EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 694 (February 2025) A&A, 694 (2025) A109 Full HTML
Open Access
Issue
A&A
Volume 694, February 2025
Article Number A109
Number of page(s) 13
Section Astrophysical processes
DOI https://doi.org/10.1051/0004-6361/202451024
Published online 07 February 2025

© The Authors 2025

Licence Creative CommonsOpen Access article, published by EDP Sciences, under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This article is published in open access under the Subscribe to Open model. Subscribe to A&A to support open access publication.

1. Introduction

Most black hole transients (BHTs) are low mass X-ray binaries showing recurrent outbursts (e.g. McClintock & Remillard 2006; Belloni et al. 2011; Corral-Santana et al. 2016). During these events their X-ray luminosity increases by several orders of magnitude and mass ejecta occur, both in the form of winds and jets (see, e.g. Fender & Muñoz-Darias 2016).

Compact, relativistic jets emitting synchrotron radiation (e.g. Blandford & Königl 1979) are detected during the hard state of the outburst, when a hard power-law characterises the X-ray spectrum (e.g. Gallo et al. 2003). These jets dominate the radio band and can be detected up to the infrared and optical regimes (e.g. Corbel et al. 2001; Fender 2001). The jet is quenched during the soft state (e.g. Russell et al. 2011), which is characterised by a soft, thermal X-ray spectrum. Conversely, accretion disc winds are primarily observed in the X-ray band during this state (e.g. Neilsen & Lee 2009; Ueda et al. 2009; Ponti et al. 2012; Parra et al. 2024). These hot winds are associated with high values of the ionisation parameter (ξX ≥ 103 erg cm s−1; e.g. Díaz Trigo & Boirin 2016), and densities in the range of n ∼ 1013 − 1015 cm−3 have been proposed (e.g. Schulz et al. 2008; Kallman et al. 2009; see also Kosec et al. 2024 for a theoretical study).

Over the past few years, a new ingredient has been added to the above picture: the detection of cold (i.e. low-ionisation) winds in several BHTs (e.g. Muñoz-Darias et al. 2016; Mata Sánchez et al. 2018). This is inferred from the presence of conspicuous P-Cygni line profiles, blue-shifted absorption troughs, broad emission line wings and asymmetries in optical and near-infrared (NIR) emission lines. Optical signatures of wind-type outflows have been primarily detected during the hard state, sometimes simultaneously with the jet, in prominent emission lines, such as Hα, He I line at 5876 Å (He I-5876 hereafter) and He I-6678 (e.g. Muñoz-Darias et al. 2018, 2019; Cúneo et al. 2020; Mata Sánchez et al. 2022). These winds are characterised by a much lower ionisation parameter (ξV ≲ 10−3ξX; see, e.g. Muñoz-Darias & Ponti 2022), while numerical simulations suggest that their densities are comparable to those of hot winds (n ∼ 1013 cm−3; Koljonen et al. 2023).

In the NIR, evidence for winds is limited to a few systems. Wind signatures are mainly observed in emission lines of the Paschen and Brackett series (Paβ, Paγ, Brγ), as well as in He I-10830. NIR winds have been detected during both the hard and soft states, pointing to a scenario in which wind-type ejecta are present throughout most of the outburst (e.g. Sánchez-Sierras & Muñoz-Darias 2020; Panizo-Espinar et al. 2022; Sánchez-Sierras et al. 2023).

Interestingly, both hot and cold winds have been detected almost exclusively in intermediate to high inclination systems, leading to the commonly accepted idea that these outflows have a preferentially equatorial geometry (Ponti et al. 2012; Panizo-Espinar et al. 2022). The detection of wind signatures at different wavelengths and phases of the outburst (see, e.g. Castro Segura et al. 2022; Fijma et al. 2023; Sánchez-Sierras et al. 2023), suggests that the detectability of the winds is also influenced by the physical properties of the ejecta. This points towards a multi-phase nature of the winds, which is supported by the similarity of their kinetic properties in simultaneous X-ray and optical data (Muñoz-Darias & Ponti 2022).

In this paper we present optical and NIR spectroscopy of the BHT GX 339–4. This system was first discovered in 1973 (Markert et al. 1973) and has since shown frequent outbursts, with a recurrence period of ∼2.5 years. It hosts a K-type sub-giant star feeding a MBH ≈ 4 − 11 M⊙ black hole with an orbital period of Porb = 1.76 d (Hynes et al. 2003; Muñoz-Darias et al. 2008; Heida et al. 2017; Zdziarski et al. 2019). Its orbital inclination is thought to fall within the range 40° ≲i ≲ 60° (Zdziarski et al. 2019), making GX 339–4 a low or intermediate inclination system (see also Muñoz-Darias et al. 2013). Previous spectroscopic studies of this source at limited spectral resolution revealed that its optical and NIR spectrum is dominated by emission lines of low ionisation elements (H I, He I, He II). Single-peaked, round-top Hα profiles were observed in the hard state optical spectra presented by Soria et al. (1999) and Wu et al. (2001), and were interpreted as a possible signature of a dense outflow. This conclusion is further supported by Rahoui et al. (2014), who also detected broad Paβ line wings during the hard state.

In January 2021, GX 339–4 entered into a new outburst (e.g. Garcia et al. 2021; Sguera et al. 2021). Here, we analyse two epochs of optical and NIR spectroscopy collected during the soft state phase. We also consider two archival spectra obtained during the hard state of two previous outbursts (2013 and 2015) to study the influence of the X-ray state in the optical and NIR spectral properties. These simultaneous optical and NIR spectroscopic observations represent the highest-resolution outburst data on this source currently available in the literature. The high signal-to-noise ratio of our data enables us to investigate outflow-related signatures across a significant number of emission lines during both the hard and soft states.

