Investigating High Mass X-ray Binaries at hard X-rays with INTEGRAL

TL;DR

This study uses INTEGRAL data to analyze the X-ray properties of 58 high mass X-ray binaries, revealing their variability, duty cycles, and subclass distinctions, including bridging sources.

Contribution

It provides a comprehensive quantitative overview of HMXB subclasses using hard X-ray observations and variability analysis, highlighting intermediate sources.

Findings

Identified duty cycles and luminosity ranges for HMXB subclasses.

Discovered intermediate properties bridging SgHMXBs and SFXTs.

Quantified variability patterns across the sample.

Abstract

The INTEGRAL archive developed at INAF-IASF Milano with the available public observations from late 2002 to 2016 is investigated to extract the X-ray properties of 58 High Mass X-ray Binaries (HMXBs). This sample consists of sources hosting either a Be star (Be/XRBs) or an early-type supergiant companion (SgHMXBs), including the Supergiant Fast X-ray Transients (SFXTs). INTEGRAL light curves (sampled at 2 ks) are used to build their hard X-ray luminosity distributions, returning the source duty cycles, the range of variability of the X-ray luminosity and the time spent in each luminosity state. The phenomenology observed with INTEGRAL, together with the source variability at soft X-rays taken from the literature, allows us to obtain a quantitative overview of the main sub-classes of massive binaries in accretion (Be/XRBs, SgHMXBs and SFXTs). Although some criteria can be derived to distinguish them, some SgHMXBs exist with intermediate properties, bridging together persistent SgHMXBs and SFXTs.

Tables

Table 1. Table 1: Results of our survey of a sample of HMXBs.

Source	sub-class	DC18-50keV	Av. Luminosity (18-50 keV)1	DR1-10keV
		[%]	[erg s-1]	(Fmax/Fmin)
SMC X-1	SgHMXB	49.05	1.7E+38	7.7
3A 0114+650	SgHMXB	14.63	2.1E+36	$-$
Vela X-1	SgHMXB	79.22	1.3E+36	1.7
1E 1145.1-6141	SgHMXB	31.95	3.0E+36	$-$
GX 301-2	SgHMXB	94.47	2.8E+36	2.6
H 1538-522	SgHMXB	30.15	9.2E+35	32.9
IGR J16207-5129	SgHMXB	0.39	1.1E+36	9.3
IGR J16320-4751	SgHMXB	21.32	5.9E+35	14.7
IGR J16393-4643	SgHMXB	0.40	3.4E+36	3.2
OAO 1657-415	SgHMXB	59.78	5.8E+36	10.0
4U 1700-377	SgHMXB	73.09	1.1E+36	12.0
IGR J17252-3616	SgHMXB	4.65	2.9E+36	17.3
IGR J18027-2016	SgHMXB	0.54	5.2E+36	375
IGR J18214-1318	SgHMXB	0.06	3.4E+36	$-$
XTE J1855-026	SgHMXB	9.64	4.2E+36	$-$
H 1907+097	SgHMXB	20.13	8.1E+35	546
4U 1909+07	SgHMXB	24.84	7.1E+35	11.5
IGR J19140+0951	SgHMXB	14.18	5.2E+35	769
LMC X-4	giant HMXB	47.23	1.2E+38	3.4
Cen X-3	giant HMXB	62.79	4.0E+36	5.0
IGR J08408-4503	SFXT	0.09	3.0E+35	6750
IGR J11215-5952	SFXT	0.64	1.6E+36	$>$480
IGR J16328-4726	SFXT	0.28	1.7E+36	300
IGR J16418-4532	SFXT	1.22	6.1E+36	308
IGR J16465-4507	SFXT	0.18	2.9E+36	37.5
IGR J16479-4514	SFXT	3.33	3.6E+35	1667
IGR J17354-3255	SFXT	0.01	3.0E+36	$>$929
XTE J1739-302	SFXT	0.89	4.8E+35	$>$2040
IGR J17544-2619	SFXT	0.54	5.6E+35	1.67$\times {10}^6$
SAX J1818.6-1703	SFXT	0.81	2.9E+35	$>$1364
IGR J18410-0535	SFXT	0.53	3.8E+35	1.1$\times {10}^4$
IGR J18450-0435	SFXT	0.35	1.5E+36	513
IGR J18483-0311	SFXT	4.63	5.2E+35	899
H 0115+634	Be/XRB	9.55	1.5E+37	1.4$\times {10}^5$
RX J0146.9+6121	Be/XRB	0.11	1.1E+35	$-$
EXO 0331+530	Be/XRB	25.10	2.4E+37	1.07$\times {10}^6$
X Per	Be/XRB	76.96	2.5E+34	10
1A 0535+262	Be/XRB	12.34	4.4E+36	2.7$\times {10}^4$
GRO J1008-57	Be/XRB	8.87	2.4E+36	181
4U 1036-56	Be/XRB	0.35	7.5E+35	60
IGR J11305-6256	Be/XRB	0.41	1.9E+35	$-$
IGR J11435-6109	Be/XRB	2.68	1.4E+36	$-$
H 1145-619	Be/XRB	1.07	1.2E+35	250
XTE J1543-568	Be/XRB	0.14	2.7E+36	8
AX J1749.1-2733	Be/XRB	0.17	8.1E+36	$-$
GRO J1750-27	Be/XRB	4.88	2.9E+37	$>$10
AX J1820.5-1434	Be/XRB	0.15	2.1E+36	$-$
Ginga 1843+009	Be/XRB	3.39	5.8E+36	5660
XTE J1858+034	Be/XRB	5.34	8.8E+36	$-$
4U 1901+03	Be/XRB	10.44	1.2E+37	1000
KS 1947+300	Be/XRB	9.41	6.8E+36	800
EXO 2030+375	Be/XRB	28.99	7.8E+36	$>$2784
SAX J2103.5+4545	Be/XRB	11.14	2.0E+36	6364
IGR J16318-4848	other HMXB	35.17	7.4E+35	3.3
3A 2206+543	other HMXB	6.41	2.5E+35	250
Cyg X-1	other HMXB	99.88	2.5E+36	3.7
Cyg X-3	other HMXB	93.49	1.0E+37	4.9
SS 433	other HMXB	14.97	8.5E+35	5.0
XTE J1743-363	symbiotic	0.13	1.1E+36	6.2

