XMM-Newton Survey of Local O vii Absorption Lines in the Spectra of Galactic X-Ray Sources

Yang Luo; Taotao Fang; Renyi Ma

doi:10.3847/1538-4365/aab270

1. Introduction

The presence of the hot interstellar medium (ISM) has been predicted for a long time (for a review, see Spitzer 1956; Ferriere 2001; Putman et al. 2012). This hot gas (∼10⁶ K) could be a result of stellar feedback (supernova explosions or stellar wind from massive stars) or a shock-heated intergalactic medium (IGM) accreting to the Galactic center, therefore studying the hot ISM could have profound implications on the theory of galaxy formation and evolution.

The current understanding of Galactic hot gas initially came from various broadband observations of the diffuse soft X-ray background. At the soft X-ray band, other than the extragalactic component, the solar wind charge exchange, and the local hot bubble component, a contribution from the background is expected from the large-scale hot plasma present in the Galactic halo (e.g., Lallement 2004; Cappelluti et al. 2012; Henley & Shelton 2013; Galeazzi et al. 2014). Insights into the properties of the hot ISM also come from the study of z = 0, metal absorption lines in extragalactic active galactic nuclei (AGNs). The detection of z = 0, highly ionized absorption lines (e.g, Ne ix, O vii and O viii) toward bright extragalactic sources (e.g., Nicastro et al. 2002, 2016; Fang et al. 2003; Rasmussen et al. 2003), and their Galactic origins (Fang et al. 2006; Bregman & Lloyd-Davies 2007; Nicastro et al. 2016), also suggested the existence of the hot ISM.

However, current data cannot give a precise description of the spatial distribution of the hot gas. Evidence from the disk-morphology of the X-ray emission of nearby galaxies is suggestive of the gas confined in the disk (Tüllmann et al. 2006; Li & Wang 2013). However, semi-analytic calculations and numerical simulations for disk galaxy formation predict the existence of extended hot gaseous halos around massive spiral galaxies due to the accretion of the IGM (e.g., White & Frenk 1991; Mo & Mao 2002; Maller & Bullock 2004; Fukugita & Peebles 2006; Crain et al. 2010; Ntormousi & Sommer-Larsen 2010; Marinacci et al. 2011; Stinson et al. 2012; Fang et al. 2013; Feldmann et al. 2013; Sokołowska et al. 2016). The hot ISM may extend out to at least the Magellanic System, as suggested by the pressure confinement of high-velocity cloud or satellite galaxies stripping stream (Sembach et al. 2003; Collins et al. 2005; Grcevich & Putman 2009; Lehner et al. 2012; Gatto et al. 2013). Detections of diffuse X-ray halos around massive normal spiral galaxies also suggested the existence of hot, X-ray emitting gas beyond the disk (Anderson & Bregman 2011; Bogdan et al. 2013a, 2013b; Dai et al. 2012; Hodges-Kluck & Bregman 2013; Anderson et al. 2016).

In the past few years, it has become well established that a large amount of hot gas is present along the sightline toward background XRBs. Located in the Galactic disk, the XRBs provide a potential probe of the ISM in the disk. The detection of X-ray absorption lines of highly ionized species (e.g, Ne ix, O vii and O viii) toward the XRBs leads to the assumption that the hot gas is located in the Galactic thick disk (e.g., Futamoto et al. 2004; Wang et al. 2005; Yao & Wang 2005; Yao et al. 2009; Pinto et al. 2013). This hot ISM derived from XRB observations has an exponential scale height of several kiloparsecs (Yao & Wang 2005; Yao et al. 2009). The hot gas near the Galactic plane has a density around 1 × 10⁻³ cm⁻³ (Yao et al. 2009; Hagihara et al. 2010; Sakai et al. 2014) and a temperature around 2 × 10⁶ K (Yao & Wang 2007). By comparing the amount of absorption detected along the sightline toward the XRBs versus those detected toward the distant AGNs, it has been concluded that most of the hot gas is constrained within the disk (Yao et al. 2008).

However, the circumstellar medium local to XRBs could also have produced such absorption lines. Miller et al. (2004) analyzed the absorption lines from GX 339–4 and XTE J1650–500 and they found these highly ionized absorption lines could not be produced in the ISM, but in a photoionized warm absorber intrinsically. Later work on the time variability of the absorption lines showed that in some targets the line strength varies between observations taken at different times (Cackett et al. 2008; Luo & Fang 2014; Petrucci et al. 2014). These lines are likely photoionized by the XRB luminosity with an ionization parameter log(ξ) in the range of 1–3 (Cackett et al. 2008; Luo & Fang 2014). Some of the highly ionized lines also are found to be blueshifted, and could be modeled with outflowing photoionized gas near the accretion disk or its corona (e.g, SAX J1808.4–3658 observed by Pinto et al. 2014). Studying the origins of these highly ionized absorption lines would benefit our understanding of both the circumstellar medium intrinsic to the XRB and the ISM, as well as the hot gas content in our Galaxy in general.

In order to fully understand the origin of the absorption lines detected in the X-ray spectra of the XRBs and therefore the properties of the hot ISM, a large sample of targets with detections of hot gas produced by absorption lines is required. In this work, we perform a comprehensive analysis of the highly ionized absorption lines in the XRBs, with a focus on the highly ionized O vii absorption line at 21.6 Å. We have performed a comprehensive analysis of all the available data in the XMM-Newton archive with a concentration on O vii lines. Recent work by Fang et al. (2015) provided an all-sky survey of O vii toward background AGNs from archival XMM-Newton observations, and they find a 100% sky-covering fraction of O vii lines, which suggest a uniform distribution of the absorbers. Here, we present the second paper in this series.