2. Observations and data reduction

Four epochs of spectroscopy were obtained with the X-shooter spectrograph (Vernet et al. 2011) at the Very Large Telescope (VLT) in Cerro Paranal, Chile. The instrument consists of three spectroscopic arms, operating simultaneously in the UVB (3000–5600 Å), VIS (5500–10 200 Å) and NIR (10 200–24 800 Å) spectral ranges. Two observations were performed on April 11 (MJD 59315) and April 15 (MJD 59319) 2021 (hereafter S2021a and S2021b, respectively). We also analysed two X-shooter archival spectra of the same source, which were collected during the 2013 and 2015 outbursts (hereafter H2013 and H2015). For each observation, 8 to 64 exposures in nodding configuration were obtained, with individual exposure times in the range of ∼40 − 1200 s and resolutions of ∼30 − 70 km/s. In order to optimize sky subtraction, the source was observed using two nod positions. The details of the observations are provided in Table 1. The absolute orbital phase for each epoch, based on the ephemeris by Heida et al. (2017), is also reported in the table.

Table 1.

Journal of observations.

The data were reduced using the X-shooter Pipeline version 3.6.1 (Freudling et al. 2013), which provides sky-subtracted, wavelength-calibrated mono-dimensional spectra. For each observing epoch, flux calibration was performed using a standard star observed close in time, either on the same or preceding night. In order to improve the flux calibration throughout the entire band covered by the spectra, a correction routine was implemented to compensate for the effects of wavelength-dependent light loss due to variable atmospheric seeing. To achieve this, a Gaussian was fitted at each wavelength to the spatial profile of the trace in the rectified and merged 2D spectra. This process was repeated for both the source and the standard star, effectively producing wavelength-dependent solutions for the seeing. We calculated the integrated area of the profiles within the region truncated by the fixed slit width (Aob, λ for the source and Astd, λ for the standard star). This allowed us to derive a correction factor for the flux lost due to slit truncation, defined as fλ = Astd, λ/Aob, λ, which was then multiplied by our target spectra. We find fλ values close to unity, with a median of ∼0.9. As expected, the points that significantly deviate are those at the joints between the different X-shooter arms.

All the spectra are Doppler-shifted to the barycentric rest frame. MOLECFIT (Smette et al. 2015; Kausch et al. 2015, version 1.5.9) was used for correcting telluric absorption in the optical and NIR ranges. The spectra were dereddened using the dust_extinction python package (Gordon et al. 2022). The dereddening was performed using the F19 routine, which is based on the model by Fitzpatrick et al. (2019). We adopted the extinction law AV = 3.1E(B − V) by Cardelli et al. (1989), along with a previously reported colour excess E(B − V) = 1.2 ± 0.1 (Zdziarski et al. 1998).

Finally, we note that our analysis is complemented by daily averaged Swift-XRT data obtained in Windowed Timing mode. We used count rates in the 2–10 keV range and hardness ratios (4–10 keV/2–4 keV) corresponding to the 2013, 2015, and 2021 outbursts. These data were used solely to approximately trace the X-ray behaviour of the source at the time of our spectroscopic observations (see below) and were directly derived from the Swift-XRT data products generator1 (Evans et al. 2007, 2009). The data analysis presented in the next sections was performed using custom routines developed under Python 3.9.

3. Analysis and results

The evolution of the 2021 outburst across the hardness-intensity diagram (HID; e.g. Homan et al. 2001), is shown in Fig. 1 (see also Stiele & Kong 2023). The outburst transitioned from the hard to the soft state and returned to quiescence following the typical hysteresis pattern of BHTs (e.g. Fender et al. 2004). The approximate location of our spectroscopic observations is marked in the diagram. We note that the spectra S2021a and S2021b were collected during the most X-ray luminous stages of the outburst. The position of the spectroscopic epochs H2013 and H2015 in the HID was determined using data from their corresponding outbursts. In particular, H2013 was collected during the decay of a failed outburst (Belloni et al. 2013), whereas H2015 was observed during the final stages of the 2015 complete outburst. Their location in the HID diagram confirms the classification of S2021a and S2021b as soft state (Stiele & Kong 2023; Liu et al. 2023), and H2013 and H2015 as hard state observations (Belloni et al. 2013; Tomsick et al. 2015).

thumbnail Fig. 1.

Swift-XRT HID during the 2021 outburst of GX 339–4. The epochs S2021a and S2021b are marked with up- and down-pointing triangles (overlapping in the figure), correspondingly. The hard state epochs, H2013 and H2015, are marked with a circle and a square, respectively.

3.1. Evolution of the continuum

The optical to NIR spectral energy distribution (SED) of the four epochs is presented in Fig. 2. The spectral continuum of the two soft state epochs (S2021a and S2021b) is almost identical. Likewise, the SED in the two hard state epochs is similar, despite corresponding to different outbursts. We note that the hard state spectra exhibit an enhanced NIR continuum when compared to those of the soft state. To further investigate this, the hard state SEDs were fitted with a broken power law given by:

thumbnail Fig. 2.

Fit to the dereddened continuum of the four epochs. The fit excludes regions with strong telluric absorptions. The symbols, coded as in the previous figures, mark the break location.

f ν ( ν ) = { a · ( ν ν 0 ) α OPT ν ≥ ν 0 a · ( ν ν 0 ) α NIR ν < ν 0 , $$ \begin{aligned} f_{\nu }(\nu ) = {\left\{ \begin{array}{ll} a\cdot \left(\frac{\nu }{\nu _0}\right)^{\alpha _{\mathrm{OPT}}}&\nu \ge \nu _0\\ a\cdot \left(\frac{\nu }{\nu _0}\right)^{\alpha _{\mathrm{NIR}}}&\nu < \nu _0, \end{array}\right.} \end{aligned} $$(1)

where a, ν0, αOPT and αNIR are free parameters. Conversely, a single power law was employed for fitting the soft state SEDs (i.e. α = α OPT = α NIR ) $ \alpha= \alpha_{\mathrm{OPT}}=\rm\alpha_{\mathrm{NIR}}) $. The fit excludes regions heavily affected by telluric absorption and diffuse interstellar bands (DIBs). Emission lines have also been masked. The region 520-600 THz was excluded from the fit to the H2013 SED due to a significant flux drop at the frequency connection between the UVB and VIS X-shooter arms (possibly due to the disabling of the atmospheric dispersion correctors of the instrument during this epoch).

