Multi-wavelength Observations of AT2019wey: a New Candidate Black Hole Low-mass X-ray Binary

Low-mass X-ray binaries (LMXBs) contain an accreting neutron star (NS) or black hole (BH) in orbit with a low-mass (≲2 M_⊙) companion star. Most of the known BH LMXBs were discovered by X-ray all-sky monitors (ASMs) during X-ray outbursts induced by instabilities in the accretion processes. The most sensitive X-ray ASM to date, the Monitor of All-sky X-ray Image (MAXI; Matsuoka et al. 2009), has a transient triggering threshold of 8 mCrab (1 mCrab = 2.4 × 10⁻¹¹ erg s⁻¹ cm⁻² over 2–10 keV) sustained for 4 days (Negoro et al. 2016). Due to the relatively shallow sensitivity of ASMs, the sample of LMXBs is biased toward nearby sources that exhibit bright X-ray outbursts.

Prior to 2020, the most sensitive all-sky X-ray imaging survey was carried out in 1990/1991 by ROSAT at 0.1–2.4 keV (Truemper 1982; Voges et al. 1999). It cataloged X-ray sources brighter than ∼10 μ Crab, providing the deepest X-ray all-sky reference at the time (Boller et al. 2016). Three decades after ROSAT, the dynamic X-ray sky is being surveyed by the eROSITA (0.2–10 keV; Predehl et al. 2021) and the Mikhail Pavlinsky ART-XC (4–30 keV; Pavlinsky et al. 2021) telescopes on board the Spektrum-Roentgen-Gamma (SRG) mission (Sunyaev et al. 2021). This planned four-year survey obtaining full-sky images created every six months is a powerful X-ray time-domain facility. The first eROSITA All-Sky Survey (eRASS1; 2019 December–2020 June) was sensitive to point sources down to ∼0.8 μCrab (Predehl et al. 2021).

On 2020 March 18, SRG discovered a new X-ray (∼1 mCrab) transient, SRGA J043520.9+552226 (=SRGE J043523.3+552234; Mereminskiy et al. 2020). It coincided with an optical (r ∼ 17.5) transient, AT2019wey, first reported by ATLAS (Tonry et al. 2019). This transient, bright at both X-ray and optical wavelengths and located at low Galactic latitude, (b = 53) was not present in the Palomar Observatory Sky Survey or the ROSAT catalog. We conducted an extensive follow-up campaign, revealing that AT2019wey is a Galactic LMXB with unique properties.

Yao et al. (2020a, hereafter Paper I) presented X-ray observations of AT2019wey from 2019 January to 2020 November, suggesting that AT2019wey is an LMXB with a BH or NS accretor. In this work, we present multi-wavelength observations of AT2019wey. We conclude that the compact object is probably a BH and the companion star must be of low mass (<1 M_⊙). We therefore call AT2019wey a candidate BH LMXB. This class of objects and the classification of their X-ray states is reviewed in McClintock & Remillard (2006), Remillard & McClintock (2006), Belloni et al. (2011), Zhang (2013), and Tetarenko et al. (2016).

The paper is organized as follows. The association between the optical and X-ray transients is outlined in Section 2. We present optical and ultraviolet (UV) photometry in Section 3, optical and near-infrared (NIR) spectroscopy in Section 4, and radio observations in Section 5. We discuss the nature of the source in Section 6, and summarize our findings and conclusions in Section 7.

Throughout this paper, times are reported in UT. Optical magnitudes are reported in the AB system. We adopt the reddening law of Cardelli et al. (1989) with R_V = 3.1.

We constructed the optical light curve using the forced-photometry services of ZTF ¹² (Masci et al. 2019) and ATLAS ¹³ (Smith et al. 2020). We obtained Gaia photometry from the Gaia alerts page. ¹⁴ See Table 5 for the ZTF photometry.