Full text

††volume: 346††journal: High Mass X-ray Binaries: illuminating the passage from massive binaries to merging compact objects††editors: L. Oskinova, E. Bozzo & T. Bulik, eds.

1 Introduction

High Mass X-ray Binaries (HMXBs) host a compact object (most frequently a neutron star [hereafter, NS]) accreting matter from an O or B-type massive star. In the great majority of these systems the mass transfer to the accretor occurs by means of the stellar wind, while in a limited number of HMXBs (SMC X-1, LMC X-4, Cen X-3) it happens through Roche Lobe overflow (RLO). Before the launch of the INTEGRAL satellite ([Winkler et al. 2003, Winkler et al. 2003], [Winkler et al. 2011, Winkler et al. 2011]), two types of HMXBs were known, depending on the kind of companion, either an early type supergiant star (SgHMXBs) or a Be star (Be/XRBs).

Nowadays the scenario has significantly changed, with a number of Galactic HMXBs tripled and new sub-classes of massive binaries (the “highly obscured sources” and the “Supergiant Fast X–ray Transients”, SFXTs) discovered thanks to the observations of the Galactic plane performed by the INTEGRAL satellite. The first type includes HMXBs where the absorbing column density due to the local matter is more than one order of magnitude larger than the average in HMXBs (reaching 1024 cm*-2* in IGR J16318–4848).

The SFXTs are HMXBs that undergo short (usually less than a few days) outbursts made of brief (typical duration of $\sim$2 ks) and bright X–ray flares (peak LX$\sim$1036 erg s*-1*), while most of their time is spent below LX$\sim$1034 erg s*-1*. The physical mechanism producing this behavior is debated: the main models involve different ways to prevent accretion onto the NS (invoking opposite assumptions on the NS magnetic field and spin period), coupled with different assumptions on the donor (clumpy and/or magnetized) wind parameters (see [Martínez-Núñez et al. 2017, Martínez-Núñez et al. (2017)], [Walter et al. (2015), Walter et al. (2015)] and [Sidoli 2017, Sidoli (2017)] for recent reviews, and references therein).

