A Broadband Look at the Old and New ULXs of NGC 6946

We reprocessed the NuSTAR observations using the NuSTAR Data Analysis Software (NuSTARDAS; v1.7.1) routine nupipeline. We extracted source and background spectra and light curves from the NuSTAR science data with the NuSTARDAS task nuproducts, using circular source extraction regions with radius 45'' and background regions of radius 60'' on the same chip. Spectra were grouped into 20 counts per bin to allow for χ² statistics to be used in spectral model fitting. A 3σ upper limit for the initial non-detection of ULX-4 was found using the FTOOLS task sosta, using a 45'' radius source region and a 60'' radius background region.

The XMM-Newton data were reduced using v16.1.0 of the XMM-Newton Science Analysis System (SAS) software and up-to-date CALDB as of 2018 June, producing calibrated event files with epproc and emproc and removing periods of background flaring—several observations were heavily affected by background flaring so they were removed from this analysis. We extracted data products using 20'' radius circular source regions and 40'' radius circular background regions located on the same chip at a similar distance from the readout node. We selected FLAG == 0 && PATTERN < 4 events for the EPIC-pn camera, and PATTERN < 12 for the EPIC-MOS cameras. In all cases, spectra were grouped into 20 counts per bin as for NuSTAR. Redistribution matrices and auxiliary response files were generated with the tasks rmfgen and arfgen, respectively. 3σ flux upper limits for non-detections were determined using the SAS task eregionanalyse, again using 20'' radius circular source regions and 40'' radius circular background regions.

Previous Chandra observations of NGC 6946 were reduced using the task chandra_repro of v4.7.7 of the Chandra Interactive Analysis of Observations (CIAO) software. The srcflux task was used to find the 0.3–10 keV 3σ flux upper limits of non-detections, using a 3'' radius source region surrounded by a 20'' radius annulus for the background.

Swift monitored NGC 6946 closely during 2017, and we use all observations from that year. X-ray source products were generated using the FTOOLS task xrtpipeline. The spectrum for ULX-3 was taken using a 45'' radius source region and a 70'' radius background region located outside of the galaxy, and the spectrum was grouped to 20 counts per bin as for NuSTAR. 3σ flux upper limits for non-detections in Swift observations were determined using the FTOOLS task sosta, using a 20'' radius source region (to avoid contamination by a nearby source) and a 70'' radius background region as for ULX-3. Magnitude lower limits for optical/ultraviolet non-detections were determined using the FTOOLS task uvotsource.

Optical photometry was performed on five observations of NGC 6946 taken using the WFPC2 and WFC3/IR cameras on board the HST during three different epochs before 2017, which cover the region of the galaxy in which ULX-4 is contained. We use observations in the F547M, F606W, and F814W optical bands with WFPC2 (proposal IDs 8591, 8597, and 8599, respectively; the F656N band is also observed, but the resolution is insufficiently good to characterize individual sources in this band) and the F110W and F128N near-infrared bands with WFC3/IR (proposal ID 14156). We retrieved preprocessed images from the Hubble Legacy Archive created from multiple exposures combined using the MultiDrizzle routine.

We corrected the HST astrometry using the USNO star catalog and the Image Reduction and Analysis Facility tools ccfind, ccmap and ccupdatewcs, and combined the 90% confidence error circle of the resulting HST position errors in quadrature with the XMM-Newton source position error, resulting in a 09 90% error circle around the source position, which we used to identify potential counterparts. We performed aperture and point-spread function (PSF)-fitting photometry on these potential optical counterparts using the DAOPHOT II/ALLSTAR software package (Stetson 1987), using the values from aperture photometry where a good PSF fit could not be found, and placed magnitude limits based on a combination of the read noise, dark current, and sky background where a source could not be detected at all.

