Hyper Wide Field Imaging of the Local Group Dwarf Irregular Galaxy IC 1613: An Extended Component of Metal-poor Stars

The resulting catalog has g and i photometry for an enormous number of sources within the HSC field of view, but only a fraction of these sources are resolved stars associated with IC 1613. Many of the sources, especially toward faint magnitudes, are barely resolved (or unresolved) distant background galaxies. A central challenge in resolved stellar populations is separating the stars of interest from these background galaxies.

As a simple but effective means of star/galaxy separation, we use the ratio of PSF to cmodel flux. True stars scatter around a ratio of unity, while galaxies show fainter PSF fluxes than their true fluxes and so are found at lower values. This statistic has been successfully used for star/galaxy separation with a similar data set by Carlin et al. (2017).

Figure 2 plots the PSF to cmodel flux ratio as a function of magnitude, showing that stars are well-separated from galaxies at g ≲ 24.5 and i ≲ 24, but less so at fainter magnitudes. Hence some galaxy contamination is inevitable at the faintest magnitudes. To balance this contamination with a reasonable sample of stars, we classify sources as stars if they are within 1σ of the expected stellar flux ratio of unity (where σ is the per-object uncertainty on the PSF/cmodel flux ratio). There is an asymmetric extension toward f_PSF/f_cmodel < 1 at bright (i < 23) magnitudes. We have confirmed that most of the sources in this feature are RGB stars in the main body of IC 1613, suggesting that the feature is likely due to the elevated background in the crowded inner regions. We also only consider sources detected at 5σ or above, corresponding to a depth of g ∼ 27 and i ∼ 26.5 in the uncrowded regions of the image (Figure 3).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 shows the spatial distribution of the point sources selected in this manner. A red circle of radius 2 kpc (95) is overplotted on the scatter plot to visually set the scale. The galaxy center, as calculated in Section 3, is also marked. It is clear that crowding affects the photometry in these inner regions of the galaxy where few stars are detected. In the next subsection we discuss artificial star tests to determine the regions in which the photometry is reliable. We also note that several stellar overdensities are apparent in the IC 1613 field (e.g., at R.A. = 15.85 deg, decl. = +2.4 deg). Inspection of the color–magnitude diagrams (CMDs) corresponding to these overdensities do not reveal any obvious stellar population that could be associated with IC 1613 or, for instance, a very faint dwarf galaxy at the distance of IC 1613. A detailed, quantitative search for dwarf galaxies associated with IC 1613 will be presented in a future publication.

Zoom In Zoom Out Reset image size

We used artificial star tests to quantify and correct for the incompleteness in point sources as a function of magnitude and position. Fake stars were added into the processed images using the Synpipe software (Huang et al. 2018). The input catalog contained stars with magnitudes 16 < i < 28, linearly weighted so that there are ∼3 times as many objects at i = 28 than at i = 16, and uniformly distributed between $-1\lt g-i\lt 2.5$. A total of 2000 artificial stars were inserted per ∼112 × 112 patch (i.e., ∼16 stars arcmin⁻²). These fake stars were added into images using the measured PSF and noise characteristics of each individual image, and the resulting images were processed in the same way as for the real observations.

Figure 5 shows the completeness as a function of magnitude for both g and i bands. Taking bins of 0.5 mag in each filter, the completeness was computed as the number of output fake stars in each bin divided by number of input fake stars in the same bin. The photometry is more than 80% complete for magnitudes brighter than 26 mag in both filters. Due to crowding, completeness also depends on the distance from the galaxy center in a magnitude-dependent manner. Completeness as a function of position and magnitude is shown in Figure 6. The data are almost 80% complete for both bands for regions that lie outside ∼12' from the center of IC 1613. Hence we exclude this region (<12') that is strongly affected by crowding from our analysis.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7 shows the CMD of point sources within the radial range of 12'–40' (2.5–8.4 kpc) from the center of IC 1613. First we discuss the features not associated with IC 1613: well-measured point sources extend as faint as g ∼ 27 mag, but a large fraction of the sources fainter than 26 mag are unresolved background galaxies (Figure 2). Except for the upper giant branch of IC 1613, the majority of the points redder than $g-i\gtrsim 1.5$ mag are foreground stars of the Milky Way disk.

Zoom In Zoom Out Reset image size

There is also a narrow plume of sources that extends upwards from g ∼ 23 at $g-i\sim 0.5$. This feature appears to be a well-defined main sequence (MS) observed at a narrow range in heliocentric distance. We note that IC 1613 is in the same direction as previous detections of tidal debris from the Sagittarius dwarf spheroidal (e.g., Koposov et al. 2012). The distance of the Sagittarius point sources along the line of sight to IC 1613 is ∼25 kpc (Belokurov et al. 2014). Adopting this distance, and ages and metallicities of the Sagittarius stream found in the literature (e.g., Marconi et al. 1998; Bellazzini et al. 1999), we confirmed the observed feature to be consistent with that expected from Sagittarius. Hence, we conclude that this additional feature in the CMD is indeed debris from the Sagittarius dwarf galaxy along the line of sight to IC 1613.

