The X-Ray Binary Population of the Nearby Dwarf Starburst Galaxy IC 10: Variable and Transient X-Ray Sources

Images for visual inspection were generated using both Gaussian-smoothed and adaptively smoothed B (0.3–8 keV), S (0.3–1.5 keV), and H (2.5–8 keV) band data. Smoothing is necessary to create an image from sparse data. The adaptive smoothing maps created with ciao smooth for the B-band images were applied to make the S and H band images, and all three bands used the exposure maps generated for 1.5 keV photons (roughly the peak of ACIS sensitivity). To illustrate our search for transient sources, we present all 10 B-band images in Figure 1, which shows a number of prominent transient events and also illustrates the effects of shifting spacecraft roll angle and aim-point. The individual variable sources can be identified with the aid of Figure 2, which is a representative color image constructed using the visualization and analysis program SAOimage DS9⁷ from the deep merged data set 57082 (Table 1).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Light curves for all of the 110 unique sources were constructed by processing the master catalog with a Unix script, which computed the following parameters for each group of rows bearing the same SrcNumber identifier: number of occasions observed (N_P), number of detections in the broadband (N_B), minimum count rate (${R}_{B}^{\min }$), maximum count rate (${R}_{B}^{\max }$), average count rate (${R}_{B}^{\mathrm{mean}}$), root-mean-square ${R}_{B}^{\mathrm{rms}};$ variability range (Range = ${R}_{B}^{\max }$ − ${R}_{B}^{\min }$), variability ratio ($\mathrm{Ratio}={F}_{B}^{\max }/{F}_{B}^{\min }$), relative variability (${\sigma }_{\mathrm{Var}}=\mathrm{Range}/\sqrt{{(\mathrm{error}{R}_{B}^{\max })}^{2}+{(\mathrm{error}{R}_{B}^{\min })}^{2}}$), and total number of net broadband counts (${\mathrm{Counts}}_{B}^{\mathrm{Total}}$) obtained by summing the values from each detection. In computing the above parameters, we included all B-band detections and B-band upper limits, except that ${F}_{B}^{\max }$ (and its associated error bar) are required for a positive detection. The complete set of parameters characterizing each source are provided in the online journal as supplemental data accompanying Table 2, which shows only the variable objects, whose selection is described below.

Light curves were plotted for all X-ray sources in the master catalog; selected examples are shown in Figures 3–5. The plots display detections in all three energy bands and upper limits for observations with no B-band detection. All 110 light curves were visually examined alongside the ACIS images to verify the detections and to note any abnormalities such as sources falling very close to the edge of a CCD or having a close neighbor. In this way, we flagged eight sources as "probably false" (objects detected only once at the extreme edge of the CCD) and four sources as being duplicates (multiple detections of a real source being falsely identified as separate sources when detected in subsequent observations). These eventualities are expected and the number of cases was within the expected range. wavdetect expects to generate one false source per CCD (i.e., ∼10 in total) and faint off-axis sources are subject to larger positional errors. We note that most of the sources were detected in at least two independent observations. Specifically, 43 sources were detected more than once in the B band, 36 only once in B-band (some of which were also detected in a separate observation in S or H only), and 25 were only ever detected in the S or H band. The probability of a source with multiple detections being false is extremely small given the low surface density of sources.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The light-curve parameters described above (and given in Table 2) were used to identify variable and transient sources by sorting on the σ_Var and ratio parameters. Assuming the usual statistical definitions, a light curve with σ_Var > 3 is variable at the 99% confidence level. There are 21 such objects, listed in Table 2, ranked in descending order by σ_Var > 3. Our criterion for detecting a variable or transient source is at least one B-band detection with a count rate of at least 3σ above the lowest upper limit; thus, the only real sources are included in Table 2.

Complete data for each variable source are provided in tabular form as "data-behind-the figure" accompanying this paper in the online journal.

