IDENTIFICATIONS OF FIVE INTEGRAL SOURCES VIA OPTICAL SPECTROSCOPY

Since its launch on 2002 October 17, the International Gamma-Ray Astrophysics Laboratory (INTEGRAL) has discovered hundreds of new hard X-ray sources. According to the most recent census, of the ∼500 sources detected in hard X-rays at energies greater than 20 keV, 214 had not been well studied (or even detected in most cases) before (Bodaghee et al. 2007). These new sources are called "IGR" sources, for INTEGRAL Gamma-Ray sources. Of these IGR sources, 50 had been identified as active galactic nuclei (AGNs), 32 as high-mass X-ray binaries (HMXBs), six as low-mass X-ray binaries (LMXBs), and 15 as sources such as cataclysmic variables (CVs), supernova remnants, and anomalous X-ray pulsars, leaving 111 unclassified. Since this last census, much work has been done to identify the ∼50% of unclassified IGR sources (corresponding to ∼25% of all sources detected by INTEGRAL). Many have been identified as AGNs, LMXBs, and HMXBs, as well as relatively rare systems such as heavily absorbed (i.e., "obscured") supergiant HMXBs, supergiant fast X-ray transients, and intermediate polar (IP) CVs (e.g., Masetti et al. 2009; Chaty et al. 2008).

INTEGRAL is particularly well suited for finding such systems. Part of INTEGRAL's Core Program involved scans of the Galactic plane (Winkler et al. 2003). One expects to find HMXBs here, as they consist of a neutron star or black hole accreting from a short-lived massive star, which would not be expected to move far from its birthplace in star-forming regions of the Galactic plane. Other instruments observing at lower energies could not easily find these because of the obscuring dust and gas in the plane. Furthermore, objects such as obscured HMXBs, which consist of a compact object embedded in the stellar wind of a supergiant star, suffer obscuration not only from the intervening medium, but also from local absorption (e.g., Rodriguez et al. 2003; Walter et al. 2006). In addition, INTEGRAL has found disproportionately many IP CVs according to the previously known population statistics, since they emit at these high energies more than nonmagnetic CVs or polar CVs (Barlow et al. 2006). Currently, less than 10% of known CVs are magnetic. The majority of these are polars, so called since they show polarization of optical flux. These are systems in which the white dwarf has a magnetic field strong enough to synchronize the orbital period of the binary with the spin of the white dwarf. IPs have a weaker magnetic field that is not strong enough to cause synchronization, but they do display variability associated with the rotation of the white dwarf. Of the at least 15 CVs detected by INTEGRAL, the vast majority are these relatively rare magnetic systems, with most of those being IPs.

Identification of these unclassified sources requires observations of the source in the optical and/or infrared. The IBIS imager on INTEGRAL is unique among hard X-ray/soft γ-ray detectors in that it is able to locate point sources with an accuracy on the order of arcminutes (Gros et al. 2003). To identify a unique optical or infrared counterpart, however, the error circle must be reduced to the level of (sub)arcseconds. Thus, from the INTEGRAL position, soft X-ray telescopes, such as the Chandra X-Ray Observatory, XMM-Newton, and Swift observe the region. This typically allows identification of the optical counterpart, which can then be studied to yield an identification (e.g., Tomsick et al. 2006).

We targeted INTEGRAL sources that had also been observed in soft X-rays, which allowed the position to be narrowed down from the INTEGRAL error circle (on the order of arcminutes) to less than 10''. IGR J18214−1318, IGR J19267+1325, and IGR J18308−1232 were observed by Chandra, providing positions with 90% confidence uncertainties of 064 (Tomsick et al. 2008a, 2008b, and J. A. Tomsick 2009, in preparation). IGR J22292+6647 was observed by Swift (Landi et al. 2007), which reduced the position uncertainty to 36. IGR J16426+6536 has one XMM-Newton source with a 1σ uncertainty of 8'' (Ibarra et al. 2008a) and two ROSAT sources within the INTEGRAL error circle. Optical/infrared counterparts for IGR J18214−1318 and IGR J19267+1325 were reported with the X-ray observations. We searched the USNO-B1.0 and Two Micron All Sky Survey (2MASS) catalogs for optical and infrared counterparts to the other sources. IGR J18308−1232 and IGR J22292+6647 both have one optical/infrared counterpart. The X-ray sources associated with IGR J16426+6536 have three possible optical/infrared counterparts, and we observed the brightest optical source in the XMM-Newton error circle.