The upper panel of Figure 2 shows the ZTF, ATLAS, and Gaia light curves of AT2019wey. Over the first two weeks, the light curve rose to r_ZTF = 17.3 mag. After that, the light curve displayed small amplitude (≲0.3 mag) variability for more than 300 days. The lack of photometry between MJD ∼ 58980 to MJD ∼ 59040 is due to the source being in the day sky. On September 9 and 13 we undertook continuous observations as part of the ZTF "deep-drilling" program (Kupfer et al. 2021). On each day, ≈130 r-band exposure frames were obtained.

Zoom In Zoom Out Reset image size

On 2020 July 23 (MJD 59053), we obtained high-speed photometry in the SDSS g and i band using the Caltech HIgh-speed Multi-color camERA (CHIMERA; Harding et al. 2016) on the 200-inch Hale telescope of the Palomar Observatory. We operated the detectors using the 1 MHz conventional amplifier in frame-transfer mode with a frame exposure time of 1 s, and obtained 3300 frames in each filter. We reduced the data with a custom pipeline. ¹⁵ Figure 3 shows the CHIMERA light curve. AT2019wey appears to exhibit intra-night variability of ∼0.1 mag.

Zoom In Zoom Out Reset image size

We ran a periodicity search on the CHIMERA and the ZTF deep-drilling data sets using the analysis of variance (AOV) method (Schwarzenberg-Czerny 1998). ¹⁶ We used a frequency grid from 16 day⁻¹ to 500 day⁻¹ for the ZTF data and a frequency grid from 48 day⁻¹ to 40,000 day⁻¹ for the CHIMERA data. To see how the observational cadence affects the periodogram, we used the Lomb–Scargle algorithm (see a recent review by VanderPlas 2018) to compute the window function.

We define "significance" of a period as the maximum value in the periodogram divided by the standard deviation of values across the full periodogram. A possible period at 0.055 day (1.3 hr) at a significance of 9.2 can be seen in the ZTF periodogram (upper panel of Figure 4). We note that the 1.3 hr peak is mainly caused by the sinusoidal-like structure observed on September 19, not the dip-like structure observed on September 23. Since the data on September 19 and 23 do not follow the same trend as a function of phase (see lower panel of Figure 4), we consider the possible period at 1.3 hr to be spurious. No period above 8σ can be identified from the CHIMERA periodogram.

Zoom In Zoom Out Reset image size

We obtained UV observations of AT2019wey with the UltraViolet/Optical Telescope (UVOT; Roming et al. 2005) on board the Neil Gehrels Swift Observatory (Gehrels et al. 2004) from 2020 April to September. The UVOT data were processed using HEASoft. We extracted the photometry with uvotsource using a 3'' circular aperture. Background counts were estimated in a 10'' source-free circular aperture. AT2019wey was only marginally detected in April. Therefore, for the April data sets, we undertook photometry on co-added images.

In October 2020, we obtained U-band photometry using the Spectral Energy Distribution Machine (SEDM, Blagorodnova et al. 2018; Rigault et al. 2019) on the robotic Palomar 60 inch telescope (P60, Cenko et al. 2006). Data reduction was performed using the FPipe pipeline (Fremling et al. 2016). The UVOT and SEDM photometry are presented in Table 6 and a shown in the middle panel of Figure 2.

We identify the following features at redshift z = 0 in all of our spectra: Balmer absorption lines, Ca II H and K lines, the Na I D doublet, diffuse interstellar band (DIB) λ5780, λ6283 absorption features, and the Balmer jump (Figure 5, 6). He II λ4686 emission seems to be detected in the spectra obtained on July 31, August 14, and September 20. We conclude that AT2019wey is a transient of Galactic stellar origin.

