Hard-state Accretion Disk Winds from Black Holes: The Revealing Case of MAXI J1820+070

Disk winds are observed in accreting black holes (BHs) across the full range of masses (Fabian 2012; Díaz Trigo & Boirin 2016; Ponti et al. 2016). In stellar-mass BHs, X-ray winds have been established as a fundamental property of their most radiatively efficient outburst phases, the so-called soft states, likely impacting on the accretion process. However, these highly ionized (hot) winds are scarcely observed during the typically dimmer hard states, where most of the BHs of the universe exist and kinetic feedback from jets dominates (Miller et al. 2006; Neilsen & Lee 2009; Neilsen et al. 2011; Ponti et al. 2012; Fender & Muñoz-Darias 2016). The disappearance of the wind in the hard state is a matter of strong debate and has been suggested to be related to different physical processes, such as the details of the wind-launching mechanism, photoionization instabilities and over-ionization of the ejecta (e.g., Chakravorty et al. 2013; Bianchi et al. 2017; Gatuzz et al. 2019). In addition, the presence of a continuous, state-independent wind has been proposed as a viable mechanism to explain the high efficiency of angular momentum removal inferred from fits to X-ray light curves (Tetarenko et al. 2018). Low-ionization (cold) disk winds, on the other hand, have been detected via optical/infrared observations in the luminous and violent outbursts of V404 Cyg (Muñoz-Darias et al. 2016 (hereafter MD16), Muñoz-Darias et al. 2017; Rahoui et al. 2017; Mata Sánchez et al. 2018; see also Casares et al. 1991) and V4641 Sgr (Chaty et al. 2003; Lindstrøm et al. 2005; Muñoz-Darias et al. 2018). These are also thought to severely impact on accretion properties, with mass outflow rates significantly exceeding the accretion rate (MD16, Casares et al. 2019).

MAXI J1820+070 (hereafter J1820+070) was initially discovered as an optical transient by the All-Sky Automated Survey for Supernovae (and named ASASSN-18ey; Tucker et al. 2018) and then as an X-ray transient by the Monitor of All-sky X-ray Image (MAXI; Matsuoka et al. 2009). It was classified as a BH candidate based on its multi-wavelength properties (Baglio et al. 2018; Bright et al. 2018; Kawamuro et al. 2018; Kennea et al. 2018; Shidatsu et al. 2018; Tucker et al. 2018). The system is among the brightest BH transients ever observed (Corral-Santana et al. 2016), with an X-ray flux of ∼4 Crab and an optical magnitude of g ∼ 11.2 at outburst peak (Shidatsu et al. 2019). In this Letter, we report on an intensive optical spectroscopic follow-up campaign obtained with a suite of the largest telescopes providing high signal-to-noise data.

We carried out spectroscopy during 37 epochs, which cover in great detail the different phases of the outburst (see Table 1).

2.1. Gran Telescopio Canarias (GTC)

2.2. Very Large Telescope (VLT)

2.3. The Southern African Large Telescope (SALT)

2.4. Telescopio Nazionale Galileo (TNG)

2.5. Keck Observatory

The long-slit spectra were reduced, extracted, and wavelength- and flux-calibrated using standard IRAF tools or the PySALT package (Crawford et al. 2016) tailored for SALT data. The X-shooter spectra were processed and combined using version 3.2.0 of the EsoReflex pipeline. We used molly and custom software under python to perform the analysis. In order to increase the signal-to-noise ratio we henceforth focus on average spectra computed for each observing epoch (Table 1).

