Extensive Broadband X-Ray Monitoring During the Formation of a Giant Radio Jet Base in Cyg X-3 with AstroSat

Establishing the true nature of the connection between the accretion disk and radio jets in X-ray binaries at different mass accretion rate has been considered one of the challenging problems in astrophysics and several attempts have been made so far in this direction (Fender et al. 2004, 2009; Dinçer et al. 2014). Such attempts successfully establish the long-term radio/X-ray correlations from weeks to years timescales in both transient and persistent/semi-persistent sources like GX 339-4 (Corbel et al. 2013), Cyg X-1 (Stirling et al. 2001), GRS 1915+105 (Muno et al. 2001; Fender & Belloni 2004), and Cyg X-3 (Zdziarski et al. 2016). However, there are a handful of attempts where inner disk activity was closely monitored before, during, and after the superluminal radio ejection from X-ray transients. Miller-Jones et al. (2012) made a remarkable attempt during the 2009 outburst of the Galactic micro-quasar H 1743-322. They monitored the formation and ejection of superluminal radio jets using VLBA at 8.4 GHz between 2009 May 28 and June 6 almost on a daily basis simultaneously with the RXTE/PCA observations to monitor accretion flow properties. They found that radio emission was highly quenched before the major radio flare and the ejection event occurred during the state transition from hard-intermediate to soft state, which is marked by the disappearance of type-C quasi-periodic oscillations (QPOs) from the power density spectra (Casella et al. 2005). Until now, the changes in X-ray spectro-temporal properties during the formation of a superluminal jet have been monitored with the frequency of days to weeks, except GRS 1915+105, which is only source that has radio and sensitive X-ray data on a scale of days (Yadav 2006). However, depending upon the size of the radio jet base, it is possible that the decoupling may occur in the first few minutes to hours and may leave a signature in X-ray spectro-timing properties. Unfortunately, no studies exist where the change in X-ray properties of the inner accretion flow are traced with the accuracy of a timescale of minutes. One key reason for the lack of such studies is that superluminal radio ejections are rare, transient, and their time of occurrences are completely unpredictable within a timescale of days to weeks. Although a few studies with microquasars have shown that major radio flares occur during the transition from the hard-intermediate to the soft spectral state (Fender et al. 2009), the occurrence of such a transition is highly uncertain within a timescale of weeks. Therefore, it is difficult to determine the time of radio jet formation more accurately than a few days.

Cyg X-3 is a Galactic, high-mass, wind-fed X-ray binary (Parsignault et al. 1972), and on many occasions it shows unique giant radio flares with the peak radio flux density up to 20 Jy (Leach et al. 1975; Molnar et al. 1988; Schalinski et al. 1995; Mioduszewski et al. 2001). Additionally, radio and X-ray variability have been observed from this source, and their correlation has been studied (Zdziarski et al. 2016). Based on hardness and X-ray and radio flux, five X-ray spectral states have been identified (Koljonen et al. 2010): hypersoft state (HPS), flaring soft X-ray state (FSXR), flaring intermediate state, flaring hard X-ray state, and quiescent state. QPOs of the order of mHz are detected from Cyg X-3 during the hard state (Manchanda & Rao 1993; Koljonen et al. 2011; Pahari et al. 2017). The latest giant radio flare, which reached nearly 20 Jy on 2017 April 4–5, was first predicted by Trushkin et al. (2017b). Monitoring Cyg X-3 using RATAN-600 radio telescope in three bands: 4.6, 8.2, and 11.2 GHz on a daily basis and comparing the Swift/BAT hard X-ray flux, they showed that before the giant flare, the source entered into the HPS where both hard X-ray and radio flux are highly quenched (Koljonen et al. 2017). On 2017 April 1–2, the radio flux started increasing rapidly, and on April 4–5 it reached the peak value of ∼20 Jy in all three radio bands. During this period, 15–50 keV Swift/BAT flux increased by a factor of ∼2–3. A similar transition from the ultrasoft to the hard X-ray state was also observed previously with INTEGRAL (Beckmann et al. 2007). Interestingly, AstroSat observed the source continuously on April 1–2 when the X-ray state transition and major radio flare formation were taking place.

