SUZAKU OBSERVATION OF THE BLACK HOLE BINARY 4U 1630–47 IN THE VERY HIGH STATE

Because 4U 1630–47 suffers from heavy absorption (N_H ∼ 8 × 10²² cm⁻²) by the interstellar medium, we have to take into account the effects by dust scattering, which become more significant at lower energies. A part of the direct emission from the object is scattered out by interstellar dust, while photons emitted toward slightly different angles are scattered in the line of sight, thus making the dust-scattering halo around the point source. When one cannot integrate all the emission including both direct and halo components, the scattered-in and scattered-out photons do not cancel each other. The situation becomes more complex for a variable source because there is a time delay between the scattered-in and direct photons due to their different path lengths. Hence, we have to take into account these effects in the spectral analysis of a heavily absorbed object unless the spectrum is constant and the integrated image region is sufficiently large to fully cover the dust-scattering halo.

Following the work by Ueda et al. (2010), who applied the same procedure to Suzaku data of GRS 1915+105, we first estimate the fraction of the dust-scattering halo of 4U 1630-47 contained in our XIS spectra. According to Smith et al. (2002), who studied the detailed image profile of the dust-scattering halo of the bright Galactic source GX 13+1 observed with Chandra, the surface density of the halo profile can be approximately represented by a power law with r^−1.7 below r = 25 and r^−2.8 above r = 25. We then run the FTOOL xissimarfgen by assuming this profile and find that about 65% of the total photons of the dust-scattering halo are included in the VHS spectra within our integration region. In the spectral fit, we utilize the Dscat model, a local model on XSPEC developed by Ueda et al. (2010). It calculates the fractions of the scattered-in and scattered-out components as a function of energy, by referring to the cross section of the dust scattering given by Draine (2003).¹⁰ The free parameters of the Dscat model are the hydrogen column density (N_H) and the fraction of the halo (scattered-in) component contained in the XIS spectra (hereafter "scattering fraction"), which is fixed at 0.65. This is not applied to the spectra of the HXD, which has a sufficiently large field of view. For simplicity, we neglect the effect of time variability. The maximum time delay of the scattered-in component to the direct one is estimated to be ≈1.5 days at the outer integration radius of 18 by assuming that the scatterers are located at the half distance of D = 10 kpc. The MAXI light curve (Figure 1) shows that the soft X-ray flux was constant within 10% for 1.5 days before the Suzaku observation. We confirm that our conclusions are not affected (see Section 6.2).

In the spectral fit presented below, we add to an intrinsic model both the wabs and Dscat models, which take into account the interstellar photoelectric absorption (Morrison & McCammon 1983) and dust-scattering effects, respectively. (The hydrogen column density of the Dscat model is linked to that of the wabs model.) The estimated contribution of the halo component in the XIS0 spectrum is plotted by the magenta curve in the upper limit of Figure 5. We find that the net effects by dust scattering are ≈83% at 2 keV and ≈66% at 3 keV compared with the case when neglected. Thus, consideration of the dust-scattering effects is critical when we discuss the intrinsic source spectrum at low energies.

To begin with, we try simple phenomenological models to fit the XIS+HXD spectra in the VHS. We find that a single power-law model gives a very poor fit (χ²/dof = 4386/1032), with the best-fit photon index of Γ ≃ 3.2. The model leaves systematic hard X-ray excess above ≈20 keV that monotonically increases with energy and soft excess below ≈2 keV. The residuals indicate that the overall spectral shape is approximated by a power law flatter than Γ ≃ 3.2, on which an additional soft component is required.