While SgHMXBs and Be/XRBs differ in the type of their companion star, the boundaries between SgHMXBs and SFXTs are based only on their X–ray phenomenology (persistent vs transient X–ray emission), since they both harbour an early-type supergiant donor. In fact, unlike SgHMXBs, SFXTs display a large dynamic range that can reach six orders of magnitude in X–ray luminosity, between quiescence and the outburst peak (as in IGR J17544-2619; [in’t Zand 2005, in’t Zand 2005], [Romano et al. 2015, Romano et al. 2015]). For other SFXTs the range of flux variability is typically comprised between 102 and 104.

The present paper summarizes our systematic analysis of the INTEGRAL observations of HMXBs, spanning fourteen years of operations, from 2002 to 2016. The main aim of this work is to obtain an overall, quantitative, characterization of the different sub-classes of HMXBs at hard X–rays (above 18 keV) and to put this phenomenology into context of other known properties (like pulsar spin period, orbital geometry) and soft X–ray behavior (1–10 keV), as described in the literature. We refer the reader to [Sidoli & Paizis 2018, Sidoli & Paizis (2018)] for more details on this work.

2 The INTEGRAL archive and the selection of the sample

Our investigation is based on observations performed by IBIS/ISGRI on-board the INTEGRAL satellite and it is focussed on the energy range 18–50 keV. We built an INTEGRAL local archive of all public observations (see [Paizis et al. 2013, Paizis et al. 2013] and [Paizis et al. 2016, Paizis et al. 2016] for the technical details). For all known HMXBs we extracted the long-term light curves of the sources (bin time of $\sim$2 ks, the typical duration of an INTEGRAL observation, called “Science Window”) spanning fourteen years (from late 2002 to 2016). We retained in our final sample only the sources which were detected (above 5 sigma) in at least one INTEGRAL observation (i.e. one single Science Window), within 12*∘* from the centre of the field-of-view. These selection criteria translated into a sensitivity threshold of a few 10*-10* erg cm*-2* s*-1* (18–50 keV) for our survey, and into a total exposure time of $\sim$200 Ms for the final HMXB sample.

The final list of sources includes 58 HMXBs, classified in the literature as SgHMXBs (18 sources), SFXTs (13 sources) and Be/XRBs (20 sources); the remaining 9 massive binaries are two pulsars accreting from early type giant stars (LMC X–4 and Cen X–3), three black hole (candidate) systems (Cyg X-1, Cyg X-3, SS 433) plus two peculiar massive binaries (IGR J16318-4848 and 3A 2206+543). Then, we also included a symbiotic binary XTE J1743-363, that is a different kind of wind-fed system, to compare it with massive binaries. The complete list of sources is reported in Table 1, together with their sub-class, as reported in the literature.

3 INTEGRAL results

Duty Cycles (18-50 keV). The long-term light curves for our sample of HMXBs were used to calculate the source duty cycle in the energy range 18-50 keV (DC${}_{18-50~{}keV}$), defined as the percentage of detections (at $\sim$2 ks time bin) or, in other words, the ratio between the exposure time when the source is detected and the total exposure time at the source position. Table 1 (third column) lists the values obtained. Even in case of a persistent SgHMXB, the duty cycle can be lower than 100%, because of source variability leading the source flux below the IBIS/ISGRI threshold of detectability on the adopted time bin. Eclipses or off-states also reduce the source duty cycle in persistent sources (e.g. in Vela X-1, [Kreykenbohm et al. 2008, Sidoli et al. 2015]). The advantage of the adoption of a long-term archive, analysed here in a systematic way, is that we are confident that the source duty cycles are close to the real source activity, above the INTEGRAL sensitivity. We refer the reader to Sidoli & Paizis (2018) for a detailed discussion of the possible observational biases.

Cumulative Luminosity Distributions. The hard X-ray light curves were used to extract the Cumulative Luminosity Distributions (CLDs). We adopted a single average conversion factor of 4.5$\times 10^{-11}$ erg cm*-2* count*-1* from IBIS/ISGRI count-rates to X-ray fluxes (18–50 keV) and assumed the source distances reported by [Sidoli & Paizis 2018, Sidoli & Paizis (2018)].

