Detection of gamma rays of likely jet origin in Cygnus X-1

Detection of gamma rays of likely jet origin in Cygnus X-1

Key Words.:
acceleration of particles – accretion, accretion disks – gamma rays: general – radiation mechanisms: non-thermal – Stars: individual: Cygnus X-1 – X-rays: binaries

Abstract

Context:

Aims:Probe the high-energy (60 MeV) emission from the black hole X-ray binary system, Cygnus X-1, and investigate its origin.

Methods:We analysed 7.5 yr of data by Fermi-LAT with the latest Pass 8 software version.

Results:We report the detection of a signal at 8  statistical significance spatially coincident with Cygnus X-1 and a luminosity above 60 MeV of 5.510 erg s. The signal is correlated with the hard X-ray flux: the source is observed at high energies only during the hard X-ray spectral state, when the source is known to display persistent, relativistic radio emitting jets. The energy spectrum, extending up to 20 GeV without any sign of spectral break, is well fitted by a power-law function with a photon index of 2.30.2. There is a hint of orbital flux variability, with high-energy emission mostly coming around the superior conjunction.

Conclusions:We detected GeV emission from Cygnus X-1 and probed that the emission is most likely associated with the relativistic jets. The evidence of flux orbital variability points to the anisotropic inverse Compton on stellar photons as the mechanism at work, thus constraining the emission region to a distance 10 cm from the black hole.

1 Introduction

Cygnus X-1 is an X-ray binary (XRB), a system in which the compact object accretes matter from the companion star. The former has been identified as a black hole (BH) with (14.81.0) M (Orosz et al., 2011). The latter is the early-type O9.7Iab supergiant HDE 226868 (Walborn, 1973), with a mass of (19.21.9) M (Orosz et al., 2011). However, this value has been questioned by Ziółkowski (2014), who suggested a range of 25–35 M. Cygnus X-1 is the only high-mass XRB for which the compact object has been identified to be a BH.

Located at a distance of 1.86 kpc (Reid et al., 2011; Xiang et al., 2011), it is one of the brightest X-ray sources, thus considered an optimal candidate for the study of the accretion and ejection processes onto a BH system. The spectrum of the BH X-ray binaries can be roughly described as the sum of two components: a blackbody-like emission coming from the geometrically thin, optically thick accretion disk, and a power-law tail whose origin is still under debate. The dominance of one or the other component defines the two main spectral states the system can display: the soft state (SS) and the hard state (HS). The two main states are joined by short-lived (typically of a few days; Grinberg et al. 2013) intermediate states (IS), and the complete sequence of states is well-represented in a hardness intensity diagram (HID) by the q-shaped track trajectory (Fender et al., 2004).
The SS is dominated by the thermal emission peaking at 1 keV and a steep power-law at higher energies with photon index  2–3. In the HS instead, the blackbody component is much less luminous, with a 0.1 keV temperature and most of the energy is emitted in a hard tail component characterised by a 1.5 photon index and an exponential cutoff at a few hundred keV. The canonical explanation for this hard X-ray emission is inverse Compton scattering of disk photons by hot ( keV) thermal electrons in the inner region of the accretion flow, usually referred to as “corona” (Shapiro et al., 1976; Sunyaev & Truemper, 1979). However, Aharonian & Vardanian 1985 proposed that this emission has a non-thermal origin related to the development of electromagnetic cascades initiated by particles accelerated to relativistic energies in regions close to the BH, i.e. in the accretion disk. In this scenario the authors showed that the resulting photon spectrum has a spectral break at most at 1 MeV. In addition, the HS generally displays relatively persistent relativistic jets emitting synchrotron radiation at GHz radio frequencies, whereas in the SS, where the disk comes up to the BH last stable circular orbit, such an emission is strongly quenched. A two-cluster non-linear correlation between the radio and the X-ray fluxes, with slopes of and 1, respectively, suggests that there is a close coupling between the X-ray and the radio emitters (Gallo et al., 2003, 2012). The existence of such a correlation was used to prove a possible synchrotron origin of the X-ray power-law tail (Markoff et al., 2003).

As a persistent source, Cygnus X-1 retains always a strong power-law spectral component; even in its SS, i.e. its spectrum is never fully disk-dominated. Whereas in the HS it shows a mildly relativistic () radio jet (Stirling et al., 2001; Gallo et al., 2003) which carries a significant fraction of the system X-ray luminosity (Gallo et al., 2005), in the SS there is evidence for a factor 3–5 weaker unresolved compact jet (Rushton et al., 2012). The constant mean level of the radio emission is of 10–15 mJy, with a flat spectrum and no evidence for a cutoff (Fender & Hendry, 2000) up to IR frequencies, where the emission is totally dominated by the supergiant, making impossible the measurement of the spectral break. Another peculiarity of the Cygnus X-1 HS is that above the hard X-ray tail which cuts off at 100 keV (Wilms et al., 2006), an additional harder (with a 1.6 photon index) non-thermal component emerges extending up to a few MeV (Cadolle Bel et al., 2006; Rodriguez et al., 2015). This soft gamma-ray radiation was recently shown to be polarised with a polarisation fraction increasing with energies and an average value of (7615)% at a position angle of (423) for energies above 230 keV (Laurent et al., 2011; Jourdain et al., 2012). The most plausible explanation for this is that the jet synchrotron emission extends itself up to MeV energies (Jourdain et al., 2012; Zdziarski et al., 2012), requiring the existence of a parent population of ultra-relativistic electrons. In addition, such a high-level of polarisation would imply that the phenomenon is persistent on time-scales of weeks to months. However, the origin of this emission is still controversial (Zdziarski & Gierliński, 2004) and it was also suggested to originate in the corona (Romero et al., 2014). Therefore, the only proof that there is non-thermal gamma-ray jet emission would undoubtedly be the detection of GeV emission.

Cygnus X-1 spends most of its time in the HS, although the fraction of time observed in the SS is not constant with time. The latter increased from 10% between 1996 and 2000, to 34% between 2000 and mid 2006 (Wilms et al., 2006), most probably due to an overall increase of the stellar radius. In this work we adopt the state definition described in Grinberg et al. (2013) who used the data of the available all-sky monitors: RXTE-ASM, MAXI, Swift-BAT, and Fermi-GBM. They showed that publicly available Swift-BAT (15–50 keV) data can be used to distinguish SS from the HS+IS: the source is in the SS when the Swift-BAT daily count rate is smaller than 0.09 cts cm. However, without soft coverage it is not possible to distinguish between the HS and the IS.