Next, we fit the spectra with an MCD plus a power-law model (diskbb + pow in the XSPEC terminology, hereafter referred to as "Model A"). This is a conventional model often adopted to fit the spectra of BHBs in the HSS when a power-law component is rather weak. We find that this model well reproduces the Suzaku spectra in the VHS with χ²/dof = 1266/1030, even though the power-law component dominates the total flux at all energies from 1.2 keV to 200 keV, unlike the case of the HSS. We obtain a power-law photon index of $\Gamma =2.79_{-0.02}^{+0.01}$ and innermost disk temperature and radius of $T_{\rm in} = 1.41_{-0.01}^{+0.02}$ keV and r_in = (24.8 ± 0.5)ζ₇₀d₁₀ km from the MCD component, where ζ₇₀ = (cos i/cos 70°)^−1/2 and d₁₀ = D/(10 kpc). The model description and best-fit parameters are given in Tables 1 and 2, respectively. Figure 5(a) shows the residuals between the data and model in units of χ. The best-fit unabsorbed spectrum of Model A is shown in Figure 6(a), where different components are separately plotted. The absorbed 1.2–200 keV flux is estimated to be 1.7 × 10⁻⁸ erg  cm⁻² s⁻¹, which corresponds to the absorbed luminosity of 2.0 × 10³⁸ d₁₀ erg  s⁻¹ assuming isotropic emission. This high luminosity and the dominant steep power-law component confirm that the object was in the VHS.

Zoom In Zoom Out Reset image size

Table 2. Fitting Results

Component	Parameter	Model A	Model B	Model C	Model D
wabs	NH(1022 cm−2)	9.38 ± 0.06	$8.32_{-0.06}^{+0.07}$	8.46 ± 0.07	8.39 ± 0.07
diskbb	kTin(keV)	$1.41_{-0.01}^{+0.02}$	$1.16_{-0.01}^{+0.02}$	1.12 ± 0.02	⋅⋅⋅
	rin(ζ70d10 km)a	24.8 ± 0.5	$48.2_{-1.1}^{+1.2}$	$42.2_{-1.4}^{+1.2}$	⋅⋅⋅
nthcomp	$kT_{\rm in}^{\rm int}({\rm keV})$	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	$1.30_{-0.02}^{+0.04}$
or	Γth	⋅⋅⋅	⋅⋅⋅	$3.01_{-0.04}^{+0.05}$	$2.89_{-0.04}^{+0.05}$
dkbbfth	kTe(keV)	⋅⋅⋅	⋅⋅⋅	300−227	$53_{-13}^{+10}$
	Norm	⋅⋅⋅	⋅⋅⋅	$1.6_{-0.3}^{+0.2}$	$0.073_{-0.014}^{+0.008}$
	rtran(rin)	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	$5.5_{-0.9}^{+0.8}$
	Derived τb	⋅⋅⋅	⋅⋅⋅	$0.08_{-0.01}^{+0.22}$	$0.41_{-0.07}^{+0.12}$
PL	ΓPL	$2.79_{-0.02}^{+0.01}$	$2.83_{-0.03}^{+0.02}$	(2.1)c	(2.1)c
or	Norm	21.2 ± 0.9	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
simpl	fPL	⋅⋅⋅	$0.29_{-0.05}^{+0.02}$	$0.037_{-0.07}^{+0.09}$	$0.024_{-0.015}^{+0.009}$
rfxconvd	Ω/2π	⋅⋅⋅	$0.66_{-0.07}^{+0.25}$	$0.87_{-0.19}^{+0.59}$	$1.16_{-0.33}^{+1.62}$
	log(ξ)	⋅⋅⋅	$3.30_{-0.07}^{+0.06}$	$3.70_{-0.22}^{+0.15}$	$3.30_{-0.12}^{+0.05}$
	χ2/dof	1266/1030	1231/1028	1202/1026	1189/1026
Photon fluxe	$F_{{\rm disk} + {\rm th}}^{\rm photon}$	⋅⋅⋅	⋅⋅⋅	21.1	21.0
Inner radiusf	$r_{\rm in}^{*}(\zeta _{70}d_{10}\, {\rm km})$	⋅⋅⋅	⋅⋅⋅	$51.5_{-1.3}^{+1.6}$	$41.0_{-1.7}^{+0.7}$

Notes. Errors represent 90% confidence limits (statistical errors only). Energy range is extended to 0.01–1000 keV. ^aCalculated using only diskbb normalization. ^bCalculated using Equation (1) in Tamura et al. (2012). ^cFixed in the spectral fitting. ^dThe solar abundances are assumed for all heavy elements. The inclination angle is fixed at 70°. The reflected component is relativistically blurred by rdblur with fixed $R_{\rm in}^{\rm rdblur} = 6 R_{\rm G}$, $R_{\rm out}^{\rm rdblur} = 10^5 R_{\rm G}$, and β = −3. ^eUnabsorbed photon flux of the sum of the disk and thermal Compton component in the range of 0.01–100 keV after excluding the reflected component. ^fThe apparent innermost disk radius is calculated by using Equation (A.1) in Kubota & Makishima (2004) for the slab geometry.

