The Nearby Dwarf Irregular Galaxy IC 1613 and Its Complex Bubble Region: Chandra and VLA Observations

We used the Advanced Charge-Coupled Device (CCD) Imaging Spectrometer (ACIS; Garmire et al. 2003) on Chandra to observe IC 1613 on 2005 September 4 (Chandra Observation number 5905). Table 1 provides a brief summary. The aimpoint for the observation was approximately 15 east of the nucleus of the galaxy, to ensure coverage of the bubble complex in the northeast quadrant (nucleus: 01:04:47.8, +02:07:04.0 (NED ⁷ ; ned.ipac.caltech.edu); aimpoint: 01:04:54, +02:08:34.6, all J2000.0)(q.v., Figure 1). The observation was obtained using the back-illuminated S3 CCD in Very Faint mode, to have the best chance of detecting low-energy emission from the bubble complex. The exposure lasted approximately 49,994 s; after correcting for the deadtime, the effective exposure time was 49,327 s. The Chandra Point Source Catalog reports 49,324.46 s.

Zoom In Zoom Out Reset image size

The offset in aimpoint means the entire galaxy is not covered by the observation, given that its size has been measured to be ∼16' ×20' (Ables 1972) and perhaps as large as ∼20' × 30' based on the discovery of carbon stars ∼15' from the center (Albert et al.2000). Essentially, the ACIS observation covered an ∼8' swath down the center of IC 1613 (Figure 1). Much of the optically brightest portion of the galaxy lies on the upper two-thirds of CCD 7 = S3. The galaxy cluster covers a portion of IC 1613 at the south end of CCD-7. We include CCDs 6 and 7 in our analysis, treating each separately where appropriate in the analysis.

The Chandra Interactive Analysis of Observations (CIAO) software (version 4.9) and the associated calibration files (version 4.7) were used. We filtered the event file to retain events from CCDs 7 and 6 (=S2) to provide as much coverage of the galaxy as possible. We accumulated source-free background areas offset south from the galaxy center and away from the galaxy cluster. We extracted a light curve using 50 s bins to test for the presence of soft background flares; no flares were detected.

Point sources were detected using wavdetect at 1'', 2'', and 4'' scales (Freeman et al. 2002) using the recommended 10⁻⁶ false-source threshold (one false source per CCD). The detected sources were merged into a final source list after eliminating duplicate detections. Source counts were extracted using apertures centered on the wavdetect positions and of sufficient radius to enclose 95% of the detected X-rays. We applied a signal-to-noise cut at 3.0, separating the sources into "detected" and "weak." We examined a random sample of the weak sources to verify that the detection routine had separated them correctly. We also ran the detection software on three bands: soft (0.3–1 keV), medium (1–2 keV), and hard (2–8 keV). This approach provides crude spectral information ("color"); we will discuss the colors in a later section.

All detected sources were also compared to the limiting count sensitivity of the CCDs as determined by the CIAO routine lim_sens. The limiting sensitivity task convolves the exposure map across the field of view with the effective area that any given X-ray would sense. Consequently, sources well off the optical axis must necessarily be brighter to be detected. Such sources all fell below 10–15 counts. Detections with extracted counts lower than the limiting sensitivity value at the location of the detected source were dropped. That limit is also the point at which the source count distributions of the ChaMP project turned over (their Figure 16; Kim et al. 2007).

Source counts were converted to fluxes using the PIMMS tool ⁸ . We adopted a 3 keV thermal bremsstrahlung spectrum absorbed by the known column toward IC 1613 and converted the counts to fluxes in the 0.5–8 keV band. Table 2 lists the source properties, which are further discussed in Section 3.

The VLA was used on 1995 Jan 2 and 3 for 0.7 hr to survey IC 1613 at 6 cm (4.86 GHz) in the C array configuration. The observing strategy was chosen to yield comparable surface brightness limits as part of a survey of nearby galaxies that includes M33, NGC 300, NGC 6946, and NGC 7793 (Gordon et al. 1993; Pannuti et al. 2000; Lacey et al. 1997; Pannuti et al. 2002). The observation of IC 1613, with a beam size of 45x43, results in a spatial resolution of ∼10 pc. The optical size of the galaxy is larger than the 9' field of view at 6 cm, although the most prominent optical and IR emitting regions are well within the radio field of view. To eliminate bandwidth smearing (chromatic aberration), 4IF spectral line mode, with eight channels with a total bandwidth of 25 MHz for each intermediate frequency (IF) pair, was used for the 6 cm observations.