Cygnus X-1 BH is in a 5.6 d orbit. In this paper we adopted the most updated ephemerides in Gies et al. (2008), with a phase 0 corresponding to superior conjunction T=52872.788 HJD, i.e. when the companion is between the observer and the BH (see the schematic diagram of the binary in Figure 1). The orbital period is observed at all wavelengths: optical, infrared (Gies & Bolton, 1982), X-ray (Brocksopp et al., 1999), and also at radio frequencies (Pooley et al., 1999), suggesting that such a modulation could be the result of absorption by the stellar wind (Brocksopp et al., 2002). As confirmation, Grinberg et al. (2015) show that the absorption column density in the HS is strongly modulated with a maximum around superior conjunction. The existence of the radio modulation supports the idea that radio emission comes from a continuous jet versus discrete ejections. Another type of periodical behaviour is observed in Cygnus X-1 both at X-ray and radio frequencies, a superorbital modulation of 140 d (Brocksopp et al., 1999; Pooley et al., 1999), although such a value is rather unstable and it has been recently showed to be doubled (Lachowicz et al., 2006; Rico, 2008; Zdziarski et al., 2011). The superorbital modulation is possibly related to the precession of the disk-jet system (Brocksopp et al., 1999), or alternatively to a variable mass accretion rate (Brocksopp et al., 2001), and its period possibly varies when an X-ray spectral state change occurs (Rico, 2008).

Figure 1: Schematic diagram of Cygnus X-1. The orbital phase =0 corresponds to the superior conjunction of the compact object. The orbit of the BH is indicated by the ellipse with an eccentricity of 0.018 (Orosz et al., 2011) . Neither the inclination of the orbit with the line of sight or the longitude of the ascending node were considered. The stellar radius, assumed to be 16.4R, as well as the periastron phase and argument were taken from Orosz et al. (2011). AU stands for astronomical unit.

HE emission from BH XRB is theoretically predicted invoking either leptonic or hadronic processes (e.g. Bosch-Ramon & Khangulyan 2009 for a review) and generally tied to the existence of the radio jets, where particles can be accelerated up to relativistic energies. So far, the only microquasar (i.e. XRB displaying relativistic jets) firmly detected at high energies ( MeV) is Cygnus X-3 (Tavani et al., 2009; Fermi LAT Collaboration et al., 2009) and its gamma-ray emission is related to the formation/existence of the radio jets (Piano et al., 2012; Corbel et al., 2012). However Malyshev et al. (2013) showed there is a 4-level evidence of gamma-ray signal for Cygnus X-1, above 100 MeV, in 3.8 yr of Fermi-LAT data, only when the source is in the HS. In addition to this steady emission there are claims of isolated 1-2-day long flaring events reported by AGILE above 100 MeV (Sabatini et al., 2010, 2013) and of a flare of  d duration reported by the MAGIC collaboration above 100 GeV (Albert et al., 2007). In particular, AGILE detected three episodes of significant transient emission while it was in its pointing operational mode. These 1–2 d long events occurred on 2009, October 16 (0.38–0.56 orbital phase) with an integral flux of (2.320.66)  ph cm s between 0.1 and 3 GeV (Sabatini et al., 2010); on 2010, March 24 (Bulgarelli et al., 2010) with a flux above 2.50  ph cm s above 100 MeV; and on 2010, June 30 with a (1.450.78)  ph cm s average flux in the same energy range (Sabatini et al., 2013). Whereas the first two episodes happened when the source was in the HS, the last one occurred during a hard-to-soft state transition, but coincident with the source entering in the SS and a couple of days before of an anomalous radio flare (Negoro et al., 2010; Rushton et al., 2010; Wilson-Hodge & Case, 2010). An independent analysis of 3.6 yr of Fermi-LAT data confirmed evidence of flaring activity on 1–2 d timescales contemporaneous, but not coincident, with AGILE at 3–4 level (Bodaghee et al., 2013). Such episodes show integral fluxes typically lower than the ones reported by AGILE, but still compatible within the large statistical uncertainties. The reported evidence (at 4 post-trial) of very-high-energy (VHE, 100 GeV) emission was detected by MAGIC on 2006, September 24 for 80 minutes (corresponding to an orbital phase of 0.9) when the source was in the HS (Albert et al., 2007), but it occurred exactly one day before of a hard X-ray flare observed by INTEGRAL (Malzac et al., 2008). Further long observational campaigns (100 hr) were carried out by MAGIC meant to catch additional short flaring episodes similar to the September-2006 one, but with no success (Fernández-Barral et al., 2015). VERITAS did not report any VHE signal from Cygnus X-1 too (Guenette, 2009).
In this work we search for both steady and variable emission from Cygnus X-1 at high energies, above 60 MeV, by using 7.5 yr of data by Fermi-LAT.

2 Observations and analysis

Fermi-LAT is an electron–positron pair production telescope, featuring solid state silicon trackers and cesium iodide calorimeters, designed to be sensitive to photons from 20 MeV up to  GeV (Atwood et al., 2009).