The best-fit spectral indices are reported in Table 2. The errors are primarily driven by uncertainties in interstellar reddening. The SEDs of the soft state epochs are best modeled by a power law with spectral indices of α = 1.6 ± 0.3 and α = 1.7 ± 0.3 for S2021a and S2021b, respectively. A slight deviation from this slope is observed at the red end in the NIR, hinting at a possible NIR excess, which is discussed in Sect. 4.1. As a test, a broken power-law fit was also applied, yielding consistent values for αOPT and αNIR. Therefore, we prefer to retain the single power-law fit.

Table 2.

Best-fit indices obtained from fitting the optical to NIR SEDs.

In the hard state, the best-fit spectral indices for H2013 are αOPT = 0.1 ± 0.2 and αNIR = −0.43 ± 0.08. For H2015, we find αOPT = 0.8 ± 0.3 and αNIR = −0.08 ± 0.08. As expected from visual inspection, the spectral indices in the hard state continua significantly differ from those measured in the soft state, suggesting that distinct radiative processes may be at work (see Sect. 4.1).

3.2. Evolution of the emission lines

The normalised spectra of all epochs were obtained by fitting a low-order spline to the data after masking the emission lines and strong telluric absorptions. The normalised spectra of the S2021a (red) and H2015 (cyan) epochs are shown in Fig. 3. For completeness, the H2013 and S2021b spectra are also reported in Appendix A. Several emission features can be observed, with line shapes and intensities varying across the epochs. The soft state spectra appear widely populated, with spectral features corresponding to the Balmer, Paschen and Brackett series up to high orders, He I and He II transitions and also signatures of O III and of O II, C III and N III (Bowen blend). Some lines appear particularly intense, such as Hα and He I-10830. Conversely, only a few emission lines are clearly detected in the hard state, such as some Balmer, Paschen, He I and He II transitions, being Hα, Hβ, He II-4686 and He I-10830 the most intense features. Fig. 3 highlights the relative strength of the lines during the soft state epoch S2021a over those in the hard state (H2015), with the Brackett transitions almost indiscernible from the continuum in the latter case.

thumbnail Fig. 3.

UVB, VIS and NIR normalised spectra of the soft state S2021a (red) and hard state H2015 (cyan) epochs.

In order to quantify the relative strength of the most prominent emission lines with respect to their underlying continuum, we determined their equivalent width (EW). Features contaminated by adjacent lines, DIBs or strong telluric absorptions were excluded from the calculation. We also computed the V/R ratio (defined as the integrated flux ratio between the blue and red parts of the line, measured from the systemic velocity of the source), which is sensitive to line asymmetries. For completeness, we calculated the full width at zero intensity (FWZI), as this parameter may also be useful in future studies. The results are presented in Table B.1.

The analysis confirms key features of the emission lines, such as their being stronger during soft state epochs, as evidenced by higher EW values. The most dramatic EW changes occur in the NIR, where emission lines are either undetected or show a substantial drop during the hard state compared to the soft state. The V/R ratio analysis shows that, during soft state epochs, most lines are asymmetric, with the red part dominating over the blue (V/R < 1). A similar trend is observed during the hard state epoch H2013; however, both red- and blue-dominated lines are seen in H2015. The FWZI shows a significant dispersion, ranging from several hundred to ∼2000 km s−1, depending on the line and epoch. In particular, the optical hydrogen lines detected during the soft state are narrower than those observed in the hard state or the NIR hydrogen lines in the soft state.

3.2.1. Soft state: main lines and their properties

A selection of the most prominent emission lines observed across the epochs is shown in Fig. 4. The two columns on the left include features present in the soft state spectra. Although these spectra were collected only four days apart during the same accretion state and with almost identical flux, the emission lines exhibit differences in both intensity and shape.

thumbnail Fig. 4.

Line profiles of the most intense features across all four epochs, normalised with respect to their adjacent continuum. The velocities are given in the observer’s rest frame.

The low-ionisation features Hα, Hβ, He I-5876 and He I-10830 appear asymmetric and skewed towards redder wavelengths in both epochs. In particular, they are characterised by a red-shifted peak in S2021a, while a shallower blue peak is also present in S2021b. During both epochs, Hα wings meet the continuum at ∼ ± 600 km s−1, while its EW is ∼10 Å (Table B.1). The V/R ratio is significantly lower than unity for most emission lines, indicating that the blue part of the line is weaker than the red part. This trend is consistently observed in lines that are thought to be good wind tracers, such as Hα, He I-6678 and He I-7065 (see, e.g. Muñoz-Darias et al. 2019).

The higher ionisation features, such as the Bowen blend, are strong in both soft state spectra. He II-4686 stands out especially in S2021a, with an EW of ∼6 Å. In both S2021a and S2021b, the line exhibits a strongly asymmetric profile, with a prominent red peak and a shallower blue peak.

In the NIR, H I lines are broader than in the optical, meeting the continuum at ∼ ± 700 km s−1 (compared to ∼ ± 500 km s−1 in the optical). Pa9 exhibits a broad double-peaked profile, whose wings meet the continuum at approximately ±1000 km s−1. Double-peaked profiles are common in S2021b (e.g. in He I-4922, He I-6678, He I-7065, and high-order Paschen and Brackett emission lines), whereas Pa9 is the only double peaked feature in S2021a. Finally, the profiles of Paβ and Brγ share common features, including a prominent red-shifted peak. However, Paβ is contaminated by the adjacent He I-12784 line, preventing a proper characterisation of the line wings. Brγ exhibits asymmetric wings, the blue one reaching ∼ − 1000 km s−1, while the red one meets the continuum at ∼ + 700 km s−1 in S2021a.