The data were calibrated following the usual procedures in AIPS, produced and maintained by NRAO, and imaged using IMAGR. 3C 48 was observed as the primary calibrator and B0106 + 013 was used as the phase calibrator. The spectral channels were summed and an I polarization continuum map was cleaned using IMAGR in AIPS with Robust = −5 and uniform weighting, which produces a smooth background, well-suited to identify the isolated compact radio sources in IC1613. The resulting image rms sensitivity was 70 μJy for the map. Sources were identified using the AIPS subroutine STFND and measured with JMFIT after the primary beam correction was applied. Slices were taken through the sources with the routine SLICE to ensure that the sources were not located on positive or negative bowls of radio emission. Detected sources are listed in Table 4.

Because we are interested in compact objects, we did not need to add single-dish observations to get the zero spacings. The zero spacings are sensitive to the largest-scale structures such as the diffuse disk component, which adds unnecessary noise to the compact sources. At this resolution and sensitivity, most of the detected objects in IC 1613 are either not resolved or slightly resolved; thus, structural morphology cannot be used to discriminate between SNRs, HII regions, and background objects.

As described above, we detected 31 discrete X-ray sources within the D₂₅ radius of IC 1613 at or above the 3σ count level over the energy range of 0.3–8 keV. To calculate a flux for each source, we adopted a foreground column density of N_H = 3 × 10²⁰ cm⁻² and a thermal bremsstrahlung model with a temperature of 3 keV. Table 2 lists the properties of these sources while Figure 2 shows the detected sources.

Zoom In Zoom Out Reset image size

We also calculated the limiting flux sensitivity—the counts needed to exceed a flux threshold. For the IC 1613 field, a flux threshold of ≈4σ requires a minimum of ∼18.5 counts leading to a minimum flux (CCD 7) of ∼4.3 × 10⁻¹⁵ erg s⁻¹ cm⁻² or an unabsorbed luminosity of ≈7.8 × 10³⁴ ergs s⁻¹. This threshold means that our six detected sources between 10 and 18 counts have poor flux measurements. The limiting flux sensitivity is about a factor of 2 above the limiting sensitivity of Kim et al. (2007).

We note that there are 16 objects reported by Liu (2011), using the same data, that we do not detect. There are also 17 sources listed in the Chandra Source Catalog (CSC) that we do not detect. We consider the work of Liu first: those objects are shown in Figure 2 and listed in Table 4. Both studies used the same values in the CIAO wavdetect task for the false source threshold and the wavelet detection scales. What leads to the differences between the two studies?

For Liu sources 27, 38, 35, 45, 39, 41, 36, and 44, the signal-to-noise ratio falls below our initial signal-to-noise (S/N) acceptance screening of >3.0. Liu sources 6 and 19 lie outside of the D₂₅ radius for IC 1613. Source 34 (CCD-7) lies on the exact edge of a CCD. Removing all those objects leaves sources 3, 28, 32, 40, and 42 as possible mis-detected objects. In the intervening time from the publication of Liu (2011), the limiting sensitivity task became available within the CIAO software package. Those sources lie below our calculated limiting count sensitivity value at the source's location on CCDs 6 or 7. The overall upshot of the differences: our analysis uses more exposure time on the background to define it more precisely, combined with an improved understanding of the characteristics of the point-spread function behavior with energy and off-axis angle.

Finally, we detect one source not listed in Liu (2011): source 15. This source lies near the limiting count sensitivity value, so a slight change in the adopted background, the minimum S/N threshold, or the false source acceptance threshold would exclude it. Several re-runs of the wavdetect task were tried using variations of the task's parameters. In each case, we detected source 15. The source lies near the ACIS aimpoint, so it remains puzzling that this source was not detected by Liu (2011) and we cannot offer an explanation at this time.