Given the large distance to IC 10, we are able to extract spectra for relatively few sources. Consequently, the median energy for the combined events is reported in Table 2 as an indication of spectral hardness. We plotted a hardness-intensity diagram for all 110 point sources in Figure 6, using cumulative events as a proxy for intensity, and the median event energy for the hardness axis. This parameter space serves to illustrate some key properties of the objects in our catalog. Figure 6 shows that large amplitude variability is seen at all intensity levels, that most bright sources are moderately variable (σ_var > 3), and that below 100 counts, our selection method (Section 3.1) is sensitive only to large amplitude variables. Thus we are not picking up random fluctuations in faint sources as variables. Figure 6 also shows that the distribution of median event energy is bimodal, with a hard/soft split at 1.4 keV. This was revealed by a kernel density estimator (density in the R package, with bandwidth = 0.1 keV), which we ran both with and without weighting by the errors. Weighting increased the strength of the bimodality, as seen by comparing the solid and dashed lines in the right-hand panel of Figure 6. We performed a bootstrap resampling (10⁴ trials) using the error-weighted values to generate 95% confidence intervals for the density plot. The kmeans clustering test identifies a similar hard-soft split, which can be attributed to different extinction values for IC 10 versus foreground galactic objects.

Zoom In Zoom Out Reset image size

Energy quantiles (Hong et al. 2004) were computed from the event lists for each unique source in the master catalog. Quantiles were computed for each observation and also for the combined event list, which provides the maximum number of counts and hence delivers better S/N values. These values and the accumulated number of events in the source extraction regions is listed in the supplementary table in the online journal. In our analysis, we use two of the quantiles defined by Hong et al. (2004): the Slope quantile QDx, which is related to the steepness of the spectrum, and the Curvature quantile QDy, which is constructed from the ratio of the quartiles Q₂₅/Q₇₅. Full details are available in Hong et al. (2004) and in the source code.¹⁰

The QDx, QDy parameter space shown in Figure 7 was designed to explore, in a single analysis, the spectral shape of low count sources that differ widely in hardness. Panel A shows spectral model grids generated for the back-illuminated ACIS-S3 detector for two classes of sources: power law (photon index Γ = 0–4) and thermal bremsstrahlung (kT = 0.2–10 keV), both of which with absorption varying between ${N}_{H}=(0.01-10)\times {10}^{22}$ cm⁻². These models are intended to represent accretion-powered and stellar coronal sources, respectively. The quantile values for the variable sources selected in Table 2 are plotted in panel B, with labels to facilitate discussion. Error bars are in panel C, and those sources with optical counterparts (Table 4) are highlighted in panel D.

Zoom In Zoom Out Reset image size

Pulse-height spectra were extracted for all sources having greater than 200 net counts in the master catalog, yielding seven objects (there are 13 sources with >100 net counts). In this article, we include only those appearing in Table 2 after excluding IC 10 X-1 and X-2 (which have been reported elsewhere), which leaves five objects for the analysis in this paper. For every such object, we used the Ciao script specextract to generate source and background spectra and response files (RMF, ARF) for each observation; and also for the merged event list containing all observations in which the source was detected. The resulting spectra are plotted in Figure 8. The spectra were examined in XSPEC¹¹ and were fitted by the following models (combined with the phabs absorption model): power law (PL), Raymond–Smith (RS), Mekal (M), blackbody (BB), and thermal bremmstrahlung (TB). These models were selected as appropriate for X-ray spectra of accretion, magnetic flare, and thermal origin. In cases where the absorption was unconstrained by the model, we froze it at 5 × 10²¹ cm⁻² for the column density toward IC 10. The resulting model parameters and goodness of fit statistics are presented in Table 3. Given the low S/N ratios, only single-component fits were considered. We caution that both quantiles (in common with all flux ratio-based color systems) and single-component spectral fits can be non-unique. For example, spuriously low N_H values can occur when soft components remain unaccounted for (Brassington et al. 2010). This is a motivation to obtain deeper data sets, where such components can be resolved.

Zoom In Zoom Out Reset image size

A list of positional counterparts from Massey et al. (2007) was generated by the procedure described in Laycock et al. (2014). A limiting visual magnitude of V > 23.8 was used for sources without counterparts, this being the faintest magnitude of any object in the catalog. The distance-independent X-ray to optical flux ratios were found using

$$\begin{eqnarray}&&{\mathrm{log}}_{10}\displaystyle \frac{{f}_{x}}{{f}_{v}}={\mathrm{log}}_{10}({f}_{x})+\displaystyle \frac{V}{2.5}+5.37,\end{eqnarray} \tag{ 1 }$$

where the observed flux f_x was computed from the net-count rate assuming an absorbed power law with a photon index of 1.5 and N_H = 5 × 10²¹ cm⁻². The values in Table 4 are the maximum measured values, remembering that f_x is variable.