In this work, we analyze the simultaneous observations of Cyg X-3 using the Soft X-ray Telescope (SXT) and Large Area X-ray Proportional Counter (LAXPC) instruments on board AstroSat. Based on light curve, hardness, and broadband energy spectra, we found that during the first 10–12 hr, the source was in the HPS. After ∼50 ks, an unusual, flat power-law component appeared in the spectrum of Cyg X-3, and we termed it as the very high state (VHS). We show that the flat power-law component is consistent with the synchrotron emission from the radio jet monitored simultaneously with the RATAN telescope. This observation provides an opportunity to closely monitor the X-ray properties during the formation of giant radio flare in Cyg X-3.

AstroSat continuously observed Cyg X-3 between 2017 April 1 14:47:24 and 2017 April 2 16:43:39, covering 14 consecutive satellite orbits. The total observation duration is 93.3 ks. For broadband spectroscopic purpose, we use simultaneous observations from the SXT and LAXPC instruments.

SXT is a focusing telescope with a CCD camera that can perform X-ray imaging and spectroscopy in the 0.3–8.0 keV energy range (Singh et al. 2014). Level 1 imaging mode data are processed through a pipeline software to produce level 2 event files and then a good time interval correcter and SXT event merger script⁹ is used to create merged event files from all the clean events with the corrected exposure time. We use XSELECT V2.4d in HEASOFT 6.22.1 to extract light curves and spectra using source regions between 1 and 13 arcmin. An off-axis auxiliary response file (ARF) appropriate for the source location on the CCD is generated from the provided on-axis ARF using sxtmkarf tool (see footnote 9). Blank-sky SXT observations are used to extract background spectrum.

LAXPC consists of three identical X-ray proportional counters with the absolute time resolution of 10 μs in the energy range 3.0–80.0 keV (Misra et al. 2016; Yadav et al. 2016a, 2016b; Agrawal et al. 2017; Antia et al. 2017). LAXPC data are analyzed using the LAXPC software.¹⁰ Details of the response and background spectra generation can be found in Antia et al. (2017).

In the top left panel of Figure 1, we show the Swift/BAT daily light curve in 15–50 keV where the AstroSat observation span is marked by two vertical lines. The light curve shows that the Swift/BAT count rate sharply rises by a factor of ∼2 within the AstroSat observation period. The bottom panel shows the 11.2 GHz radio light curve obtained from the RATAN-600 telescope (Trushkin et al. 2017a). The radio flux density was at a low, persistent level of ∼50–100 mJy until MJD 57845 and then it rises to ∼1 Jy within the AstroSat observation window. Therefore, the AstroSat observation during the onset of the giant radio flare simultaneous with the X-ray hardening is extremely valuable to monitor the disk/jet connections.

Zoom In Zoom Out Reset image size

The 3–80 keV background-subtracted LAXPC light curve and 0.3–8 keV SXT light curve are shown in the top right panel of Figure 1. We note that the peak count rate of the binary modulation of ∼4.8 hr observed from both LAXPC and SXT light curves increases significantly (∼50%) after ∼60 ks. To check the soft and hard X-ray behavior of the source, we plot the 3–10 keV, 17–60 keV background-subtracted light curves and their ratios in the bottom left panel of Figure 1. With the 5% background systematics, the source count rate in the 17–60 keV energy range is close to the non-detection level up to ∼60 ks and then increases significantly up to ∼150 counts s⁻¹. Additionally, the hard-to-soft ratio becomes significant after ∼60 ks. This implies the absence of a hard component in the spectra during the first 60 ks. This part of the observation belongs to the HPS. The hard component appears from 60 ks onward during VHS. We check the flux–flux plot in 3–6 keV and 15–60 keV using LAXPC10 observations shown in the bottom right panel of Figure 1. During the HPS, the 3–6 keV count rate strongly correlates with the 15–60 keV count rate with the Spearman rank correlation coefficient (SRCC) of 0.82 (two-tailed p-value is <0.0000004), while both light curves are not correlated during the VHS with the SRCC of 0.17 (two-tailed p-value is ∼0.0006). The non-correlation indicates that the origin of hard and soft X-ray emission during the VHS may be independent of each other.