To build a list of detected point sources from the CSC, we entered the RA and Dec of IC 1613 (Table 1) into the catalog query and requested all CSC sources within 8 arc min of that position ⁹ . An 8-arc min circle easily covers CCDs 6, 7, and 8. The resulting list included 51 sources. The detected sources in this paper were present on the CSC list, but 17 additional point sources were present in the catalog that were not detected in this paper (Table 3). Of those sources, curiously, only four were in common with the undetected Liu sources. Figure 3 shows the fluxes as computed in this paper compared to the fluxes from the CSC. Overall, there is a 1:1 correspondence. The most glaring exception is the resolved SNR (Schlegel et al.2019)—the CSC's extraction circle does not include all of the events.

Zoom In Zoom Out Reset image size

In general, as listed in Table 5, the undetected CSC sources fall below our initial S/N cut. We checked the S/N ratio by converting the fluxes from the CSC to count rates assuming the known column toward IC 1613 and a 5 keV thermal bremsstrahlung. Sources with count rates less than ≈1.5 × 10⁻⁴ counts sec⁻¹ in the IC 1613 field generate too few counts to rise statistically above the background. Our limiting sensitivity map is broadly comparable with that available from the CSC. We use "broadly comparable" because we used 0.5 arcsec pixels in calculating our map while the CSC used 3.96 arcsec pixels. The trend with off-axis angle is very comparable; values at specific locations in our map average to values on the CSC map. We discuss the colors of the CSC sources in Section 4.2 below.

Some number of the detected X-ray sources are expected to be background active galaxies shining through IC 1613. The results from the Chandra deep surveys or the ChaMP project are useful to estimate the number of possible background sources (Campana et al. 2001; Kim et al. 2004, 2007).

With the limiting flux sensitivity map, we used the relation given by Campana et al. (2001) for the number N of background sources greater than a flux density S per square degree, which may be expressed in CGS units as

$$\begin{eqnarray}&&N(\gt S)=360{\left(\displaystyle \frac{S}{2\times {10}^{-15}}\right)}^{-0.68}{\deg }^{-2}.\end{eqnarray} \tag{ 1 }$$

If we consider an entire ACIS chip (with a field of view of 81 × 81) plus the south half of chip 6, and given our flux limit of ≈4 × 10⁻¹⁵ erg s⁻¹ cm⁻², we estimate that ≈6 of our detected discrete X-ray sources are background objects.

Eleven radio sources were identified in IC 1613 at 6 cm. Three of the sources, IC1613-1, IC1613-5, and IC1613-7, were also identified by Dickel et al. (1985) in a 20 cm and 6 cm survey of IC 1613. ¹⁰ Those authors reported only one source at 6 cm: the SNR S8. The Dickel VLA observations were centered on S8, but the data were observed in continuum mode, which means that the sources suffered from bandwidth smearing (chromatic aberration), which worsens farther from the phase pointing center.

Table 3 lists the 6 cm flux densities for the sources we detect. We also include the 20 cm sources of Dickel et al. (1985), yielding a complete-to-date catalog of the known radio sources in IC 1613 (see Figure 8). Dickel et al. (1985) did not list uncertainties for their 20 cm fluxes. We estimate those uncertainties at 15% of the 20 cm flux density. With those values, we then calculate spectral indices for the sources detected at 6 and 20 cm: IC1613-1, IC1613-5, and IC1613-7. IC1613-7 is the previously known radio and optical SNR S8 (α: 12^h 28^m 10^s.94, δ: 44^o 06' 48''.34 (J2000.0)) (Sandage 1971) and is discussed in Schlegel et al. (2019).

All three sources for which the spectral indices were calculated are nonthermal. The spectral index, α (where S_ν ∝ ν^α ), is used to distinguish between thermal sources and nonthermal sources. To discriminate between thermal and nonthermal radio sources, we use the definitions of spectral index given in Pannuti et al. (2000) and Pannuti et al. (2002). A spectral index of − 0.2 ≤ α < 0.2 is used as a criterion for identifying a thermal spectrum, e.g., a candidate H ii region. A spectral index of α ≥ 0.2 is used as a criterion for identifying a nonthermal spectrum, a candidate SNR, or a background object. As expected, the known SNR is a nonthermal source; the other two sources are probable background sources based on the correspondence with sources in the NVSS and FIRST surveys (Condon et al. 1998; Becker et al. 1995).

The beam sizes of the 6 cm and 20 cm observations are well-matched, allowing for direct comparison of the flux densities and calculation of the spectral indices. The error in the spectral index was calculated by differentially propagating the errors in the 20 cm and 6 cm flux densities, and the errors were added in quadrature. The average error in the position of the sources is ∼10.

