DISCOVERY OF AN X-RAY-EMITTING CONTACT BINARY SYSTEM 2MASS J11201034−2201340

After the detection of 1FGL J1119.9−2205 (also named 2FGL J1120.0−2204 and 3FGL J1119.9−2204), the Swift XRT took 27 exposures of this field for a total exposure time of ∼67 ks between 2010 and 2013. Hui et al. (2015) studied all of the Swift observational data and found two millisecond pulsar candidates within the 95% error ellipse of 3FGL J1119.9−2204. On the other hand, several uncataloged X-ray point sources outside the error ellipse were also detected, but their properties were not further investigated. 2MASS J11201034−2201340, which has relatively faint X-ray emission, is one of the outliers unrelated to the γ-ray source 3FGL J1119.9−2204.

The energy range of XRT is 0.2–10 keV, the pixel scale is 236, and the FWHM of the point-spread function (PSF) is roughly 7'' in 1.5 keV (or a half-power diameter of 18'' ). All the data, including two target IDs (41371 and 49351), were used to determine the X-ray positions of the point sources. We extracted photon events and X-ray images from the standard products of all the XRT observations using xselect version 2.4. The point sources were detected using the detect task of the multi-mission X-ray image analysis program XIMAGE, for which the signal-to-noise threshold was set to 3.0. The position and corresponding uncertainty were determined using the xrtcentroid task. We found an X-ray point source located at R.A. = 11^h20^m1065 and decl. = −22°01'307 with a 90% uncertainty of ∼ 76. Figure 1(a) is the Swift image of the field containing 2MASS J11201034−2201340 and 3FGL J1119.9−2204.

Zoom In Zoom Out Reset image size

2FGL J1120.0−2204 was observed by the XMM-Newton observatory on 2014 June 14 for a total exposure time of ∼70 ks (ObsID 0742930101). All three detectors of the European Photon Imaging Camera (EPIC) were used in this observation. The MOS1 and MOS2 detectors were operated in full-frame mode with a timing resolution of 2.6 s and an on-axis PSF FWHM of 6'' (half-energy width 136). The pn detector was operated in timing mode with an extreme timing resolution of 0.03 ms. In this mode, all X-ray photons are compressed in one dimension. However, this observation was performed to investigate the timing properties of the millisecond pulsar candidate 2FGL J1120.0-2204. The pn data were useless because the target is far from the aim point of the observation. The optical/UV monitor also was used in this observation but our target was outside the field of view and the optical/UV monitor data were unavailable. We applied the pipeline task emproc of the XMM Science Analysis Software (XMMSAS version 15.0.0) program to the MOS data using the latest instrumental calibration database. Events with patterns $\lt 12$ were adopted in this research, and the flaring background was filtered out when the entire count rate was $\gt 3.5$ counts s⁻¹. We performed source detection using the maximum likelihood fitting with the aid of the XMMSAS task edetect_chain and the signal-to-noise threshold set to $4\sigma$.

We found a source located at R.A. = 11^h20^m1032 and decl. = −22°01'356 with a 90% uncertainty of 2''. The detection likelihood is 34.5, which corresponds to a 8σ significance level. This source is likely a point source because the likelihood of source extent is not determined. Figure 1(b) shows the XMM-Newton image of the region containing 2MASS J11201034−2201340 where a point source is clearly seen.

The field around the X-ray source was surveyed in the optical and infrared bands by USNO-B-1.0, the Two Micron All Sky Survey (2MASS), and the Wide-field Infrared Survey Explorer (WISE). All three catalogs are well organized by the the NASA/IPAC Infrared Science Archive (IRSA). Each catalog has only one source that is within the PSF of the XMM-Newton: USNO-B1.0 0679−0311979 at R.A. = 11^h20^m1036 and decl. = −22°01'3414, 2MASS J11201034−2201340 at R.A. = 11^h20^m1035 and decl. = −22°01'3404, and WISE J112010.35−220134.1 at R.A. = 11^h20^m1035 and decl. = −22°01'3416. 2MASS J11201034−2201340 is 53 from the Swift determined X-ray source and 15 from the position determined by XMM-Newton. Hence, 2MASS J11201034−2201340 is likely to be the optical counterpart of the X-ray source because the angular separation between the optical and X-ray positions reside well within the uncertainties of Swift and XMM-Newton.