Table 4.  Optical Counterparts

Src	Massey	S	r95	V	B–V	U–B	V–R	R–I	${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$	Hα
		arcsec	arcsec	mag	mag	mag	mag	mag
1	J002008.68+591340.7	0.15	0.43	22.384	0.700	−0.087	0.762	0.855	0.77	n
8	J002012.68+591501.4	0.1	0.32	22.974	1.242	⋯	0.735	1.144	1.45	n
13	J001958.69+591527.7	0.37	0.53	20.473	1.525	⋯	0.932	0.962	−0.33	n
20	J002029.08+591651.7	0.30	0.28	21.722	0.905	−1.094	0.137	0.754	2.43	⋯
20	J002029.12+591651.8	0.17	0.28	22.478	0.017	−0.916	0.851	0.805	2.74	y
28	J002046.55+591731.7	0.61	0.53	18.793	1.521	1.327	0.957	0.958	−0.53	⋯
28	J002046.36+591732.4	0.55	0.80	20.199	1.786	99.999	1.191	1.395	−0.53	n
29	J002009.14+591738.3	0.32	0.39	21.777	1.059	−0.820	0.536	1.052	−0.09	n
29	J002009.15+591738.4	0.37	0.39	21.732	0.932	−0.588	0.494	1.049	−0.1	n
46	J002020.94+591759.3	0.28	0.31	19.954	1.211	−0.259	1.043	0.820	0.54	y
50	J002047.28+591935.7	0.89	0.66	22.539	1.823	⋯	1.292	1.564	0.6	⋯
52	J002033.29+592135.4	1.16	0.75	17.277	1.551	1.365	0.946	⋯	−1.5	⋯
53	J002013.70+592236.1	0.49	0.96	21.404	1.872	⋯	1.354	1.603	−0.2	⋯
90	J002028.42+592024.0	0.34	0.54	21.348	1.805	⋯	1.294	1.579	0.37	y
18	a	⋯	⋯	⋯	⋯	⋯	⋯	⋯	⋯	y

Note. Positional optical counterparts from the catalog of Massey et al. (2007) for variable X-ray sources listed in Table 2. Matching criterion was Separation $S\lt \sqrt{{r}_{95}^{2}+1}$. Hα column refers to the presence or absence of Hα emission in Gemini GMOS narrow-band images, which did not cover all sources. Column ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ was computed from column V and ${R}_{B}^{\max }$ (Table 2) via Equation (1).

^aObject 18 is coincident with ring-like Hα structure.

Download table as:  ASCIITypeset image

In the following analysis, sources with optical counterparts believed to lie in IC 10 have their apparent magnitudes corrected for the foreground extinction assuming a value of A_v = 2.85 and distance modulus μ = 24. Narrow-band Hα and ${\rm{H}}\alpha \,{continuum}$ images from the Gemini Observatory were examined in order to look for sources of Balmer-line emission as described in Laycock et al. (2014). Preliminary spectroscopic follow-up was reported by Balchunas et al. (2011), though spectral type and radial velocity (RV) are only available for three of the variable sources.

Four sources have variability ratio > 10 and σ_var > 3, these are Src 46 (IC 10 X-2), Src 90, Src 28, and Src 50 (listed in order of decreasing amplitude). The light curves for these sources appear in Figure 3. There are a further two sources (Src 53 and Src 29) that we include as transients because each was detected in a single observation, with a broadband count rate exceeding the lowest upper limit by three σ (i.e., N_B = 1 and σ_Var < 3, but not reaching Ratio > 10) listed in Table 2. Both were in the ACIS-S3 field of view 8–10 times, so that the null detections and corresponding upper limits place meaningful limits on the duty cycle. Light curves for these two objects are include in Figure 4. Only one transient, Src 46, which is IC 10 X-2, Laycock et al. (2014) provided enough counts to produce pulse-height spectra and appears to be an X-ray binary with a supergiant Be companion and similarities to Supergiant Fast X-ray Transients (SFXTs). For reference, we include its light curve in Figure 3 and it appears on the quantile diagram (Figure 7) at (−0.18, 1.17), which is consistent with a very hard power-law spectrum. The case of IC 10 X-2 highlights the need for optical spectra, since its photometric colors are unremarkable unless one knows the appropriate correction, and its nature as an extragalactic HMXB required the unambiguous Doppler-shift to place it firmly in IC 10.