We used 7.5 yr of Pass 8 Fermi-LAT data from August 4, 2008 (MJD 54682) to February 2, 2016 (MJD 57420). The recently released Pass 8 data benefits for a wider energy range (from 60 MeV to 500 GeV), better energy resolution, improved point spread function (PSF), and significantly increased effective area. In addition, more accurate Monte Carlo simulations of the detector led to a reduction of the systematic uncertainty in the LAT instrument response functions (IRF). The data were reduced and analysed using Fermipy1, a set of python tools which automatize the Pass 8 analysis with the FERMI SCIENCE TOOLS v10r0p5 package2. Standard FERMI SCIENCE TOOLS were also used to cross-check the recently released python package. We selected LAT photons from the “P8R2_SOURCE” class in the widest possible energy range between 60 MeV and 500 GeV and within a 30 acceptance cone centered on the position of Cygnus X-1 (RA=19h58m21.676s, Dec=+3512m5.78s, van Leeuwen 2007). We chose 60 MeV as the lowest energy threshold because that is the minimum energy of the public Galactic diffuse model. The “SOURCE” event class was chosen to maximize the effective area, knowing that Cygnus X-1 is expected to be pointlike. Thus, we used the corresponding “P8R2_SOURCE_v16” IRF. Within any photon class, Pass 8 subdivides the events into quartiles according to the quality of either the direction (PSF event-type) or the energy reconstruction (EDISP event-type). In this work, we analysed the four PSF event-types separately and later we combined them by means of a joint likelihood fit. In order to minimize the contamination by the albedo gamma rays from the Earth, we excluded those photons having reconstructed directions with angles with respect to the local zenith larger than 90, 85, 75, and 70 for the four PSF quartiles, respectively: the tighter the better the PSF is. Since our analysis is not very sensitive to the contamination from the Earth Limb, and, in addition, our data sample includes data taken during the period of the Galactic-Center biased pointing strategy, we did not apply any cut on the rocking angle. The PSF event-type analysis has the best achievable angular resolution among the possible Fermi-LAT analysis, e.g. at 1 GeV it is 0.5, being a value between 0.8 and 0.3 which are the angular resolutions of the analysis that uses the four PSF event-type in one single likelihood and of the one of the quartile with the best PSF (PSF3), respectively3. We cross-checked those results repeating the analysis with the standard conversion-type (FRONT+BACK) selection of events, and a conservative maximum zenith of 90. The results obtained with the two selections are compatible within the expected improvements of the first approach.

We created 14 14 regions of interest (ROI) in Galactic coordinates. To model the diffuse background, we used the templates for the Galactic diffuse emission (gll_iem_v06.fits) and an isotropic component (iso_P8R2_SOURCE_V6_PSFx_v06.txt with x=0,1,2,3) including the extragalactic diffuse emission and the residual background from cosmic rays4. We built a model of point-like gamma-ray background sources within 22 starting with the third LAT source catalogue (3FGL, Acero et al. 2015), which is based on 4 yr of Pass 7 Reprocessed data. We let all the spectral parameters of all the sources, except those located more than 7 away from the center, free to vary in a maximum likelihood fit (using gtlike). We also let free the flux normalisation of the Galactic diffuse and isotropic components and of the extremely bright 3FGL sources (significance 100) lying between 7 and 14 from the center.

Source detection significance is determined, fixing the source position to its nominal value given by van Leeuwen (2007), using the Test Statistic value, TS =(L0/L1) which compares the likelihood ratio of models including, e.g., an additional source, with the null hypothesis of background only (Mattox et al., 1996). The TS maps were computed for a power-law test source with a photon index of 2.5 and obtained with all the background sources fixed. The TS maps presented in this work were obtained above 1 GeV where the angular resolution is 0.5.

For the spectral analysis we splitted the 0.06-500 GeV energy range into 7 logarithmically spaced bins. The spectral energy distribution (SED) was computed by fitting the source normalisation factor in each energy bin independently while keeping its photon index fixed to the value found in the overall, full energy range, fit. The spectral parameters of the background sources were fixed to those previously found in the overall fit. For each spectral point we required a TS of at least 4, when this condition was not fulfilled, upper limits (UL) at 95% confidence level (CL) were computed.

The source localisation was performed above 1 GeV with a two-step algorithm: first it looks for the maximum peak in a reduced TS map of 4 4 centered around the considered source, and then it redefines the source position by performing a full likelihood fit in the vicinity of the peak found in the first step5, with the normalisation parameters of bright (TS100) background sources let free to vary.

The light curve, i.e. integral flux as function of the observation time, is the only result which was not produced with the Fermipy software package, but with the standard FERMI SCIENCE TOOLS and with the standard conversion-type selection of events and a maximum zenith angle cut of 90. The energy threshold for the light curve is 100 MeV in order to have a direct comparison with the previously published results (Sabatini et al., 2010, 2013; Bodaghee et al., 2013). We computed a maximum likelihood fit for each temporal bin of 1  d, and then we estimated either its 0.1–20 GeV integral flux or 95% CL UL depending on the strength of the signal, with a threshold of TS=9.

3 Results

The TS map above 1 GeV obtained by using the background model including all the 3FGL sources is not flat. Besides a clear excess in the center, coincident with Cygnus X-1, the TS map shows 7 excess spots with a TS larger than 25 in the full energy range, between 60 MeV and 500 GeV (see Figure 2). We modelled these excess spots as power-law point-like sources, in particular:

Figure 2: 5 5 TS map centered at the nominal position of Cygnus X-1 above 1 GeV obtained when including only the 3FGL sources in the background model.
  • J1942+40: the LAT excess located at RA=19h:42m:7s and Dec=+40:14m:7s most probably comes from the direction of the open cluster NGC 6819 where several X-ray sources were detected by the XMM-Newton observatory (Gosnell et al., 2012). In the full energy range the source has a TS of 55.

  • J1949+34: a LAT excess with a TS of 35 in the full energy range which is located at RA=19h:49m:7s and Dec=+34:15m:44s.

  • J1955+33: an excess located at RA=19h:55m:10s, Dec=+33:18m:34.8s, and with a TS of 90 in the full energy range.

  • J2005+34: a LAT excess centered in RA=20h:05m:19.7s, Dec=+34:18m:23.7s with a TS of 49 above 60 MeV.

  • J2006+31: a clearly identified (TS=115) new LAT source, outside of the Galactic Plane, it is centered in RA=20h:06m:12.8s, Dec=+31:02m:38.3: is spatially coincident with the 164 ms period radio pulsar PSR J2006+3102 (Nice et al., 2013). We found that the LogParabola function provides a better (at more than 3  level) fit of this source spectrum. Just for this specific case we included a non-power-law spectrum in our background model.

  • J2009+35: a LAT excess with a TS of 48 (above 60 MeV) located at RA=20h:09m:57.8 and Dec=+35:44m:48.6s.

  • J2017+35: a LAT excess with a TS of 65, above 60 MeV, located at RA=20h:17m:25s and Dec=35:26m:5s.