The paper is organized as follows. In Section 2, we first describe the selection of our sample and the procedure of data reduction, and then we analyze the absorption lines and present our main results in this section. We discuss the distribution of the absorption line in Section 3, as well as the origins of these lines. We also comment on several targets with newly detected O vii lines in this section. The last section is the summary.

2. Observation and Data Analysis

2.1. Target Selection and Data Reduction

Numerous X-ray absorption lines were detected in the X-ray spectra of the XRBs. These absorption lines sample the ISM at a variety of temperature ranges, providing rich information on the multi-phase distribution of the ISM (e.g., Ferriere 2001; Yao & Wang 2005; Costantini et al. 2012; Pinto et al. 2013). Here, we focus on highly ionized, He-like O vii for the following reasons. First, as one of the most abundant elements in the universe, oxygen is a very important tracer of the metals in the ISM. Second, under collisional ionization, the peak temperature of O vii ionization fraction is in the range of 10^5.5–10^6.5 K, providing an effective way to probe the hot, ionized ISM. Therefore, in our work, we focused on the absorption line from O vii K_α transition with a rest-frame wavelength of λ_rest = 21.6019 Å (Verner et al. 1996; Yao et al. 2009; Liao et al. 2013).

Currently, there are three grating-based, high-resolution X-ray spectrometers suitable to our study of highly ionized metal absorption lines, namely the Low and High Energy Transmission Grating Spectrometer⁴ (LETG and HETG) on board the Chandra X-ray Observatory and the Reflection Grating Spectrometer⁵ (RGS) on board the XMM-Newton X-ray telescope. RGS has a relatively higher collecting area than both LETG and HETG. For RGS, there are two units, RGS1 and RGS2. However, one CCD assembly in each RGS unit has an operation failure due to electronics problems. These are RGS1 CCD7 and RGS2 CCD4, roughly covering the wavelength ranges 11–14 Å and 20–24 Å, respectively, in the first order. Therefore, in this work, we will focus on the RGS1 data only.

Based on Chandra observation, Luo & Fang (2014) studied a sample of Ne ix lines toward XRBs. Due to the small effective area and strong galactic absorption, the detected O vii photon counts from Chandra are small. In our selection of all XRB observations, the number of XRBs observed by Chandra is smaller than that of RGS, therefore we could not create a large well-defined O vii sample from Chandra. At a wavelength of 19.0 Å, the RGS1 has one instrumental feature that makes it unlikely for a good fit of the O viii K_α line. At a wavelength of 18.6 Å we found some weak detections of the O vii K_β line in part of our sample. Due to the small line oscillation strength, the line EWs and their significances could not be well constrained. We have tried to jointly fit the O vii K_α and K_β, and the joint fit could not give a better improvement of the line EWs. Therefore, in our work, we will concentrate on the O vii K_α line from RGS1 observation.

We have compiled a list of XRBs from the XMM-Newton archive. We inspected the RGS data of these targets and selected the ones with relatively high photon statistics. We define our selection criteria as follows: the counts per resolution element (CPRE) at wavelengths near O vii must be at least 20 photons. For each source, we define the CPRE, which is the photon counts within a bin size of the spectral resolution (50 mÅ), as

$\begin{eqnarray}&&\mathrm{CPRE}={F}_{\lambda }{AT}{\rm{\Delta }}\lambda ,\end{eqnarray} \tag{ 1 }$

where F_λ is the photon flux, A is the effective area of the detector, T is the observation time, and Δλ is the width of one resolution element. We also ignore those targets with O vii K-shell emission lines (consisting of resonance, intercombination, and forbidden lines at 21.60, 21.80, and 22.10 Å, respectively) to avoid contamination. This leads to a sample of 33 sources, with 29 low-mass X-ray binaries (LMXBs) and 4 high-mass X-ray binaries (HMXBs). In Figure 1 we plot the all-sky Hammer–Aitoff projection of our targets in the Galactic coordinates. Most of our targets are located in or near the Galactic disk.

**Figure 1.** An all-sky Hammer–Aitoff projection of targets in our sample.
Download figure:
Standard image High-resolution image

We have reduced the spectra with the Science Analysis System (SAS) version 13.0.⁶ RGS data are processed with the SAS task rgsproc. Events with flags of BAD_SHAPE, ON_BADPIX, ON_WINDOW_BORADER, and BELOW_ACCEPTANCE are all rejected. We also did not keep the cool pixels (keepcool = no). We produced the light curves for the background in CCD9 following the XMM-SAS guide in order to remove time intervals that are contaminated with soft proton flares and spurious events. For each observation, only the first-order spectrum was extracted.

2.2. Absorption Line Analysis

Many of these sources were observed multiple times. To avoid unreal artifacts that could be introduced in the co-added exposures, for each target we simultaneously fitted all the exposures, each with its own response. Our analysis focuses on the segment spectrum around 20–22 Å and models the spectra continuum with the Galactic neutral absorption, plus a power-law model, using XSPEC ver 12.7 (Arnaud 1996). For each target, while we have adopted different continua for each exposure, the parameters of the interstellar absorption were always tied together. We also excluded wavelength regions of known detector features. Absorption lines were fitted with a Voigt line profile model developed in Buote et al. (2009). This line model has three free parameters: column density, Doppler-b parameter, and the velocity shift of the central wavelength. We refer the reader to Buote et al. (2009) and Fang et al. (2010) for details. Since we are interested in the absorption lines produced by the local hot gas, we limited the line-shift velocity within 500 km s⁻¹. The Doppler-b parameter is also limited in the range of 20–300 km s⁻¹. The lowest temperature that can still produce a significant fraction of O vii is around 5 × 10⁵ K under collisional ionization equilibrium; this corresponds to a thermal velocity of ∼20 km s⁻¹. The upper limit is adopted by assuming the absorbing gas shall not escape from our Galaxy. For a better constraint of the line EWs, we estimate the statistical significance of the lines using Monte Carlo simulations. Briefly, we make a simulated spectra based on our fitted absorption line models and run 1000 simulations on each targets. We fit the simulated spectra, record the measured EWs, and finally obtain the line significance (more details see Fang et al. 2010). We performed the fit by minimizing the C-statistic, which yields less biased best-fitting parameters.