Src 28 was observed eight times yielding three B-band (and one S-band only) detections with a variability ratio of 15.6, a 9.3σ variation above the most constraining upper limit. Quantile values for the four observations were all consistent with the merged value. Lying at [−0.9, 1.1] in Figure 7 the source is consistent with a soft power law (γ ∼ 2.5) or approximately a few keV thermal plasma. In both cases, the absorption is small (≪10²¹ cm⁻²), suggesting a foreground galactic location for the object (subject to the caveats in Section 3.2.) There are two optical counterparts (Table 4 from Massey et al. 2007) and the source is not in the field of the Gemini images, so no follow-up spectrum was obtained. Only one of these stars is consistent with the X-ray detection having the smallest error circle. Both stars have color indices suggestive of spectral type M. Taking the most likely counterpart, whose B-V is already at the un-reddened value for M0V, we calculate distance modulus μ ≃ 18.79–9 = 9.8. The distance is then ∼900 pc, yielding peak L_x = 3.7 × 10³⁰ erg s⁻¹, in the range of isolated dMe stars, which is supported by the distance-independent ratio ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ = −0.53 being lower than seen for active binaries. Note that in the absence of a spectral type or RV value, IC 10 membership cannot be entirely ruled out. If located at 660 kpc, and de-reddened by A_V = 2.85 the star would have an extreme absolute magnitude M_V ≳ −8, which favors the galactic dMe interpretation.

Src 29 was detected in obsid 08458 with 22 net counts, which placed it a factor of five (or 3.6σ) above its most sensitive non-detection. The detection is in the B and H (and not S) bands. The quantile value [−0.13, 1.2] lies in the hard power law region (Γ ∼ 0.5) of Figure 7 on the contour for N_H ∼ 10²² cm⁻². There are two optical counterparts in M07 with near identical photometry (V = 21.7, B − V ≃ +0.93), Gemini Hα line and continuum images indicate that two stars are blended in the Massey et al. (2007) images, with one catalog position on top of the brighter star, and the other midway between the two. The source lies in a region containing clumpy diffuse Hα emission, but neither star is a strong Hα point-source. A GMOS spectrum of the brighter star (which is near the center of the X-ray error circle) shows a strong continuum without obviously identifiable features other than telluric lines. This rules out a foreground M-dwarf and suggests a reddened early-type star in IC 10. Applying the distance modulus and extinction correction assumed throughout this paper gives ${M}_{V}=21.7-2.85-24=-5$, and ${(B-V)}_{o}\simeq 0$. The star is also blue in U–B and V–R color indices compared to the other stars in Table 4, and once de-reddened looks like a B spectral type. This object is likely an HMXB in IC 10, based on X-ray and optical properties.

Src 50 was observed eight times, yielding six B-band detections and two upper limits. The most constraining upper limit comes from one of the pair of 45 ks exposures made in 2006 November providing the evidence for a large variability ratio (ratio = 12, σ_Var = 5.7). The quantile values varied with time, the spectrum being harder in obsids 11081, 11082, 11084. The merged value plotted in Figure 7 lies at [−0.52, 1.05], which is dominated by photons from its softer apparition in 2006. The model grid indicates a power-law (Γ ∼ 1) spectrum with typical IC 10 absorption (∼5 × 10²¹ cm⁻²). There is an optical counterpart in Massey et al. (2007) but it lies outside the Gemini field and no follow-up spectrum has been obtained. If the object is in IC 10, then its de-reddened photometry gives ${M}_{V}=22.5-2.8-24.01\,=-4.3,B-V=1.823\,-0.86$ $=0.9$.