We also observed the field with the Lulin One-Meter Telescope (LOT), located at Central Taiwan, using the ALTA U42 CCD with SDSS r'-band and g'-band filters. This configuration has a pixel scale of 032 and a limiting magnitude of ∼19.5 at the detection level of 5 σ. Figure 1(c) shows the cropped r'-band image. Only one source is detected in the region of the PSF of XMM-Newton.

To investigate the temporal variation of 2MASS J11201034−2201340, we first collected the multi-epoch photometric data obtained by the Catalina Real-Time Transient Survey (CRTS) to search for periodicity. The CRTS data set contains the observations made by the Siding Springs Survey 0.5 m Schmidt telescope (218 exposures) and by the Catalina Sky Survey 0.7 m Schmidt telescope (76 exposures). We found a significant periodicity with a period of 0.144 day by applying the Lomb–Scargle periodogram (Scargle 1982; Horne & Baliunas 1986). Figure 2 shows the light curve from the CRTS and the power spectrum. One-year aliases due to the observational window function also were detected in the power spectrum.

Zoom In Zoom Out Reset image size

To check the robustness of the detected result and the possibility that the true periodicity is twice the detected value, we observed the field using the R and V bands of the 0.4 m Super Light Telescope (SLT) at the Lulin Observatory on 2015 April 24–27. The SLT was equipped with an Andor iKon-L 936 CCD with a field of view of 28 × 28' and a pixel scale of 082. The exposure time for each frame was 5 minute, and the observation lasted ∼5 hr on April 24, 27, and 2.2 hr on April 26. The R band had 117 exposures and the V band had 115. For comparing the phase alignment of the CRTS data, the SLT data, and the data from other wavelengths, the observation times were corrected from the Earth to the solar barycenter. The CCD reduction package of the Image Reduction and Analysis Facility was used for standard image reduction, including bias, dark current, and flat-field corrections. We then compared the instrumental magnitude of 2MASS J11201034−2201340 with that of four comparison stars in the same field to obtain differential photometric data. The light curves clearly showed variability on a timescale of several hours. After folding both the light curves and the B − V color curve using 0.144 and 0.288 day, respectively, we found that the color seemed to vary with a period of 0.288 day (see Figure 3). The true period of $0.28876208\pm 5.6\times {10}^{-7}$ day is twice that detected by the Lomb–Scargle periodogram from the CRTS light curve. The uncertainty was estimated by a combination of the limited time span and photon statistical error. The uncertainty caused by limited time span and the strength of the signal was estimated using Equation (3) in Levine et al. (2011) and the statistical error was estimated using a 10⁴ times Monte Carlo simulation.

Zoom In Zoom Out Reset image size

From the archived multiepoch photometry data of WISE observations of this field, we found 52 W1-band (3.4 μm) and 42 W2-band (4.6 μm) significant detections and folded them with the best-determined periodicity. Figure 3 shows that there is still modulation and it is coherent in the mid-infrared bands.

We further investigated the physical properties of this system using currently available broadband SED data. For the optical band, we used the photometric B-, R-, and I-band data in the USNO-B 1.0 catalog, where $B1=16.4$, $B2=16.03$, $R1=14.36$, $R2=14.96$, and I = 14.55. To describe the spectral behavior of a contact binary system, it is better to use one of the flux minimum in the folded light curve that has all of the photons from the primary component. However, the amount of current data is insufficient because only one or two measurements were made for individual bands. Therefore, we took the average of the detected magnitudes for the B and R bands, for which there were two measurements, and applied a typical photometric uncertainty of 0.3 mag (Monet et al. 2003). The near-infrared data were from 2MASS in the J, H, and Ks bands, where $J=13.875\pm 0.03$, $H=13.354\pm 0.03$, and ${Ks}=13.276\pm 0.04$. The mid-infrared data were from WISE in the W1 and W2 bands, where $W1=13.286\pm 0.025$ and $W2=13.331\pm 0.031$.

The SLT did not observe standard stars, and so calibration in the V and R bands was not possible. Instead, to estimate the apparent magnitude of 2MASS J11201034−2201340, we used the cataloged star TYC 6090-207-1, which is a bright star with a coordinate of R.A. = 11^h19^m56983 and decl. = −22°04'4394 and magnitudes of $B=13.09\pm 0.26$ and a $V=12.20\pm 0.16$ (Høg et al. 2000). By scaling of the brightness of 2MASS J11201034−2201340 and TYC 6090-207-1, we estimated the V-band magnitude of 2MASS J11201034−2201340 as 15.37 ± 0.16.