Download table as:  ASCIITypeset image

Hereafter we apply more physically motivated models, where the power-law component is modeled as Comptonization of seed photons emitted from the disk. Essentially the same models are employed to fit the Suzaku spectrum of the BHB GX 339–4 in the VHS by Tamura et al. (2012). First, we utilize the simpl model (Steiner et al. 2009) multiplied to the diskbb component. The simpl model is an empirical model in which a fraction of the photons in an input seed spectrum are Compton-scattered into a power-law component. We find that this model alone does not reproduce the XIS+HXD spectra, yielding χ²/dof = 1386/1030.

To improve the fit, we further consider the reflection of the Comptonized component from the accretion disk by using the rfxconv model (Kolehmainen et al. 2011). This is a convolution model that combines the reflected spectra of Ross & Fabian (2005) from a constant density ionized disk with the ireflect model (a convolution version of the pexriv model; Magdziarz & Zdziarski 1995). To consider general relativistic smearing effect, we utilize the model rdblur (Fabian et al. 1989). Following Tamura et al. (2012), we fix the iron abundance to be solar, the inner and outer radius R_in = 6 R_g, R_out = 10⁵ R_g, and the power-law index of emissivity β = −3, which are very difficult to determine from the fit. We confirm that other choices of the parameters between R_in = 6 R_g–10² R_g or R_out = 10³ R_g–10⁵ R_g give almost the same results within the errors.

We refer this model including the disk reflection component as "Model B," whose description in the XSPEC terminology is given in Table 1. We find that Model B gives a much better fit with χ²/dof = 1231/1028 than that without a reflection component. The best-fit parameters are summarized in Table 2. The fitting residuals and unabsorbed best-fit model are plotted in Figures 5(b) and 6(b), respectively.

The simpl model used in Model B assumes no high energy cutoff, and hence should be regarded to represent Comptonized emission by nonthermal electrons. However, previous studies of BHBs in the VHS suggest that the Comptonizing corona contains both thermal and nonthermal electrons (e.g., Gierliński et al. 1999; Kubota et al. 2001; Gierliński & Done 2003). It is not clear whether it consists of two separate components or is a hybrid plasma whose energy distribution is Maxwellian at low energies but a power law at high energies. For simplicity, here we consider the former case and leave the detailed modeling of Comptonization by a hybrid plasma for future work. We note, however, that while a hybrid electron distribution can fit the spectrum, such a homogeneous source model cannot explain the result by Axelsson et al. (2014), who show that the spectrum of the fastest variability is different to that of the time-averaged spectrum. Accordingly, we add a thermal Comptonization model (Zdziarski et al. 1996; Życki et al. 1999) (Model C in Table 1) to Model B. As in Model B, we include the disk reflection component of the Comptonized emission (both thermal and nonthermal ones).

The nthcomp model has four parameters, the seed photon temperature, which we tie at the disk temperature of diskbb, the electron temperature, the photon index Γ_th, and its normalization. We set the upper limit of the electron temperature to be 300 keV, as the nthcomp model does not include full relativistic corrections. Since we cannot constrain both Γ_th in nthcomp and Γ_pl in simpl, we fix Γ_pl = 2.1, a typical value for a nonthermal power-law component seen in the HSS (Gierliński et al. 1999). We note that the choice of Γ_pl is not sensitive to the overall fitting result as our data is available up to only 200 keV.

In the nthcomp model, the relation among the optical depth of corona, the electron temperature, and the photon index is given by Equation (1) of Tamura et al. (2012)

$$\begin{equation} \tau = \frac{1}{2} \cdot \left[ \sqrt{\frac{9}{4} + \frac{3}{\Theta _{\rm e} \cdot \big(\big(\Gamma _{\rm th} + \frac{1}{2} \big)^2 - \frac{9}{4} \big) }} - \frac{3}{2} \right], \end{equation} \tag{ 1 }$$

where Θ_e = kT_e/m_ec². It is derived from Equation (A1) of Zdziarski et al. (1996) by assuming a slab geometry. We cannot directly estimate the innermost disk radius using the diskbb normalization alone, since the nthcomp model does not conserve the photon number unlike simpl. We therefore use Equation (A1) of Kubota & Makishima (2004),