3.2.2. Hard state: main lines and their properties

The relative intensity of the spectral features is lower in the hard state compared to the soft state and they consistently exhibit broader wings (see Table B.1). Hα and He I-10830 are the strongest spectral features. In H2013 their profiles appear flat-top, while in H2015 they show a red-skewed asymmetry with a red-shifted peak, as commonly observed in the soft state epochs. As for the higher ionisation features, He II-4686 emerges as a strong double-peaked line in H2015, with a peak-to-peak separation of ∼490 km s−1 and an extended blue wing that reaches ∼ − 1000 km s−1. In the NIR, Pa9 shows extended wings up to ∼1000 km s−1 in H2015, as observed also in the soft state. During H2013, Paβ is flat-top, similarly to the most prominent optical lines. Most H and K-band emission lines are significantly weaker than in the soft state or completely absent.

4. Discussion

We have presented four epochs of optical and NIR spectroscopy obtained during three different outbursts of GX 339–4. Both the continuum and spectral lines exhibit variations across the epochs, with large differences between the hard and the soft states. In the former, the SED shows an enhanced NIR flux, coinciding with the disappearance of most emission lines in the wavelength range corresponding to the H and K bands. During both accretion states, the analysis of the spectral lines does not reveal the presence of P-Cygni profiles. However, we observe line profiles which are often associated with outflows, such as flat-top and asymmetric profiles as well as broad line wings. These signatures are consistent with those observed in other BHTs (e.g. Cúneo et al. 2020; Mata Sánchez et al. 2024).

Some selected recent studies of newly discovered BHTs benefit from a larger number of spectroscopic epochs (e.g. Muñoz-Darias et al. 2019; Panizo-Espinar et al. 2022), allowing for statistical tests to compare them (see also Mata Sánchez et al. 2023). Our work, by contrast, includes four epochs taken at different outburst stages of the prototypical system GX 339–4. While this limits statistical analyses, the high resolution and sensitivity of our data allowed us to study numerous emission lines across each epoch, revealing consistent patterns in their profiles. These findings further contribute to our understanding of this fundamental system, as well as to the impact that outflows (both jets and winds) have on the optical and NIR spectra of BHTs.

4.1. The spectral energy distribution

The optical-NIR SED (Fig. 2) exhibits significant variations between the soft and hard state epochs. In the soft state spectra, the best-fit spectral indices are in the range of 1 ≲ α ≲ 2. This is consistent with a thermal origin for the radiation (e.g. Hynes 2005). However, in the NIR frequency range, the SED appears to deviate from a simple power law at the red end. A similar behaviour has been observed in at least one other source, the BHT MAXI J1820+070, which shows a comparable infrared excess during a soft state epoch (Koljonen et al. 2023). This was modeled by including the presence of a wind (or an atmosphere) extending above the disc, which could act as a reprocessing site for accretion disc radiation, yielding the infrared excess. Our soft state spectra suggest the presence of wind-type ejecta, and thus, the above interpretation might also apply to GX 339–4. However, we caution the reader that the evidence for this excess is relatively weak, as it is found at the red end of our coverage. More data (e.g. mid-infrared) would be needed to properly study and discuss this feature.

The hard state SED is significantly different from that of the soft state. In particular, the NIR continuum is clearly enhanced, which, together with the weakness (H2015) or total absence (H2013) of emission lines in the H and K bands, suggests the presence of an additional component contributing to these frequency bands. As a matter of fact, the optically thin synchrotron emission generated by the compact jet can be observed up to the optical/infrared range (e.g. Corbel et al. 2001; Fender 2001) and it can dilute the line profiles, hampering their detection. In the NIR range, its contribution is usually well approximated by a power law with spectral index −1 ≤ α ≤ −0.5 (e.g. Russell et al. 2013). This scenario is supported by the broken power law of αNIR ∼ −0.5 for H2013. In H2015, the flatter spectral index (αNIR ∼ −0.1) might be attributed to the presence of an additional (bluer) component, perhaps associated to the accretion disc.

The contribution of the synchrotron component to the optical-NIR regime (e.g. Corbel & Fender 2002; Shidatsu et al. 2011) and thus the weakening of the emission lines at these wavelengths (Rahoui et al. 2012) have been observed in previous outbursts of GX 339–4. The thus produced NIR excess has been found to be particularly bright in GX 339–4 in the hard state (e.g. Corbel & Fender 2002; Homan et al. 2005; Buxton et al. 2012), with a dramatic drop/rise during transitions from/to the hard state (Cadolle Bel et al. 2011; Dinçer et al. 2012; Saikia et al. 2019). As discussed above, the jet component can comfortably account for the observed NIR SED of GX 339–4 during the hard state. However, we note that other physical components, such as the so-called hot flow (e.g. Kosenkov et al. 2020), might also contribute.

4.2. Accretion disc winds

Optical and NIR wind signatures have been detected in several BHTs, although only a few detections have been made in low-to-intermediate inclination systems (e.g. Panizo-Espinar et al. 2022). Our spectroscopy reveals the presence of flat-top profiles during the hard state epoch H2013 in Hα, He I-10830 and Paβ (Fig. 5). Low-velocity components associated with asymmetric emission from the outer disc or the companion star can generate these line profiles (e.g. Casares et al. 2006). However, flat-top profiles can also be produced by accretion disc winds, as discussed in theoretical studies (e.g. see Murray et al. 1995 for a study on active galactic nuclei) and demonstrated using simple toy models (Mata Sánchez et al. 2023). Indeed, flat-top profiles have been previously reported in a number of BHTs (e.g. Cúneo et al. 2020; Panizo-Espinar et al. 2022; Mata Sánchez et al. 2024).

thumbnail Fig. 5.

Normalised hard state spectra centred on the most prominent line profiles. The left panel shows the flat-top profiles of Hα, He I-10830 and Paβ during epoch H2013. The right panel highlights the difference between the red-skewed, single-peaked profiles of Hα and He I-10830 and the double peak of He II-4686 during H2015.