To examine the source variability, we plot the soft color (the ratio of count rate between 6–10 keV and 3–6 keV) and the hard color (the ratio of count rate in 10–40 keV and 3–10 keV) as a function of time in the left panel of Figure 2. The modulations in the timescale of the binary orbital period is seen in both soft and hard colors during the HPS and VHS. We confirm this by phase folding the data over the binary orbital period. The hardness intensity diagram (HID) plotted in the right panel of Figure 2 shows that the second half of the observation (shown in red) is significantly harder and brighter than the first half (shown in black), which motivates naming it as "VHS." For comparison, we also plot the HID of the FSXR (in blue), which was observed by AstroSat on 2016 August 12.

Zoom In Zoom Out Reset image size

To confirm the HPS to VHS transition as indicated by the model-independent approach, we separately fit SXT+LAXPC joint spectra in the energy range 0.6–70.0 keV extracted during the first 40 ks (HPS) and last 40 ks (VHS). We use the diskbb model to fit the HPS emission (Koljonen et al. 2010). However, the disk is embedded in a plasma medium (Manchanda 2002; Zdziarski et al. 2010) accreted from the companion, and therefore the bremsstrahlung emission from the collision between the inflowing and outflowing plasma (Koljonen et al. 2013) has been taken care of by the bremss model. To account for the interstellar medium absorption and the local, partial absorption by the accreted wind and hot medium, we use tbabs and pcfabs models, respectively (Koljonen et al. 2010). Fitting with this continuum model, we found weak/no residuals above 20 keV in the HPS but a strong residual in the VHS that resembles a power-law-like emission. To account for this hard X-ray residual, we use the power-law model in XSpec. The Fe emission line and an absorption feature are modeled using a Gauss and a gabs model, respectively. Best-fit parameters along with 1σ errors are provided in Table 1. Best-fit spectra during the HPS and the VHS along with their model components and residuals are shown in the top left and top right panels of Figure 3. In the bottom left panel, we plot source+background and background spectra during the VHS in 10–60 keV to show that the flat power-law-like emission observed in the VHS is due to the source only. From Table 1, we note that during the VHS, the hard X-ray excess has a photon power-law index of ${1.43}_{-0.03}^{+0.06}$. Properties of such a flat power law were not reported from any spectral states in Cyg X-3 and are unlikely to be associated with the coronal emission. To make this point clear, we plot the energy spectra of the FSXR state (in gray) along with the best-fit HPS (shown by pluses) and VHS (shown by circles) in the bottom right panel of Figure 3. Comparing three spectra, we note that the FSXR spectrum has a very distinct and clear coronal emission component in the energy range 10–60 keV which is absent from the HPS and VHS. While the FSXR spectral fitting details will be discussed elsewhere, here it is used for comparison purposes. Since the time of giant radio flare formation coincides with the VHS, the most likely origin of the flat power-law component is the synchrotron emission from the jet base.

Zoom In Zoom Out Reset image size

Table 1.  Best-fit Parameters Using SXT and LAXPC Joint Spectral Fitting during the Hypersoft and the Very High State