The search for the origin of these excesses goes beyond the goal of this paper. Their location was estimated above 1 GeV and has a statistical uncertainty of 0.2. In addition, we found that the centroid of the LAT source associated with the SNR G73.9+0.9 is offset by 0.24 with respect to the position given in the 3FGL catalogue (3FGL J2014.4+3606). The new centroid is located at RA=20h:13m:33.8s and Dec=36:11m:54.0s. The LogParabola spectral model suggested by the new Pass 8 analysis in Zdziarski et al. (2016b) is not significantly favoured with respect to a power-law function, which was used in this work.

Figure 3: 2 2 TS maps centered at the nominal position of Cygnus X-1 above 1 GeV. On the left: the Fermi-LAT data subsample corresponding to Cygnus X-1 being in the HS; On the right: the data subsample corresponding to the SS of the source. The white cross indicates the nominal position of Cygnus X-1.

Once the new background sources are included in our model, a point-like source at the position of Cygnus X-1 is popping up at TS=53 in the full energy range, between 60 MeV and 500 GeV. Above 1 GeV the source is still detected at TS of 31. Among the new background sources included in our model the ones at more than 3 away from the Cygnus X-1 position do not have any effect on the source significance estimation, whereas the new background excesses lying within a 3 radius from the nominal Cygnus X-1 position decrease the TS of the signal from 65 to 53, once included in the background model. Applying the localisation algorithm in the Fermipy package, we fitted the position of the gamma-ray excess above 1 GeV to RA=19h:58m:56.8s and Dec=+35:11m:4.4s, 0.05 offset from the nominal position, but still compatible with Cygnus X-1 within the statistical uncertainties of 0.2. Besides the spatial coincidence, we found a strong correlation of the gamma-ray excess with the X-ray spectral states. We divided our sample in HS(+IS) and SS by using the public Swift-BAT data (daily Swift-BAT count rate lower/larger than 0.09 cts cm, Grinberg et al. 2013), which has a complete temporal overlap with Fermi-LAT. We did not account for any soft X-ray information in our selection criteria, aiming to clearly identify all the IS intervals, because, given the short duration of the IS, their eventual inclusion in the HS does not alter our result. The bottom panel in Figure 6 shows the X-ray fluxes as a function of the time interval considered in this work (August 2008 - February 2016), and the two X-ray spectral states are emphasised by the different-colored-shadowed bands. In particular, the HS and SS intervals in MJD are listed in Table 1. We detected Cygnus X-1 when it was in the HS with a TS of 49 above 60 MeV and an energy flux integrated over the entire energy analysis range of (7.71.3) MeV cm s. On the other hand, there is no significant LAT excess in coincidence with Cygnus X-1 when it is in the SS (TS=7) and the UL on its energy flux above 60 MeV at 95% CL is of 5.4 MeV cm s. The two corresponding TS maps for energies larger than 1 GeV are shown in Figure 3. Given the comparable exposure time for the two considered LAT subsamples of 3.6 yr and 3.7 yr, for the HS and SS, respectively, we do not need to normalize with respect to it and we can confirm that Cygnus X-1 is detected at high significance only when the source is in the HS, as previously claimed by Malyshev et al. (2013), and recently confirmed by Zdziarski et al. (2016a). Some high-energy emission in the SS, although significantly fainter than in the HS, cannot be excluded. Spectral and timing results were then computed only for the HS subsample. We checked for a possible dependence on the orbital period of the flux. Given the low significance of the signal we divided the HS data sample in only two bins, one centered on the superior conjunction (), and one on the inferior conjunction (). Cygnus X-1 is clearly detected in the full energy range with a TS of 31 only around the superior conjunction. In the second bin, around the inferior conjunction, no significant signal from the Cygnus X-1 position is seen (TS=10). Figure 4 shows the corresponding two TS maps above 1 GeV. The energy flux above 60 MeV is (7.61.7) MeV cm s for the phase interval around superior conjunction. The low statistics does not allow us to make any strong conclusion on a possible flux dependence on the orbital position. Nevertheless, the low significance on the inferior conjunction phase bin, for the same exposure time, can be considered as a hint of the orbital modulation of the flux.

HS SS
54682–55375 55391–55672
55672–55790 55797–55889
55889–55945 55945–56020
56020–56086 56086–56330
56718–56753 56338–56718
56759–56839 56839–57009
56848–56852 57053–57103
57009–57053 57265–57325
57103–57265
57325–57420
Table 1: Days (MJD) intervals for the HS and SS.
Figure 4: Phase-dependent TS map of a 2 2 region centered on Cygnus X-1 (above 1 GeV) when the source is in the HS on the left around the superior conjunction and on the right around the inferior conjunction .

The energy spectrum of Cygnus X-1 is well fitted by a power-law function with a photon index =(2.30.1) and a normalisation factor of =(5.80.9)  MeV cm s at the decorrelation energy of 1.3 GeV, and it extends from 60 MeV to 20 GeV. The obtained SED is illustrated in Figure 5. We also tried to fit the LAT data with a broken power law, but the obtained improvement is not statistically significant (TS2). The photon indices of the inferior and superior conjunction energy spectrum are compatible within 1  with the overall HS ones. The flux normalisation at 1.3 GeV for the superior and inferior conjunction, computed both with 2.3 photon index, is of =(5.71.3)  MeV cm s, and =(3.71.3)  MeV cm s, for the superior and inferior conjunction, respectively.

Figure 5: SED of Cygnus X-1 in the HS, extending from 60 MeV to 20 GeV. It is best fitted by a power-law function with a photon index =(2.30.1).

We performed a timing analysis looking for the orbital period of 5.6 d, or eventually the super-orbital period. We applied the Lomb-Scargle test of uniformity (Lomb, 1976), to our daily light curve for the HS data sample, where significant signal is detected, using the most probable value for the integral flux regardless the TS values. Between the maximum and minimum period sampled (2 and 1000 d), no periodicity is observed being the most significant one compatible at 3% with the null hypothesis of no periodicity.