The spectrum H2015 shows single-peaked, red-skewed profiles in Hα and He I-10830. Line asymmetries can be attributed to asymmetric emission from the accretion disc (i.e. a hot spot; Stover 1981). However, the double-peaked shape of He II-4686 (Fig. 5) and high order Paschen lines are consistent with arising in a Keplerian and symmetric accretion disc, arguing against a geometrical explanation for the remaining asymmetric profiles. Therefore, we propose that these single-peaked lines might be generated in a wind. This phenomenon naturally produces blue-shifted absorptions that might overlap with the disc line profile, resulting in asymmetric, red-skewed shapes.

Figure 5 illustrates the differences between the emission line profiles during the two hard state epochs. These differences are not entirely surprising, as variability in the properties of wind signatures on timescales ranging from minutes to days has been observed in other sources (e.g. Mata Sánchez et al. 2018; Muñoz-Darias et al. 2019). Such variability is perhaps more expected when comparing two epochs from different outbursts, even though they occur in a similar hard state. Different line properties are thought to result from variations in the ionisation and density conditions of the wind, which can change due to several factors, such as irradiation from the central source and inhomogeneities along the line of sight.

During the soft state epochs, broad line wings are observed in intense NIR lines (i.e. Pa9, Br11, Br10, Brγ). Broad line wings are a common feature in BHTs (e.g. Muñoz-Darias et al. 2016, 2018, 2019; Sánchez-Sierras et al. 2023; Panizo-Espinar et al. 2022) and can be produced by an expanding envelope. Additionally, most of these lines are asymmetric. Again, the presence of a hot spot could be invoked to explain this feature, however, we observe that asymmetric, red-skewed profiles coexist with double-peaked lines such as Pa9 (Fig. 4), weakening a geometric explanation. The orbital phase difference between the two soft state epochs is ΔΦ = 0.23 ± 0.01. The presence of similar line profiles with an enhanced red peak at phases separated by a quarter of the orbital period argues against a hot spot origin, unless it is significantly extended in orbital phase. Unfortunately, we cannot calculate the relative orbital phases for the hard state epochs due to large uncertainties in the projected orbital solution (Heida et al. 2017). Nevertheless, the fact that red-skewed asymmetries are also observed in H2015 (Fig. 4), coupled with the absence of any instances of blue-skewed profiles in the four epochs analysed in this study, further argues against a phase-dependent origin for this feature.

In the optical range, wind signatures are limited to the presence of red-skewed, asymmetric profiles. Although a geometrical explanation cannot be fully ruled out, it could hardly account for the simultaneous observation of broad line wings in the NIR. Conversely, the wind scenario has the potential to simultaneously explain the observed features, including broad line wings, red-skewed asymmetry, and flat-top profiles across all inspected accretion states.

The possible presence of outflow signatures in the optical spectrum of GX 339–4 was explored by Soria et al. (1999) and Wu et al. (2001), who identified single-peaked and asymmetric Hα lines, simultaneous with double-peaked He II-4686 profiles. In these works, the single-peaked lines were interpreted as a hint for the presence of an outflow, either a dense wind originated in the disc or an evaporating corona. Broad Paβ wings were also identified by Rahoui et al. (2014), who suggested a wind origin to explain them. The higher resolution and broad coverage of our data allowed for a detailed analysis of multiple emission lines, revealing wind signatures across both the optical and NIR spectra in both hard and soft states. Thus, the spectral features identified in this study are consistent with those observed previously in this system and in other low-to-intermediate inclination BHTs (see also e.g. Panizo-Espinar et al. 2022).

4.3. The role of orbital inclination and ionisation state in the wind detectability

According to the current paradigm, which describes the geometry of accretion disc winds in black hole transients as equatorial (e.g. Ponti et al. 2012), the detection of P-Cygni profiles in low-to-intermediate inclination systems may be challenging. The reason being that the blue-shifted absorption component caused by the wind would occur at lower velocities, due to projection effects (e.g. Panizo-Espinar et al. 2022; see also Higginbottom et al. 2019 for a theoretical study). This can make the (wind-related) absorption feature overlap with the (disc) emission line profile, leading to the suppression of the blue peak, either completely (producing skewed profiles, e.g. Muñoz-Darias et al. 2019; Panizo-Espinar et al. 2022), or partially (leaving a red bump, e.g. Muñoz-Darias et al. 2020). The predominant features associated with outflows in these systems would instead consist of broad emission wings, asymmetric red-skewed profiles, and low-velocity blue-shifted absorptions. Taking this into account, the potential outflow signatures that we have identified in GX 339–4 are in agreement with the intermediate inclination proposed for this system (Zdziarski et al. 2019).

During the hard state of GX 339–4, wind-related features are only detected in low-ionisation transitions (e.g. Hα, He I-10830), in agreement with observations of other BHTs. Only a few systems show optical wind signatures in high ionisation lines (e.g. Charles et al. 2019; Jiménez-Ibarra et al. 2019; Mata Sánchez et al. 2024). These are interpreted as indicative of a hot and dense outflow. During the soft state epochs, we observed an enhanced red peak in both low ionisation lines and He II-4686. Assuming these features are produced by a wind, they suggest changes in the physical conditions of the ejecta during different accretion states, being more highly ionised and denser during the soft state.

The flux ratio Iratio = He II-4686/Hβ is a good tracer of the ionisation state (e.g. Muñoz-Darias et al. 2016) and was calculated for each epoch (see Fig. 6). We find Iratio > 2 for the soft state spectra, whereas Iratio ∼ 1 − 1.5 is found during H2013 and H2015. This is consistent with more highly ionised material during the soft state epochs. The Balmer ecrement (BD), which is defined as the flux ratio between Hα and Hβ, is a proxy for optically thinner gas (e.g. nebular phases, Muñoz-Darias et al. 2016). The BD of each epoch has been calculated, and its values reported in Fig. 6. We find BD ∼ 1 − 1.5 in both accretion states, which is roughly consistent with previous reports (e.g. Rahoui et al. 2014). Such low BD values are typically associated with emission lines originating in dense regions, e.g. the accretion disc. We do not find evidence for nebular phases, characterised by a BD ≳ 3 (e.g. Mata Sánchez et al. 2018).

thumbnail Fig. 6.