Images from the Digitized Sky Survey (DSS) showed a clear counterpart for all sources except for IGR J18214−1318. A clear image of this source is seen in Figure 1, from the medium-resolution spectrometer TÜBİTAK Faint Object Spectrometer and Camera (TFOSC) which is mounted on the Russian–Turkish 1.5 m telescope (RTT150) located at Turkish National Observatory (TUG), Antalya, Turkey. The camera is equipped with a 2048 × 2048, 15μm pixel (039 pixel⁻¹) Fairchild 447BI CCD chip. We took three 300 s observations of the field in B, V, R, and I filters on 2008 August 22, and only detected the source in the I band. After the standard bias and flat correction, we obtained the instrumental I magnitude of the counterpart using DAOPHOT in MIDAS. We carried out point-spread function photometry for the corrected image to obtain the instrumental magnitude, and calibrated it by comparing to the magnitudes of the reference stars, which were obtained from the USNO-B1.0 catalog.

Zoom In Zoom Out Reset image size

Between 2008 June 28 and June 30, we carried out spectroscopy of these targets with the Ritchey–Chrétien spectrograph on the 4 m Mayall Telescope at Kitt Peak National Observatory. We provide an observing log in Table 1. We used a slit width of 15, and rotated the slit to the parallactic angle. The wavelength range covered from 4750 Å to 9500 Å, with a dispersion of ∼3.4 Å pixel⁻¹ and a resolution of 13.8 Å.

Table 1. Observations: Kitt Peak Optical Spectroscopy

IGR Name	Optical Counterpart	R.A. (J2000)	Decl. (J2000)	Start Time (UT)	Exposure Time (s)
J16426+6536	USNO-B1.0 1555–0172189	16h43m0399	+65°32'512	2008 Jun 29, 4.0 hr	3600
J18214−1318	USNO-B1.0 0766–0475700	18h21m1972	−13°18'388	2008 Jun 28, 9.3 hr	3600
J18308−1232	USNO-B1.0 0774–0551687	18h30m4987	−12°32'192	2008 Jun 30, 8.8 hr	1800
J19267+1325	USNO-B1.0 1033–0440651	19h26m2700	+13°22'044	2008 Jun 28, 10.6 hr	3 × 600
J22292+6647	USNO-B1.0 1567–0242133	22h29m1390	+66°46'519	2008 Jun 30, 11.2 hr	827

Download table as:  ASCIITypeset image

After applying the flat field correction, subtracting the bias and dark current, and removing cosmic rays, we extracted the spectra with standard procedures in Image Reduction and Analysis Facility (IRAF). We observed arc lamps for wavelength calibration immediately preceding and following each source. Checking against background sky lines, we find agreement to within ∼3 Å. We also observed spectrophotometric standard stars Feige 110 and BD +33°2642 for flux calibration. Conditions were variable throughout each night, making the flux calibration somewhat uncertain. We report fluxes using a single observation of Feige 110 for calibration because it falls roughly halfway between the other observations in flux, presumably providing a good indication of the average conditions. Note that using the other observations can change the flux measurements by up to ∼15%. All lines and their parameters are reported in Table 2.