Figure 6: Light curve of Cygnus X-1 obtained by the analysis of the Fermi-LAT data between August 4, 2008 and February 2, 2016 in 1 d bins. From top to bottom: the integral fluxes in the 0.1–20 GeV energy range, the TS corresponding to the above integral fluxes, the daily light curves in soft and hard X-rays from MAXI(cts s cm in 2–20 keV), RXTE-ASM (cts s in 3–5 keV divided by 10) and Swift-BAT (10 cts s cm in 15–50 keV), and the daily integral fluxes at 15 GHz by AMI. In the 0.1–20 GeV range, days with TS9 are shown in the top panel as filled black circles, whereas the purple arrows represent the 95% CL UL for the bins with TS9. Dashed black vertical lines show the previously reported gamma-ray flares by AGILE. The horizontal green dot-dashed line in X-ray panel indicates the threshold level of the Swift-BAT count rate used in this work to separate the SS from the HS(+IS): 0.0910 cts s cm. The shadowed gray bands identify the HS(+IS), whereas the red ones the SS.

We also looked for short flux variability on daily timescales as claimed by the AGILE collaboration. Figure 6 shows the light curve of the 7.5 yr of data with a one-day binning, in the 0.1–20 GeV energy range, where 20 GeV is the limit to which Cygnus X-1 energy spectrum extends. No hint of variability at daily timescale is observed and the distribution of the daily TS is consistent with the expected distribution. For completeness, Table 2 lists the 9 days showing signal with TS9. They are not clustered around previously reported daily variabilities, being none of them coincident with the AGILE claims (Sabatini et al., 2010, 2013; Bulgarelli et al., 2010) and just one (MJD 55292) with a 3 event detected by Bodaghee et al. 2013. Nevertheless, possible differences on the exposure times and the effective area degradation due to large off-axis angles of the Cygnus X-1 daily observations by the two gamma-ray detectors AGILE and Fermi-LAT could explain this apparent contradiction, as it shown for the case of AGL J2241+4454 (Munar-Adrover et al., 2016).

Date TS Fermi-LAT flux X-ray State
(yyyy mm dd) (MJD) (10 photons cms)
2009-03-05 54895 10.3 4.82.0 HS
2010-02-02 55229 10.5 6.22.3 HS
2010-04-06 55292 12.2 3.11.1 HS
2012-12-31 56292 9.2 5.32.2 SS
2014-02-10 56698 9.7 7.73.1 SS
2014-03-01 56717 10.5 6.32.5 SS
2014-03-06 56722 9.4 6.02.5 HS
2014-08-08 56877 10.2 6.72.5 SS
2015-05-26 57168 9.6 4.51.8 HS
Table 2: Days (MJD) with a significance 3  in the 1-day bin light curve (0.1–20 GeV) shown in Figure 6.

4 Discussion

We established a new statistically significant LAT source above 60 MeV, spatially coincident with the prototype BH microquasar Cygnus X-1, and the previous marginal detections reported by (Malyshev et al., 2013; Sabatini et al., 2010, 2013). The use of the more sensitive Pass 8 analysis software and 7.5 yr of data by Fermi-LAT allowed us to obtain the first high-significance detection of a BH binary at high energies, as well as to study a possible flux variability. The correlation between the HE flux and the hard X-ray one, together with the hint of flux orbital modulation, strongly support the identification of the HE source with the microquasar. In particular, Cygnus X-1 is detected only when in the HS. During these periods, the emission is more significant (TS=31) when the source is around the superior conjunction (0.250.75), while becomes fainter at inferior conjunction (TS=10). The overall HS emission is well described by a 2.3 power-law function with a luminosity of the GeV emission of  erg s, few orders of magnitude smaller than the total power carried by the jets ( erg s; Gallo et al. 2005; Russell et al. 2007).

Gamma-ray emission from XRB, and in particular from microquasars, has been predicted by several authors and associated to either the corona or the relativistic radio jets. Both leptonic (e.g. Atoyan & Aharonian 1999; Georganopoulos et al. 2002) and hadronic (Romero et al. 2003) mechanisms have been proposed in the literature to explain such high-energy radiation (see Bosch-Ramon & Khangulyan 2009 for a discussion on different processes). Leptonic models invoke inverse Compton scattering on seed photons, where the target photon field depends on the production region, mainly on the distance from the BH. If particles are accelerated close to the BH (Kafatos et al., 1981), the main target photons are the thermal ones from the accretion disk. When particles are accelerated along the relativistic jets, the seed photons can be either thermal photons from the accretion disk or synchrotron soft photons produced by the same population of electrons (Synchrotron-Self-Compton, SSC), or the photons from the companion star (with a black-body peak emission at 2.7kT10 eV). The existence of synchrotron emission from the jet is supported by the hint of strong polarisation in the 0.2–1 MeV tail (Laurent et al., 2011; Jourdain et al., 2012; Rodriguez et al., 2015), the luminosity of which is  erg s. The inverse Compton scattering on stellar photons would be the dominant mechanism of high-energy radiation if the emission is not originated at the base of the jet. At a distance of few times 10 cm (see also Romero et al. 2014), the energy density of the stellar radiation field () becomes dominant with respect to the other two photon fields. In particular, , where =710 erg s is the star luminosity (Orosz et al., 2011), is the orbital distance assumed to be 3 cm, and is the distance from the BH along the jet, whereas and , with the luminosity in the 1–20 keV energy range spanning from to erg s, depending on the model used to fit the soft part of the spectrum (Di Salvo et al., 2001). Particles could also be accelerated outside the binary system in shocks formed when the jets interact with the surrounding medium, as it is likely to be the case in the microquasar SS 433 (Bordas et al., 2015). In particular, Cygnus X-1 jets are thought to inflate a ring-like (5 pc in diameter) structure, detected at radio frequencies (Gallo et al., 2005), and extending 10 cm away from the BH. This value is assumed as the maximum extension of the relativistic jets.