The basic information of our sample and the fitted line properties are listed in Table 1. We show the source name and type in columns 1 and 2, respectively. Column 3 is the galactic neutral hydrogen column density adopted from Kalberla et al. (2005). Columns 4, 5, and 6 are the galactic latitude, longitude, and distance, respectively. All of the distances are adopted from the literature. We list the total exposure time and CPRE for each source in columns 7 and 8, respectively. Columns 9, 10, and 11 are the O vii column density, velocity shift of the line center, and the Doppler-b parameter, respectively. The line equivalent width (EW) and 1σ statistical uncertainty are given in column 12. We show the line significance in column 13 and C-statistic and the degree of freedom in column 14. The targets are listed in descending order based on their CPRE, which indicates the quality of the spectra. For targets for which we cannot constrain either the Doppler-b parameter or the shift velocity of the line center, we set the velocity shift of the line center at 0 km s⁻¹ or the Doppler-b parameter at 300 km s⁻¹. In our sample, we find one target with a detected absorption line with a shift velocity larger than 500 km s⁻¹. Such a large shift suggests that the absorption line is produced by the circumstellar medium local to the source. We also list this high-velocity line in the table, plus with the 3σ upper limits of EW for their lines at 0 km s⁻¹. We adopt 1σ errors throughout the paper unless otherwise mentioned.