Model	Parameters	Hypersoft	Very High
		State	State
tbabs	${N}_{{\rm{H}},\mathrm{ISM}}$ (1022 cm−2)	${3.17}_{-0.05}^{+0.04}$	${3.63}_{-0.03}^{+0.04}$
pcfabs	${N}_{{\rm{H}},\mathrm{loc}}$ (1022 cm−2)	${13.5}_{-0.6}^{+0.5}$	${14.1}_{-0.5}^{+0.4}$
	fsc	${0.65}_{-0.01}^{+0.02}$	${0.71}_{-0.02}^{+0.02}$
diskbb	kTin (keV)	${1.39}_{-0.02}^{+0.03}$	${1.11}_{-0.02}^{+0.03}$
	$\sqrt{{N}_{\mathrm{dbb}}}$	${16.4}_{-2.6}^{+3.1}$	${31.8}_{-4.4}^{+2.1}$
bremss	kTplasma (keV)	${4.58}_{-0.31}^{+0.42}$	${3.87}_{-0.16}^{+0.19}$
	Nbremss	${0.59}_{-0.18}^{+0.21}$	${3.62}_{-0.53}^{+0.62}$
powerlaw	Γ	⋯	${1.43}_{-0.03}^{+0.06}$
	Npowerlaw	⋯	${0.04}_{-0.01}^{+0.01}$
gauss	EGauss (keV)	${6.94}_{-0.03}^{+0.04}$	${7.08}_{-0.12}^{+0.04}$
	Weq (eV)	${96}_{-8}^{+11}$	${59}_{-3}^{+5}$
gabs	Egabs (keV)	${2.96}_{-0.01}^{+0.01}$	${2.95}_{-0.02}^{+0.01}$
	${\sigma }_{\mathrm{gabs}}$ (keV)	${0.07}_{-0.01}^{+0.01}$	${0.06}_{-0.01}^{+0.02}$
	Depthgabs (keV)	${0.04}_{-0.01}^{+0.01}$	${0.03}_{-0.01}^{+0.01}$
F${}_{0.3\mbox{--}3}$	(10−10 erg s−1 cm−2)	${8.68}_{-0.43}^{+0.22}$	${13.6}_{-0.62}^{+0.64}$
F3–6	(10−10 erg s−1 cm−2)	${53.4}_{-1.55}^{+1.29}$	${55.8}_{-1.43}^{+1.69}$
F6–10	(10−10 erg s−1 cm−2)	${15.8}_{-0.56}^{+1.13}$	${28.7}_{-1.15}^{+1.14}$
F10–20	(10−10 erg s−1 cm−2)	${2.32}_{-0.09}^{+0.13}$	${7.91}_{-1.17}^{+1.09}$
F20–80	(10−10 erg s−1 cm−2)	${0.35}_{-0.24}^{+0.12}$	${9.43}_{-2.32}^{+3.17}$
	${\chi }^{2}$/dof	981/708 (1.39)	1221/854 (1.43)

Note. ${N}_{{\rm{H}},\mathrm{ISM}}$ and ${N}_{{\rm{H}},\mathrm{loc}}$ are the line of sight, interstellar medium, and local absorption column density, respectively; f_sc is the scattering fraction; and kT_in and N_dbb are the inner disk temperature and diskbb normalization, respectively. Γ and N_powerlaw are the photon power-law index and the power-law normalization. kT_plasma and N_bremss are plasma temperature and the bremss normalization. E_Gauss and W_eq are the Fe emission line energy and the equivalent width, respectively. E_gabs and ${\sigma }_{\mathrm{gabs}}$ are the energy and width of the Gaussian absorption line in keV while Depth_gabs is the line depth. ${F}_{0.3-3}$, ${F}_{3-6}$, ${F}_{6-10}$, ${F}_{10-20}$, and ${F}_{20-80}$ are fluxes in the energy ranges 0.3–3, 3–6, 6–10, 10–20, and 20–80 keV, respectively.

Download table as:  ASCIITypeset image

3.1. Consistency Check on the Radio Flux Density from the X-Ray Spectra

If the synchrotron emission from the radio jet is the origin of the flat X-ray power law, then an extrapolation of this power law to the radio frequency should indicate the expected radio flux density. To check the consistency, we consider the LAXPC spectrum of the VHS in 25–70 keV and fit it with a power law. Such a choice of energy range ensures that no other model components contribute significantly in this band. The fitted spectrum and its residual are shown in the top and middle panels of Figure 4. The best-fit power-law index and power-law normalization are ${1.49}_{-0.03}^{+0.04}$ and ${0.065}_{-0.005}^{+0.006}$, respectively. Both parameters are consistent with those from the power-law model used during the broadband spectral fitting of VHS (see Table 1). When we extrapolate the fitted power-law model to the radio band, shown in the bottom panel of Figure 4, we find that the radio flux density predicted by the X-ray power-law model is in the range of ∼185–511 mJy at 11.2 GHz. The radio flux density observed by the RATAN-600 radio telescope at 11.2 GHz is 40 ± 2 mJy on MJD 57845.23 and 1650 ± 70 mJy on MJD 57846.21. The VHS spectrum analyzed here was observed between MJD 57845.34 and 57845.69. Therefore, our observations lie exactly between two radio observations, and the X-ray spectral model fit predicted radio flux density is highly consistent with the trend shown in the top left panel of Figure 1.