To provide confirming evidence that the radio sources are associated with IC 1613, we compared IC 1613 to an Hα image (from NED archive; original image taken 1995 Dec 19 at KPNO), as both thermal radio HII regions and nonthermal SNRs produce Hα emission. Since IC 1613 has a low redshift velocity of −240 km s⁻¹ (average value from NED), many background sources will have much higher velocities and the Hα emission would be redshifted out of the narrowband filter, discriminating between radio sources within the galaxy and background objects.

Unfortunately, 11 of the 15 sources from Table 3 fall outside of the spatial area covered by the Chandra observation (Figure 8); of the remaining four sources, two have X-ray counterparts. The first is the the SNR S8 (Schlegel et al. 2019).

The second is interesting in its own right: the source at 01:04:44, +02:04:37 (J2000.0) was originally detected by Dickel et al. (1985) at 20 cm. An infrared counterpart exists: the VLA source is consistent with 2MASX J01044429 +0204367, a source for which the NED has very little data beyond an object listed in each of the NVSS and FIRST surveys (Condon et al. 1998; Becker et al. 1995). Visually, the source appears to be a faint galaxy, potentially elongated. In spite of the paucity of data, we may conclude that the counterpart is a background galaxy or AGN.

But the Chandra wavdetect algorithm finds two sources at this location (Chandra 4 and 5) of nearly equal flux separated by (ΔRA = 0^s .2, Δδ = 19, or roughly 4 Chandra ACIS pixels in decl.). To determine whether the wavdetect result provided the correct interpretation, we generated an artificial source using CIAO's ChaRT and the MARX software from the MIT Chandra team. ¹¹ We did not attempt to include the background, deeming the brightness of sources 4 and 5 sufficient to overwhelm the background. Figure 4 shows the result: the PSF is sufficiently broad at the position of the detected sources 4 and 5 that we are almost certainly looking at a single X-ray source. The observation was carried out as a single pointing, so aspect uncertainties are unlikely to explain the apparent doubling. No other source at comparable off-axis angles shows apparent doubling. Possibly, a small fluctuation within the galaxy's distribution could lead to an apparent double source. However, to properly verify that the source is not double would require an on-axis pointing. In the remainder of this paper, we treat the two "sources" as one.

Zoom In Zoom Out Reset image size

We adopted the quantile approach to a color–color diagram (Hong et al. 2004). The quantile method determines the energy below which falls a specified percentage of events. Colors are determined by ratios or differences of the resulting energies. Instead of an orthogonal grid as built for a hardness ratio, the interpretative grid has a distorted appearance that represents the true spectral energy information available from the instrument for a given spectral model.

In Figures 5(a) and (b), we present the quantile grids for the bremsstrahlung and power-law models, respectively (values in Table 6). For both grids, moving quasi-vertically in the grid crosses lines of equal N_H with values of 0.001 (bottom), 0.005, 0.01, 0.05, 0.1, 0.5, and 1.0 × 10²² cm⁻² (top). The bremsstrahlung temperature increments quasi-horizontally with values of 0.2 (left), 0.5, 1.0, 2.0, 5.0, and 10.0 keV (right). The power-law photon indices include grid values of 0.5 (right), 1.0, 1.5, 2.0, 2.5, 3.0, and 5.0 (left).

Zoom In Zoom Out Reset image size

The foreground N_H is ∼3 × 10²⁰ cm⁻² and corresponds approximately to a line of constant $3\times \displaystyle \frac{{{\rm{Q}}}_{25}}{{{\rm{Q}}}_{75}}$ of 1. Each grid also contains two "typical" point sources: an AGN and an XRB. The spectrum of the AGN was built using an average power-law index (PLI) of 1.8. The spectrum was then fit with both power-law and thermal bremsstrahlung models absorbed by a column N_H = 1 × 10²¹ cm⁻² to recover the modeled spectral shape parameters as well as an approximate gauge using the other model. We adopted a higher column to mimic the typically higher columns toward AGN. Similarly, the XRB spectrum was built using a 7 keV bremsstrahlung model, then fit with power-law and bremsstrahlung models. For the AGN, the recovered values were PLI ∼1.85 and bremsstrahlung T ∼ 4 keV. For the XRB, the recovered values were ∼1.9 and 7.2 keV, respectively.