Table 1. The X-Ray Targets Sample

Source	Type	N_H	l	b	Distance	Exp.	Counts	Log(N_Ovii)	Velocity	Doppler-b	EW(K_α)	S/N	C/dof
		(cm⁻²)			(kpc)	(ks)		(cm⁻²)	(km s⁻¹)	(km s⁻¹)	(mÅ)	(σ)
Sco X–1	LMXB	1.48E+21	359.094	23.784	${2.8}_{-0.3}^{+0.3}$ [6]	36.8	16612.7	${15.8}_{-0.4}^{+1.2}$	${15.4}_{-93.4}^{+98.9}$	${42.8}_{-22.8}^{+274.2}$	7.9 ± 0.8	9.9	972/662
Cygnus X–2	LMXB	1.87E+21	87.328	−11.316	${13.4}_{-2}^{+1.9}$ [6]	199.9	3089.7	${16.3}_{-0.5}^{+1.2}$	${145.5}_{-108.3}^{+108.4}$	${78.9}_{-50.9}^{+217.4}$	18.3 ± 1.5	12.2	1190/971
XTE J1817–330	LMXB	1.41E+21	359.817	−7.996	${2.5}_{-1.5}^{+1.5}$ [6]	18.0	1379.4	${16.2}_{-0.2}^{+0.2}$	${2.3}_{-304.2}^{+271.3}$	${300.0}_{-300.0}^{+0.0}$	34.2 ± 5.4	6.3	219/206
4U 1820–30	LMXB	1.29E+21	2.788	−7.914	${7.6}_{-0.4}^{+0.4}$ [24]	76.4	1185.2	${16.8}_{-0.9}^{+0.9}$	${56.4}_{-110.3}^{+108.0}$	${75.4}_{-35.2}^{+224.6}$	20.5 ± 2.1	9.8	347/327
GX 339–4	LMXB	3.74E+21	338.939	−4.326	${10}_{-4}^{+5}$ [1]	443.1	1017.9	${16.7}_{-0.1}^{+0.1}$	$-{71.3}_{-80.3}^{+81.5}$	${202.2}_{-43.3}^{+60.7}$	42.5 ± 3.5	12.1	1384/1354
SAX J1808.4–3658	LMXB	1.13E+21	355.385	−8.148	${2.57}_{-0.15}^{+0.15}$ [2]	58.8	910.6	${15.9}_{-0.1}^{+1.6}$	$-{118.1}_{-216.3}^{+186.5}$	${300.0}_{-280.0}^{+0.0}$	18.5 ± 2.5	7.4	289/206
Swift J1753.5–0127	LMXB	1.74E+21	24.898	12.186	${5.4}_{-2.5}^{+2.5}$ [11]	118.1	798.8	${15.9}_{-0.2}^{+0.1}$	$-{185.4}_{-259.9}^{+220.2}$	${300.0}_{-280.0}^{+0.0}$	18.5 ± 2.7	6.9	572/520
EXO 0748–676	LMXB	1.01E+21	279.975	−19.811	${8}_{-1.2}^{+1.1}$ [15]	382.9	691.8	${16.0}_{-0.7}^{+1.7}$	${24.5}_{-170.5}^{+204.2}$	${46.2}_{-26.2}^{+253.8}$	9.7 ± 2.3	4.2	1877/1433
Cygnus X–1	HMXB	7.21E+21	71.335	3.067	${2.1}_{-0.25}^{+0.25}$ [16]	194.4	595.0	${15.8}_{-0.3}^{+1.7}$	$-{494.5}_{-5.5}^{+379.6}$	${251.4}_{-231.4}^{+48.6}$	14.4 ± 3.5	4.1	1629/1472
MAXI J0556–332	LMXB	2.91E+20	238.940	−25.183	20 [5]	71.7	576.4	${16.1}_{-0.7}^{+1.6}$	${84.6}_{-274.6}^{+259.5}$	${68.2}_{-48.2}^{+231.8}$	13.5 ± 3.4	4.0	322/369
LMC X–3	HMXB	4.32E+20	273.576	−32.082	49.97 [17]	146.1	564.0	${16.5}_{-0.7}^{+1.2}$	${92.9}_{-137.2}^{+139.5}$	${85.6}_{-46.5}^{+214.4}$	20.5 ± 3.1	6.6	960/955
MAXI J1910–057	LMXB	2.27E+21	29.903	−6.844	⋯	51.2	529.1	${15.9}_{-0.2}^{+1.5}$	$-{205.4}_{-250.2}^{+254.5}$	${300.0}_{-271.4}^{+0.0}$	21.3 ± 3.6	5.9	225/120
4U 1636–54	LMXB	2.76E+21	332.915	−4.818	${6}_{-0.5}^{+0.5}$ [9]	245.1	390.9	${16.7}_{-0.7}^{+1.0}$	$-{170.8}_{-164.4}^{+171.0}$	${113.0}_{-59.2}^{+187.0}$	27.7 ± 4.5	6.2	1064/1067
4U 1728–16	LMXB	1.96E+21	8.513	9.038	10 [10]	32.1	234.0	${15.9}_{-0.9}^{+1.6}$	$-{364.7}_{-135.3}^{+417.8}$	${55.1}_{-35.1}^{+244.9}$	10.1 ± 4.7	2.1	452/401
4U 1254–690	LMXB	2.16E+21	303.482	−6.424	${13}_{-3}^{+3}$ [13]	214.0	215.9	${15.9}_{-0.4}^{+1.8}$	$-{35.3}_{-340.4}^{+416.9}$	${203.5}_{-183.5}^{+96.5}$	17.6 ± 5.3	3.3	685/664
Aql X–1	LMXB	2.84E+21	35.718	−4.143	${5.2}_{-0.8}^{+0.7}$ [6]	95.1	149.1	${16.0}_{-0.3}^{+1.4}$	${248.4}_{-520.0}^{+251.6}$	${299.4}_{-279.4}^{+0.6}$	25.0 ± 8.1	3.1	405/386
4U 1957+11	LMXB	1.21E+21	51.309	−9.331	22 [21]	44.8	132.4	${16.4}_{-1.0}^{+1.3}$	$-{281.5}_{-218.5}^{+584.1}$	${69.5}_{-49.5}^{+230.5}$	16.1 ± 6.3	2.6	164/162
4U 1543–62	LMXB	2.43E+21	321.755	−6.337	7 [3]	49.1	113.2	${16.5}_{-1.1}^{+1.0}$	$-{189.3}_{-310.7}^{+307.9}$	${71.3}_{-51.3}^{+228.7}$	18.6 ± 7.4	2.5	190/207
Swift J1357.2–0933	LMXB	2.95E+20	328.839	50.210	1.5 [19]	27.8	108.5	${16.0}_{-1.0}^{+1.7}$	${48.5}_{-548.5}^{+451.5}$	${90.8}_{-70.8}^{+209.2}$	14.0 ± 7.4	1.9	153/163
GS 1826–238	LMXB	1.71E+21	9.272	−6.088	${7.5}_{-0.5}^{+0.5}$ [6]	228.2	104.7	${16.2}_{-0.4}^{+0.3}$	$-{437.4}_{-62.6}^{+535.8}$	${300.0}_{-280.0}^{+0.0}$	31.6 ± 10.0	3.2	290/309
4U 2129+12	LMXB	6.28E+20	65.013	−27.312	10.4 [7]	86.4	98.1	${16.4}_{-1.4}^{+1.0}$	$-{133.4}_{-366.6}^{+433.4}$	${42.4}_{-22.4}^{+257.6}$	14.1 ± 8.7	1.6	297/242
4U 1735–44	LMXB	2.61E+21	346.054	−6.994	${9.4}_{-1.4}^{+1.4}$ [14]	20.1	82.0	${16.1}_{-0.4}^{+0.3}$	$-{75.0}_{-345.8}^{+475.0}$	${300.0}_{-193.7}^{+0.0}$	26.4 ± 11.3	2.3	170/163
4U 0614+09	LMXB	4.48E+21	200.877	−3.364	${2.2}_{-0.7}^{+0.8}$ [20]	23.4	74.0	${16.5}_{-1.5}^{+1.2}$	${206.8}_{-572.2}^{+293.3}$	${72.3}_{-52.3}^{+227.7}$	17.4 ± 9.2	1.9	232/229
SMC X–1	HMXB	3.32E+20	300.415	−43.559	65 [23]	36.5	73.2	${16.5}_{-0.8}^{+1.2}$	${318.8}_{-571.4}^{+181.2}$	${101.6}_{-81.6}^{+198.4}$	22.7 ± 9.1	2.5	209/199
GRO J1655–40	LMXB	5.33E+21	344.982	2.456	${3.2}_{-0.2}^{+0.2}$ [6]	113.8	70.7	${15.8}_{-0.8}^{+1.7}$	$-{230.5}_{-269.5}^{+297.2}$	${61.1}_{-41.1}^{+238.9}$	15.8 ± 14.2	1.1	351/331
XTE J1650–500	LMXB	4.34E+21	336.718	−3.427	${2.6}_{-0.7}^{+0.7}$ [17]	23.9	63.2	${16.1}_{-1.0}^{+1.6}$	${228.1}_{-435.1}^{+272.0}$	${300.0}_{-280.0}^{+0.0}$	32.5 ± 14.6	2.2	189/198
Ser X–1	LMXB	4.14E+21	36.118	4.842	${11.1}_{-1.6}^{+1.6}$ [4]	64.2	57.4	${17.1}_{-1.5}^{+0.6}$	${159.4}_{-492.4}^{+340.6}$	${95.8}_{-75.8}^{+204.2}$	27.5 ± 11.9	2.3	253/278
XB 1832–330	LMXB	9.24E+20	1.531	−11.371	${9.6}_{-0.4}^{+0.4}$ [8]	63.6	54.0	${16.7}_{-1.1}^{+1.0}$	$-{212.6}_{-287.5}^{+459.9}$	${102.1}_{-82.1}^{+198.0}$	25.7 ± 10.5	2.5	151/181
4U 0513–40	LMXB	4.21E+20	244.510	−35.036	${12.1}_{-0.3}^{+0.3}$ [6]	23.7	28.7	⋯	⋯	⋯	<56.1	⋯	⋯
4U 0513–40*	⋯	⋯	⋯	⋯	⋯	⋯	⋯	${16.5}_{-0.5}^{+0.4}$	$-{595.0}_{-5.0}^{+340.1}$	${300.0}_{-280.0}^{+0.0}$	44.1 ± 15.0	2.9	90/92
LMC X–4	HMXB	1.17E+21	276.335	−32.529	49.97 [12]	43.7	28.7	${16.8}_{-1.8}^{+0.6}$	${0.2}_{-391.9}^{+383.7}$	${97.1}_{-77.1}^{+202.9}$	25.5 ± 15.5	1.6	100/92
4U 0919–54	LMXB	6.16E+21	275.852	−3.845	${5}_{-0.7}^{+0.8}$ [18]	39.7	26.3	${15.8}_{-0.8}^{+1.6}$	${0.0}_{-500.0}^{+500.0}$	${97.5}_{-77.5}^{+202.5}$	3.4 ± 15.2	0.2	73/103
MXB 1659–29	LMXB	1.82E+21	353.827	7.266	${11.5}_{-1.5}^{+1.5}$ [6]	30.6	24.8	${16.4}_{-1.4}^{+1.2}$	${237.0}_{-537}^{+263.0}$	${95.1}_{-75.1}^{+204.9}$	21.1 ± 15.7	1.3	83/82
4U 1746–371	LMXB	2.63E+21	353.531	−5.005	${11}_{-0.8}^{+0.9}$ [22]	71.3	22.9	${17.1}_{-1.3}^{+0.4}$	$-{272.7}_{-227.3}^{+301.8}$	${141.0}_{-120.9}^{+159.0}$	38.3 ± 17.5	2.2	93/94