If particles are accelerated to relativistic energies close to the BH, they could create electromagnetic cascades and originate gamma rays. Those gamma-rays will suffer heavy absorption due to photon-photon collision. Following the approach of Aharonian et al. 1985, we can constrain the minimum region size for GeV photons to escape avoiding pair production on  1 keV X-ray photons. Considering a spherical accretion geometry, and for a distance of 1.86 kpc (Orosz et al., 2011) and a de-absorbed flux at 1 keV of 1.6 erg cm s (Di Salvo et al., 2001), the emission region size must be larger than R10 cm. Given that the radius of the corona, meant as the inner part of the accretion flow, is of 20–50 R5–10cm (Poutanen & Coppi, 1998), where R is the gravitational radius, we can exclude that the observed GeV emission is produced in the inner regions of the accretion flow. This absorption also disfavours the advection-dominated-accretion-flows (ADAF) models, predicting gamma-ray emission in the HS, when the ADAF flows are expected to be present (Mahadevan, 1997). The GeV emission should be produced outside the corona, and most likely it is associated with the jets. This conclusion is also strengthened by the detection of the system only in the HS, when persistent jets have been detected at radio frequencies. Further constraints on the production region can be obtained if the hint of flux dependence on the orbital phase peaking at around superior conjunction, as reported in this work, is finally confirmed. First of all, it sets an upper limit on maximum distance of the production region of a few times the size of the system (R):  cm, thus confirming that the GeV emission cannot come from the region where the jets interact with the ring-like structure, but from the jets themselves. In addition, such a flux variability is expected if the production mechanism of this GeV emission is anisotropic inverse Compton scattering (Jackson, 1972; Aharonian & Atoyan, 1981; Zdziarski & Pjanka, 2013; Khangulyan et al., 2014) on stellar photons. If no additional sources of variability are assumed, no variability is expected if either SSC or inverse Compton with the thermal accretion photons is the dominant mechanism. Since the energy density of the stellar photons dominates over the other possible photon fields only at distances  cm, the flux orbital modulation, if confirmed, constrains the GeV emitter to be located within a range 10 cm. This region is compatible with the results obtained by hydrodynamic simulations of stellar winds interacting with Cygnus X-1-like jets carrying a total power of 10 erg s (Perucho & Bosch-Ramon, 2008; Yoon et al., 2016; Bosch-Ramon & Barkov, 2016).

The energy of the parent population of electrons is at least several tens of GeV, and the inverse Compton scattering occurs mostly in the Thompson regime. A moderate magnetic field of B G (where is the acceleration efficiency: where is the acceleration timescale, and ) would be enough to accelerate the inverse Compton emitting electron population up to a few tens of GeV (Khangulyan et al., 2008), enough to produce HE photons via inverse Compton, in the large stellar photon field. Under the assumption that the same population of electrons that produces the GeV emission by inverse Compton scattering on stellar photons at  cm also emits synchrotron radiation at lower energies, the maximum magnetic field strength in this region is limited by the ratio between the luminosity of the observed X-ray emission, , and the one of the detected high-energy radiation . Otherwise, the synchrotron X-ray flux would exceed the X-ray observations:

(1)

as we considered the luminosity between 20 and 100 keV of 2.210 erg s (Cadolle Bel et al., 2006). At  cm the maximum magnetic field strength is of 2 kG, decreasing down to 700 G up to  cm.

At  GeV, the energy spectrum should already be affected by gamma-ray absorption due to pair creation in the stellar photon field. The created secondary pairs will mostly radiate inverse Compton emission around the pair production energy threshold ( GeV) leading, for typical primary gamma-ray spectra, to the formation of a bump in the SED in that energy range. The ULs at the highest energies reported in this paper indicate that the spectrum does not harden above  GeV. If gamma rays are indeed produced at energies  GeV, then significant inverse Compton cascading seems unlikely, which would imply that, either gamma-ray absorption is not attenuated by electromagnetic cascading in the GeV emitter, or the emission is produced at the upper end of the inferred emitter Z-range, where this absorption is expected to be minor (Bosch-Ramon et al., 2008).

In analogy with Cygnus X-3, the only other microquasar firmly established at energies above 100 MeV, gamma-ray emission of Cygnus X-1 is related to the existence of relativistic jets. In both cases the GeV emission is most likely produced by inverse Compton scattering on stellar photons (Fermi LAT Collaboration et al., 2009; Dubus et al., 2010). However, contrarily to Cygnus X-1, the conditions required to detect gamma rays from Cygnus X-3 are that the source is in the SS and showing significant emission (0.2–0.4 Jy) with rapid variations in the radio flux from the radio jets (Corbel et al., 2012). The latter is probably related to strong shocks (probably due to discrete jet ejections) occurring when the source undergoes state transitions in and out of the ultra-SS. Whereas the nature of Cygnus X-3 HE emission is transitional, the gamma-ray detection of Cygnus X-1 seems persistent within the limited statistics, whenever the jet is present, i.e. when the source is in the HS, and shows a radio flux at 15 GHz larger than 10 mJy6 . However, the daily flux variations weaker than ph cm s or variations on shorter timescales, as reported by the MAGIC collaboration (Albert et al., 2007), cannot be excluded (since Cygnus X-1 is not always in the field-of-view of the Fermi-LAT on a timescale of an hour). If confirmed, it could be associated to discrete jet ejections as well, like in the case of HE emission from Cygnus X-3.

The detection of the spectral break of the HE emission from Cygnus X-1 may be possible by combining 10 yr of Fermi-LAT data and the future generation of imaging atmospheric Cherenkov telescopes (IACT), the Cherenkov Telescope Array (CTA). Under the assumption of Pass 8 sensitivity, 10 yr of Fermi-LAT data will allow to constrain an eventual cutoff below 100 GeV, whereas CTA will be more fsensitive above this energy in 200 hr of observation. CTA could still detect a different emission component, possibly of hadronic origin (Pepe et al., 2015). The sensitivity curve of CTA for 200 hr of observations with the North array is included in the broad-band SED shown in Figure 7. Moreover, CTA will be an optimal instrument to probe the short-term flux variability of Cygnus X-1 hinted by MAGIC that is showing a flux which one order of magnitude larger that the ULs obtained for the steady emission. Figure 7 illustrates the sensitivity curve of the CTA North array7 scaled to a few hours of observations, corresponding to the maximum observation time for a specific source during 1 d.