We fitted the XMM-Newton spectra of ULX-1 and ULX-2 from observation 0794581201 with an absorbed power-law model, using two tbabs absorption components, one frozen to the Galactic value of N_H = 1.84 × 10²¹ cm⁻² and the other allowed to vary. In the case of ULX-1, a power-law model was not sufficient to produce a good fit, with χ²/dof = 131.9/90 and the fit showing significant soft residuals at ∼1 keV. These soft residuals are known to be a common feature in the spectra of ULXs and bright X-ray binaries (e.g., Bauer & Brandt 2004; Carpano et al. 2007; Middleton et al. 2015b), and found in other ULXs to be a combination of emission and absorption features related to powerful outflowing winds (including in NGC 6946 ULX-3; Pinto et al. 2016). We used an additional Gaussian component to empirically fit these soft residuals, which resulted in a very large statistical improvement and provided an acceptable fit (χ²/dof = 84.5/87). Using the best-fitting model for each source, we calculated the absorbed 0.3–10 keV flux of both objects. We do not correct for absorption since extending steep power laws, which are empirical rather than physical models, to low energies is likely to overestimate the true flux of the object beneath the absorption—unabsorbed fluxes using such models can be a factor of 2 or 3 higher than those for more physically motivated models, and the amount of absorption is itself model-dependent, so we make fewer assumptions by just considering the observed, absorbed flux. For comparison, we also fitted both spectra with an absorbed multicolor disk model. Although this was able to produce a statistically acceptable fit, the Gaussian representing the soft residuals in ULX-1 contributes to a physically unreasonable portion of the spectrum, broadening to fit the majority of the soft emission, and the residuals for ULX-2 are even less well characterized than for a power-law fit. Therefore, a steep power law seems to be the best empirical model for both spectra. We present the spectral fit results for these two sources in Table 3 and the best-fitting model plots in Figure 2.

Zoom In Zoom Out Reset image size

Table 3.  The ULX-1 and ULX-2 Spectral-fitting Results for XMM-Newton Observation 0794581201

Source	NHa	Γ/Tin	pb	Tin	Eline	σline	χ2/dof	F0.3–10keVc
	(1021 cm−2)	(-/keV)		(keV)	(keV)	(keV)		(10−14 erg s−1)
ULX-1	tbabs∗tbabs∗	(powerlaw			+ Gauss)
	${4.2}_{-0.6}^{+0.7}$	4.5 ± 0.3	⋯	⋯	⋯	⋯	131.9/90	⋯
	${2.3}_{-0.7}^{+0.8}$	${3.6}_{-0.3}^{+0.4}$	⋯	⋯	${0.90}_{-0.07}^{+0.04}$	${0.13}_{-0.04}^{+0.05}$	84.5/87	9.8 ± 0.4
	tbabs∗tbabs∗	(diskbb			+ Gauss)
	0.6 ± 0.4	0.31 ± 0.03	⋯	⋯	⋯	⋯	179.7/90	⋯
	<8.2	0.6 ± 0.1	⋯	⋯	0.7 ± 0.1	${0.29}_{-0.05}^{+0.06}$	99.1/87	⋯
ULX-2	tbabs∗tbabs∗	powerlaw
	${2.9}_{-1.0}^{+1.1}$	${3.6}_{-0.3}^{+0.4}$	⋯	⋯	⋯	⋯	44.9/42	7.8 ± 0.5
	tbabs∗tbabs∗	(diskbb		+ diskbb)
	<0.3	0.43 ± 0.04	⋯	⋯	⋯	⋯	62.6/42	⋯
	${3.3}_{-0.2}^{+0.3}$	${0.17}_{-0.04}^{+0.06}$	⋯	${0.7}_{-0.1}^{+0.2}$	⋯	⋯	38.1/40	⋯
	tbabs∗tbabs∗	diskpbb
	<0.8	0.7 ± 0.1	<0.56	⋯	⋯	⋯	52.6/41	⋯

Notes.

^aThe column density for the absorption component allowed to vary, with the first component frozen to the Galactic value of N_H = 1.84 × 10²¹ cm⁻². ^bThe radial dependence of disk temperature, with a hard lower limit of 0.5. ^cThe absorbed flux in the 0.3–10 keV band.

Download table as:  ASCIITypeset image

Both sources are found to be very soft and at a low enough flux to be a little below the ULX threshold luminosity (both have L_X ∼ 6–7 × 10³⁸ erg s⁻¹), consistent with these sources being persistently bright Eddington threshold objects that only occasionally reach the ULX luminosity regime. ULX-1 is consistent with a steep power-law spectral shape with soft residuals, as previously found in Earnshaw & Roberts (2017), where it was suggested that it is a super-Eddington accreting source viewed at a high accretion rate and/or at a high inclination, with most of the high-energy central emission obscured and down-scattered by the surrounding wind and outer disk. While the most typical examples of ultraluminous supersoft sources have spectra that are almost entirely dominated by thermal emission (e.g., Urquhart & Soria 2016), the very steep power-law tail may come from a very small amount of visible central emission, making ULX-1 a source on the borderline between the soft ultraluminous regime and the thermal-dominated ultraluminous supersoft sources. ULX-2 lies on an EPIC-pn chip gap, so we only have data from the EPIC-MOS cameras. Its spectrum, while similarly steep to ULX-1 and statistically well fitted by a simple absorbed power law, appears on visual inspection to have a two-component shape to its residuals not dissimilar to that found in the soft ultraluminous accretion regime in ULXs (e.g., Gladstone et al. 2009). However, likely owing to the low signal-to-noise and limited bandpass, fitting the spectrum with two thermal components rather than a power law does not provide a significant improvement in the quality of fit. Still, it is possible that this source is also a super-Eddington accreting system like ULX-1, only with typical luminosity below the standard ULX definition.