Note. [1] Jonker & Nelemans (2004), [2] Trimble (1973), [3] Sala & Greiner (2006), [4] Kuulkers et al. (2003), [5] Hynes et al. (2004), [6] in't Zand et al. (2001), [7] Caraveo et al. (2001), [8] Zurita et al. (2008), [9] Helton et al. (2008), [10] Ziółkowski (2005), [11] Mason & Cordova (1982), [12] Pietrzyński et al. (2013), [13] Galloway et al. (2006), [14] Savolainen et al. (2009), [15] in't Zand et al. (2003), [16] Nowak et al. (2008), [17] Kaplan et al. (2007), [18] Wang & Chakrabarty (2004), [19] Rau et al. (2011), [20] Kong et al. (2000), [21] Hessels et al. (2007), [22] Paerels et al. (2001), [23] Muno et al. (2001), [24] Tetzlaff et al. (2011).

Download table as: ASCII Typeset image

Of the total of 33 targets, 16 show an O vii detection significance larger than the 3σ threshold, 9 are at the 2σ–3σ level. Details about the detections at different significance levels are listed in Table 2. Only one target has a weak absorption line presented at a significance less than 1σ. The difference between HMXB and LMXB mostly comes from the mass of the companion star and the way in which accretion of matter occurs. The spectra of HMXB is sensitive to the properties of the companion wind. In our sample, we have four HMXBs, two (LMC X–3 and Cygnus X–1) of which show a O vii detection significance larger than 3σ. 4U 0513–40 has a shift velocity larger than 500 km s⁻¹, and its detection significance is 2.9σ. In the spectrum of EXO 0748–676, O vii K-shell emission is clearly shown near 21.8 Å (Cottam et al. 2002; van Peet et al. 2009), whereas one absorption line at 21.6 Å is also detected here with a significance of 4.1σ. Of the targets with a significance larger than 3σ, four targets are new detections: Sco X–1, Swift J1753.5–0127, MAXI J0556–332, and MAXI J1910–057.

Table 2. Target Detection

	>3σ	>2σ	>1σ	Total
All	16	25	32	33
LMXB	14	22	28	29
HMXB	2	3	4	4

Note. Target detection in our line measurements. Columns 2, 3, and 4 list the targets detected with significances larger than 3σ, 2σ, and 1σ, respectively. The last column is the total number of targets in each group.

Download table as: ASCII Typeset image

For a more careful examination of the line significances, we have tested the detections using another Monte Carlo simulation method. We first fit and simulate the continuum spectrum, and then we fit the simulated data with a continuum model and record the C-statistic. We add the absorption model, and record the change of the C-statistic. We have performed 10,000 simulations for 3 test targets to check how many times the changed C-statistic as given by adding a O vii line provides a result better than the measurement. These three test targets, Sco X–1, EXO 0748–676 and GS 1826–238, are selected due to their different statistics and we expect this small sample to provide a very interesting view on the overall significance of the total sample. We find that the line significances for these three targets are >6σ, 2.8σ, and 3.6σ, respectively. These results are in general consistent with the significances we obtained in the first Monte Carlo simulations.