Table 2. Spectral Lines of IGR Sources

Line	Quantitya	J16426+6536	J18214−1318b	J18308−1232	J19267+1325	J22292+6647
		(AGN)	(HMXB)	(CV)	(CV)	(AGN)
Hα	EW	290 ± 4c	<4.3	26 ± 1	64 ± 1	760 ± 20c
	Flux	13.8 ± 0.2c	<0.1	7.7 ± 0.2	28.4 ± 0.2	27.2 ± 0.8c
	λfit	8683 ± 3c	...	6562 ± 3	6564 ± 3	7307 ± 3c
Hβ	EW	46 ± 2	...	9.7 ± 1.3	49 ± 2	...
	Flux	2.0 ± 0.1	...	3.1 ± 0.4	15 ± 1	...
	λfit	6430 ± 3	...	4860 ± 3	4859 ± 3	...
Hγ	EW	25 ± 3	...	...	...	...
	Flux	1.0 ± 0.1	...	...	...	...
	λfit	5746 ± 4	...	...	...	...
${\rm [{O}\;\hbox{{\scriptsize }{\sc iii}}]\, \lambda 4959}$	EW	7.2 ± 1.1	...	...	...	...
	Flux	0.32 ± 0.05	...	...	...	...
	λfit	6560 ± 3	...	...	...	...
${\rm [{O}\;\hbox{{\scriptsize }{\sc iii}}]\, \lambda 5007}$	EW	32 ± 1	...	...	...	...
	Flux	1.4 ± 0.1	...	...	...	...
	λfit	6624 ± 3	...	...	...	...
${\rm O\;\hbox{{\scriptsize }{\sc i}}\, \lambda 7772,7774,7775}$	EW	...	...	...	2.2 ± 0.4	...
	Flux	...	...	...	1.2 ± 0.2	...
	λfit	...	...	...	7774 ± 3	...
${\rm He\;\hbox{{\scriptsize }{\sc i}}\, \lambda 4921}$	EW	...	...	...	4 ± 1	...
	Flux	...	...	...	1.1 ± 0.4	...
	λfit	...	...	...	4920 ± 4	...
${\rm He\;\hbox{{\scriptsize }{\sc i}}\, \lambda 5015}$	EW	...	...	...	5 ± 1	...
	Flux	...	...	...	1.5 ± 0.3	...
	λfit	...	...	...	5015 ± 3	...
${\rm He\;\hbox{{\scriptsize }{\sc i}}\, \lambda 5876}$	EW	...	...	4.6 ± 0.6	12 ± 1	...
	Flux	...	...	1.4 ± 0.2	5.0 ± 0.2	...
	λfit	...	...	5874 ± 3	5876 ± 3	...
${\rm He\;\hbox{{\scriptsize }{\sc i}}\, \lambda 6678}$	EW	...	...	3.8 ± 0.5	8.2 ± 0.4	...
	Flux	...	...	1.2 ± 0.2	3.8 ± 0.2	...
	λfit	...	...	6677 ± 3	6678 ± 3	...
${\rm He\;\hbox{{\scriptsize }{\sc i}}\, \lambda 7065}$	EW	...	...	2.9 ± 0.5	6.1 ± 0.4	...
	Flux	...	...	0.91 ± 0.15	3.1 ± 0.2	...
	λfit	...	...	7063 ± 3	7065 ± 3	...
${\rm He\;\hbox{{\scriptsize }{\sc ii}}\, \lambda 5412}$	EW	...	...	...	5.1 ± 0.9	...
	Flux	...	...	...	1.8 ± 0.3	...
	λfit	...	...	...	5412 ± 3	...

Notes. ^aEW is the equivalent width in Å and line flux is measured in units of 10⁻¹⁵erg cm^-2 s⁻¹. Errors are at the 68% confidence level and upper limits are at the 90% confidence level. Note that the line flux is also subject to the 15% systematic uncertainty in overall flux level, which is not included here. The wavelength from the Gaussian fit is λ_fit, and the error is dominated by the ∼3 Å uncertainty in wavelength calibration. ^bThe fitted wavelengths of the Paschen and He I absorption lines are λ_P9 = 9234, λ_P10 = 9016, λ_P11 = 8865, λ_P12 = 8753, λ_P13 = 8669, λ_P15 = 8545, λ_P16 = 8501, λ_P17 = 8469, $\lambda _{He {\sc I}}=8780$. ^cHα is blended with [N ii] lines.

Download table as:  ASCIITypeset image

3.1. IGR J16426+6536

IGR J16426+6536 has one XMM-Newton slew source at α(J2000) = 18^h17^m22^s, δ(J2000) = −25°08'38'' (Ibarra et al. 2008a) and two ROSAT sources within the INTEGRAL error circle. There is one USNO-B1.0 source within one of the ROSAT error circles and none in the other. The XMM-Newton source has two USNO-B1.0 sources within its 8'' (1σ) error circle. We observed the brighter of the two sources (USNO-B1.0 1555-0172189) in the XMM-Newton error circle.