$$\begin{eqnarray} F_{\rm dkbbfth}^{\rm photon} &=& 0.0165\left[ \frac{r_{\rm in}^2\,{\rm cos}\,i}{\left(D/10 \ {\rm kpc} \right)^2} \right] \left(\frac{T_{\rm in}}{1 \ {\rm keV}} \right)^3 \nonumber \\ && {\rm photons \ s^{-1} \ cm^{-2}}, \end{eqnarray} \tag{ 2 }$$

to derive the true innermost radius by considering the photon-number conservation. Note that we replace "$F_{\rm disk}^{\rm p} + F_{\rm thc}^{\rm p}2\,{\rm cos}\,i$" in Equation (A1) of Kubota & Makishima (2004) with $F_{\rm dkbbfth}^{\rm photon}$ to assume the slab corona geometry.

We find that Model C gives a better fit than Model B, yielding χ²/dof = 1202/1026. The improvement from Model B is significant at 99.9995% level from an F test, i.e., the data require the presence of another Comptonized component at high confidence, although the electron temperature of the nthcomp component is not well constrained, with a lower limit of 73 keV. The best-fit parameters are summarized in Table 2. Figure 5(c) plots the residuals, while Figure 6(c) shows the unabsorbed best-fit model spectrum. It is seen that the thermal Compton component dominates below ∼50 keV.

In reality, it is expected that the hot Comptonizing corona surrounds only an inner part of the disk. In this picture, Model C is unphysical because it does not take into account any energetic coupling between the disk and corona; the summed emissivity of the disk blackbody and thermal Comptonization components in an inner region is larger than the extrapolation from the outer disk that is not Comptonized (see Done & Kubota 2006; Tamura et al. 2012). This would lead an incorrect estimate of the innermost disk radius. To overcome this situation, we replace diskbb and nthcomp to dkbbfth (Done & Kubota 2006) (Model D in Table 1). The dkbbfth model assumes that the energy released by gravity is dissipated locally, which is emitted purely from the standard disk at r > r_tran but is split into the disk and corona at r_in < r < r_tran. Thus, the inner part of the disk covered by the corona is less luminous than it would have been if the corona did not exist. The dkbbfth model has five parameters, the disk temperature, the outer radius of the corona r_tran, the photon index of the Comptonized component, the temperature of the corona, and the normalization. We set all these parameters to be free in the fitting. Since this model only takes into account thermal Comptonization, we further add the simpl model to describe nonthermal Comptonization. There we set the disk emission to be the seed photons of simpl and fix the photon index at 2.1. We note that the number of free parameters in Model D is the same as in Model C.

The fit shows a significant improvement (χ²/dof = 1189/1026) from Model C. Thus, we conclude that Model D gives the best description of the Suzaku spectra among the three physically motivated models examined above. The best-fit parameters are summarized in Table 2 and the fitting residuals are plotted Figure 5(d). The best-fit unabsorbed model is shown in Figure 6(d). The dash–dotted curve shows the intrinsic disk emission that would be observed when all the luminosity of the dkbbfth component were emitted as the standard disk.

We performed near-infrared photometric observations of 4U 1630–47 in the J (1.25 μm), H (1.63 μm), and K_s (2.14 μm) bands on 2012 October 1 (one day before the Suzaku observation) by using the SIRIUS camera (Nagayama et al. 2003) on the 1.4 m IRSF telescope at the South African Astronomical Observatory. We obtained 25 object frames with an exposure of 15 s per frame, and thus the net exposure time was 375 s. The seeing in FWHM in the J band was ∼15 (3.5 pixels).

We perform standard data reduction (i.e., dark subtraction, flat-fielding, sky subtraction, and combining dithered images) with the IRSF pipeline software on IRAF (the Image Reduction and Analysis Facility, distributed by the National Optical Astronomy Observatory), version 2.16. We combine all the object frames obtained in one night to maximize signal-to-noise ratio.

The near-infrared counterpart has been identified by Augusteijn et al. (2001), but we are not able to measure the flux by aperture photometry since 4U 1630–47 is contaminated by nearby stars in the infrared images. Thus, we perform point-spread functionphotometry using the DAOPHOT (Stetson 1987) package in IRAF.