Zoom In Zoom Out Reset image size

From March to September, the hydrogen profile clearly changed (Figure 6). Figure 7 presents the velocity profiles of Balmer lines in the March 23 and the September 12 spectra. On March 23, we observed a relatively narrower (FWHM ∼ 1200 km s⁻¹) emission component in the middle of a rotationally broadened (FWHM ∼ 2700 km s⁻¹) shallow absorption trough. At the same epoch, we also observed broad Hβ and Hγ absorption features with FWHM ∼ 2000–3000 km s⁻¹. There was a marginal detection of narrow emission cores redshifted by ∼300–400 km s⁻¹ from the line center of the absorption troughs. On September 12, we observed flat-topped Hα in emission (∼400 km s⁻¹), while the Hβ and Hγ profiles were similar to the Hα profile on March 23. The variable Balmer features are discussed further in Section 6.4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The reddening of AT2019wey can be constrained to 0.8 < E(B − V) < 1.2 (Appendix C.1) using the equivalent width (EW) of the interstellar absorption lines. We find a lower limit to the distance of AT2019wey of D > 1 kpc using the velocities of the Na I doublet in the ESI spectrum (Appendix C.2). In addition, since AT2019wey is in the Galactic anti-center direction, the distance to AT2019wey is likely less than ∼10 kpc. Taken together, we conclude that the distance of AT2019wey is between ∼1 kpc and ∼10 kpc.

The NIR spectrum of AT2019wey is shown in Figure 8. Hydrogen emission lines of Paγ, Paβ, and Brγ are clearly distinguished. We tentatively attribute the emission lines around 1083 nm to double-peaked He I. No absorption lines or molecular bands from the secondary star can be identified. With an FWHM of ≈200–300 km s⁻¹, the velocities of NIR emission features are much narrower than the Hα line, hinting at different formation locations in the accretion disk.

Zoom In Zoom Out Reset image size

Figure 10 shows that on the L_radio–L_X diagram, the position of AT2019wey is above the region occupied by the majority of NS binaries and is closer to BH binaries. Therefore, the bright radio luminosity favors a BH accretor.

Zoom In Zoom Out Reset image size

We separate the multi-wavelength light curve of AT2019wey into five stages (see the bottom panel of Figure 2): (i) before MJD ∼ 58814, the source was in quiescence, (ii) from MJD ∼ 58814 to MJD ∼ 58880, the optical light curve exhibited a fast-rise linear-decay outburst, after which it settled onto a r-band flux of f_ν,r ∼ 315 μ Jy; around the same time, the X-ray flux rose to ∼1 mCrab, and stayed in the LHS, (iii) from MJD ∼ 58880 to MJD ∼ 59010, the optical and X-ray light curves stayed almost flat, (iv) From MJD ∼ 59010 to MJD ∼ 59080, AT2019wey exhibited a multi-wavelength brightening and the X-ray remained in the LHS (Paper I), (v) from MJD ∼ 59081 to MJD ∼ 59180, the source entered into the HIMS (Paper I). The optical stayed around f_ν,r ∼ 400 μ Jy and X-ray stayed around ∼20 mCrab (Paper I).

6.2.1. UV/optical–X-ray Correlation

During stage (iv), the X-ray and radio fluxes increased by a factor of ≳10 but in the optical/UV the increase was modest, between a factor of 1.3 and 2. During stages (iii) and (v), the source was stable and representative luminosities can be found in Table 3. For these two stages, following Russell et al. (2006), we link the UV/optical and X-ray luminosities as

$$\begin{eqnarray}&&{L}_{\mathrm{UV}/\mathrm{opt}}={{AL}}_{{\rm{X}}}^{\beta },\end{eqnarray} \tag{ 1 }$$

and find β ∼ 0.08 in r band, β ∼ 0.12 in g band, and 0.12 ≲ β ≲ 0.34 in the UV bands. Russell et al. (2006) derived A = 10^13.1±0.6, β = 0.61 ± 0.02 for a sample of 15 BH LMXBs and A = 10^10.8±1.4, β = 0.63 ± 0.04 for a sample of 8 NS LMXBs. As can be seen from Figure 11, over the distance range of 1 ≲ D ≲ 10 kpc, the inferred luminosities of AT2019wey are suggestive of an accreting BH system.