This spectrum has redshifted emission lines, including broad Balmer lines and narrow forbidden lines (Figure 2(a)), indicating that this source is a Seyfert 1 AGN. This is congruous with its position out of the Galactic plane, with Galactic coordinates l = 96.64°, b = +37.65°. To further classify the source, we use the scheme described in Winkler (1992), namely, the ratio of the flux in Hβ λ4861 to that in [O iii] λ5007, where these are the laboratory wavelengths. After dereddening the spectral lines using the Galactic absorption from Schlegel et al. (1998) along with the extinction law from Cardelli et al. (1989), we find F(4861)/F(5007) = 1.5 ± 0.1, which makes this a Seyfert 1.5 galaxy. Calculating the redshift from both Hβ and [O iii] λ5007 and averaging, we find z = 0.323 ± 0.001. Though, we did not observe all possible sources within the INTEGRAL error circle, the identification of the one we did observe as an AGN allows us to conclude that this is very likely the counterpart of the INTEGRAL source. We note that Masetti et al. (2009) came to similar conclusions from their spectroscopic analysis. Although we identify it as a Seyfert 1.5, while they find that it is a narrow-line Seyfert 1, there were several months between the observations, and this level of variability is not unheard of (e.g., Tran et al. 1992).

Zoom In Zoom Out Reset image size

With a redshift of z = 0.323, this source is at a luminosity distance of D_L = 1690 Mpc, using H₀ = 70 km s^-1 Mpc⁻¹, Ω_M = 0.3, and Ω_Λ = 0.7. We calculate the mass of the supermassive black hole using the relations described in Wu et al. (2004) and Kaspi et al. (2000). To use the relations, we recalculate the luminosity distance with the cosmological parameters used in those papers, H₀ = 75 km s^-1Mpc⁻¹, Ω_M = 1, and Ω_Λ = 0. This gives D_L = 1380 Mpc, from which we find the luminosity in Hβ (dereddened as described above), L_Hβ = 4.9 × 10⁴¹ erg s^-1. Combining this with the rest-frame FWHM velocity of the broad-line region (from Hβ), v_FWHM = 2000 km s^-1, gives a black hole mass of $M_{\rm BH}=9 \times 10^6 \mbox{$\,M_{\mathord \odot }$}$.

3.2. IGR J18214−1318

IGR J18214−1318 was observed by Chandra, which found one source at α(J2000) = 18^h21^m1976, δ(J2000) = −13°18'389, coinciding with USNO-B1.0 0766–0475700. The X-ray spectrum yields a column density of N_H = (11.7^+3.0_−2.7) × 10²² cm^-2 and Γ = 0.7^+0.6_−0.5. With a Galactic hydrogen column density of N_H = 2.4 × 10²² cm^-2, this implies a local absorption of N_H = (7–15) × 10²² cm^-2, which was suggested to be from the wind of a high-mass star (Tomsick et al. 2008a). This source was also observed by Swift, which found a position and photon index consistent with that obtained by Chandra, but a lower hydrogen column density of N_H = (3.5^+0.8_−0.5) × 10²² cm^-2, indicating that N_H is variable in this source (Rodriguez et al. 2009).

We find that the optical spectrum has a very reddened continuum (Figure 2(b)). From this, we infer that the source is relatively distant, since significant reddening would be expected for a distant source in the Galactic plane (l = 17.69°, b = +0.48°). Convolving the spectrum with the filters from Bessell (1990), we find V = 22.2 ±  0.2, R = 19.3 ±  0.2, I = 16.6 ±  0.2, where the errors are from the 15% systematic uncertainty in flux. We note that the I-band magnitude from the photometric observation made at TUG, I = 16.92 ± 0.17, agrees within the errors. We first consider the possibility that the source is a CV. Combining the apparent magnitudes with typical absolute magnitudes for CVs (M_V ∼ 9 and (V − R)₀ ∼ 0 (Masetti et al. 2009)) and a range of extinctions from A_V = 5–15 mag results in distances that range from d = 4–440 pc. This distance range is far too close to account for the observed amount of extinction (assuming that the optical extinction is interstellar). Furthermore, the spectrum lacks emission lines, while a CV would be expected to have strong emission lines. These arguments allow us to rule out the possibility that the source is a CV. Applying the same procedure to check for the possibility of an LMXB, we combined the apparent magnitude, the typical absolute magnitude for an LMXB (M_V ∼ 0 and (V − R)₀ ∼ 0; van Paradijs & McClintock 1995), and the same range of extinctions. In this case, the distances range from d = 300–28,000 pc, which are not inconsistent with the amount of extinction, but the source has a hard X-ray spectrum which would be very unusual for an LMXB (Muno et al. 2004).