We obtain the near-infrared magnitudes of 17.9 ± 0.4 mag in the H band and 16.0 ± 0.2 mag in the K_s band, while we are unable to constrain the J-band flux because of the severe source confusion. These magnitudes are almost the same as those reported in the 1998 outburst (Augusteijn et al. 2001). We estimate the extinction in each band as A_H = 7.8 ± 1.3 mag and A_K = 5.0 ± 0.9 mag by using the relation between the hydrogen column density and optical extinction A_V, N_H = (1.79 ± 0.03) × A_V[mag] × 10²¹[cm⁻²] (Predehl & Schmitt 1995), and the conversion factor from A_V into A_H or A_K given by Rieke & Lebofsky (1985). The extinction-corrected IRSF fluxes ($m_H^{\rm cor} = 10.1$ mag, $m_K^{\rm cor} = 11.0$ mag) are plotted as the red points in Figure 7. Note that the extremely blue color ($m_H^{\rm cor} - m_K^{\rm cor} = -0.9$ mag), which is hard to explain by any stellar or continuum source, most probably comes from the errors in the photometry, N_H, and/or the N_H-to-A_V conversion. We find that the last one can produce the largest uncertainty. Figure 3 of Predehl & Schmitt (1995) shows that there is a deviation by a factor of ∼two in the relation between N_H and A_V for some objects. Accordingly, we consider an extreme case where the A_V-to-N_H ratio is twice smaller than the above as a lower limit of the extinction correction. This gives the corrected fluxes of $m_H^{\rm cor} = 13.8$ mag and $m_K^{\rm cor} = 13.4$ mag, which are plotted as the blue points in Figure 7. Thus, the extinction-corrected magnitudes are estimated to be in the range of $13.8\ {\rm mag} > m_H^{\rm cor} > 10.1\ {\rm mag}$ and $13.4\ {\rm mag} > m_K^{\rm cor} > 11.0\ {\rm mag}$.

Zoom In Zoom Out Reset image size

In Figure 7, the intrinsic thermal emission from the standard disk based on Model D (Section 3.5) that would be observed without Comptonization (i.e., the red dashed line in Figure 6(d)) is plotted by the green solid curve. As noticed, even the lower limits of the observed near-infrared fluxes are larger than the expected fluxes from the MCD component. This suggests that the fluxes are dominated by other components. We estimate the absolute magnitudes to be $-1.2\ {\rm mag} > M_H^{\rm cor} > -5.3$ mag and $-1.6\ {\rm mag} > M_K^{\rm cor} > -4.2$ mag for a distance of 10 kpc. Thus, if the near-infrared fluxes completely come from the companion star, it would be an early-type star (O7–B2), supporting the suggestion by Augusteijn et al. (2001), or a K-type giant star.

These magnitudes should be regarded as the upper limit of the true luminosity of the companion star, however, as we also expect other origins that could be responsible for the near-infrared fluxes, including (1) the irradiation of the outer disk by the X-ray emission, (2) synchrotron radiation from the jets, and/or (3) cyclo-synchrotron emission from the hot flow (Veledina et al. 2013). Unfortunately, the large uncertainties in the extinction-corrected fluxes make it very difficult to discriminate these possibilities. We note that the observed (uncorrected) near-infrared fluxes in the VHS are almost the same as those reported by Augusteijn et al. (2001) when the source was in the HSS with a similar X-ray flux level. Since jet ejection is believed to be quenched in the HSS, this comparison suggests that the synchrotron emission from the jets, if any, is not a dominant source of the near-infrared fluxes in the VHS.

Using Suzaku, we have obtained so far the best-quality broadband X-ray spectra of 4U 1630–47 in the VHS, covering the 1–200 keV band with good energy resolution by X-ray CCD below 10 keV. The observation epoch corresponds to when the hard X-ray flux above 15 keV was peaked during the 2012 September–October outburst. In the spectral analysis, we carefully take into account the effects by dust scattering.