Balmer decrement (BD) against the ionisation ratio (Iratio) for each observing epoch.

To summarise, we consistently observe spectral features classically associated with the presence of winds across the four optical-to-NIR spectroscopic epochs analysed in this work. Broad emission line wings (e.g. Pa9, Brγ) reaching velocities up to ∼1000 km s−1 are detected, compatible with the typical velocities of a few thousand km s−1 commonly observed in BHT winds (e.g. Muñoz-Darias et al. 2018; Mata Sánchez et al. 2024). The red-skewed and flat-top line profiles observed also agree with the cold wind signatures detected in MAXI J1348–630, arguably the best example of winds observed so far in a low- to intermediate-inclination BHT (Panizo-Espinar et al. 2022).

The detection of wind signatures across several emission lines during all the spectroscopic epochs analysed here, combined with the profiles of Hα and Paβ reported in previous studies (Soria et al. 1999; Wu et al. 2001; Rahoui et al. 2014), supports the presence of wind-type outflows in GX 339–4. In this context, the absence of P-Cygni line profiles in the spectra can be attributed to the orbital inclination of the system, which is thought to be lower than that of the sources showing these features.

5. Conclusions

We have analysed four epochs of simultaneous optical and infrared spectroscopy of the black hole transient GX 339–4 with emphasis on searching for outflow-related features. We do not detect P-Cygni line profiles as those observed in other systems of this class. However, we detect other potential wind signatures, such as asymmetric and flat-top line profiles, as well as broad emission line wings. During the hard state, the wind signatures are mainly observed in the optical. Additionally, signatures of the jet are evident in the form of an enhanced NIR continuum, coinciding with the disappearance of most H and K bands’ emission lines. During the soft state, the strongest wind signatures are observed in the NIR spectral range. We discuss that the absence of P-Cygni profiles can be attributed to the moderately low inclination of the system. Additional optical and NIR spectroscopy of GX 339–4 and other low-to-intermediate inclination systems in outburst is encouraged to delve deeper into the optical and infrared observational properties of black hole transients.

Acknowledgments

We acknowledge support by Spanish Agencia estatal de investigación via PID2020-120323GB-I00 and PID2021-124879NB-I00. Based on observations collected at the European Southern Observatory under ESO programmes 105.200K.001, 105.200K.002, 291.D-5047(A) and 295.D-5027(C). This work made use of data supplied by the UK Swift Science Data Centre at the University of Leicester. M.A.P. acknowledges support through the Ramón y Cajal grant RYC2022-035388-I, funded by MCIU/AEI/10.13039/501100011033 and FSE+. This material is based upon work supported by Tamkeen under the NYU Abu Dhabi Research Institute grant CASS. GP acknowledge financial support from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program HotMilk (grant agreement No. 865637). GP acknowledges financial support from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program HotMilk (grant agreement No. 865637), from Bando per il Finanziamento della Ricerca Fondamentale 2022 dell’Istituto Nazionale di Astrofisica (INAF): GO Large program and from the Framework per l’Attrazione e il Rafforzamento delle Eccellenze (FARE) per la ricerca in Italia (R20L5S39T9). MCB is supported by the INAF-Astrofit fellowship. VC acknowledges support from the German DLR under contract number 50OR2405. We acknowledge Astropy, (http://www.astropy.org) a community-developed core Python package for Astronomy (Astropy Collaboration 2013, 2018).