Prestwich et al. (2003) describe an X-ray color–color analysis of the discrete sources based on separating each source's counts into soft, medium, and hard bands. We followed their approach, defining the color bands using the same energy ranges: soft (S) equal to 0.3–1.0 keV, medium (M) as 1.0–2.0 keV, and hard (H) 2.0–8.0 keV. This then leads to two colors: M–S and H–M, each normalized by the total counts.

Figure 6 shows the result for IC 1613; the values are listed in Table 7. The majority of the sources are scattered throughout the regions labeled as "LMXBs," "HMXBs," and "Absorbed." The known SNR lies just outside of the SNR circle.

Zoom In Zoom Out Reset image size

A very soft source, number 9, lies just inside the bottom edge of the SNR circle. It exhibits an upper limit on both the hard and medium energy bands; consequently, it is a candidate supersoft source. Supersoft sources are generally considered to be nuclear-burning white dwarfs (Kato 2010); whether that is the case here remains to be determined. Assuming zero local absorption, the luminosity of an accreting white dwarf of ≈0.5M_⊙ would roughly match the detected X-ray luminosity, based on figures in Kato (2010). Larger masses are possible if local absorption exists.

We may also compare our color–color analysis with the CSC-extracted values. Figure 7(a) plots the (H–M)/B and (M–S)/B from our analysis versus the "same" quantities for the CSC. We restrict the comparison to the sources in common that have measured values in all bands (i.e., we ignore any source with an upper limit). We see a clear, systematic difference in both plots—this difference is attributed to the differing definitions of the bands. The CSC defines "soft" as 0.5–1.2 keV versus our definition of 0.5–1 keV; "medium" as 1.2–2 keV versus our 1–2 keV; and "hard" as 2–7 keV versus our 2–8 keV.

Zoom In Zoom Out Reset image size

Defining the "medium" band more restrictively will diminish the counts in that band, and for (M–S)/B, will shift them to S, hence making the resulting value more negative. In Figure 7(a), this results in a leftward shift. "Correcting" the CSC (M–S)/B will then require a rightward shift, as shown in the figure. A similar situation is also the case for the (H–M)/B band.

Figure 7(b) shows the color–color result side-by-side with our analysis—the points all fall roughly in the same general vicinity, but if the interpretation circles are used literally, then the CSC points all become LMXBs—whereas we would argue for there being some absorbed objects. To properly know which interpretations are correct, we need detailed identifications of the counterparts to each of the point sources.

A relatively long observation can detect variability if the variability of each discrete source is of a relatively short duration. We used the CIAO routine glvary (and its associated setup tasks) to examine the variability of all of the sources. Only four sources emerged as possibly variable: numbers 8, 11, 16, and 30. An examination of the fractional effective area for each of these sources demonstrated that, while variable, the effective areas were all consistently variable, mostly due to the intentional dithering of Chandra. That eliminates changes in effective area as an explanation. We then examined each source in detail.

Source 16 is immediately dismissed as variable because it shows a very small degree of change, well within the uncertainties both before and after the change. In contrast, Source 8 shows a drop about halfway through the observation, from a count rate of 0.00194 ± 0.00020 counts s⁻¹ to a rate of 0.00165 ± 0.00020 counts s⁻¹.

Sources 11 and 30 are more difficult to assess—each has a short-term variation during the time range. Both lie on the glvary boundary between "probably not a variable" and "may be a variable." Neither lies near a chip boundary. Both have low counts, so the statistical variation could be dismissed as small number statistics.

Of the sources in IC 1613 during the Chandra observation, only source 8 appears to warrant some attention for variability. As the Chandra observation of IC 1613 was relatively short, we do not discuss possible variability further in this paper.

If we match the possibly variable sources with counterparts, source 8 lies in the HMXB circle of Figure 6, source 11 lies on the boundary of the LMXB and Absorbed circles, and source 16 lies near the center of the LMXB circle. Longer and deeper X-ray, optical, IR, and radio observations are necessary to confirm or refute any possible counterpart identifications.