Figure 7: Spectral energy distribution of Cygnus X-1 from X-rays up to TeV energies when in the HS, except for the MAGIC upper limits which were obtained by combining both SS and HS. Soft X-ray ( keV) data are taken from Di Salvo et al. (2001) (by BeppoSAX ), hard X-ray (10 keV-2 MeV) INTEGRAL from Figure 3 in Rodriguez et al. (2015), and HE (30 MeV-20 GeV) results from this work and from the previously published ones in Malyshev et al. (2013). At higher energies the differential UL on the steady emission obtained by the MAGIC collaboration (Albert et al., 2007), under the assumption of a 3.2 power law spectrum, are shown. The two gray curves are CTA-North differential sensitivities scaled for 5 and 200 hr of observation were taken from the CTA webpage https://portal.cta-observatory.org/Pages/CTA-Performance.aspx. No statistical errors are drawn.
Acknowledgements.
We would like to thank Luigi Tibaldo and Matthew Wood for the fruitful discussion about the Fermi analysis. We would like also to acknowledge Tiziana Di Salvo, Guy Pooley and Gabriela Vila for providing us multi-wavelength historical data, and Marc Ribó and Pol Bordas for their valuable comments. R.Z. acknowledges the Alexander von Humboldt Foundation for the financial support and the Max-Planck Institut fur Kernphysik as hosting institution. The Spanish funding agency MINECO supported the work of O.B. and A.F.B. through the project FPA2012-39502, which includes ERDF funds, as well as the one of EdOW and D.G.F. and V.B.R. through the projects AYA2015-71042-P, AYA2013-47447-C3-1-P, respectively. The Catalan DEC supported EdOW through the grant SGR2012–1073m and D.G.F. and V.B.R. through the grant SGR2014-86. In addition, D.G.F. and V.B.R. acknowledge support from MINECO under grant MDM-2014-0369 of ICCUB (Unidad de Excelencia María de Maeztú) and BES-2014-069376. EdO and V.B-R. also acknowledges financial support from MINECO and European Social Funds through a Ramón y Cajal fellowship. Finally, O.B. and A.F.B. are thankful for the support of the grant SEV-2012-0234 (Centro de Excelencia Severo Ochoa). This research has been supported by the Marie Curie Career Integration Grant 321520.

Footnotes

  1. http://fermipy.readthedocs.org/en/latest/
  2. See the Fermi Space Science Center (FSSC) Web site for details of the Science Tools: http://fermi.gsfc.nasa.gov/ssc/data/analysis/software
  3. https://www.slac.stanford.edu/exp/glast/groups/canda/lat_Performance.htm
  4. http://fermi.gsfc.nasa.gov/ssc/data/access/lat/BackgroundModels.html
  5. http://fermipy.readthedocs.org/en/latest/advanced/localization.html
  6. from public AMI-Large Array data. http://www.mrao.cam.ac.uk/guy/cx1/
  7. Taken from https://portal.cta-observatory.org/Pages/CTA-Performance.aspx