These considerations allow us to conclude that it must be either an HMXB or a symbiotic star system. In Figure 3, spectral shapes typical of both are interpolated from the colors in Ducati et al. (2001) and plotted with extinctions of A_V = 5, 7.5, 10, 12.5, and 15, using the extinction relations from Cardelli et al. (1989). The infrared magnitudes from Deep Near Infrared Survey of the Southern Sky (DENIS) and 2MASS (reported in Tomsick et al. 2008a) along with the optical spectrum for the source are overlaid. These show that the data are consistent with a B0V star or an O9I star with an extinction of A_V = 12.5–15. The model K2III star spectra provide a worse fit to the data. The J-band magnitude requires an extinction of A_V ∼ 7.5, while the optical data are consistent with A_V = 10–12.5. Further evidence supporting an HMXB is shown in Figure 4, which is a close-up of the near-infrared region of the spectrum. Several Paschen absorption lines and one neutral helium line are observed (although P14 is not observed). These lines are indicative of an O or B star (e.g., Carroll & Ostlie 2006; Munari & Tomasella 1999). As a check, we indicate the positions of the Ca II triplet with the dot-dashed lines, and confirm that the absorption is more consistent with Paschen lines. In addition, other Paschen lines (P12, P11, P10, P9) are clearly detected. We also note that the upper limit on the equivalent width of Hα is consistent with Hα equivalent widths of OB stars (Leitherer 1988). Thus, we firmly identify the optical source as a high-mass star.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The Galactic hydrogen column density combined with the relationship between visual extinction and hydrogen column density detailed in Predehl & Schmitt (1995) (A_V = 0.56N_H[10²¹ cm^-2] + 0.23) gives a visual extinction of A_V = 13.7, which is consistent with the extinction inferred from colors for stars of type B0V and O9I. Considering a range of spectral types and using absolute magnitudes from Cox (2000) with A_V = 13.7, we find distances that range from d = 3–7 kpc for B0V to O5V stars and d = 9–10 kpc for A0I to O9I stars, giving an overall distance range of d = 3–10 kpc. However, using the general relation of 2A_V kpc⁻¹ (Tielens 2005), the lower limit of visual extinction (A_V = 12.5) implies a distance of at least 6 kpc, suggesting that the most likely distance range is d = 6–10 kpc. In addition, the observed variation in hydrogen column density argues in favor of a supergiant (and, therefore, a distance near the upper end of the range), since HMXBs with supergiant companions are known to show variations in column density (Prat et al. 2008).

3.3. IGR J18308−1232

3.4. IGR J19267+1325

3.5. IGR J22292+6647

We have firm identifications of five IGR sources. IGR J18308−1232 and IGR J19267+1325 are CVs. They have distances on the order of a couple hundred parsecs, which is typical of other CVs detected by INTEGRAL (e.g., Barlow et al. 2006; Masetti et al. 2006). The column densities from X-ray observations imply extinctions that are rather large given the distances we find. For example, IGR J19267+1325 is found to have 1.4 mag of extinction at a distance of 250 pc, while typical average extinction is A_V = 2 mag kpc^-1 (Tielens 2005). However, more reasonable distances and extinctions are feasible within the uncertainties. This could also mean that some of the X-ray absorption is intrinsic to the source. The identification of IGR J19267+1325 as a magnetic CV is not surprising, and, in fact, IGR J18308−1232 was also identified as a magnetic CV by Masetti et al. (2009). As a result of the hard X-rays emitted by magnetic systems, the majority of CVs detected by INTEGRAL have been magnetic (Barlow et al. 2006) even though such systems only comprise ∼10% of the CV population as a whole. INTEGRAL observations of CVs, therefore, unambiguously show that the magnetic field plays an important role in the hard X-ray emission of these systems.

We conclude that IGR J18214−1318 is an obscured HMXB. It has an extremely reddened continuum, indicating much absorption along the line of sight, which implies a fairly large distance. It has absorption lines that are typical of a high-mass star, and the observed colors and flux are consistent with an extincted high-mass main-sequence or supergiant star (and the variable hydrogen column density may argue for the latter). In addition, the X-ray spectrum shows significant local absorption, which could come from a wind from a high-mass star. An infrared spectrum may be one way to further classify the companion star. The infrared suffers less extinction than the optical, and more features would be observable. A higher resolution spectrum would also be valuable in identifying the luminosity and exact spectral class.