Zoom In Zoom Out Reset image size

Table 3. X-ray and Optical Luminosity of AT2019wey at Different Stages of the Multi-wavelength Evolution

Stage	Band	Luminosity	Comments
(iii)	r and g	4.0 × 1034 & 6.1 × 1034	Averaged between MJD ∼ 58880 and MJD ∼ 59010
(iii)	X-ray	1.0 × 1035	Averaged between MJD ∼ 58951 and MJD ∼ 58967
(v)	r and g	4.9 × 1034 & 8.4 × 1034	Averaged between MJD ∼ 59080 and MJD ∼ 59153
(v)	X-ray	(1.3–1.7) × 1036	Range of values from minimum (MJD∼59082) to maximum (MJD ∼ 59112)

Note. Luminosity is given in units of (D/5 kpc)² erg s⁻¹. X-ray column density corrected luminosity is given in 2–10 keV, assuming N_H = 5 × 10²¹ cm⁻². Optical luminosity has been corrected for extinction, adopting E(B − V) = 0.9.

Download table as:  ASCIITypeset image

6.2.2. Possible Mechanisms for the Optical Emission

The spectral energy distribution (SED) of AT2019wey is shown in Figure 12. The X-ray data are presented in Paper I and we briefly summarize the X-ray spectra in Section 6.3.1. In Section 6.3.2, based on radio data, we conclude that jet emission is unlikely to be the dominant mechanism in the optical. In Section 6.3.3, we show that the UV/optical emission during stage (iii) originates from the intrinsic emission of a truncated accretion disk. In Section 6.3.4, we show that the UV/optical emission during stage (v) arises from X-ray reprocessing.

Zoom In Zoom Out Reset image size

6.3.1. The X-ray SED

Briefly speaking, the X-ray spectrum observed in stage (iii) can be described by an absorbed power-law with photon index Γ = 1.8. In stages (iv) and (v), the X-ray spectrum can be fitted with a combination of disk–blackbody (diskbb, Shakura & Sunyaev 1973; Mitsuda et al. 1984) and power-law components (Paper I). On August 14, 21, and 28, the fitted models have Γ ∼ 1.9 and inner-disk temperature T_disk ∼ 0.21 keV ∼ 2.4 × 10⁶ K. The inner-disk radius is

$$\begin{eqnarray}&&{R}_{{\rm{in}}}\sim (360-470){\left(\displaystyle \frac{\cos i}{1}\right)}^{-1/2}\left(\displaystyle \frac{D}{5\,{\rm{kpc}}}\right)\,{\rm{km}}.\end{eqnarray} \tag{ 2 }$$

On September 20, the soft X-ray flux reached a local maximum in the HIMS, where the PL softened to Γ = 2.3 and the inner-disk temperature increased to T_disk ∼ 0.29 keV ∼ 3.4 × 10⁶ K, while the inner-disk radius remains at ∼400 km. The fitted T_disk and R_in are typical for thermal emission from a truncated accretion disk observed in the LHS and HIMS of BH LMXBs (Done et al. 2007). Denoting the innermost stable circular orbit radius as R_ISCO = 6GM/c² and the Schwarzschild radius as R_S = 2GM/c², then R_in ∼ 15R_S ∼ 5R_ISCO for a 10 M_⊙ non-spinning black hole.

6.3.2. The Radio SED

6.3.3. UV/Optical Emission in the Dim LHS

In Figure 13, we show the UV/optical data and the best-fit X-ray model in the dim LHS (stage iii) in orange. The low level of X-ray flux (compared to that in the UV/optical) suggests that there is not enough X-ray flux to illuminate the outer accretion disk. As a result, the UV/optical probably comes from the intrinsic thermal emission of an accretion disk.

Zoom In Zoom Out Reset image size

To obtain a constraint on the outermost annulus of the accretion disk, we compute a set of simple blackbody models (upper panel of Figure 13). We adopt the 11,000 K blackbody as an approximation of the outer disk annulus and compute a set of diskbb models to obtain a lower limit to the inner-disk radius (and an upper limit to the inner-disk temperature). The dotted line in the lower panel of Figure 13 suggests T_in < 4.8 × 10⁵ K and R_in > 3.3 × 10³ km ∼ 38R_ISCO ∼ 114R_S.