References

  1. Astropy Collaboration (Robitaille, T. P., et al.) 2013, A&A, 558, A33 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  2. Astropy Collaboration (Price-Whelan, A. M., et al.) 2018, AJ, 156, 123 [Google Scholar]
  3. Belloni, T. M., Motta, S. E., & Muñoz-Darias, T. 2011, Bull Astron. Soc. India, 39, 409 [NASA ADS] [Google Scholar]
  4. Belloni, T. M., Motta, S., & Casella, P. 2013, ATel, 5417, 1 [NASA ADS] [Google Scholar]
  5. Blandford, R. D., & Königl, A. 1979, ApJ, 232, 34 [Google Scholar]
  6. Buxton, M. M., Bailyn, C. D., Capelo, H. L., et al. 2012, AJ, 143, 130 [NASA ADS] [CrossRef] [Google Scholar]
  7. Cadolle Bel, M., Rodriguez, J., D’Avanzo, P., et al. 2011, A&A, 534, A119 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 [Google Scholar]
  9. Casares, J., Cornelisse, R., Steeghs, D., et al. 2006, MNRAS, 373, 1235 [NASA ADS] [CrossRef] [Google Scholar]
  10. Castro Segura, N., Knigge, C., Long, K. S., et al. 2022, Nature, 603, 52 [NASA ADS] [CrossRef] [Google Scholar]
  11. Charles, P., Matthews, J. H., Buckley, D. A. H., et al. 2019, MNRAS, 489, L47 [Google Scholar]
  12. Corbel, S., & Fender, R. P. 2002, ApJ, 573, L35 [NASA ADS] [CrossRef] [Google Scholar]
  13. Corbel, S., Kaaret, P., Jain, R. K., et al. 2001, ApJ, 554, 43 [NASA ADS] [CrossRef] [Google Scholar]
  14. Corral-Santana, J. M., Casares, J., Muñoz-Darias, T., et al. 2016, A&A, 587, A61 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  15. Cúneo, V. A., Muñoz-Darias, T., Sánchez-Sierras, J., et al. 2020, MNRAS, 498, 25 [CrossRef] [Google Scholar]
  16. Díaz Trigo, M., & Boirin, L. 2016, Astron. Nachr., 337, 368 [Google Scholar]
  17. Dinçer, T., Kalemci, E., Buxton, M. M., et al. 2012, ApJ, 753, 55 [CrossRef] [Google Scholar]
  18. Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2007, A&A, 469, 379 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  19. Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2009, MNRAS, 397, 1177 [Google Scholar]
  20. Fender, R. P. 2001, MNRAS, 322, 31 [NASA ADS] [CrossRef] [Google Scholar]
  21. Fender, R., & Muñoz-Darias, T. 2016, in Lecture Notes in Physics, eds. F. Haardt, V. Gorini, U. Moschella, A. Treves, & M. Colpi (Berlin Springer Verlag), 905, 65 [NASA ADS] [CrossRef] [Google Scholar]
  22. Fender, R. P., Belloni, T. M., & Gallo, E. 2004, MNRAS, 355, 1105 [NASA ADS] [CrossRef] [Google Scholar]
  23. Fijma, S., Castro Segura, N., Degenaar, N., et al. 2023, MNRAS, 526, L149 [NASA ADS] [CrossRef] [Google Scholar]
  24. Fitzpatrick, E. L., Massa, D., Gordon, K. D., Bohlin, R., & Clayton, G. C. 2019, ApJ, 886, 108 [Google Scholar]
  25. Freudling, W., Romaniello, M., Bramich, D. M., et al. 2013, A&A, 559, A96 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  26. Gallo, E., Fender, R. P., & Pooley, G. G. 2003, MNRAS, 344, 60 [Google Scholar]
  27. Garcia, J., Tomsick, J., Harrison, F., Connors, R., & Mastroserio, G. 2021, ATel, 14352, 1 [NASA ADS] [Google Scholar]
  28. Gordon, K., Larson, K., McBride, A., et al. 2022, https://doi.org/10.5281/zenodo.6397654 [Google Scholar]
  29. Heida, M., Jonker, P. G., Torres, M. A. P., & Chiavassa, A. 2017, ApJ, 846, 132 [Google Scholar]
  30. Higginbottom, N., Knigge, C., Long, K. S., Matthews, J. H., & Parkinson, E. J. 2019, MNRAS, 484, 4635 [NASA ADS] [CrossRef] [Google Scholar]
  31. Homan, J., Wijnands, R., van der Klis, M., et al. 2001, ApJS, 132, 377 [NASA ADS] [CrossRef] [Google Scholar]
  32. Homan, J., Buxton, M., Markoff, S., et al. 2005, ApJ, 624, 295 [NASA ADS] [CrossRef] [Google Scholar]
  33. Hynes, R. I. 2005, ApJ, 623, 1026 [NASA ADS] [CrossRef] [Google Scholar]
  34. Hynes, R. I., Steeghs, D., Casares, J., Charles, P. A., & O’Brien, K. 2003, ApJ, 583, L95 [NASA ADS] [CrossRef] [Google Scholar]
  35. Jiménez-Ibarra, F., Muñoz-Darias, T., Casares, J., Armas Padilla, M., & Corral-Santana, J. M. 2019, MNRAS, 489, 3420 [Google Scholar]
  36. Kallman, T. R., Bautista, M. A., Goriely, S., et al. 2009, ApJ, 701, 865 [NASA ADS] [CrossRef] [Google Scholar]
  37. Kausch, W., Noll, S., Smette, A., et al. 2015, A&A, 576, A78 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  38. Koljonen, K. I. I., Long, K. S., Matthews, J. H., & Knigge, C. 2023, MNRAS, 521, 4190 [NASA ADS] [CrossRef] [Google Scholar]
  39. Kosec, P., Rogantini, D., Kara, E., et al. 2024, ApJ, 972, 32 [NASA ADS] [CrossRef] [Google Scholar]
  40. Kosenkov, I. A., Veledina, A., Suleimanov, V. F., & Poutanen, J. 2020, A&A, 638, A127 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  41. Liu, H., Bambi, C., Jiang, J., et al. 2023, ApJ, 950, 5 [NASA ADS] [CrossRef] [Google Scholar]
  42. Markert, T. H., Canizares, C. R., Clark, G. W., et al. 1973, ApJ, 184, L67 [NASA ADS] [CrossRef] [Google Scholar]
  43. Mata Sánchez, D., Muñoz-Darias, T., Casares, J., et al. 2018, MNRAS, 481, 2646 [Google Scholar]
  44. Mata Sánchez, D., Muñoz-Darias, T., Cúneo, V. A., et al. 2022, ApJ, 926, L10 [CrossRef] [Google Scholar]
  45. Mata Sánchez, D., Muñoz-Darias, T., Casares, J., Huertas-Company, M., & Panizo-Espinar, G. 2023, MNRAS, 524, 338 [CrossRef] [Google Scholar]
  46. Mata Sánchez, D., Muñoz-Darias, T., Armas Padilla, M., Casares, J., & Torres, M. A. P. 2024, A&A, 682, L1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  47. McClintock, J. E., & Remillard, R. A. 2006, in Compact stellar X-ray sources, 39, 157 [NASA ADS] [CrossRef] [Google Scholar]
  48. Muñoz-Darias, T., & Ponti, G. 2022, A&A, 664, A104 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  49. Muñoz-Darias, T., Casares, J., & Martínez-Pais, I. G. 2008, MNRAS, 385, 2205 [Google Scholar]
  50. Muñoz-Darias, T., Coriat, M., Plant, D. S., et al. 2013, MNRAS, 432, 1330 [Google Scholar]
  51. Muñoz-Darias, T., Casares, J., Mata Sánchez, D., et al. 2016, Nature, 534, 75 [Google Scholar]
  52. Muñoz-Darias, T., Torres, M. A. P., & Garcia, M. R. 2018, MNRAS, 479, 3987 [CrossRef] [Google Scholar]
  53. Muñoz-Darias, T., Jiménez-Ibarra, F., Panizo-Espinar, G., et al. 2019, ApJ, 879, L4 [CrossRef] [Google Scholar]
  54. Muñoz-Darias, T., Armas Padilla, M., Jiménez-Ibarra, F., et al. 2020, ApJ, 893, L19 [CrossRef] [Google Scholar]
  55. Murray, N., Chiang, J., Grossman, S. A., & Voit, G. M. 1995, ApJ, 451, 498 [NASA ADS] [CrossRef] [Google Scholar]
  56. Neilsen, J., & Lee, J. C. 2009, Nature, 458, 481 [NASA ADS] [CrossRef] [Google Scholar]
  57. Panizo-Espinar, G., Armas Padilla, M., Muñoz-Darias, T., et al. 2022, A&A, 664, A100 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  58. Parra, M., Petrucci, P. O., Bianchi, S., et al. 2024, A&A, 681, A49 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  59. Ponti, G., Fender, R. P., Begelman, M. C., et al. 2012, MNRAS, 422, L11 [NASA ADS] [CrossRef] [Google Scholar]
  60. Rahoui, F., Coriat, M., Corbel, S., et al. 2012, MNRAS, 422, 2202 [CrossRef] [Google Scholar]
  61. Rahoui, F., Coriat, M., & Lee, J. C. 2014, MNRAS, 442, 1610 [Google Scholar]
  62. Russell, D. M., Miller-Jones, J. C. A., Maccarone, T. J., et al. 2011, ApJ, 739, L19 [NASA ADS] [CrossRef] [Google Scholar]
  63. Russell, D. M., Markoff, S., Casella, P., et al. 2013, MNRAS, 429, 815 [NASA ADS] [CrossRef] [Google Scholar]
  64. Saikia, P., Russell, D. M., Bramich, D. M., et al. 2019, ApJ, 887, 21 [NASA ADS] [CrossRef] [Google Scholar]
  65. Sánchez-Sierras, J., & Muñoz-Darias, T. 2020, A&A, 640, L3 [EDP Sciences] [Google Scholar]
  66. Sánchez-Sierras, J., Muñoz-Darias, T., Motta, S. E., et al. 2023, A&A, 680, L16 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  67. Schulz, N. S., Kallman, T. E., Galloway, D. K., & Brandt, W. N. 2008, ApJ, 672, 1091 [Google Scholar]
  68. Sguera, V., Bazzano, A., Fiocchi, M., et al. 2021, ATel, 14354, 1 [NASA ADS] [Google Scholar]
  69. Shidatsu, M., Ueda, Y., Tazaki, F., et al. 2011, PASJ, 63, S785 [NASA ADS] [CrossRef] [Google Scholar]
  70. Smette, A., Sana, H., Noll, S., et al. 2015, A&A, 576, A77 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  71. Soria, R., Wu, K., & Johnston, H. M. 1999, MNRAS, 310, 71 [NASA ADS] [CrossRef] [Google Scholar]
  72. Stiele, H., & Kong, A. K. H. 2023, MNRAS, 522, 268 [CrossRef] [Google Scholar]
  73. Stover, R. J. 1981, ApJ, 248, 684 [NASA ADS] [CrossRef] [Google Scholar]
  74. Tomsick, J. A., Fuerst, F., & Belloni, T. 2015, ATel, 7962, 1 [NASA ADS] [Google Scholar]
  75. Ueda, Y., Yamaoka, K., & Remillard, R. 2009, ApJ, 695, 888 [NASA ADS] [CrossRef] [Google Scholar]
  76. Vernet, J., Dekker, H., D’Odorico, S., et al. 2011, A&A, 536, A105 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  77. Wu, K., Soria, R., Hunstead, R. W., & Johnston, H. M. 2001, MNRAS, 320, 177 [NASA ADS] [CrossRef] [Google Scholar]
  78. Zdziarski, A. A., Poutanen, J., Mikolajewska, J., et al. 1998, MNRAS, 301, 435 [NASA ADS] [CrossRef] [Google Scholar]
  79. Zdziarski, A. A., Ziółkowski, J., & Mikołajewska, J. 2019, MNRAS, 488, 1026 [Google Scholar]