Src 53 was detected only in obsid 08458 with 27 net counts, at a count rate a factor of 6.6 (3.7σ) above the most constraining non-detection obsid 07082, which was of equal exposure-time (45 ks) and pointing, and occurred only a day before. A point source is clearly visible in the image of obsid 08458 and absent in obsid 07082. The detection was made in the S and B bands only. The very rapid variability and soft spectrum implied by the S-band detection are indicative of a magnetic stellar flare but not conclusive. The quantile value [−0.8, 0.5] lies outside our model grids and has large errors so we can say little about its spectral shape. There is an optical counterpart (V = 21.4, Table 4) but it lies outside the Gemini field and no follow-up spectrum has yet been obtained.

Src 90 was covered by the ACIS-S field of view nine times, but was detected only once, in obsid 11082 in the S, B, and H bands. It is clearly visible as a point source in the adaptively smoothed image. The observed count rate was a factor of 18 (4.7σ) above the most constraining upper limit. The energy quantile (Figure 7) lies at [−0.67, 1.23] within the power-law model grid at a location consistent with γ ∼ 2, N_H ∼ 10²¹ cm⁻². There is an optical counterpart (4), which we identify as an Hα emission line star in Gemini narrow-band imaging. We obtained a Gemini GMOS optical spectrum of this star Balchunas et al. (2011), which shows a red continuum with molecular bands characteristic of spectral type M. Prominent lines of Hα, Hβ in emission enable us to measure the radial velocity RV = −165 km s⁻¹, RV = −171 km s⁻¹ for the respective lines. Thus the star belongs to the Milky Way since the mean RV of IC 10 is −340 km s⁻¹. Photometry from Massey et al. (2007) (V = 21.348, B-V = 1.8) indicates that the color index is consistent with early M. For example, spectral type M0Ve ($B-V=+1.5,{M}_{V}=9)$ would require only E(B-V) ∼ 0.3, A_V ≲ 1. The distance modulus is then $\mu \simeq 21.4-1-9=11.4$, for a distance of ∼2 kpc, implying L_X ≃ 1.2 × 10³¹ erg s⁻¹. The distance-independent flux ratio during the flare reaches ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ = +0.3. Stellar flares in dMe stars reach L_X ≃ 10³¹, with the BY Dra binary type being particularly luminous Singh et al. (1996). The luminosity, quantile, and duty cycle point to a hard and energetic event on a magnetically active M-dwarf. Such interlopers are frequent bi-catch of X-ray surveys, and we have seen one in the foreground of the SMC (Laycock & Drake 2009).

Many of the brighter sources in our study show significant variability (σ_Var > 3) as expected given that astrophysical X-ray emission mechanisms are dynamic in nature. The light curves of these candidates are presented in Figure 4 for those varying by a factor of 5–10, and in Figure 5 for those not exceeding a factor of 5. All of these objects have sufficient counts in the merged event list to provide energy quantiles with reasonable S/Ns. In Figure 7, we plot the quantiles for all unique objects in the master catalog having at least 10 counts in their merged event list, with the candidate variable sources picked out in larger points. A group of 10 candidate variables clusters along the N_H contour corresponding to ∼5 × 10²¹ cm⁻² within the power-law model grid region indices in the range of Γ = 1–2. Based on this evidence, they are consistent with being HMXBs in IC 10. In the following section, we present individual analyses for the remaining variability candidates listed in Table 2, beginning with the five candidate variables to exceed 200 net counts. For these objects, we performed spectral fitting in XSPEC testing PL, BB, TB, and NEI models combined with the photo-absorption model phabs. The best fit for each of these sources is displayed in Table 3.

Source 1 was observed in half of the ACIS-S observations and is detected in four of the five. The are enough counts for the quantile diagram to place it firmly in the XRB locus. The optical counterpart V = 22.4, B − V = 0.7 if adjusted for distance (μ = 24) and reddening (A_V = 2.85) gives ${M}_{V}=-4.5,{(B-V)}_{0}=-0.16$, which implies an early-B spectral type. Thus the object is an HMXB, subject to optical spectral confirmation of the precise subtype.