Swift UVOT (Roming et al. 2004) images of IC 1613 show a large number of UV-bright sources, particularly in the northeast portion of the galaxy. Figure 8(a) displays the summed UVOT data in the filter UWM2 (∼2100 Å). A total of eight X-ray sources are definitely detected in at least one of the UV bands: sources 1, 2, 4 + 5, 9, 10, 18, and 26 (the SNR); sources 9, 14, and 15 must be investigated further, as possible multiple counterparts are found in the detection region. In general, however, there are few UV counterparts.

Zoom In Zoom Out Reset image size

The same statement may be made about the GALEX F- and N-UV images. We do not show these images, because the correspondence is very close to that shown with the Swift UVOT image discussed above (Figure 8(a)). The list of sources with counterparts in the Swift image matches the list for GALEX counterparts.

We also compared the X-ray-detected sources with images from the Infrared Array Camera (IRAC; Fazio et al. 2004) and the Multiband Imager Photometer for Spitzer (MIPS; Rieke et al. 2004) on board Spitzer.

We use IRAC because the spatial resolution is more than twice that of the Wide-field Infrared Sky Explorer (WISE; Wright et al. 2010). The IRAC frames for IC 1613 have been calibrated so fluxes were directly available. We use MIPS because the spatial coverage matches the Chandra ACIS field of view very well.

The chip offsets in the IRAC camera leave 10 Chandra sources without IR coverage for the 3.6 μ and 5.8 μ bands and four (different) sources without 4.5 μ and 8.0 μ coverage (q.v., Table 8). As a result, we use the WISE 3.4 μ and 4.6 μ bands to fill in some of the missing IRAC measures. We describe the correction from WISE to IRAC below.

In contrast to the UV sources (Section 5.1), nearly every X-ray source has at least one counterpart in one of the IR bands. We show the MIPS data with X-ray sources overlaid in Figure 8(b). Correspondences between the X-ray and their IR counterparts are listed in Table 8; fluxes are listed in Table 9. That every X-ray source has a MIPS counterpart with a significantly detected flux implies that the IRAC data would very likely have shown IR counterparts had the field of IC 1613 been fully covered.

We did make use of AllWISE data (Kirkpatrick et al. 2014) for those sources that fell outside of the IRAC field of view. The AllWISE data allow us to fill in missing 3.6 μ or 4.5 μ values. The AllWISE magnitudes were corrected to the IRAC bands using the relations in Monson et al. (2017): [3.6] = W1 + 0.038 ± 0.010; [4.5] = W2 + 0.027 ± 0.010.

Table 10. Spectral Fits for Cluster MCXC J0105.0 + 0201 ^a

					NH		Model	Flux b
Model	DoF	cstat	χ2/DoF	kTbr	(1022 cm−2)	Abundance	Norm	Absorbed	Unfolded
bremss, CCD-7	424	586.7	1.33	${1.37}_{-0.42}^{+1.03}$	${0.14}_{-0.06}^{+0.07}$	⋯	1.30(80)[−4]	6.76[−14]	1.24[−13]
CCD-8	⋯	⋯	⋯	⋯	⋯	⋯	1.93(16)[−4]	1.7[−13]	2.51[−13]
apec, CCD-7	423	732.4	1.67	${1.54}_{-0.26}^{+0.36}$	${0.21}_{-0.07}^{+0.08}$	<0.14	${3.7}_{-0.8}^{+1.8}$[−4]	8.2[−13]	8.8[−13]
CCD-8	⋯	⋯	⋯	⋯	⋯	⋯	${9.1}_{-2.0}^{+3.2}$[−4]	6.7[−13]	8.1[−13]

Notes.

^a Numbers in () = uncertainty in last digit(s); numbers in [] = 10^XX ; ^b All fluxes in ergs s⁻¹ cm⁻² in the 0.3–8 keV band.

Download table as:  ASCIITypeset image

We plot three different approaches to classification using IR colors. First, Figure 9(a) shows a color–color plane with 3.6 μ–4.5 μ versus 5.6 μ–8.0 μ magnitude differences. The boxes labeled "AGNs" and "Stars" were defined by Allen et al. (2004) and Stern et al. (2007). In this plot, sources 21, 22, 25, and 26 plus sources 7 and 29, which lie on the box edge, would be identified as candidate AGN. The remaining sources would be identified as stars, with sources 8 and 15 of questionable identification because both lie well away from the boundaries of either box.