We began by fitting the combined XMM-Newton and NuSTAR spectrum of ULX-3, with the cross calibration fixed to unity, with an absorbed power-law model as for ULX-1 and ULX-2. Like ULX-1, its residuals are dominated by the likely wind-related feature at ∼1 keV, so we fitted this feature with a Gaussian component as before, which yields a statistically acceptable fit to the spectrum (χ²/dof = 332.8/289). With a photon index of Γ = 2.51 ± 0.05, ULX-3 is the softest ULX with a broadband X-ray spectrum detected by NuSTAR studied so far. We test for a spectral turnover at high energies as seen in other ULXs, a key indicator of super-Eddington accretion. However, we find that there is no significant improvement in fit with a cutoff power-law model over a power law without a cutoff (Δχ² = 1.2 for 1 dof). We simulated 1000 spectra with the cutoff power-law parameters we measure, and found that for a spectrum this soft, NuSTAR is unable to significantly detect a cutoff and correctly constrain its parameters. We show the best-fitting model parameters for the different models in Table 4 and show the spectrum and residuals in Figure 3.

Zoom In Zoom Out Reset image size

Table 4.  The Joint ULX-3 Spectral-fitting Results for XMM-Newton Observation 0794581201 and NuSTAR Observation 90302004004, and the Results for Swift Observation 00010130008 and NuSTAR Observation 90302004002

NHa	Γ	Ecutoff	Tin,bb	Tin,pbb	pb	Eline	σline	χ2/dof	F0.3–10keVc
(1021 cm−2)		(keV)	(keV)	(keV)		(keV)	(keV)		(10−13 erg s−1)
XMM-Newton + NuSTAR epoch 2
tbabs∗tbabs∗	(powerlaw					+ Gauss)
1.2 ± 0.1	2.71 ± 0.04	⋯	⋯	⋯	⋯	⋯	⋯	489.8/292	8.7 ± 0.1
0.8 ± 0.1	2.51 ± 0.05	⋯	⋯	⋯	⋯	${0.90}_{-0.06}^{+0.04}$	${0.16}_{-0.03}^{+0.04}$	332.8/289	9.0 ± 0.1
tbabs∗tbabs∗	(cutoffpl					+ Gauss)
0.6 ± 0.3	${2.4}_{-0.2}^{+0.1}$	>9.95	⋯	⋯	⋯	${0.85}_{-0.11}^{+0.06}$	${0.20}_{-0.04}^{+0.07}$	331.6/288	⋯
tbabs∗tbabs∗			(diskbb	+ diskpbb		+ Gauss)
0.7 ± 0.3	⋯	⋯	${0.24}_{-0.02}^{+0.03}$	2.1 ± 0.2	<0.52	${0.97}_{-0.04}^{+0.03}$	0.10 ± 0.01	324.8/286	⋯
Swift + NuSTAR epoch 1
tbabs∗tbabs∗	powerlaw
2 ± 2	2.8 ± 0.2	⋯	⋯	⋯	⋯	⋯	⋯	48.6/49	8.5 ± 0.8
tbabs∗tbabs∗	cutoffpl
<2.8	${2.2}_{-0.3}^{+0.6}$	${8}_{-3}^{+63}$	⋯	⋯	⋯	⋯	⋯	45.2/48	⋯
tbabs∗tbabs∗			(diskbb	+ diskpbb)
<6.2	⋯	⋯	${0.4}_{-0.2}^{+0.1}$	${2.1}_{-0.7}^{+0.6}$	<0.9	⋯	⋯	43.1/46	⋯

Notes.

^aThe column density for the absorption component allowed to vary, as in Table 3. ^bThe radial dependence of disk temperature, as in Table 3. ^cThe absorbed flux in the 0.3–10 keV band.