The CLDs of four sources are shown in Fig. 1, representative of the behavior of a persistent SgHMXB (Vela X-1), a SFXT (SAX J1818.6-1703) and of two transient Be/XRBs (SAX J2103.5+4545 and EXO 0331+530). Their shape appears different: a lognormal-like distribution is evident in Vela X-1, a powerlaw CLD in the SFXT, while a more complex behavior is present in the Be/X-ray transients.

Since the timescale of the SFXT flare duration is similar to the bin time of the INTEGRAL light curves, the SFXT CLDs are distributions of the SFXT flare luminosities ([Paizis & Sidoli 2014, Paizis & Sidoli 2014]). The difference among supergiant systems (SgHMXBs vs SFXTs), between lognormal and powerlaw-like luminosity distributions were already found by [Paizis & Sidoli 2014, Paizis & Sidoli (2014)] from the analysis of the first nine years of INTEGRAL observations of a sample of SFXTs, compared with three SgHMXBs.

This behavior might be ascribed to a separate physical mechanism producing the bright X–ray flares in SFXTs: in the framework of the quasi-spherical settling accretion regime ([Shakura et al. 2012, Shakura et al. 2012]), hot wind matter, captured within the Bondi radius, accumulates above the NS magnetosphere; magnetic reconnection at the base of this shell (between the magnetized, captured, wind matter and the NS magnetosphere) has been suggested to enhance the plasma entry through the magnetosphere, opening the NS gate. This allows the sudden accretion of the shell material onto the NS and the emission of the SFXT flares ([Shakura et al. 2014, Shakura et al. 2014]). The detection of a $\sim$kG magnetic field from the companion of the SFXT IGR J11215–5952 supports this scenario ([Hubrig et al. 2018, Hubrig et al. 2018]).

Transient Be/XRBs can show two types of outbursts, the “normal” and the “giant” ones ([Stella et al. 1986, Negueruela et al. 1998, Negueruela et al. 2001, Negueruela et al. 2001b, Okazaki et al. 2001, Reig 2011, Kuhnel et al. 2015]). The first type happens periodically and is produced by the higher accretion rate when the NS approaches the decretion disc of the Be star, at each passage near periastron. The second type of outburst can occur at any orbital phase, is more luminous than the normal one and is thought to be produced by major changes in the Be decretion disc structure. We ascribe the bimodal behaviour evident in the CLD of SAX J2103.5+4545 shown in Fig. 1 to the two different luminosities reached during the two types of outbursts: low (high) luminosity in normal (giant) one, respectively. Other Be/XRBs show more complex shapes, multi-modal distributions (like in the case of EXO 0331+530 shown in Fig. 1), indicative of multi peaks within the same outburst, or outbursts reaching different peak luminosities.

The CLDs of all HMXBs of our sample are reported by Sidoli & Paizis (2018; their Fig. 1-4). In their normalized version, these functions allow the reader to obtain in one go, not only an easy comparison between all kind of HMXBs, but also to quantify the time spent by each HMXB in any given luminosity state, above the instrumental sensitivity.

Average Luminosity (18-50 keV). An average luminosity (18-50 keV) was calculated for each source, over the INTEGRAL detections (at 2 ks timescale; see Table 1, forth column). Note that this definition implies that, for transient sources, this is an average luminosity * in outburst*.

4 Other HMXB properties from the literature

Dynamic Ranges (1-10 keV). Other source properties were collected from the literature, in order to put the INTEGRAL behavior into a wider context: distance, pulsar spin and orbital period, eccentricity of the orbit, maximum and minimum fluxes in soft X–rays (1–10 keV, corrected for the absorption). These latter were investigated since the instruments observing the sky at soft X–rays are much more sensitive than INTEGRAL and can probe the true quiescent state in transient sources, together with their variability range between quiescence and outburst peak.

When the published soft X–ray fluxes were not available in the 1–10 keV range, we extrapolated them using WebPIMMS and the appropriate model found in the literature. Then, we calculated their ratio (the dynamic range “DR${}_{1-10~{}keV}$” = Fmax / Fmin, reported in Table 1, last column). When only a single value for the soft X-ray flux was found, the dynamic range was not calculated (“$-$” in Table 1) and the flux was ascribed to the “minimum flux”. Note that we considered only spin-phase-averaged fluxes for X-ray pulsars, and out-of-eclipse minimum fluxes for eclipsing systems, to obtain the intrinsic range of X–ray variability.