Similar SED shapes have been observed in the LHS of a few BH LMXBs, including XTE J1118+480 (R_in = 300R_S; Yuan et al. 2005) and Swift J1753.5−0127 (R_in > 100R_S; Froning et al. 2014). The observed SED of AT2019wey in the dim LHS fits into the advection-dominated accretion flow (ADAF; Narayan & Yi 1994, 1995) model of a hot accretion flow around a BH, which is predicted at low accretion rates (see reviews by Done et al. 2007; Yuan & Narayan 2014; Poutanen & Veledina 2014). If so, the X-ray PL comes from a high-temperature flow in the central regions close to the BH, while the UV/optical thermal component comes from a geometrically thin, optically thick accretion disk truncated far from the ISCO (Yuan & Narayan 2014).

6.3.4. UV/Optical Emission in the HIMS

The hydrogen lines in AT2019wey display both broad absorption and emission components (Section 4.1). This behavior is reminiscent of some LMXBs and CVs, where the hydrogen absorption and emission lines are thought to arise from different layers of the viscous accretion disk (Horne & Marsh 1986; La Dous 1989; Warner 1995). In a few BH LMXBs, such as GRO J1655−40 (Soria et al. 2000), GRO J0422+32 (Callanan et al. 1995), XTE J1118+480 (Dubus et al. 2001; Torres et al. 2002), and Swift J1753.5−0127 (Rahoui et al. 2015), double-peaked Hα was observed. The single-peaked hydrogen line profile of AT2019wey is similar to that observed in MAXI J1836−194 (Russell et al. 2014), suggesting a binary system viewed close to face-on. This is in agreement with the low inclination (i ≲ 30°) constraint from modeling the X-ray reflection spectrum (Paper I).

In Section 6.3.4 we have shown that in the HIMS, the UV/optical emission comes from the reprocessing of inner-disk and coronal emission. Irradiation of the outer disk may form a thin temperature-inversion layer on the disk surface (Tuchman et al. 1990). This naturally explains the enhanced Balmer emission lines observed during stage (iv) and stage (v).

Most BH LMXBs show strong He II emission during their outbursts (Zurita et al. 2002; Kaur et al. 2012; Jiménez-Ibarra et al. 2019; Russell et al. 2014). A lack of significant He II was observed in the optical spectra of AT2019wey. This might also be present in the 2009 outburst of XTE J1752−223 (Torres et al. 2009) and the 2021 outburst of XTE J1859+226 (Bellm 2021, Bellm, E. C. et al., 2021 in preparation). We note that the He II recombination line was also not significantly detected in the outburst spectra of a few CVs (Morales-Rueda & Marsh 2002). A possible explanation for this is that the number of photons with energies between 54 eV (the ionization potential of He⁺) and 280 eV (the ionization potential of the carbon K-edge) is not large enough (Patterson & Raymond 1985).

We conducted an archival search of optical photometry at the position of AT2019wey. The source was not detected by historical optical surveys, including the Palomar Observatory Sky Survey I (POSS-I, Minkowski & Abell 1963), the Second Palomar Observatory Sky Survey (POSS-II, Reid et al. 1991), SDSS, and the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS, PS1) DR1 (Flewelling et al. 2020; Waters et al. 2020), the intermediate Palomar Transient Factory (iPTF; Law et al. 2009; Rau et al. 2009), and the ZTF. We list 5σ upper limits in Table 4.

AT2019wey was not detected in any archival radio database. The NRAO VLA Sky Survey (NVSS, Condon et al. 1998) provides an upper limit of 2 mJy at 1.4 GHz in 1993–1996. The Karl G. Jansky Very Large Array Sky Survey (VLASS, Lacy et al. 2020) provides a 3σ upper limit of 0.40 mJy at 2–4 GHz in March 2019.