Download table as:  ASCIITypeset image

We used the best-fitting power-law model to calculate the absorbed 0.3–10 keV flux of ULX-3 during this observation, which we find to be (9.1 ± 0.2) × 10⁻¹³ erg cm⁻² s⁻¹, corresponding to a luminosity of (6.5 ± 0.1) × 10³⁹ erg s⁻¹. This is broadly consistent with previously calculated luminosities for this source when accounting for lower distances to the galaxy used previously (e.g., Middleton et al. 2015a), and it therefore continues to be a persistent ULX.

While there is no statistical case for a more complex model, we do consider the scenario that this ULX has a similar spectrum to other ULXs observed with NuSTAR in the past and fit it with a typical diskbb+diskpbb model—the soft diskbb component originating in an outer thin disk/soft outflowing wind, and the harder diskpbb component originating in the geometrically thick inner disk (Walton et al. 2014; Middleton et al. 2015a). We find that fitting the spectrum in this manner yields an acceptable fit (χ²/dof = 324.8/286), but also hints at a hard excess as often seen in hard ULXs at energies >20 keV (Walton et al. 2014, 2015b; Mukherjee et al. 2015), which cannot be entirely accounted for by a maximally broadened-disk component at high energies. This hard excess feature is not significant in these observations of ULX-3, but if it is the case that this source is spectrally similar to other ULXs with an excess present, this may be an additional reason why we do not see evidence of a turnover in the spectrum, as the entire spectrum is sufficiently soft that the strength of a turnover feature would be counteracted by a steep power-law excess at the energies we observe.

There is no simultaneous XMM-Newton observation for the first NuSTAR observation, though there is a Swift observation the following day, so we fit these observations together to get our best possible picture of the broadband spectrum during the first NuSTAR epoch. The amount of Swift data is insufficient to determine whether the soft residuals that we fitted with a Gaussian in the XMM-Newton data are contributing to the soft emission in the first observation as well, so the best comparison we can make is to the power-law-only fit to the second observation. We find that both the flux and spectral hardness are consistent with the second epoch, with no evidence for any significant variability on a ten-day timescale. While the NuSTAR data appears to hint at a spectral turnover at ∼8 keV on a visual inspection of the residuals, a cutoff power law offers only a small improvement in χ² over a regular power-law model (Δχ² = 3.4 for 1 dof). We show the best-fitting model parameters in Table 4 and show the spectrum and residuals in Figure 4.

Zoom In Zoom Out Reset image size

Our weaker constraints on the soft part of the spectrum in the first epoch limit our ability to comment on any variability of ULX-3 between the two observations. However, we can investigate the short-term variability of ULX-3 by creating an EPIC-pn power spectrum for the XMM-Newton observation, averaging the periodograms of 70 segments of length 601 s (i.e., 8192 time bins at the 73.4 ms time resolution of the EPIC-pn instrument). We find the power spectrum to exhibit red noise, as expected to be found in accreting systems (see Figure 5), but find no evidence for any quasi-periodic oscillations (QPOs) at ∼10⁻² Hz like those reported for earlier observations in Rao et al. (2010). We can rule out the presence of a 8.5 mHz QPO, as previously observed, at the 5σ level for this observation. We also do not significantly detect any other QPOs in the power spectrum. However, the fractional rms variability of ULX-3 is 40% ± 4% over the 1–100 mHz range, showing that ULX-3 continues to be highly variable and its spectrum remains bright and steep as it has previously been observed, indicating that the QPOs are possibly a transient feature in an otherwise persistent accretion state.

Zoom In Zoom Out Reset image size

ULX-4 is a new transient source. It was not detected in the first NuSTAR observation, but was strongly detected by both telescopes in the second joint XMM-Newton and NuSTAR observation.

As for the other sources, we fitted the combined XMM-Newton and NuSTAR spectrum of ULX-4 with an absorbed power-law model. We found that no absorption component is required in addition to the Galactic column, so our models for this source only contain a single tbabs component frozen to the Galactic value, which is somewhat unusual for ULXs, which usually show evidence of a local absorption column. This source has a very hard (Γ ∼ 1) power-law-shaped spectrum, and the fit is greatly improved (Δχ² = 15.8/1 dof) by the presence of a cutoff at ∼11 keV. We checked the significance of this cutoff by simulating a power-law spectrum 10,000 times and finding the improvement in χ² of a cutoff power-law model compared to a power law due to random fluctuations in the spectrum. We found that 0.02% of simulations had Δχ² ≥ 15.8, making the cutoff that we detect in the source spectrum significant to ∼3.5σ. The spectrum can also be well fitted with a very hot multicolor disk blackbody model with kT_in = ${4.3}_{-0.5}^{+0.6}$ keV, with no need for broadening using a diskpbb model—although over the portion of the spectrum we see, this is functionally very similar to the cutoff power-law model. We find the time-averaged 0.3–10 keV flux to be 3.2 × 10⁻¹³ erg cm⁻² s⁻¹, placing this new source just inside the ULX regime at L_X = (2.27 ± 0.07) × 10³⁹ erg s⁻¹. We show the joint XMM-Newton and NuSTAR spectrum in Figure 6, and the spectral-fitting results for ULX-4 in Table 5.