References

  1. Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23
  2. Aharonian, F. A. & Atoyan, A. M. 1981, Ap&SS, 79, 321
  3. Aharonian, F. A., Kririllov-Ugriumov, V. G., & Vardanian, V. V. 1985, Ap&SS, 115, 201
  4. Aharonian, F. A. & Vardanian, V. V. 1985, Ap&SS, 115, 31
  5. Albert, J., Aliu, E., Anderhub, H., et al. 2007, The Astrophysical Journal, Letters, 665, L51
  6. Atoyan, A. M. & Aharonian, F. A. 1999, MNRAS, 302, 253
  7. Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, The Astrophysical Journal, 697, 1071
  8. Bodaghee, A., Tomsick, J. A., Pottschmidt, K., et al. 2013, The Astrophysical Journal, 775, 98
  9. Bordas, P., Yang, R., Kafexhiu, E., & Aharonian, F. 2015, The Astrophysical Journal, Letters, 807, L8
  10. Bosch-Ramon, V. & Barkov, M. V. 2016, ArXiv e-prints [\eprint[arXiv]1604.06360]
  11. Bosch-Ramon, V. & Khangulyan, D. 2009, International Journal of Modern Physics D, 18, 347
  12. Bosch-Ramon, V., Khangulyan, D., & Aharonian, F. A. 2008, A&A, 489, L21
  13. Brocksopp, C., Fender, R. P., Larionov, V., et al. 1999, MNRAS, 309, 1063
  14. Brocksopp, C., Fender, R. P., & Pooley, G. G. 2002, MNRAS, 336, 699
  15. Brocksopp, C., Groot, P. J., & Wilms, J. 2001, MNRAS, 328, 139
  16. Bulgarelli, A., Pittori, C., Lucarelli, F., et al. 2010, The Astronomer’s Telegram, 2512
  17. Cadolle Bel, M., Sizun, P., Goldwurm, A., et al. 2006, A&A, 446, 591
  18. Corbel, S., Dubus, G., Tomsick, J. A., et al. 2012, MNRAS, 421, 2947
  19. Di Salvo, T., Done, C., Życki, P. T., Burderi, L., & Robba, N. R. 2001, The Astrophysical Journal, 547, 1024
  20. Dubus, G., Cerutti, B., & Henri, G. 2010, MNRAS, 404, L55
  21. Fender, R. P., Belloni, T. M., & Gallo, E. 2004, MNRAS, 355, 1105
  22. Fender, R. P. & Hendry, M. A. 2000, MNRAS, 317, 1
  23. Fermi LAT Collaboration, Abdo, A. A., Ackermann, M., et al. 2009, Science, 326, 1512
  24. Fernández-Barral, A., Blanch, O., de Oña Wilhelmi, E., et al. 2015, ArXiv e-prints [\eprint[arXiv]1510.03101]
  25. Gallo, E., Fender, R., Kaiser, C., et al. 2005, Nature, 436, 819
  26. Gallo, E., Fender, R. P., & Pooley, G. G. 2003, MNRAS, 344, 60
  27. Gallo, E., Miller, B. P., & Fender, R. 2012, MNRAS, 423, 590
  28. Georganopoulos, M., Aharonian, F. A., & Kirk, J. G. 2002, A&A, 388, L25
  29. Gies, D. R. & Bolton, C. T. 1982, The Astrophysical Journal, 260, 240
  30. Gies, D. R., Bolton, C. T., Blake, R. M., et al. 2008, The Astrophysical Journal, 678, 1237
  31. Gosnell, N. M., Pooley, D., Geller, A. M., et al. 2012, The Astrophysical Journal, 745, 57
  32. Grinberg, V., Hell, N., Pottschmidt, K., et al. 2013, A&A, 554, A88
  33. Grinberg, V., Leutenegger, M. A., Hell, N., et al. 2015, A&A, 576, A117
  34. Guenette, R. e. a. 2009, ArXiv e-prints [\eprint[arXiv]0908.0714]
  35. Jackson, J. C. 1972, Nature Physical Science, 236, 39
  36. Jourdain, E., Roques, J. P., Chauvin, M., & Clark, D. J. 2012, The Astrophysical Journal, 761, 27
  37. Kafatos, M., Shapiro, M. M., & Silberberg, R. 1981, Comments on Astrophysics, 9, 179
  38. Khangulyan, D., Aharonian, F., & Bosch-Ramon, V. 2008, MNRAS, 383, 467
  39. Khangulyan, D., Aharonian, F. A., & Kelner, S. R. 2014, The Astrophysical Journal, 783, 100
  40. Lachowicz, P., Zdziarski, A. A., Schwarzenberg-Czerny, A., Pooley, G. G., & Kitamoto, S. 2006, MNRAS, 368, 1025
  41. Laurent, P., Rodriguez, J., Wilms, J., et al. 2011, Science, 332, 438
  42. Lomb, N. R. 1976, Ap&SS, 39, 447
  43. Mahadevan, R. 1997, The Astrophysical Journal, 477, 585
  44. Malyshev, D., Zdziarski, A. A., & Chernyakova, M. 2013, MNRAS, 434, 2380
  45. Malzac, J., Lubiński, P., Zdziarski, A. A., et al. 2008, A&A, 492, 527
  46. Markoff, S., Nowak, M., Corbel, S., Fender, R., & Falcke, H. 2003, A&A, 397, 645
  47. Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, The Astrophysical Journal, 461, 396
  48. Munar-Adrover, P., Sabatini, S., Piano, G., et al. 2016, ArXiv e-prints [\eprint[arXiv]1607.03006]
  49. Negoro, H., Kawai, N., Kawasaki, Y. U. K., et al. 2010, The Astronomer’s Telegram, 2711
  50. Nice, D. J., Altiere, E., Bogdanov, S., et al. 2013, The Astrophysical Journal, 772, 50
  51. Orosz, J. A., McClintock, J. E., Aufdenberg, J. P., et al. 2011, The Astrophysical Journal, 742, 84
  52. Pepe, C., Vila, G. S., & Romero, G. E. 2015, A&A, 584, A95
  53. Perucho, M. & Bosch-Ramon, V. 2008, A&A, 482, 917
  54. Piano, G., Tavani, M., Vittorini, V., et al. 2012, A&A, 545, A110
  55. Pooley, G. G., Fender, R. P., & Brocksopp, C. 1999, MNRAS, 302, L1
  56. Poutanen, J. & Coppi, P. S. 1998, Physica Scripta Volume T, 77, 57
  57. Reid, M. J., McClintock, J. E., Narayan, R., et al. 2011, The Astrophysical Journal, 742, 83
  58. Rico, J. 2008, The Astrophysical Journal, Letters, 683, L55
  59. Rodriguez, J., Grinberg, V., Laurent, P., et al. 2015, The Astrophysical Journal, 807, 17
  60. Romero, G. E., Torres, D. F., Kaufman Bernadó, M. M., & Mirabel, I. F. 2003, A&A, 410, L1
  61. Romero, G. E., Vieyro, F. L., & Chaty, S. 2014, A&A, 562, L7
  62. Rushton, A., Dhawan, V., Fender, R., et al. 2010, The Astronomer’s Telegram, 2714
  63. Rushton, A., Miller-Jones, J. C. A., Campana, R., et al. 2012, MNRAS, 419, 3194
  64. Russell, D. M., Fender, R. P., Gallo, E., & Kaiser, C. R. 2007, MNRAS, 376, 1341
  65. Sabatini, S., Tavani, M., Coppi, P., et al. 2013, The Astrophysical Journal, 766, 83
  66. Sabatini, S., Tavani, M., Striani, E., et al. 2010, The Astrophysical Journal, Letters, 712, L10
  67. Shapiro, S. L., Lightman, A. P., & Eardley, D. M. 1976, The Astrophysical Journal, 204, 187
  68. Stirling, A. M., Spencer, R. E., de la Force, C. J., et al. 2001, MNRAS, 327, 1273
  69. Sunyaev, R. A. & Truemper, J. 1979, Nature, 279, 506
  70. Tavani, M., Bulgarelli, A., Piano, G., et al. 2009, Nature, 462, 620
  71. van Leeuwen, F. 2007, A&A, 474, 653
  72. Walborn, N. R. 1973, The Astrophysical Journal, Letters, 179, L123
  73. Wilms, J., Nowak, M. A., Pottschmidt, K., Pooley, G. G., & Fritz, S. 2006, A&A, 447, 245
  74. Wilson-Hodge, C. A. & Case, G. 2010, The Astronomer’s Telegram, 2721
  75. Xiang, J., Lee, J. C., Nowak, M. A., & Wilms, J. 2011, The Astrophysical Journal, 738, 78
  76. Yoon, D., Zdziarski, A. A., & Heinz, S. 2016, MNRAS, 456, 3638
  77. Zdziarski, A. A. & Gierliński, M. 2004, Progress of Theoretical Physics Supplement, 155, 99
  78. Zdziarski, A. A., Lubiński, P., & Sikora, M. 2012, MNRAS, 423, 663
  79. Zdziarski, A. A., Malyshev, D., Chernyakova, M., & Pooley, G. G. 2016a, ArXiv e-prints [\eprint[arXiv]1607.05059]
  80. Zdziarski, A. A., Malyshev, D., de Oña Wilhelmi, E., et al. 2016b, MNRAS, 455, 1451
  81. Zdziarski, A. A. & Pjanka, P. 2013, MNRAS, 436, 2950
  82. Zdziarski, A. A., Pooley, G. G., & Skinner, G. K. 2011, MNRAS, 412, 1985
  83. Ziółkowski, J. 2014, MNRAS, 440, L61
100704
This is a comment super asjknd jkasnjk adsnkj
Upvote
Downvote
Edit
-  
Unpublish
""
The feedback cannot be empty
Submit
Cancel
Comments 0
""
The feedback cannot be empty
   
Add comment
Cancel

You’re adding your first comment!
How to quickly get a good reply:
  • Offer a constructive comment on the author work.
  • Add helpful links to code implementation or project page.