The EW of interstellar absorption lines has been observed to be correlated with the amount of reddening. To estimate the extinction of AT2019wey, we produced a summed spectrum from the LRIS and ESI spectra. We did not include DBSP spectra in this analysis since the CCD malfunction resulted in non-astrophysical structures between 5750 Å and 6200 Å in the continuum. This problem prevents EW of spectral lines from being accurately determined from DBSP spectra. The EW of DIB λ λ5780, λ6283, and Na I D lines were measured from the summed spectrum. As a result, we got EW(λ5780) = 0.56 ± 0.02 Å, and EW(λ6283) = 1.55 ± 0.02 Å. These can be converted to E(B − V) = 0.92 ± 0.02 and 1.23 ± 0.02 using relations presented by Yuan & Liu (2012). We got EW( Na I D) = 1.84 ± 0.02 Å, which can be converted to E(B − V) = 2.01 ± 0.38 using the relation in Poznanski et al. (2012).

The inferred E(B − V) values are greater than the total Galactic extinction of E(B − V) = 0.88 (Schlafly & Finkbeiner 2011). However, we note that at the measured EW, the calibration uncertainty is large. From Yuan & Liu (2012, upper panels of Figure 4) and Poznanski et al. (2012, bottom panel of Figure 9), we infer that E(B − V) toward AT2019wey should be ≳0.8.

We also attempt to infer the extinction by assuming that the 6000–10,000 Å March 23 LRIS spectrum is in the Rayleigh–Jeans (R–J) tail of a blackbody (f_λ ∝ λ⁻⁴ when h ν ≪ kT), which yields E(B − V) = 1.29 and a blackbody radius (R_bb) of

$$\begin{eqnarray}&&{R}_{\mathrm{bb}}=(4.5\times {10}^{10}\,\mathrm{cm})\left(\displaystyle \frac{D}{5\,\mathrm{kpc}}\right){\left(\displaystyle \frac{{T}_{\mathrm{bb}}}{5.0\times {10}^{4}\,{\rm{K}}}\right)}^{-1/2}\end{eqnarray} \tag{ C1 }$$

Note that this is likely an overestimate of the true extinction (and a lower limit of the outer disk radius), since the optical is only in the R–J limit when kT ≫ 2 eV (T ≫ 2 × 10⁴ K). For instance, for an extinction of E(B − V) ∼ 0.9, we have

$$\begin{eqnarray}&&{R}_{\mathrm{bb}}=(1.0\times {10}^{11}\,\mathrm{cm})\left(\displaystyle \frac{D}{5\,\mathrm{kpc}}\right){\left(\displaystyle \frac{{T}_{\mathrm{bb}}}{1.1\times {10}^{4}\,{\rm{K}}}\right)}^{-1/2}\end{eqnarray} \tag{ C2 }$$

In Appendix C.1, we find that AT2019wey should have an extinction of 0.8 ≲ E(B − V) ≲ 1.2. If this is from diffuse interstellar absorption, the distance of AT2019wey should be greater than 1 kpc using the map of STructuring by Inversion the Local Interstellar Medium (Stilism ¹⁸ ; Capitanio et al. 2017).

We are able to put a lower limit to the distance using the velocity of the Na I D doublets in the ESI spectrum, given that the lines arise from interstellar absorption by a dust cloud along the line-of-sight to AT2019wey. The velocities of D1 and D2 lines were measured to be −11.75 ± 1.13 km s⁻¹ and −9.83 ± 1.13 km s⁻¹, respectively (see Figure 14). Assuming that the velocity of the dust cloud follows Galactic rotation, we have

$$\begin{eqnarray}&&{V}_{\mathrm{obs},\ {\rm{r}}}={Ad}\sin (2l)\end{eqnarray} \tag{ C3 }$$

where A = 15.3 ± 0.4 km s⁻¹ kpc⁻¹ is the Oort A constant (Bovy 2017), l = 1512 is the Galactic longitude of AT2019wey, and d is distance to the dust cloud. Therefore, Equation (C3) gives d = 0.83 kpc.

Zoom In Zoom Out Reset image size