Zoom In Zoom Out Reset image size

Table 5.  The ULX-4 Spectral-fitting Results for XMM-Newton Observation 0794581201 and NuSTAR Observation 90302004004

Γ/Tin	Ecut	F0.3–10keV	F3–20keV	χ2/dof
(-/keV)	(keV)	(10−13 erg cm−2 s−1)
tbabs∗powerlaw
1.00 ± 0.05	⋯	3.2 ± 0.1	3.9 ± 0.3	174.2/143
tbabs∗cutoffpl
0.7 ± 0.1	${11}_{-4}^{+9}$	3.2 ± 0.1	3.5 ± 0.2	158.4/142
tbabs∗diskbb
${4.3}_{-0.5}^{+0.6}$	⋯	3.0 ± 0.1	3.2 ± 0.1	158.7/142

Download table as:  ASCIITypeset image

We examined Swift-XRT data both immediately before, between, and after the two NuSTAR observations, as well as into the remainder of 2017. Due to the short observations and the hardness of the source spectrum, ULX-4 is not detected in any of these observations and we are unable to place strong limits on the 0.3–10 keV flux during the period over which the source appears beyond that it undergoes at least a factor >1.4 increase in flux between the two NuSTAR epochs. Similarly, putting an upper limit on the 3–20 keV flux from the initial NuSTAR non-detection (F_3–20keV < 6.23 × 10⁻¹⁴ erg cm⁻² s⁻¹) gives us a factor >5 increase in flux for that energy band.

In order to confirm that the detection by XMM-Newton and NuSTAR is the first appearance of ULX-4, we plot a long-term light curve from archival XMM-Newton and Chandra observations (Figure 7), assuming a similar spectral shape to the one we observe here of Γ = 1 and Galactic absorption. This assumption gives conservative flux upper limits; should the source have previously been in a softer state, these upper limits would be lower. We find no previous detections of ULX-4—Chandra observations place the strongest limits on the 0.3–10 keV flux, with each upper limit below F_X < 3.5 × 10⁻¹⁵ erg cm⁻² s⁻¹, or L_X < 2.5 × 10³⁷ erg s⁻¹. This implies a factor ≳80 increase in flux between its detection in 2017 and when the source was previously observed by Chandra in 2016. Ten days after the second NuSTAR epoch, a further observation with Chandra does not detect the source either, implying an upper limit on the 0.3–10 keV flux of F_X < 1.1 × 10⁻¹⁴ erg cm⁻² s⁻¹, or L_X < 7.8 × 10³⁷ erg s⁻¹, a factor of ∼26 below the flux of the XMM-Newton detection. Therefore this source is transient, and the outburst that we observe lasts a maximum of 20 days.

Zoom In Zoom Out Reset image size

ULX-4 is highly variable in the XMM-Newton energy band over the course of the observation. Its light curve, which we examine in three energy bands (Figure 8), shows a low initial flux, a predominantly soft flare lasting ∼5 ks and releasing ∼10⁴³ erg of energy, followed by a second increase of flux in the final 10 ks of the observation with a corresponding softening of the spectrum (there is a gap at ∼3.4 × 10⁴ s due to removing a period of soft proton flaring rather than source behavior). The final count rate is over an order of magnitude higher than the count rate at the start of the observation in the soft bands. The NuSTAR light curve (Figure 9) shows the source to persist throughout the remaining 50 ks of the observation, though it continues to vary in both soft and hard energy bands.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Dividing the 50 ks XMM-Newton observation into five 10 ks segments, we can track the spectral and flux evolution of ULX-4 over the course of the observation (see Figure 10). The source begins in a relatively soft low-flux state (Γ = 2 ± 1, F_X = 2.6 × 10⁻¹⁴ erg cm⁻² s⁻¹ in the first 10 ks of the observation) and quickly reaches its very hard (Γ = 0.9 ± 0.3) power-law slope in the second interval. During the third interval, which encompasses most of the soft flare, the spectrum nonetheless remains hard (${\rm{\Gamma }}={0.8}_{-0.1}^{+0.2}$) and only shows signs of softening in the final interval (Γ = 1.2 ± 0.2). At the end of the observation, the ULX-4 has a flux of F_X = 5.0 × 10⁻¹³ erg cm⁻² s⁻¹. Therefore this source undergoes a factor ∼20 increase in flux and transitions into the ULX luminosity regime over the course of the observation.