The continuum is found to be well approximated by a phenomenological model consisting of an MCD and a power-law component dominating the total flux in the entire 1.2–200 keV band (Model A). However, we regard that this modeling is unnatural because (1) no low-energy turnover is included, which should exist if the power-law component is originated by Comptonization of seed photons emitted from the disk, and (2) the derived column density is significantly larger than that obtained in the HSS. We thus consider three physically motivated models where the disk photons are Compton-scattered by a hot corona surrounding it. Among them, Model D, where the Comptonizing corona is energetically coupled with the underlying standard disk below a transition radius, best reproduces the spectra. A significant reflection component from the inner disk with a solid angle of Ω/2π ≈ 1.2 is detected. This is consistent with the recent result by King et al. (2014), although the state was different from ours. Below, we discuss the inner disk geometry on the basis of the result of Model D. No significant emission or absorption lines are detected in our spectra. This fact gives tight constraints on the presence of the baryonic jets discovered by Díaz Trigo et al. (2013) four days before the Suzaku observation and that of the disk wind seen in the HSS (Kubota et al. 2007; King et al. 2014).

We find that the standard disk in the VHS is most likely truncated before reaching the ISCO. From Model D, the best physically self-consistent model considered here, we derive the innermost radius of the standard disk as $r_{\rm in} = 41.0_{-1.7}^{+0.7} \zeta _{70} d_{10}$ km in the VHS. This is significantly larger than that found in the HSS, r_in = (35.0 ± 0.3)ζ₇₀d₁₀ km, which is found to be constant during this state and hence most probably corresponds to the ISCO (Kubota et al. 2007). Since we directly compare the results obtained with the same detector (XIS0), the comparison is free from any cross-normalization uncertainties between different instruments. Also, a common model is employed to represent the interstellar absorption and dust scattering, making the relative systematic error least. When we assume an extreme case in which the soft X-ray flux producing the dust-scattering halo is 10% lower than the Suzaku epoch (see Section 3.1), we obtain an even larger innermost radius. Thus, the conclusion of disk truncation is robust against uncertainty in the correction for dust scattering.

A truncated disk in the VHS is also suggested from the BHB GX 339–4 by applying essentially the same model as our Model D (Tamura et al. 2012). Hence, disk truncation would be a common feature of the VHS, although the innermost radius is less than twice that of the ISCO in both objects. We note that the "true" disk radius, R_in, is estimated as R_in = κξ²r_in, where κ is the color-to-effective temperature ratio and ξ is a correction factor for the stress-free boundary condition at the ISCO (Kubota et al. 1998). To explain the r_in value in the VHS by the ISCO, κ must be smaller than that in the HSS. As discussed in Tamura et al. (2012), this case is quite unlikely under strong irradiation by the corona in the VHS.

Relativistic baryonic jets of 4U 1630–47 were detected with XMM-Newton on 2012 September 28 (Díaz Trigo et al. 2013) when the source was also in the VHS. The XMM-Newton spectrum shows Doppler-shifted Fe xxvi lines at 7.3 keV and 4 keV. Assuming that they are twin jets of the same physical parameters emitted to opposite directions, they estimate the jet speed, the inclination angle, and the cone angle to be 0.3–0.4c, 58°–67°, and 37–45, respectively. The parameters are quite similar to those of SS 433 (Namiki et al. 2003), the unique Galactic microquasar exhibiting steady relativistic jets with a velocity of 0.26c, which most likely contains a black hole (Kubota et al. 2010).

In contrast, our Suzaku spectra, taken only four days after the XMM-Newton observation, do not show such emission-line features. The equivalent width of the 7.3 keV emission line detected by Díaz Trigo et al. (2013) is constrained to be <3 eV (90% confidence limit), which is about 1/12 of that in the XMM-Newton spectra. This suggests that the relativistic jets detected by Díaz Trigo et al. (2013) had vanished in four days. In the case of SS 433, the length of the X-ray-emitting region of the jets is ∼10¹³ cm, corresponding to a traveling time of only ∼10³ s (Kotani et al. 1996). Thus, if we assume conditions similar to those in SS 433, the jets of 4U 1630–47 would vanish in an hour after the ejection stops. We cannot, however, exclude the possibility that a jet might still exist, such as compact jets often seen in the LHS (Fender et al. 2004).