Before constructing the SED, the measured magnitudes underwent extinction correction using the online Galactic dust extinction tool provided by IRSA. We assumed that the relationship between the extinction (A_V) and reddening ($E(B-V)$) is ${A}_{V}=3.1E(B-V)$ (Güver & Özel 2009). We estimated the extinction for individual bands according to the latest measurements provided by Schlafly & Finkbeiner (2011). Figure 4 shows the constructed SED of 2MASS J11201034−2201340. We compared the SED with various stellar spectral templates presented in Pickles (1998) and found that our SED fit that of a K2V star that has a typical surface temperature of 4960 K. This indicates that the major component of 2MASS J11201034−2201340 is a late-type star.

Zoom In Zoom Out Reset image size

To determine the physical nature of this contact binary system using the available data, we fit the observed V-band and R-band light curves using the Wilson–Devinney code (Wilson & Devinney 1971; Wilson 1979, 1990, 1994) to obtain a photometric solution. The surface temperature of the more massive star (T₁) was fixed at 4960 K as suggested by the SED fitting. The gravity darkening parameters were fixed at ${g}_{1}={g}_{2}=0.32$, according to the prediction made by Lucy (1967), and the bolometric albedos were fixed at ${A}_{1}={A}_{2}=0.5$ (Ruciński 1969) on the basis of the assumption that both components have convective envelopes. The limb-darkening coefficients for bolometric and individual bands (${x}_{1\mathrm{bol}}$, ${x}_{2\mathrm{bol}}$, ${y}_{1\mathrm{bol}}$, ${y}_{2\mathrm{bol}}$, x_1V, x_2V, y_1V, y_2V, x_1R, x_2R, y_1R, and y_2R) were estimated using the values provided by van Hamme (1993; see Table 1). The remaining free parameters were the mass ratio (q), the inclination angle of the orbital plane (i), the surface temperature of the less massive component (T₂), the phase shift of the V and R bands, the dimensionless potential of the more massive star (${{\rm{\Omega }}}_{1}={{\rm{\Omega }}}_{2}$), and the relative luminosity (L₁ and L₂) of each star in the V and R bands.

Table 1.  Photometric Solutions for 2MASS J11201034−2201340

Parameters	Photometric Elements	Errors	Photometric Elements	Errors
	Without Cool Spot		Cool Spot Solutions
${g}_{1}={g}_{2}$	0.32	assumed	0.32	assumed
${A}_{1}={A}_{2}$	0.50	assumed	0.50	assumed
${x}_{1\mathrm{bol}},{x}_{2\mathrm{bol}}$	0.299, 0.299	assumed	0.299, 0.299	assumed
${y}_{1\mathrm{bol}},{y}_{2\mathrm{bol}}$	0.396, 0.396	assumed	0.396, 0.396	assumed
${x}_{1V},{x}_{2V}$	0.581, 0.581	assumed	0.581, 0.581	assumed
${y}_{1V},{y}_{2V}$	0.249, 0.249	assumed	0.249, 0.249	assumed
${x}_{1R},{x}_{2R}$	0.356, 0.356	assumed	0.356, 0.356	assumed
${y}_{1R},{y}_{2R}$	0.413, 0.413	assumed	0.413, 0.413	assumed
T1	4960K	fixed	4960K	fixed
Phase Shift	−0.0153	±0.0005	−0.0155	±0.0004
q	0.291	±0.006	0.317	±0.004
${{\rm{\Omega }}}_{1}={{\rm{\Omega }}}_{2}$	2.4253	±0.0126	2.4735	±0.0111
${{\rm{\Omega }}}_{\mathrm{in}}$	2.4464	⋯	2.5026	⋯
${{\rm{\Omega }}}_{\mathrm{out}}$	2.2645	⋯	2.3060	⋯
T2	5125 K	±20 K	5124 K	±16 K
i	894	±3.3	885	±2.7
${L}_{1}/({L}_{1}+{L}_{2})(V)$	0.7180	±0.0051	0.7019	±0.0046
${L}_{1}/({L}_{1}+{L}_{2})(R)$	0.7259	±0.0045	0.7100	±0.0042
θ (°)	⋯	⋯	86.5	±2.6
ψ (°)	⋯	⋯	92.2	±4.7
${\rm{\Omega }}(^\circ )$	⋯	⋯	11.9	±1.2
${T}_{s}/{T}_{\ast }$	⋯	⋯	0.80	±0.09
f	11.6%	$\pm 6.9\%$	14.8%	±5.7%