It is instructive to compare the X-ray luminosities for the above sources to typical HMXBs and CVs. In the Chandra band (0.3–10 keV), IGR J18214–1318 has an unabsorbed 0.3–10 keV flux of (60⁺³¹⁷₋₃₅) × 10⁻¹² erg cm^-2 s⁻¹ (Tomsick et al. 2008a). Taking a distance of 8 kpc, this results in a luminosity of L_{0.3–10 keV} = 5 × 10³⁶ erg s^-1. This is consistent with the average properties of neutron star HMXBs given in van Paradijs & McClintock (1995). While the hard X-ray spectrum and luminosity argue for a neutron star, detection of X-ray pulsations would be necessary to rule out a black hole. IGR J18308−1232 has F_{0.3–10 keV} = (2.3^+0.8_−0.4) × 10⁻¹¹ erg cm^-2s⁻¹ (J. A. Tomsick et al. 2009, in preparation), which implies a luminosity of L_{0.3–10 keV} = 1.3 × 10³² erg s^-1 with our calculated distance (220 pc). The INTEGRAL flux from 20 to 100 keV of F_{20–40 keV} = 0.8 ± 0.1 mcrab and F_{40–100 keV} = 1.0 ± 0.2 mcrab (Bird et al. 2007) implies a luminosity of L_{20–100 keV} = 9 × 10³¹ erg s^-1. We compare this luminosity to known IP CVs in Suleimanov et al. (2005) over the energy range 0.1–100 keV, assuming that the flux from the range where we lack data is of the same order as that from the Chandra range (a reasonable assumption, given the spectral shapes in the above paper). Our luminosity falls in the range of known IP CVs, although it is on the low end. IGR J19267+1325 is a similar case. Its flux in the Chandra band is F_{0.3–10 keV} = (8.9 ±  0.6) ×  10⁻¹² erg cm^-2 s⁻¹ (Tomsick et al. 2008b), which implies a luminosity of L_{0.3–10 keV} = 6.7 × 10³² erg s^-1 with our calculated distance (250 pc). The INTEGRAL flux of F_{20–40 keV} = 0.7 ± 0.1 mcrab and F_{40–100 keV} = 0.6 ± 0.2 mcrab (Bird et al. 2007) implies a luminosity of L_{20–100 keV} = 8 × 10³¹ erg s^-1. Again, this results in a total luminosity on the low end of the range of known CV luminosities. These low values could be an indication that the distances lie at the high end of our error range.

IGR J16426+6536 and IGR J22292+6647 are Seyfert 1 AGN, and IGR J16426+6536 is further classified as a Seyfert 1.5 AGN. The 127 INTEGRAL-detected Seyferts (which have an average redshift z = 0.033) and the 34 IGR Seyferts (with average redshift z = 0.035) are plotted in Figure 5. With redshifts of z = 0.323 and z = 0.113, respectively, IGR J16426+6536 and IGR J22292+6647 are considerably more distant than average. In fact, IGR J16426+6536 is an interesting case since it is the highest redshift IGR Seyfert, though there is one INTEGRAL-detected Seyfert, PKS 0637–752, with a higher redshift of z = 0.651.

Zoom In Zoom Out Reset image size

J.A.T. and S.C.B. acknowledge partial support from Chandra awards G07-8050X and G08-9055X issued by the Chandra X-Ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of the National Aeronautics and Space Administration (NASA), under contract NAS8-03060. S.C.B. is partially supported by a UC Berkeley Chancellor's Fellowship and an SSL Summer Science Fellowship. E.K. acknowledges partial support from TÜBİTAK, and Turkish National Academy of Sciences Young and Successful Scientist award (GEBİP), and European Commission FP 6 Marie Curie Transfer of Knowledge Project ASTRONS, MTKD-2006-42722. We thank Hal Halbadel and Daryl Willmarth for their help at Kitt Peak and Jules Halpern for sharing optical images. This publication 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 NASA and the National Science Foundation. This research also makes use of the USNOFS Image and Catalogue Archive operated by The United States Naval Observatory, Flagstaff Station and the Deep Near Infrared Survey of the Southern Sky (DENIS). We also used the Digitized Sky Survey, which was produced at the Space Telescope Science Institute under U.S. Government grant NAG W-2166. The images of these surveys are based on photographic data obtained using the Oschin Schmidt Telescope on Palomar Mountain and the UK Schmidt Telescope. We thank the anonymous referee for helpful suggestions.