In Fig. 2 we show the source DC${}_{18-50~{}keV}$ plotted against the minimum and maximum luminosities (1–10 keV), for different HMXB sub-classes: the scatter is huge in the upper panel where the duty cycle is plotted versus the minimum soft X–ray luminosity. The SFXTs are located in the lower left part of the plot, at low DC${}_{18-50~{}keV}$ and X–ray luminosity in quiescence, while the persistent SgHMXB mostly lie in the upper right part, at both high luminosities and large duty cycles. Be/XRBs appear located in-between them. In the lower panel, where the maximum soft X–ray luminosity is considered, the sub-classes regroup to the right, at more similar luminosities (in outbursts for SFXTs and Be/X-ray transients). A few sources, classified in the literature as SgHMXB (blue circles in Fig. 2), display a very low DC${}_{18-50~{}keV}$, similar to SFXTs. They might be either mis-classified transients or persistent sources emitting X–rays at a level just below the instrumental sensitivity, that are detected only during sporadic flaring. Note that the HMXBs almost reaching the Eddington luminosity are the RLO systems SMC X–1 and LMC X–4.

Orbital geometry. Among the many trends of source properties we have investigated for our sample (see Sidoli & Paizis, 2018), we report here on the plot showing the system eccentricity versus the orbital period (Fig. 3). Two trends are evident, above Porb$\sim$10 d: low eccentricity Be/XRBs with no correlation with the orbital period (X Per is the prototype) and a group of binaries (mostly Be/XRBs) where the eccentricity correlates with the orbital period. SgHMXBs are located at lower eccentricities and orbital periods. This plot has already been investigated in the literature ([Townsend et al. 2011, Townsend et al. 2011]). The novelty here is the inclusion of SFXTs (not considered by Townsend et al. 2011): some of them display circular orbits, while others very eccentric geometries, like IGR J08408-4503 (e=0.63 and Porb=9.54 d) and IGR J11215-5952 (e$>0.8$ and Porb=165 d). These SFXTs enable the HMXBs hosting supergiant stars to extend at larger eccentricities and orbital periods, in a parameter space that is unusual even for Be/XRBs.

5 Conclusions

We summarize the results of our systematic analysis in Fig. 4, making use of three characterizing quantities: two of them have been derived from the analysis of fourteen years of INTEGRAL observations (DC${}_{18-50~{}keV}$ and the average 18–50 keV luminosity, in outburst for transients), while the third one has been calculated from soft X–ray fluxes taken (or extrapolated) from the literature (DR${}_{1-10~{}keV}$).

We have obtained a global view of a large number of HMXBs where the different kind of sources tend to cluster mainly in different region of this 3D space, as follows:

• •

SgHMXBs (excluding the high luminosity RLO systems) in general show low DR${}_{1-10~{}keV}$ ($<$ 40), high duty cycles (DC{}_{18-50~{}keV}$$>10 per cent), low average 18–50 keV luminosity ($\sim$1036 erg s*-1*);

• •

SFXTs are characterized by high DR${}_{1-10~{}keV}$ ($>$100), low duty cycles (DC{}_{18-50~{}keV}$$<5 per cent), low average 18–50 keV luminosity in outburst ($\sim$1036 erg s*-1*);

• •

Be/XRTs display a high DR${}_{1-10~{}keV}$ ($>$100), intermediate duty cycles (DC{}_{18-50~{}keV}$$\sim10 per cent), high average 18–50 keV luminosity in outburst ($\sim$1037 erg s*-1*).

It is worth mentioning that a number of HMXBs exist that displays intermediate properties, in particular among SgHMXB, sometimes overlapping with some region of the parameter space more typical of SFXTs, bridging together the two sub-classes. This seems to indicate that these two sub-classes have no sharp boundaries, but their phenomenology is based on continuous parameters, from persistent SgHMXBs towards the most extreme SFXT (IGR J17544-2619).