Download table as:  ASCIITypeset image

Because the spectral mass ratio could not be determined using the radial velocity method, we searched for a proper initial guessing value before fitting the light curves. We estimated the sum of the weighted square deviations (Σ) with respect to different trial values of q, ranging between 0.1 and 6. A minimum of Σ was clearly seen at q = 0.3, so we chose that as the initial q values for the fitting. Moreover, the profile seemed to be asymmetric, so we introduced a cool (dark) spot in the fitting to compare against the model without a cool spot. The cool spot model included three additional parameters: the colatitude (θ), longitude (ψ), and radius (Ω) of the spot. The best-fit parameters are given in Table 1 and the model light curves are shown in Figure 5.

Zoom In Zoom Out Reset image size

The result of the fitting indicates that 2MASS J11201034−2201340 is a short-period A-type contact binary with a nearly edge-on orbital plane. It is necessary that there be a cool spot on the more massive component to explain the O'Connell effect. In this scenario, the best-fit mass ratio is q = 0.317 and a fill-out factor is estimated to be $f=14.8\%$, which suggests a low degree of over contact. From the averaged value of the phase shift, we proposed a linear ephemeris for the primary minimum defined as

$$\begin{eqnarray*}&&{T}_{{\rm{primary}}}={\rm{MJD}}\;{\rm{(TDB)}}\;{\rm{}}53791.3984(1)+0.28876208(56)\times N,\end{eqnarray*}$$

where N is the number of cycle count.

The magnitude of the secondary minimum (phase ∼0.5), which is not much different from that of the primary in the folded light curve, indicates that the less massive star is completely obscured and only the more massive star contributes to the flux. The V-band magnitude of the secondary minimum is estimated as 15.75 ± 0.16 in this measurement. Therefore, the distance can be roughly estimated as 690 ± 50 pc after considering the extinction and assuming that the more massive component is a K2V star.

Investigation of the timing and spectral properties of the X-ray emission from 2MASS J11201034−2201340 detected by XMM-Newton will add to our understanding of the nature of this system. After filtering out the flaring background, the effective exposure time was ∼57 ks. We extracted 238 X-ray photons from a 20 arcsec radius circular region centered at the detected centroid of 2MASS J11201034−2201340 (see Figure 1). Because the effective exposure time encompassed only approximately two orbital cycles of 2MASS J11201034−2201340, we needed to carefully examine background variability. We selected 10 backgrounds with the same area in the source-free region on the same chip and found that the background counts were ∼150. Therefore, only ∼90 X-ray photons that originated from the source were available for use in a marginal investigation of the timing and spectral analysis. First, we individually folded the source and all the backgrounds according to the best period determined in the optical band. Then, we obtained a mean folded light curve by averaging all the folded background light curves to obtain the fluctuation of the background in the phase domain. The clean folded light curve shown in Figure 3(f) was obtained by subtracting the background profile from the folded source light curve. Another way to compute the background contribution is by weighting a larger source-free region, e.g., a circle with an 80 arcsec radius or a large annulus around the source. Both methods yield similar results. The folded light curve showed no detection of significant X-ray variability. We used different bin sizes and applied the ${\chi }^{2}$ test to the folded profile and found that the detection significance in all cases was $\lt 1\sigma$. The lack of detection of variability means that the geometry of the X-ray emission area may differ from that of the optical area.

The high-energy emission mechanism can be investigated by X-ray spectral analysis, despite the small number of X-ray photons. The source selection criterion was the same as that for the X-ray timing analysis above, while we used an 80 arcsec radius circle around a source-free region to estimate the background spectrum. The response matrix was created using the XMMSAS task rmfgen and the ancillary response file was created using arfgen. The X-ray photons were further grouped to have at least 15 counts per spectral bin. No pile-up issue was addressed for this faint source.