The light curve for Src 5 (Figure 4) shows a large amplitude variable with a 100% duty cycle, being detected in all observations for which it was in the FoV. It also shows that virtually all the counts are in the soft band. The hardness-intensity plot (Figure 6) shows that Src 5 is, at once, both the softest variable and one of the brightest objects in the survey, pointing to a foreground object. Quantile analysis suggests a thermal spectrum with a median energy of just 0.5 keV that did not vary between detections. The ∼300 net-count spectrum was best fit by a blackbody model (Figure 8, Table 3). A power-law model can provide an acceptable fit but the spectral index of 7.7 implies that the cool thermal plasma is more physically motivated. No optical counterpart is found in Massey et al. (2007) and the source lies outside of the Gemini field.

Src 8 is a very low amplitude variable with a 100% duty cycle in the Chandra survey. The summed 863 net-count spectrum was fit with a power-law model of Γ = 1.6, and typical IC 10 absorption (Figure 8, Table 3). There is an optical counterpart in Massey et al. (2007; lies outside the Gemini field) for which we can estimate ${M}_{V}=22.974-2.85-24=-3.9$, ${(B-V)}_{o}\,=1.242-0.86\,=\,0.32$, which would fit a supergiant or Be-X interpretation.

Src 12 is a persistent source with 100% duty cycle and a light curve that suggests very slow X-ray variability. The source has the second highest number of total X-ray counts (>1000) in the survey, so spectral fitting was feasible. The quantiles lie at the upper end of the cluster of points in Figure 7, and accordingly a power-law fit to the spectrum yields Γ ≃ 2 power law with a typical absorption for the IC 10 field (Figure 8, Table 3). Turning to physically motivated models, acceptable fits are obtained for the thermal plasma models (TB, RS, Mekal), all giving temperatures consistent with kT = 4 keV and N_H = 6 × 10²¹ cm⁻² similar to the power-law fit. A blackbody fit yields a significantly lower column density and temperature (kT ≲ 1 keV). The lack of an optical counterpart in M07 provides the following constraints on distance-independent flux ratio (${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > 1.9) and absolute magnitude (M_V > −3) if located in IC 10. The object could be an AGN on the basis of the spectral index, large ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ and slow variability. A foreground dMe star seems unlikely due to the lack of flares and persistently high ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$. An XRB cannot be ruled out. Given the persistent nature of this source it will be possible to obtain much higher S/N X-ray spectra and timing in the future.

A transient X-ray source appearing to exhibit rapid variability, Src 13 was detected (at a very low flux) in only one of the deep (45 ks) observations separated by two days in 2006. Subsequently, the source flared up in a single one of the five monitoring observations of 2009/10, which are spaced approximately six weeks apart. These two independent positive detections confirm the reality of the object. The limiting fluxes for those four nulls are at or above the flux detected in the deep 2006 exposure. Both positive detections were dominated by soft-band counts, with no detection in the hard band. The quantile value is consistent with a plasma temperature of kT ≃ 1–2 keV, or a Γ ≃ 2 power law. There is an optical counterpart within the 95% confidence radius. The photometry (Table 4: $V=20.5,B-V=1.5$) would be consistent with an un-reddened M0V star at a distance of 2 kpc, or a hotter spectral type at larger distance. It is impossible to make the star closer since the observed color index imposes the limit of M_V ≤ 9. If located in IC 10, we obtain M_V = −6.4. An optical spectrum is needed to resolve the nature of this source.

Src 14 has an interesting light curve that contains an outburst of duration approximately two months during 2010. Quantile analysis reveals a persistently hard source, having the highest median photon energy (E50 = 3.64 keV) of any object in Table 2, and lying in the high-absorption hard power-law region of Figure 7. The lack of an optical counterpart down to a limiting magnitude of ≳23.8 provides us with a distance-independent flux ratio ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > +1.24 for the brightest recorded X-ray flux. If located in IC 10, the absolute magnitude would be ${M}_{V}=23.8-2.85-24=-3$ if we assume the typical extinction correction. A more luminous star would require more extinction to remain hidden. Spectral type B0V has ${M}_{V}=-3$, so the data support a Be-XRB. If a dMe star located in the Milky Way, then the spectral type would have to be late M (for M0V the luminosity distance implied by V > 23.8 is greater than 9 kpc) and the X-ray quantiles are too persistently hard. In summary, the long duration outburst, persistently hard spectrum, and high ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ point to an XRB.