Zoom In Zoom Out Reset image size

In this work, we jointly analyze observations of Cyg X-3 using simultaneous SXT and LAXPC data on board AstroSat on 2017 April 1–2 when the formation of a giant radio flare was reported using the RATAN-600 radio telescope.

During the first half of the AstroSat observation (∼0–50 ks), the X-ray spectra are unusually soft, and the source is not detected significantly above 20 keV. This is consistent with the HPS that was previously observed a few times from Cyg X-3 (Koljonen et al. 2010). The hardness ratio during the HPS is lower when compared to the FSXR state with no significant counts above 20 keV. During the second half of the observation (∼60–93 ks), the 20–60 keV count rate increases by a factor of ∼5–6 and the hardness ratio also increases by ∼50%. This makes source spectral state different than that observed during the first half of the observation. Comparing with the FSXR, we show that the second half of AstroSat observation falls between the HPS and the FSXR and we termed this here as the VHS. Interestingly, if we qualitatively compare our HID with Figure 3 from Koljonen et al. (2010), we notice that the jet line that divides the HPS from FSXR in Koljonen et al. (2010) coincides with the position of the VHS as observed by the AstroSat. Therefore, using AstroSat observations, we classify VHS as an additional state that falls between the HPS and FSXR state in the HID.

An indication that the significant hard X-ray emission (>15 keV) during the VHS originates from a physical process different than the disk blackbody comes from a significantly low SRCC (0.17) between the 3–6 keV and 15–60 keV count rates, while the same during the HPS is very high (0.82). The spectral modeling of the hard X-ray excess during the VHS indicates a power-law index of ∼1.49. The extrapolated power-law model predicts the radio flux density in the range of ∼185–511 mJy, consistent with the trend observed from the RATAN-600 telescope at 11.2 GHz. Comparing with the contemporary radio flux we conclude that such a flat X-ray power-law emission is solely due to the synchrotron emission from the slowly moving jet base. The radio flare rise time is slower in Cyg X-3 (Trushkin et al. 2017b) in comparison to the rise time seen in GRS 1915+105 (Yadav 2006), which is consistent with the slower speed of superluminal radio jets in Cyg X-3 (Marti et al. 2000). The detection of hard X-ray emission from radio jets in GRS 1915+105 may be difficult since accretion disk emission is stronger in GRS 1915+105 and superluminal radio jets in GRS 1915+105 are weaker at least by a factor of over ∼20 (Yadav 2006).

From Table 1, we note that during the HPS to the VHS transition, the inner disk temperature decreases from ∼1.39 keV to ∼1.11 keV, while the inner disk radius (which is proportional to the square root of diskbb normalization) increases by a factor of ∼2. This implies the inner disk recedes during the formation of a major plasma blob that emits a radio flare. A significant increase in the contribution of the bremsstrahlung emission from ∼13% (during HPS) to ∼36% (during VHS) of the total flux in 0.1–100.0 keV along with appearance of the hard X-ray tail implies that the plasma blob, in the presence of a strong magnetic field, is capable of emitting synchrotron radiation in radio frequency, and we detect the high-energy part of such radiation as a flat power-law component in the X-ray spectra.

We thank the referee for constructive comments that improved the quality of the manuscript. We acknowledge the strong support from Indian Space Research Organization (ISRO) during instrument building, testing, software development, and mission operation. We also acknowledge the support from the LAXPC Payload Operation Center (POC) and SXT POC at TIFR, Mumbai.