The X-ray spectral fitting was achieved using XSPEC v12.9.0. The X-ray emission from a contact binary system may come from a hot coronal plasma and may be represented with a thermal or non-thermal model. We tried two typical spectral models: non-thermal power law (XSPEC model powerlaw), and the thermal Bremsstrahlung (XSPEC model bremss) to fit the X-ray spectrum. The spectral fitting was calculated within the energy range 0.2–10 keV. We first set the ${N}_{{\rm{H}}}$ to be a free parameter and yielded a best-fit value of $2.7\times {10}^{-20}$ cm⁻², which is not very far from $3.65\times {10}^{-20}$ cm⁻² derived from Leiden/Argentine/Bonn (LAB) Survey (Kalberla et al. 2005). However, the uncertainty is $5\times {10}^{-21}$ cm⁻², which is too large to well constrain the ${N}_{{\rm{H}}}$ value. Therefore, we fixed the ${N}_{{\rm{H}}}$ value at the galactic one of $3.65\times {10}^{-20}$ cm⁻² in the following analysis. The power-law model yielded an acceptable fitting with a photon index of 2.4 ± 0.5 and a ${\chi }_{\nu }^{2}=1.09$. The uncertainties of spectral parameters are estimated within the 90% confidence interval. On the other hand, the thermal Bremsstrahlung model result in an equally good fitting with a slightly larger ${\chi }_{\nu }^{2}$ of 1.15 and a plasma temperature of ${kT}={0.8}_{-0.3}^{+1.4}$ keV. Both of the single component model can describe the X-ray spectrum of 2MASS J11201034−2201340 very well and no additional components are required. The best-fit parameters for the two models are given in Table 2, and the spectral data and corresponding best-fit models are shown in Figure 6. The X-ray flux in the 0.3–10 keV energy range was estimated as ${1.7}_{-0.5}^{+0.9}\times {10}^{-14}$ erg cm⁻² s⁻¹.

Zoom In Zoom Out Reset image size

Table 2.  The Best-fit Parameters of two Different Models for the X-ray Spectrum of 2MASS J11201034−2201340 Obtained with XMM-Newton

	Power Law	Bremsstrahlung
Γ	2.4 ± 0.5	⋯
kT (eV)	⋯	${0.8}_{-0.3}^{+1.4}$
Normalization	${4.2}_{-1.3}^{+1.2}\times {10}^{-6}$	${1.1}_{-0.6}^{+1.0}\times {10}^{-6}$
${\chi }_{\nu }^{2}$ (dof)	1.09(12)	1.15 (12)

Download table as:  ASCIITypeset image

Because the fitting of the optical folded light curve suggested the presence of a cool spot, the existence of X-ray emission suggests chromospheric activity. The result of the X-ray spectral fitting also suggests a possible non-thermal or thermal Bremsstrahlung origin for the X-ray emission. However, the X-ray orbital profile shows no significant variations, but this may be due to an insufficient number of X-ray photons. Considering the stability of star spots, it is possible that their position and size in these two observations are quite different. To investigate the origin of the X-ray emission and its connection to the star spot, simultaneous X-ray and optical observations using an X-ray telescope with a large effective area are necessary.

Although only ∼ 30X-ray photons were detected in all the Swift observations, we were able to estimate a crude X-ray spectrum and the corresponding X-ray flux by using the online tool provided by UK Swift Science Data Centre. The flux of X-rays between 0.3 and 10 keV determined by Swift was ${1.0}_{-0.5}^{+0.6}\times {10}^{-14}$ erg cm⁻² s⁻¹, which is consistent with the value obtained by a single XMM-Newton observation (${1.7}_{-0.5}^{+0.9}\times {10}^{-14}$ erg cm⁻² s⁻¹). This may indicate that no significant X-ray flare was detected during the Swift observations.

Using the distance estimated from the SED, the X-ray intensity was estimated to be $(0.7-1.5)\times {10}^{30}$ erg s⁻¹. Factoring in the relationship between the X-ray intensity and the orbital period (Chen et al. 2006), 2MASS J11201034−2201340 is within a reasonable range on the X-ray intensity versus orbital period plot (see Figure 7). Stȩpień et al. (2001) categorized W UMa-type stars into hot $[{(B-V)}_{0}\leqslant 0.6]$ and cool $[{(B-V)}_{0}\gt 0.6]$ groups. The X-ray flux seems to positively correlate with the color index for the hot group, whereas it reaches a constant value after ${(B-V)}_{0}\gt 0.6$. Judging from the SED fitting, 2MASS J11201034−2201340 is likely to be in the cool group. The ratio of the X-ray and bolometric fluxes, $\mathrm{log}({L}_{x}/{L}_{\mathrm{bol}})$, is roughly estimated to be between −3 and −3.3, which is reasonably close to the value presented in Stȩpień et al. (2001). Accurate multi-band photometric observations and standard star calibration are required to investigate the nature of this system in detail.

Zoom In Zoom Out Reset image size