Src 18 is a low amplitude (less than a factor of two) variable with 100% duty cycle in the Chandra survey. The total 435 net-count spectrum is soft, equivalent to power law Γ > 5 or cool thermal plasma for any reasonable N_H (Figure 8, Table 3). There is no optical counterpart in M07, providing a lower limit on ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > +1.12. The X-ray position lies at the center of a faint compact Hα shell identified in narrow-band Gemini images. A GMOS slit placed across the ring did not attain sufficient S/N to detect any lines.

Src 23 is a variable source with a high duty cycle (7/10) and no optical counterpart in M07, which puts a lower limit on ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > +1.2. With 203 net counts summed over the six observations where it was detected we obtained an acceptable X-ray spectral fit to the thermal bremsstrahlung spectral model with kT = 1.7 keV, N_H = 10²² cm⁻² (Figure 8, Table 3). Due to the low number of counts, all simple models obtained similar χ² values, with the power-law model confirming a soft spectrum (Γ > 5) and the blackbody fit unable to constrain the N_H to a significantly non-zero value. Indicative of a foreground galactic location.

Transient source with duty cycle <50 and no optical counterpart, if located in IC 10 the limiting magnitude V > 23.5 implies ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > 1, M_V ≃ −3 subject to a typical N_H. The quantile value is also consistent with an HMXB.

Another high duty cycle, hard, variable source without optical counterpart. As with Src 42, the object could be an HMXB or AGN with its high ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > 1.2, and limiting V > 23.5 implying M_V ≃ −3 depending on absorption.

Hard transient source with duty cycle <50 and no optical counterpart, the light curve is reminiscent of a Be-XRB. If located in IC 10 the limiting magnitude V > 23.5 implies ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > 1.23, M_V ≃ −3 subject to a typical N_H. The quantile value is also consistent with an HMXB.

This is a persistent variable source with a hard X-ray spectrum (median energy = 2.7 keV), and no optical counterpart in M07. Table 2 reports 9 broadband detections out of 10 observations; however, the X-ray light curve (Figure 5) shows a hard-band-only detection in that observation (final point in 2010), so the source has a 100% duty cycle. There are not enough counts to analyze the spectrum; however, the quantile diagram Figure 7 shows Src 42 (the point at ${QDx}\sim -0.4,{QDy}\sim 1.5$) is consistent with a heavily absorbed power law ${\rm{\Gamma }}\sim 2,{N}_{H}\sim {10}^{22}$ cm⁻². The limiting magnitude of V ∼ 23.4 constrains the optical counterpart to M_V < −3 for typical IC 10 absorption; however, the likely much higher N_H could hide an O star. The flux ratio ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > +1.2 together with the hard spectrum and variability make an HMXB the most likely interpretation, though it could conceivably be a background AGN.

Variable source on all timescales probed, and seen in all observations except for the 2003 exposure (which places a limit on its quiescent flux); though since it appears as [W05]24X, we know the source was seen by XMM-Newton in 2003. The lack of an optical counterpart limiting magnitude V > 23.5 implies ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ > 1.4, M_V ≃ −3 if located in IC 10. The energy quantile is hard and lies in the HMXB region.

This object is a large amplitude variable with three detections out of nine observations. Two of these detections are mere days apart and show a factor of five change in flux (perhaps a flare), with the fainter point only seen due to the long observation (45 ks), while a third and final detection was seen in the soft band (only) in a single one of the 2009/10 monitoring series. There is a bright optical counterpart in M07, which at V = 17.3, B − V = +1.6 is almost certainly a galactic star (the alternative being an extreme luminosity M_V = −9.6 hyper-giant if located in IC 10). Assuming spectral type M0 (based on the color index) the distance modulus is μ ≃ 17.3–9 ≃ 8.3 giving a distance of 450 pc. The distance-independent flux ratio ${\mathrm{log}}_{10}\tfrac{{f}_{x}}{{f}_{v}}$ = −1.5 supports this interpretation, but an optical spectrum is needed to confirm it.