In Figure 2, we show the histogram distribution of the O vii K_α line EW for targets detected at >1σ significance. The measured line EWs are centered around ∼20 mÅ, with a range of 10–40 mÅ. In Figure 3, we show the line detection rate, or the sky-covering fraction, as a function of the photon counts CPRE. The solid, dashed, and dotted lines are for the detections with at least 1σ, 2σ, and 3σ significance, respectively. For all cases, the detection fractions increase steadily as the CPRE increases. For targets with a CPRE more than about 400 counts, the 3σ detection fraction reaches 100%. Such a detection rate implies a uniform distribution of hot gas in all directions.

**Figure 3.** Detection fraction as a function of source CPRE. The fractions for absorption lines detected with at least 1σ, 2σ, and 3σ significance are presented as the solid, dashed, and dotted lines, respectively.
Download figure:
Standard image High-resolution image

Some of the O vii detections were reported previously. We have compared our results with previous measurements. We list our work and previous studies in Table 3. Column 2 shows the line EWs in our work, and columns 3 and 4 are previous results obtained with Chandra and XMM-Newton, respectively. Some of the previous results were initially listed in units of column density. For consistency, we have converted those results to units of mÅ with a 1σ error bar, assuming a Doppler-b parameter of 100 km s⁻¹. Our measurements are largely consistent with previous ones within the error range. Cygnus X–2 is a unique one because the O vii Kα line is not detected in Chandra. We refer the reader to Cabot et al. (2013) for a detailed analysis of the O vii line in Cygnus X–2. For XTE J1817–330, Sala et al. (2007) obtained an O vii EW of 0.7 eV (26.4 mÅ) and is consistent with our measurements, whereas the Chandra observation by Gatuzz et al. (2013) gives a higher value. Similar discrepancy was also presented in 4U 1820–30. One explanation for the discrepancy may be the time variability of the O vii line since Chandra observations were performed at a different time. Part of our sample has been analyzed by Pinto et al. (2013). Our measurements are in general consistent with theirs. One exception is GX 339–4, for which we found a much higher EW. The difference may be caused by the different data reduction procedures. For this target, all its observations were not co-added together as done previously, but they were simultaneously analyzed. Highly ionized absorption lines in LMC X–3 were systematically studied by Wang et al. (2005) using high-resolution spectrometers on board the Chandra, and our results are consistent with their measurements. In the Chandra spectra of Cygnus X–1, Schulz et al. (2002) found a weak detection at 21.43 Å and they identified this line as a blueshifted O vii Kα line, but the blueshifted velocity was detected differently, as 2350 km s⁻¹, and we found the velocity to be 494 km s⁻¹. This could be a signature of line variation.

Table 3. Comparison with Previous Measurements

Name	This Work	Chandra	XMM-Newton
	(mÅ)	(mÅ)	(mÅ)
Cygnus X–2	18.3 ± 1.5	<6.3 [1]	19.6 ± 1.3 [2]
XTE J1817–330	34.2 ± 5.4	54.0 ± 4.0 [4]	26.4 ± 7.6 [5]
4U 1820–30	20.5 ± 2.1	40 ± 7.9 [6]	23.3 ± 2.5 [7]
		44.8 ± 10.8 [3]	23.9 ± 3.6 [8]
GX 339–4	42.5 ± 3.5		23.7 ± 5.9 [9]
SAX J1808.4–3658	18.5 ± 2.5		31.5 ± 7.2 [9]
LMC X–3	20.5 ± 3.1	20 ± 6 [10]	21.0 ± 5.0 [8]
4U 1636–54	27.7 ± 4.5	⋯	19.9 ± 7.0 [9]
4U 1728–16	10.1 ± 4.7	⋯	8.6 ± 4.3 [9]
4U 1254–690	17.6 ± 5.3	⋯	15.2 ± 7.6 [9]
Aql X–1	25.0 ± 8.6	⋯	18.1 ± 7.2 [9]
4U 1957+11	16.1 ± 6.3	18 ± 18 [11]	19.0 ± 19.0 [11]
		18.7 ± 10.4 [12]
GS 1826–238	31.6 ± 10.0	⋯	18.1 ± 7.2 [9]
4U 1735–44	26.4 ± 11.3	⋯	19.9 ± 10.0 [9]
	⋯	⋯	24.7 ± 9.7 [8]
Ser X–1	27.5 ± 11.9	⋯	21.0 ± 4.4 [9]

Note. [1] Yao et al. (2009), [2] Cabot et al. (2013), [3] Futamoto et al. (2004), [4] Gatuzz et al. (2013), [5] Sala et al. (2007), [6] Yao & Wang (2006), [7] Costantini et al. (2012), [8] Miller & Bregman (2013), [9] Pinto et al. (2013), [10] Wang et al. (2005), [11] Nowak et al. (2008), [12] Yao et al. (2008).

Download table as: ASCII Typeset image

3. Discussion

3.1. Study of ISM Distribution

The distribution of the metal absorption lines in the spectra of different XRBs allows us to quantify how the intervening hot gas is distributed in the circumstellar medium and the ISM. Here, we show the dependence of the line EW on the target direction, neutral gas column density, and distance. In Figure 4, we show the relations between O vii K_α EW and galactic latitude (left panel) and longitude (right panel). In the left panel, for l in the range of 50°–180°, the mean EW is less than 20 mÅ, while toward the Galactic center, for l in the range of 0°–30°, the mean EW is larger than 20 mÅ. For the right panel, a large part of the sample has coordinates toward the galactic center, with coordinates $| b|$ in the range of 0°–10°; however, the tendency for the EWs along with b is rather weak. At higher latitude $| b| \gt 20$ , the EWs slightly increase with $| b|$ , while at lower latitude, there is a large scatter of EWs.

