges from 2.8 kpc (B1 ii) to 7.0 kpc (B1 ia). Given that the GBS is observing out of the plane, and this source is at b = −196, less luminous stars seem most likely, since they imply shorter distances from the plane. For a distance of 2.8 kpc, the distance below the plane is about 80 pc, whereas for 7.0 kpc it is 240 pc. We then suggest this is a B1 ii star at about 2.8 kpc in the Scutum-Centaurus arm.

CXB422 is well matched with HD 315956. This is classified as an F2 star (Nesterov et al. 1995). Its parallax distance and brightness suggest a dwarf to giant luminosity class, and hence spectral class F2 III–V.

Four apparently normal OB stars and two Be stars were found in our sample, all with very close matches between X-ray and optical positions. In addition, a third X-ray source (CXB296) was found near a Be star, but too far away to be a credible X-ray counterpart to it. There is every reason to believe that the other six early-type stars are all true optical counterparts to the X-ray sources.

Two of the OB stars CX31 and CXB8 have late-O spectral classes. They have quite high F_X/F_bol ratios, but still plausible for single, normal OB stars. Both have optical/IR colors consistent with normal stars, although CXB8 does have a mid-IR excess and appears to be a young object. CX31 has a rather hard X-ray spectrum (by the standards of our sample), but this is at least in part due to a higher N_H than the other OB stars in the sample, judging by its (B − V) color. It remains possible that it is a more unusual object such as a colliding wind binary or low-luminosity X-ray binary. CX863 and CXB306 are slightly later in spectral class, early-B. Their F_X/F_bol ratios are lower, and quite typical of OB stars. CXB306 has normal stellar colors. CX863 may have slightly redder colors than a single star, but this is only marginally significant.

One of the GBS predictions was that ∼9 Be X-ray binaries would be identified (Jonker et al. 2011). None have been found in the (nearby) Tycho-2 sample, but the sources predicted were expected to be further away and subject to more reddening, and so would not be expected to have Tycho-2 counterparts. Instead, the two Be stars we have found are fainter in X-rays. One, CX6, is an intermediate luminosity γ Cas system, possibly harboring a white dwarf. The other, CX33, is most likely a normal, single Be star.

There are a surprising number of A stars in our sample. To some extent, this is a selection effect since our Tycho-2 sample is effectively an optical flux-limited sample and so has a larger sample volume for earlier spectral types. In most cases, as expected, there is evidence from IR colors and/or variability for a companion star, usually inferred to be larger in radius than the A star. Figure 12 shows just the spectroscopically confirmed late B and A stars from Figure 5. We show the B3–A7 main sequence together with two sets of loops, one corresponding to main-sequence companions of a range of spectral types and the other to more luminous companions. We see that the sources appear to divide into three groups. Neither of the two B9 stars, CX514 and CXB211, show evidence for an IR excess, although a main-sequence companion cannot be ruled out. CX183, CX506, CXB116, CXB233, and CXB287 all show weak excesses, suggesting main-sequence or mildly evolved companions. Finally CX9, CX115, CX333, CX337, and CXB181 all require luminous companions ≳ 4 mag above the main sequence.

Zoom In Zoom Out Reset image size

The preponderance of evolved companions is not expected, since later type companions should not have had time to evolve off of the main sequence. It could arise in two ways. First, where a large difference in spectral types is seen, it is possible that the later-type star is still a very active pre-main-sequence star, and so is larger than it will eventually be. Second, an A star plus a late-type sub-giant or giant could be a typical Algol system.

CX115 is, indeed, a known eclipsing Algol. CX9 is also eclipsing and we propose that it is a new Algol candidate, and possibly a mass-transferring system. CX333 appears to show ellipsoidal variations suggesting a close, late-type companion in a 30 day period. It may be a long-period Algol. For the remainder of the sources, although a late-type companion is implied, there is no evidence that it is close.

We attempt to distinguish active late-type stars (i.e., FGKM stars with very strong coronal emission) from those showing more normal coronal emission. Criteria for "activity" include a high X-ray-to-optical flux ratio, hard X-ray spectra, or irregular variability.

The presence of radio emission would be a very powerful diagnostic as magnetically active stars usually follow the Güdel–Benz relation, whereas inactive stars are much weaker in the radio, while retaining significant X-ray emission (Forbrich et al. 2011, and references therein). Maccarone et al. (2012) looked for radio counterparts to GBS X-ray sources using the NVSS catalog (Condon et al. 1998). Twelve matches were found with the original CX sources of Jonker et al. (2011), but none correspond to objects in this paper. Neither are any of the CXB sources in this paper detected by the NVSS. Unfortunately, while the NVSS is the deepest radio survey covering the whole GBS area, it is not sensitive enough to detect radio emission from these objects, as the Güdel–Benz relation would predict that all of the sources considered in this paper would fall well below the 2.5 mJy sensitivity of the NVSS.

Among the late-type stars, CX7 stands out. This is at or near coronal saturation and appears to be a very young K dwarf. CX352 is also distinguished by being the only W UMa contact binary in the sample. A few other stars show signs of strong activity. CX4 is a very active giant. CX10 and 12, and CXB5 are all active G or K stars with no luminosity class, showing high F_X/F_bol, quite hard X-ray spectra, and irregular variability. CX12, at least, appears to have a later type companion as well.

Finally we note two somewhat unusual F stars, both with moderately high F_X/F_bol. CX72 exhibited an unclassified outburst, with no other peculiarities. CX296 is unusual in its lack of peculiarities since F0 stars typically have quite weak X-ray emission; indeed, in Figure 4 it appears somewhat isolated and lies an order of magnitude above other stars of the same spectral type.

Many of the late-type stars show no signs of strong activity, modest F_X/F_bol, and no variability. These include two giants, CX256 and CX632, and several dwarfs, CX26, CX205, CX275, and CX785. The remainder, CX91, CX402, CX485, CX916, and CXB422, have no luminosity classifications.

Our sample only included one Tycho-2 M dwarf, CXB93, together with its binary companion CXB36 which is not in the Tycho-2 catalog. Since the sample is selected by optical brightness, this is not surprising. The X-ray luminosities of these two stars are unremarkable, and if anything unusually low.

After checking for known doubles, a few apparently single Tycho-2 stars are significantly offset from the nearby Chandra source, and these remain as probable chance alignments. These are CX59, CX156, CX360, CX388, CX622, CXB17, and CXB296. CXB296 is a particularly interesting case since it appears to lie near a Be star, yet that star rather securely is not the detected X-ray source. In addition, CX77 seems likely to not be associated with either HD 157571 or the star close to it, bringing us to a total of eight likely coincidences. There are additionally a few objects for which a real match could not be ruled out at the 95% confidence level, but which may nonetheless be close coincidences.

We estimated in Section 2.3 that there should be about five true coincidences in our sample. While finding eight is not a substantial excess, with the high incidence of binarity among field stars, it is likely that some of these could be undetected resolved binary companions.