As previously reported by Tucker et al. (2018), the spectra are very rich in emission lines, including the Balmer series as well as several He i and He ii transitions. Our multi-epoch observations show that their shape and strength change dramatically throughout the outburst. We observe strong line asymmetries, especially during the early stages of the outburst, as well as shallow P-Cyg profiles and broad, non-Gaussian emission line wings. These features are standard accretion disk wind signatures, similar to those observed in V404 Cyg (MD16) and V4641 Sgr (Muñoz-Darias et al. 2018). Figures 1 and 2 show a suite of the different emission line behaviors present in our data, albeit we do not show every spectrum for clarity reasons (however, see Table 1 for a full summary). In particular, P-Cyg profiles are best detected in He i 5876 Å (He i–5876 hereafter), while asymmetries and broad emission wings are best seen in Hα, which is the strongest optical emission line. We note that these are also typical outflow tracers in other stellar classes. In particular, He i–5876 is one of the best optical wind markers in massive stars, and P-Cyg profiles in this line together with broad Hα wings are a common feature in O supergiants (e.g., Prinja & Fullerton 1994). Likewise, disk-dominated accreting white dwarfs can show, in addition to P-Cyg profiles in ultraviolet resonance lines, wind signatures in He i–5876 and Hα (Kafka & Honeycutt 2004).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We focused the analysis on three narrow spectral ranges covering the following lines: (i) Hα (6563 Å) and the adjacent He i 6678 Å, (ii) He i–5876, and (iii) a third region covering the Bowen blend (mainly N iii at 4641 Å) and He ii 4686 Å (B+He ii hereafter). We have searched for winds in the former two regions, while the last one was used as a proxy for the ionization state of the disk. The three bands are covered by the whole data set with the exception of He i–5876, which is absent in the (four) Keck and TNG epochs. Every spectral region was carefully normalized in each of the 37 average spectra by fitting the adjacent continuum with a first-order polynomial. This resulted in accurate continuum normalizations that were individually inspected.

In addition, we used data from MAXI to infer the X-ray state at the time of each epoch (see Table 1). We built the hardness-intensity diagram (Homan et al. 2001) using MAXI standard bands and daily averaged fluxes (Figure 3). As was shown by Shidatsu et al. (2019), the system displayed the q-shaped diagram typically observed in BH transients (e.g., McClintock & Remillard 2006; Belloni et al. 2011 for reviews). Our spectroscopic observations cover in great detail the evolution of the system throughout both the hard and soft X-ray states. We define hard and soft spectroscopic epochs as those with X-ray colors¹⁵ higher and lower than 0.4, respectively. As a proxy for the luminosity (i.e., intensity) we used the 2–20 keV MAXI count-rate, but note that a significant amount of the total flux is expected to be below 2 keV during the soft state.

Zoom In Zoom Out Reset image size

3.1. P-Cygni Profiles

3.2. The Hα Spectral Region

We have presented an optical spectroscopic campaign that is one of the most intensive and sensitive ever carried out on a BH transient in outburst. We have detected wind signatures in the form of P-Cyg profiles and broad emission line wings in several observing epochs during the hard states of the outburst of J1820+070. Such features have been observed before in the BH transients V404 Cyg and V4641 Sgr, although they undergo non-standard outbursts, when only the X-ray hard state, accompanied by strong radio flaring and high X-ray absorption is observed. J1820+070, on the other hand, displayed a standard outburst, showing both hard and soft states across the 8 months spanned by our observations (Shidatsu et al. 2019; Figure 3). It is important to note that the wind signatures seen in J1820+070 are significantly shallower than observed in the other two systems (e.g., 1%–2% below the continuum level versus ∼20%). Their detection has only been possible thanks to the combination of the exceptional brightness of the source and our systematic monitoring at very high signal-to-noise. Therefore, these could have been easily missed in other systems, suggesting that cold winds could be a common feature of BH accretion in X-ray binaries.

X-ray winds are typically observed during soft states, but are rarely ever present during BH hard states. Hence, these are mostly detected when the radio jet is quenched (Neilsen & Lee 2009; Ponti et al. 2012), albeit there are some counter-examples (e.g., King et al. 2015; see also Ponti et al. 2014; Díaz Trigo & Boirin 2016; Homan et al. 2016 for neutron star X-ray winds). Figure 3 reveals that the optical wind is almost ubiquitously detected during the hard state, extensively during outburst rise, but also during the decay toward the end of the soft-to-hard transition (Figure 1). In Figure 4 we represent the evolution of the EW of the B+He ii emission lines, which—given their higher ionization potential—can be used as a tracer for the state of the disk. All wind detections are found below EW ∼7 Å. This empirical threshold, which is not meant to be accurate and it is also sensitive to variations in the underlying continuum, excludes the whole soft state and some (typically the softest) hard-state epochs (Figure 3). A possible exception to this could be the first soft-state epoch, where a shallow blueshifted absorption might be present in He i–5876 (#19; Figure 1). The B+He ii EW from this epoch is just above the ∼7 Å threshold. Therefore, Figure 4 suggests that the non-detection of optical wind features during the soft state might be related to an over-ionization of the ejecta. This was shown to be a key factor for the detectability of the outflow in V404 Cyg (MD16). In addition, some of our most conspicuous wind detections occur at the hard-state peak (epochs #5 to #11; blue rectangle in Figure 3), a stage when radio emission is present (J. Bright et al. 2019, in preparation) and strong jet activity is typically witnessed in BH transients (e.g., Fender et al. 2004). This is also the peak of the optical outburst (g ∼ 11.2; about 0.8 mag brighter than the soft-state peak), a fact that has been interpreted as being the result of significant jet contribution to the optical regime (Shidatsu et al. 2018, 2019). Therefore, it is clear that the optical wind of J1820+070 is simultaneous with the jet.