Zoom In Zoom Out Reset image size

Assuming the sources falling within the AGN box of Figure 9(a) and the sources touching the edge of the box are all AGN, then this exercise yields 6 possible candidates of 15 sources with available IR colors. This is approximately in line with our previous estimate of background objects unrelated to IC 1613.

The second Figure (9) plots the ratio R_Shal−1 =F(5.8 μ)/F(3.6 μ) > 1 (Shalima et al. 2013) versus the X-ray flux. Using this plot, sources 7, 25, and 26 are identified as candidate AGN. Note that in neither plot is source 4 + 5 (discussed at the end of Section 3.4) identified as an AGN, yet this source was detected in the VLA observations as a nonthermal source. Note that sources 25 and 26 appear in the AGN realm based on the criteria of Allen et al. (2004) and Shalima et al. (2013). We will return to that point momentarily.

The third Figure (9) plots the ratio R_Shal−2 = log(F(8.0 μ)/F(4.5 μ)) versus R_Shal−1 = log(F(5.8 μ)/F(3.6) μ)) as in Lacy et al. (2004). This ratio is intended to separate objects with strong blue versus red continua. The AGN box is indicated. Only "clean" sources are plotted, where "clean" equals a detection in all four IRAC bands. Based on this figure, 12 of the 16 completely detected sources fall within the AGN box.

If the Shalima et al. (2013) approach also applies to the ratio of the 8.0 μ/4.5 μ fluxes, i.e., if a value > 1 indicates an AGN, then the "R_Shal−2" ratio implies 11 sources are AGN. That then leads to a background sky density about 80% higher than the Chandra-determined density. The Swift BAT survey showed significant variations across the sky (Ajello et al. 2012). From that paper, IC 1613 lies in or near a region where the AGN density is ≈1.5 × higher than a uniform distribution. We do not push this interpretation, as the background work to establish the "R_Shal−2" ratio has not apparently been done—but the interpretation is reasonably consistent with published behavior.

However, there is an issue here: AGN exhibit significant IR excesses and these have been used to identify them in IR surveys (e.g., Maitra et al. 2019 for the SMC). We note that the known SNR, source 26, lies within the AGN box of Figure 9(a) and has an R_Shal−1 > 1 as defined in the previous paragraph. Very likely this occurs because, in the infrared, SNRs are dusty emission line objects (e.g., Cherchneff 2016), a description equally applicable to AGN (e.g., Mason 2015). This raises a significant caution for identification of AGN through a galaxy, as some of the sources could be SNRs. Such a restriction would not apply for candidate AGN identified in the field. Slightly reassuringly, in Figure 9(c) the SNR lies outside of the AGN box.

Given the proximity of IC 1613, follow-up studies can and should be made of the X-ray-detected sources to identify the optical or infrared counterparts. Such a follow-up would not only be useful for investigating the IC 1613 sources, but could also aid in establishing which of the AGN identification criteria most closely identifies actual AGN. For example, analyses of the spectral indices of detected radio counterparts would likely distinguish SNRs from background AGN.

Given the relative proximity of IC 1613, there is a long list of objects that have been studied at other wavelengths. We have extracted the positions of H II regions (Smith 1975; Price et al. 1990; Hodge et al. 1990), OB stars and associations (Borissova et al. 2004; Garcia et al. 2009), Mira variables (Kurtev et al. 2001), and planetary nebulae (Magrini et al. 2005), and compared each to the detected Chandra sources. None of these objects have been detected.

The possible X-ray detection of a W-O star (Davidson & Kinman 1982; at (J2000) 01:05:01.7, +02:04:20.8) was Lozinskaya et al. (1998)'s original motivation for the ROSAT HRI observation, but they obtained an upper limit. Our Chandra observation likewise does not detect this object, with an upper limit of ≈2 × 10³⁵ erg s⁻¹ based on the counts, or an upper limit of ∼8 × 10³⁵ erg s⁻¹ based on the limiting sensitivity, at the location of the star. While our upper limit improves on the ROSAT value by an order of magnitude, the value remains several orders of magnitude above the to-date measured X-ray luminosities of ∼10^30–31 erg s⁻¹ for W-O and other Wolf–Rayet stars in the Milky Way (Oskinova 2015).