The variation of the X-ray flux can provide clues to reveal the condition for the ejection of baryonic jets from a black hole. According to the MAXI and Swift/BAT light curves (Figure 1), the Suzaku observation was performed just after the hard X-ray flux showed a sudden increase from MJD≃56,201. The same trend is confirmed in the soft band, though less clear because of the data gap of MAXI. The source was by ∼30% fainter when the jets were detected with XMM-Newton (MJD = 56,198) than in the Suzaku epoch (MJD = 56,202). Thus, phenomenologically, it would be possible that the launch of the baryonic jets somehow ceased as more accretion power was dissipated into radiation. However, we estimate the kinetic power of the jets to be >10⁴⁰ erg s⁻¹ on the basis of the jet model of SS 433 (Kotani et al. 1996), assuming that it is proportional to the luminosity of the ionized iron-K emission lines. This huge power cannot be accounted for by the observed increase in the luminosity at MJD≃56,201 (ΔL_X ∼ 10³⁸ erg s⁻¹) unless the radiation efficiency is very low. Thus, a significant decrease in the intrinsic mass accretion rate would be required. To obtain firm observational clues on the jet formation mechanism, we need more frequent monitoring observations of BHBs in high-luminosity states with high-quality X-ray spectroscopy.

The Suzaku spectra of 4U 1630–47 in the VHS do not show any significant absorption-line features of highly ionized iron ions that were detected in the HSS by Kubota et al. (2007). The upper limits for the equivalent widths of the absorption lines are 1.3 eV and 1.4 eV (90% confidence limits) at 6.72 keV and 6.99 keV, corresponding to those of Fe xxv and Fe xxvi observed in the HSS, respectively. There are two possibilities to account for this fact: (1) there is actually no disk wind in the VHS, or (2) it exists but simply becomes invisible by being almost fully ionized by the strong X-ray irradiation.

To examine the latter possibility, we perform simulations of a photoionized gas using XSTAR version 2.2.1bk. Following Kubota et al. (2007), we first reproduce the physical parameters (the ionization stage and column density) of the disk wind detected in the HSS by assuming the same spectrum and luminosity as observed. Then, we change the X-ray spectrum and luminosity to those in the VHS by keeping the other parameters the same. We find that the ion fractions of Fe xxv and Fe xxvi become ≈1/20 and ≈1/4 of those in the HSS, respectively. The predicted equivalent width of Fe xxvi is larger than the observed upper limit, suggesting that photoionization effects alone cannot explain the absence of the absorption lines in the Suzaku spectra. However, it is still plausible that a disk wind with a smaller density (and/or at a smaller distance) than that in the HSS is present in the VHS (see Chakravorty et al. 2013 for discussion of thermodynamical stability, and Begelman et al. 1983 for discussion of launch radius and density). Further observations of BHBs with good energy resolution covering different states and luminosities are necessary to establish the critical conditions to launch a disk wind.

We observed 4U 1630–47 in the VHS with Suzaku and IRSF during the 2012 September–October outburst.

The conclusions are summarized as follows.

• 1.  
  The time-averaged Suzaku spectra are well described by a physically self-consistent model consisting of thermal and nonthermal Comptonization of the inner disk emission with energetic coupling between the disk and corona. A strong relativistic reflection component from the accretion disk is required. We properly take into account the effects of dust scattering.
• 2.  
  Comparing the result of the Suzaku spectra in the HSS observed in 2006 February, we find evidence that the accretion disk is slightly truncated before reaching the ISCO in the VHS.
• 3.  
  Our spectra do not show any Doppler-shifted line emissions from the relativistic jets that were detected four days before our observation with XMM-Newton. This suggests that the jets were a transient event and vanished between these observations.
• 4.  
  We do not detect absorption-line features from highly ionized iron ions that were previously present in the high/soft state. The upper limits on the equivalent width of these lines are not compatible with over-ionization by the higher luminosity irradiation of the same wind. If a wind still remains, it has changed in launch radius and/or density from that seen in the HSS.
• 5.  
  From near-infrared data taken with IRSF one day before the Suzaku observation, we obtain 17.9 ± 0.4 mag in the H band and 16.0 ± 0.2 mag in the K_s band for the counterpart. Since the extinction-corrected fluxes are larger than the extrapolated fluxes from the MCD component, contribution from a companion star, which would be an early-type star or a K-type giant star if it dominates the total fluxes, and/or other origins such as irradiation of the outer disk by the X-ray emission are required.