**Figure 4.** O vii K_α EW against the galactic longitude (left panel) and latitude (right panel).
Download figure:
Standard image High-resolution image

In Figure 5 we show the measured O vii K_α EW as a function of the Galactic hydrogen absorption N_H. It could be found the EWs likely show little variation with N_H and there is no clear correlation between the EWs and N_H. Assuming a mean O vii EW of 20 mÅ and a Doppler-b parameter of 100 km s⁻¹, we could obtain an O vii column density of about 2 × 10¹⁶ cm⁻². Assuming solar abundance and that all oxygen of the hot gas would be in the ionization stage of O vii, the total hydrogen column density of the hot gas would be 4.1 × 10¹⁹ cm⁻². By taking the ratio of the hot gas and neutral gas column density, we find that the relative fraction with neutral hydrogen could be in the range of 0.4–0.004.

To investigate how the hot ISM distributes around the disk, we show the O vii K_α EW versus the target distance in Figure 6. In general, the EWs of the targets show a very weak and positive correlation, and also show scatters by a factor of a few or more among targets at similar distances. For targets with distances larger than 10 kpc, their EWs are mostly greater than 10 mÅ, and for those with distances larger than 50 kpc, their EWs are greater than 20 mÅ; for targets with distances less than 5 kpc, there are at least two objects with EW > 30 mÅ. However, due to large scattering, the trend between EW and distance is not obvious, and at small distances these large scatters could be a result of the intrinsic absorption contamination.

We have calculated the dependence of the line EW on the galactic neutral absorption, target distance, and the galactic latitude and longitude. Briefly, we first use software package ASURV Rev 1.2 (Isobe & Feigelson 1990; Lavalley et al. 1992), which implements the methods presented in Isobe et al. (1986) and is a particularly useful tool for a correlation analysis of censored astronomical data, i.e., non-detections or detection limits. Then, we characterize a correlation with a simple log–log linear relation, together with the dispersion around this relation from the root mean square (rms) of the data. The corresponding rms could give the probabilistic nature of the correlation and the measurement uncertainties. We listed the probability of correlation by chance for three correlation tests: Cox hazard model, generalized Kendall's tau, and Spearman's rho in Table 4 columns 1, 2, and 3, respectively. A correlation exists if the probability is less than 5%. In general, the probabilities for all correlations are much larger than 5%. In columns 4 and 5, we list the index of the fitted log–log linear relation and the corresponding rms, respectively. The index is quite close to zero, and the correlations are significantly sub-linear. With all these in consideration, we may then conclude that the EW has no correlation with the galactic neutral absorption, the target distance, or the galactic coordinates.

Table 4. Correlation Test

Correlation	Cox Hazard	Kendall Tau	Spearman Rho	Γ	rms (dex)
EW versus N_H	40%	87%	87%	0.075 ± 0.001	0.23 ± 0.06
EW versus Distance	20%	21%	19%	0.04 ± 0.12	0.26 ± 0.07
EW versus l	36%	32%	34%	−0.06 ± 0.08	0.26 ± 0.07
EW versus $\| b\|$	53%	40%	37%	−0.15 ± 0.18	0.26 ± 0.08

Download table as: ASCII Typeset image

If the hot gas probed by the O vii absorption line is uniformly distributed in the ISM, one would expect the detected O vii column density may show some correlations with physical properties such as the the distance and/or coordinates of the background targets. Assuming the absorption line is in the linear part of the curve-of-growth, such correlations can be presented as the correlations between the line EW and these physical quantities. The non-detection of the correlations may suggest that the line is in the nonlinear part of the curve-of-growth and that we do not have strong constraints on the line Doppler-b parameter. However, it is not likely caused by a non-uniform distribution of the hot gas properties, such as temperature, metallicity, or density, because of the uniform distribution of the O vii absorption lines. Additionally, the contamination of the circumstellar material and the historical supernova activity (Ferriere 2001) could also introduce absorption to the lines.

An example of studying the ISM comes from comparing spatially closed targets, since the ISM distribution along the sightline toward these targets should be similar. We searched our sample for targets close to each other through their relative l and b within 5°. First, we found that the two targets SAX J1808.4–3658 and 4U 1820–30 show comparable O vii EWs, whereas their distances differ by three times. The less distant target XTE J1817–330 has an EW much larger than these two targets, even though all three targets are spatially close to each other. Similarly, in another group, the EW for 4U 1636–54, with an error bar taken into consideration, is comparable with that of XTE J1650–500, while 4U 1636–54 has a distance twice as far away. Even though there are still too few pairs to draw definitive conclusions, this result may suggest that in some cases the intrinsic absorption dominates the ISM contribution.

3.2. Comments on Some Individual Targets

We briefly discuss several individual targets with newly detected O vii absorption lines at more than the 3σ significance level. In our sample, we find 16 targets with a line significance at more than the 3σ level, of which four targets are newly detected: Sco X–1, Swift J1753.5–0127, MAXI J0556–332 and MAXI J1910–057.

Sco X–1. This target is one of the brightest X-ray sources in the sky. With a high signal-to-noise ratio spectrum we can achieve much better constraints of the absorption line properties. Previous studies based on absorption lines mostly focused on the low-ionization absorption lines or absorption edges, and our measurement is the first report on the O vii detection (de Vries & Costantini 2009; García et al. 2011). The O vii line in Sco X–1 is detected at the 9.9σ significance level and the line-shift velocity is very low (less than 100 km s⁻¹).