Zoom In Zoom Out Reset image size

We have measured two characteristic wind velocities: ${V}_{{\rm{t}}}$ ∼ 1200 and ∼1800 km s⁻¹. Sometimes these are simultaneously observed in several features of the same spectrum (#7), but also in observations taken months apart. This suggests a rather complex wind structure, and a very appealing scenario would be that they represent different wind-launching mechanisms. However, more detailed modeling, beyond the scope of this discovery work, is needed to support such a scenario (see Muñoz-Darias et al. 2018 for a discussion on optical wind-launching mechanisms). In any case, the obtained ${V}_{{\rm{t}}}$ are similar to outflow velocities derived from (hard-state) optical winds and (soft-state) X-ray winds in other systems (e.g., MD16, Ponti et al. 2016). This raises the question of whether or not hot and cold winds are intrinsically different or just different observables of the same outflow at different stages of the outburst. In this regard, photoionization instability curves computed by Bianchi et al. (2017) show that the highly ionized soft-state wind might become either fully ionized (hot and low density) or almost neutral (cold and dense) once the system moves to the hard state (and the irradiating spectrum changes accordingly). We are not aware of any X-ray wind detection in J1820+070,¹⁷ but we note that inclination effects play a key role in the detectability of hot outflows (Ponti et al. 2012). Nevertheless, the detection of X-ray dips during the hard state (Homan et al. 2018; Kajava et al. 2019) and the shape of the hardness-intensity diagram (e.g., pronounced hard-state peak; Muñoz-Darias et al. 2013) advocates for a relatively high inclination, which would favor the detection of X-ray outflows.

Finally, it is worth mentioning that the shallowness of the P-Cyg features that we have discovered in J1820+070 might be indicative of a lower-mass outflow rate (e.g., Castor & Lamers 1979) than for the cases of V404 Cyg and V4641 Sgr, where winds have been proposed to significantly affect the outburst evolution (e.g., MD16). In particular, an outflow mass of ∼100 times the accreted mass has been estimated for V404 Cyg (Casares et al. 2019). This could explain the standard outburst displayed by J1820+070 as compared to these objects. Nevertheless, we note that mass outflow rates up to ∼10 times the accretion rate have been proposed for systems displaying regular outbursts and X-ray winds (e.g., Ponti et al. 2012).

We have detected optical accretion disk winds during the hard states of the outburst in MAXI J1820+070. Wind signatures are not detected in the soft state, but the visibility of the wind during this stage might be affected by a higher ionization of the accretion disk/ejecta. The detection of the wind has only been possible thanks to the brightness of the source and our exceptionally intensive and sensitive monitoring, implying that similar outflows are likely present in (at least) a significant fraction of BH X-ray binaries. Therefore, wind-like outflows would not be exclusive of bright, hot states but a common mass and angular momentum loss mechanism that operates through most of the accretion episode.

We acknowledge support by the Spanish MINECO under grant AYA2017-83216-P. T.M.D. and M.A.P.T. acknowledge support via Ramón y Cajal Fellowships RYC-2015-18148 and RYC-2015-17854. T.M.D. is thankful to Alex Fullerton for useful discussion on winds from massive stars. D.M.S. acknowledges support from the ERC under the European Unions Horizon 2020 research and innovation programme (grant agreement No. 715051; Spiders). P.A.C. is grateful to the Leverhulme Trust for the award of an Emeritus Fellowship. J.J.E.K. acknowledges support from the Academy of Finland grant 295114. Some of these observations were obtained with the Southern African Large Telescope under the Large Science Programme on transients, 2016-2-LSP-001 (PI: D.A.H.B.). Polish support of this SALT programme is funded by grant No. MNiSW DIR/WK/2016/07. D.A.H.B., E.J.K., and J.T. acknowledge research support from the South African National Research Foundation. D.S. acknowledges support from STFC via grant ST/P000495/1. molly software developed by Tom Marsh is gratefully acknowledged. Based on observations collected at ESO under programmes 0100.D-0292(A) and 0101.D-0158(A).