We also do not detect the cluster of galaxies 2MASX J01044836 + 0207107 centered on the cD galaxy at 01:04:48, +02:07:11.3 (J2000.0), and visible through IC 1613. The cluster is easily visible in HST images (e.g., HLA 12304_07_wfc3_uvis_f350m). That location is within the region of highest spatial resolution on CCD 7. We do detect the galaxy cluster MCXC J0105.0 + 0201—this source is discussed in Section 8.1.

The cluster of galaxies MCXC J0105.0 + 0201 (hereafter, J0105; Piffaretti et al. 2011), at a redshift of 0.197, is visible through the southern portion of IC 1613. The cluster is indicated by a green circle in Figure 1 (the middle green circle). In Figure 2, the cluster is positioned between X-ray sources 18 and 28 and straddles the boundary between CCDs 7 and 8 (marked "C"). It is source 33 in Table 2.

We extracted the events using two rectangular regions, one for the source and one for the background. The spectral extraction was carried out twice, once for each portion of the cluster on the two CCDs. Response matrices and effective area files were constructed for each CCD. We fit the resulting four files simultaneously using an absorbed thermal bremsstrahlung model for the source and a power-law model for the background. Model parameters were forced to vary together, with the sole exception of the component norms (Table 10). Note that the adopted model does not include an abundance variable.

The raw spectra of J0105 both peak near 1 keV, while the background components are generally sloped from low- to high-energy (Figure 12). The cluster spectrum merges with the background near ∼4–5 keV. We obtained a temperature of ∼1.4 keV and a column density of ∼0.14 × 10²² cm⁻². The model norms yielded a 0.3–8 keV flux of ∼6.7 × 10⁻¹⁴ erg cm⁻² s⁻¹ (CCD7) and 1.7 × 10⁻¹³ erg cm⁻² s⁻¹ (CCD8), and unfolded fluxes of ∼1.2 × 10⁻¹³ and ∼2.5 × 10⁻¹³ erg cm⁻² s⁻¹, respectively. Summing these values yields a total, unabsorbed flux of ∼3.7 × 10⁻¹³ erg cm⁻² s⁻¹ and a total, 0.3–8 keV, unabsorbed luminosity of ∼4.4 × 10⁴³ erg s⁻¹, assuming the redshift of ∼0.2 is precise and corresponds to the luminosity distance of 1000 Mpc.

Zoom In Zoom Out Reset image size

An optically thin thermal gas model (apec; Foster et al. 2010) yielded a slightly higher temperature of 1.5 keV, but within the uncertainties, both temperatures are identical. The column density is 50% higher, but again, within the uncertainties, the fitted columns are identical. Only an upper limit of 0.14 could be assigned to the abundances.

Based on the XMM-Newton survey of clusters in the COSMOS field, Finoguenov et al. (2007) plot the cluster luminosity versus redshift, with bands representing the ∼10x range in luminosity. Most of the clusters in the survey field lie near a curve of increasing luminosity with increasing redshift. On that plot, J0105 would lie on the high side of the trend for its redshift. To fall within the trend of other clusters at a redshift of 0.2, the luminosity of J0105 would need to be about 5–10x lower. We acknowledge that our flux estimate could be incorrect, given that the flux was estimated by summing over portions of the cluster spread not only over two CCDs, but also over the gap between those CCDs. We do not think, however, that our estimate is incorrect by such a large factor.

Lozinskaya et al. (1998) mentioned that a ROSAT HRI source corresponded to a mag m_v 11.4 star. Our coordinates are essentially identical to those of Lozinskaya et al. (1998), impressive given that the star lies on CCD 8, about 12' from the aimpoint (the southernmost green circle in Figure 1). The variable star GT Cet lies within 3'' of the coordinates, based on the SIMBAD database (Wenger et al. 2000), with no other object within 50''. GT Cet has been labeled a classical Cepheid (δ Cep type; Kazarovets et al. 2011).

Source 32 appears to lie outside of IC 1613 (the northernmost green circle in Figure 1). It falls very near or into the chip gap at the north edge of CCD 6. It does not exhibit a counterpart at UV wavelengths. The archive-available optical and IR data are unusable because diffraction spikes from HD 6375, a star of mag 1.7, cover its location. At present, the only information that appears to be available about this source is its color–color location (Figure 6(c)). It exhibits an upper limit in the hard band, hence it is a medium-soft to soft source. It is the leftmost point in the figure. Based on the color–color values, the source is likely a foreground star, but not an extremely bright one.