Swift J1753.5–0127. This is also a LMXB discovered with the Swift/BAT on 2005 May (Palmer et al. 2005). Mostafa et al. (2013) reported the discovery of broad emission lines of N vii and O viii in the RGS spectrum. They attributed this feature to reflection of X-ray photons off the accretion disk.

MAXI J0556–332. This is an X-ray transient. Maitra et al. (2011) reported the detection of a strong emission line near 24.8 Å, identified as N vii. So far there is no report of absorption lines produced by an outflow. We detect the O vii absorption with a 4σ significance level.

MAXI J1910–057. This source was first detected by the Swift/BAT (Krimm et al. 2012) on 2012 May and simultaneously by the MAXI telescope (Usui et al. 2012). The observed spectra are typical of an LMXB. Charles et al. (2012) indicated that it is comparable with other neutron star/black hole soft X-ray transients, but the distance is still unclear. The O vii line in this target is first reported here, with a blueshift velocity of 205 km s⁻¹, an EW of 21.3 mÅ, and a significance of 6.0.

4U 0513–40. This is an ultracompact X-ray binary in the globular cluster NGC 1851. This target was only observed by XMM-Newton once in 2003 April. Previous X-ray observations did not report any detection of a O vii absorption line and in our work we found this line to have a shift velocity of −595 km s⁻¹ and a detection significance at the 2.9σ level. With its high shift velocity, this absorption line could be a good indicator of a circumstellar medium origin.

4. Summary

Understanding the properties of the hot ionized medium has a profound impact on the study of stellar feedback and the multi-phase structure of the ISM. In this work, we perform a survey in the XMM-Newton archive for searching O vii absorption lines in the spectra of Galactic X-ray sources. We summarize our findings as follows:

1.
We analyzed 33 Galactic XRBS; 29 targets were LMXBs and 4 targets were HMXBs. Most targets are located at a distance of within 20 kpc, in the Galactic disk. Of these targets, 16 have O vii detection at more than 3σ, among which 4 are newly discovered in this work. We found one target with a shift velocity larger than 500 km s⁻¹, indicating a circumsteller medium origin.
2.
We fitted the O vii K_α transitions with a Voigt-profile-based line model. We find that most K_α lines have an EW of 20 mÅ, with a range of 10–40 mÅ.
3.
We find that the detection fractions, or the sky-covering fractions, increase steadily as the CPRE increases. For targets with a CPRE greater than about 400 counts, the 3σ detection fraction reaches 100% regardless of the direction of the XRBs. Such a detection rate implies a uniform distribution of the hot gas in all directions.
4.
The EWs of the O vii line do not correlate with the galactic neutral absorption N_H. The fraction of content of the hot medium to that of the neutral phase is estimated to be in the range of 0.004–0.4. We do not find any strong correlations between O VII EW and target distance, or the target coordinates. The EW for targets at similar distances shows large scatters and the difference could be as large as one order of magnitude. The reasons for the non-correlation could be the contribution of circumstellar material or the uncertainties on the column density measurements, as we do not have constraints on the Doppler-b parameter, so EWs may not exactly reflect the column density distribution.

The highly ionized z = 0 O vii absorption lines detected in the X-ray spectra of background AGNs, along with the detections in Galactic XRBs, could help to constrain the distribution of hot the ISM with a broad range of temperature around 10⁶ K. We briefly compare our results with the AGN O vii line survey by Fang et al. (2015), and we find there are some similarities in the line properties. The detection fractions, in both surveys, increase steadily as the CPRE increases, which suggests a uniform distribution of the O vii absorbers. The histogram of K_α lines in both cases has a median EW of 20 mÅ, and there are also no correlations between the EWs and the target coordinates.

By considering the different coordinates and distances or covering areas and absorbing depths of the AGNs and XRBs, we could find that the XRBs have a large fraction of contaminations intrinsically contributing to the O vii absorption. But how much the XRBs locally contribute to the absorption still requires further investigation. Further theoretical modeling in combination with X-ray emission studies and X-ray absorption studies will be presented in a later paper.

We thank Drs. David Buote and Daniel Wang for helpful discussions. This work is supported by the National Key R&D Program of China No. 2017YFA0402600, and NSFC grants No. 11525312, 11333004, 11443009, U1531130, and 11333004. T.F. is also supported by the Specialized Research Fund for the Doctoral Program of Higher Education (SRFDP; No. 20130121110009).

Appendix:

We present all the target spectra in Figures 7–10 in the 21–22 Å wavelength range. The blue lines in each panel are the model-fitted continua and absorption lines. The model-fitted continua in these figures are only for display purposes, because for each target the continuum changes among different observations. There is a detector feature at about 21.82 Å, so we have excluded this region in fitting the spectrum.

**Figure 7.** RGS spectra between 21 and 22 Å for each target. The blue line is the fitted continuum. This is only for demonstration purposes, because for each target the spectra were jointly fitted with the same absorption line parameters but different continuum levels. The structure at ∼21.8 Å in each panel is an instrumental feature, and is ignored in the fit.
Download figure:
Standard image High-resolution image

**Figure 8.** Same as Figure 7 but for different targets.
Download figure:
Standard image High-resolution image

**Figure 9.** Same as Figure 7 but for different targets.
Download figure:
Standard image High-resolution image

**Figure 10.** Same as Figure 7 but for different targets.
Download figure:
Standard image High-resolution image

XMM-Newton Survey of Local O vii Absorption Lines in the Spectra of Galactic X-Ray Sources

Article metrics

Permissions

Author e-mails

Author affiliations

ORCID iDs

Dates

Abstract

1. Introduction