Appendix A: H2013 and S2021b Spectra

thumbnail Fig. A.1.

UVB, VIS, NIR spectrum of the epoch H2013.

thumbnail Fig. A.2.

UVB, VIS, NIR spectrum of the epoch S2021b.

Appendix B: Table

Table B.1.

EW, V/R ratio and FWZI of the most prominent emission lines.

All Tables

Table 1.

Journal of observations.

In the text
Table 2.

Best-fit indices obtained from fitting the optical to NIR SEDs.

In the text
Table B.1.

EW, V/R ratio and FWZI of the most prominent emission lines.

In the text

All Figures

thumbnail Fig. 1.

Swift-XRT HID during the 2021 outburst of GX 339–4. The epochs S2021a and S2021b are marked with up- and down-pointing triangles (overlapping in the figure), correspondingly. The hard state epochs, H2013 and H2015, are marked with a circle and a square, respectively.
In the text
thumbnail Fig. 2.

Fit to the dereddened continuum of the four epochs. The fit excludes regions with strong telluric absorptions. The symbols, coded as in the previous figures, mark the break location.
In the text
thumbnail Fig. 3.

UVB, VIS and NIR normalised spectra of the soft state S2021a (red) and hard state H2015 (cyan) epochs.
In the text
thumbnail Fig. 4.

Line profiles of the most intense features across all four epochs, normalised with respect to their adjacent continuum. The velocities are given in the observer’s rest frame.
In the text
thumbnail Fig. 5.

Normalised hard state spectra centred on the most prominent line profiles. The left panel shows the flat-top profiles of Hα, He I-10830 and Paβ during epoch H2013. The right panel highlights the difference between the red-skewed, single-peaked profiles of Hα and He I-10830 and the double peak of He II-4686 during H2015.
In the text
thumbnail Fig. 6.

Balmer decrement (BD) against the ionisation ratio (Iratio) for each observing epoch.
In the text
thumbnail Fig. A.1.

UVB, VIS, NIR spectrum of the epoch H2013.
In the text
thumbnail Fig. A.2.

UVB, VIS, NIR spectrum of the epoch S2021b.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.