Zoom In Zoom Out Reset image size

We created a power spectrum for the XMM-Newton observation of ULX-4 by averaging the periodograms of 72 segments of length 601 s. There are no significant features in the power spectrum beyond some possible red noise becoming apparent at low frequencies (≲10⁻³ Hz), which is consistent with an accreting source. Similarly, we created a power spectrum for the NuSTAR observation, though no features are seen. We searched for pulsations using the High-ENergy Data Reduction Interface from the Command Shell (HENDRICS) software (Bachetti 2015), a package based on stingray (Huppenkothen et al. 2019) for timing analysis of X-ray data and particularly optimized to handle NuSTAR data. We ran a pulsation search, considering potential acceleration with $\dot{f}$ up to 10⁻⁸ Hz s⁻¹, but failed to detect any pulsations between 10⁻⁴ and 10 Hz. On simulating light curves of ULX-4, assuming a sinusoidal pulsation with a constant 1 s period at different pulse fractions, we find that we can place a 90% upper limit on the pulsed fraction of ∼20% for the observed light curve. Given that ULX pulsars are known to be affected by spin-up and orbital modulations (e.g., Bachetti et al. 2014; Carpano et al. 2018), it is conceivable that we would not detect a pulsation at higher pulse fractions than this with the data we currently possess. We therefore cannot rule out the potential presence of pulsations in this source during its outburst.

There is no obvious single optical counterpart to the X-ray source. Data taken with the Ultraviolet and Optical Imager instrument on the Swift satellite show no detection during any of the Swift observations of NGC 6946, including the days immediately before and after the second NuSTAR epoch. While we have no optical data precisely simultaneous with the X-ray detection, this limits the presence of an optical counterpart with magnitude m_v < 18.1 and m_b < 19.0 (M_v < −11.1, M_b < −10.2) to less than a day, if one appears at all. However optical counterparts of ULXs are mostly far fainter than this (e.g., Gladstone et al. 2013), so these limits do not place strong constraints on the sort of counterpart that may be present, though they do rule out a very bright optical transient.

The region of NGC 6946 in which ULX-4 is located has been observed with the HST on several occasions with the WFPC2 and WFC3/IR instruments—in the F547M, F606W, and F814W optical bands with WFPC2 and the F110W and F128N near-infrared bands with WFC3/IR. The F656N band is also observed with WFPC2, but the spatial resolution is insufficient for performing photometry on the individual sources. All of these observations took place before 2017, but we can still investigate potential optical/infrared counterparts before the outburst. We corrected the astrometry (see Section 2) and derived a 09 error circle around the XMM-Newton source position, which we found to be centered on a dark dust cloud. We find six potential counterparts around the edge of this error circle and a bright star cluster just outside it. We show the potential counterparts in Figure 11 in three of the HST bands and list them in Table 6—not all sources are visible in all bands. It is also feasible that the source is associated with the star cluster itself, although it is too crowded for us to characterize individual stars within it.

Zoom In Zoom Out Reset image size

We used the DAOPHOT II/ALLSTAR software (Stetson 1987) to obtain photometric data for all six sources in the bands that they are detected in, and correct for Galactic reddening using E(B − V) = 0.2942 ± 0.0028 (Schlafly & Finkbeiner 2011) and the Fitzpatrick (1999) reddening law. These sources are faint and in a crowded field, and given the presence of a dust lane, may be affected by significant local extinction. Therefore they are not detected at high significance, and these magnitudes should be considered approximations with large errors, especially at the lowest fluxes. We present a table of magnitudes in Table 6. Using a distance modulus of μ = 29.16 we convert these to absolute magnitudes. In addition, we use the F606W and F814W bands to produce a V − I color.

3.3.1. Foreground/Background Source

3.3.2. Supernova

3.3.3. Super-Eddington Accreting Source

3.3.4. Transient Outbursts

3.3.5. Micro-tidal Disruption Events (μTDEs)