Focusing on IC 1613 itself, several features that can be clearly attributed to specific stellar populations are visible in Figure 7:

• The RGB, consisting of stars older than 1 Gyr, is the most distinguishing feature in the CMD. It becomes distinct from the blob of unresolved galaxies at g ≳ 25.5. In this same magnitude range, the photometric errors are smaller than the observed width of the RGB; thus the observed thickness must be intrinsic (see Figure 7). This is primarily due to a metallicity spread among the old stellar populations in the galaxy (the expected contribution due to a range of distances along the line of sight is <0.01 mag and hence negligible).
• The red clump is another prominent feature in the CMD, located around g ∼ 25 and $g-i\sim 0.5$. The red clump occurs for intermediate-age to old stellar populations with ages of 1–10 Gyr, though its position varies little with age in this range.
• The MS is the roughly vertical feature of blue stars at −1 ≲ (g − i)  ≲ −0.5, which consists of young (less than ∼1 Gyr) stars. While it makes up only a tiny fraction of the identifiable stars in our radial range, its presence does imply recent star formation in the outskirts of IC 1613.
• The horizontal branch (HB) stars are from ancient stellar populations (age ≳10 Gyr). The HB—clearly visible in the CMD at g ∼ 25—can be separated roughly into two separate features: a red horizontal branch extending blueward from the red clump, and the blue horizontal branch (BHB) extending redward from g − i ∼ −0.5. The apparent gap in the HB is the RR Lyrae instability strip, which is clearly well-populated (see also Cole et al. 1999).

We focus our analysis in this paper on three of these tracers—the young MS stars, intermediate-to-old age RGB stars, and old HB stars. The MIST (Choi et al. 2016; Dotter 2016) isochrones based on ages and metallicities of the populations given in Cole et al. (1999) and Skillman et al. (2003) are overplotted on the CMD at the distance adopted for IC 1613. RGB (age = 10 Gyr, [Fe/H] = −1.5) and MS (age = 100 Myr, [Fe/H] = −0.8) isochrones are overplotted on the CMD and provide a good visual match to the observed populations (Figure 8). In addition, a fiducial ridgeline of an old metal-poor ([Fe/H] = −1.5) globular cluster (NGC 5272; Bernard et al. 2014) is overplotted on the CMD for the horizontal branch. This ridgeline provides a reasonable fit to the shape of the HB, and is used in our selection method (see Section 3.2) to include the most likely HB stars.

Zoom In Zoom Out Reset image size

The three stellar tracers used in this project—MS, RGB, and HB—trace out young, intermediate to old, and old populations, respectively. The weight of a star's likelihood of membership in the stellar tracer selection is calculated using its separation from the isochrone and its photometric errors as:

$$\begin{eqnarray*}&&w=\exp \left(-\tfrac{{\left(q-{q}_{\mathrm{iso}}\right)}^{2}}{2({\sigma }^{2}+{\delta }^{2})}\right),\end{eqnarray*}$$

where δ is the intrinsic width of the IC 1613 stellar locus. For the RGB and MS, q = (g − i) is the color of the point source and q_iso is the corresponding interpolated point on the isochrone for the same g-magnitude. For the HB, q = g is the g-magnitude of the point source and q_iso is the corresponding interpolated point on the isochrone for the same g − i color. σ is the combined photometric error of the point source from both of the filters.

The stellar tracers are separated by selecting all the points with weight w ≥ 0.4. The stars selected for further analysis are shown overplotted on the CMD in the right panel of Figure 8. Only stars with g brighter than 26 mag are considered. This ensures using high signal-to-noise data (see Figure 3) while also avoiding severe unresolved galaxy contamination at the fainter end (Figure 2).

We first focus on determining the overall structure of IC 1613 by finding its center and ellipticity. To calculate the center and ellipticity of IC 1613's outskirts, we repeatedly draw random samples of 15,000 (out of a total 20,274) RGB stars with r > 12' without replacement. The center and ellipticity are taken from the best-fit ellipse for the isodensity contour that is 3σ above the background level outside 25' from the center of the galaxy (e.g., Hammel & Sullivan-Molina 2018). By repeating this process for 1000 iterations, we find the coordinates of the center to be J2000 (α, δ) = (1621405 ± 000650, 212918 ± 000774). This is formally offset by about 11 ± 04 to the northeast of the literature center: (α, δ) = (1619917, 211778), from McConnachie (2012).

By fitting ellipses to different isodensity contour levels, we find that the ellipticity varies minimally from ∼0.10 at ∼13' to ∼0.04 at ∼18', compared to a central ellipticity of ∼0.11 (McConnachie 2012). Our results show that the galaxy has a nearly circular morphology in the outer regions. We assume these values for IC 1613's center and ellipticity for the rest of the paper.

We use surface density profiles to trace out the radial structure of IC 1613 for stars with g < 26. The profiles for each of three stellar tracers (MS, RGB, and HB) are given in Figure 9. Since the ellipticity is nearly zero, we use circular projected radial bins of 1' width to calculate the profiles. The surface density is corrected for completeness by applying the radial corrections shown in Figure 6.

Zoom In Zoom Out Reset image size

All profiles reach the background at ∼22'–24' (∼5.2 kpc; 5 r_h), which is much further than typical studies of old populations in dwarf galaxies, owing to our large field of view. It is interesting that the young MS population extends to almost ∼20'. To rule out BHB contamination, we compared the MS profile to an MS selection limited to g < 25, and found that it does not change this result. The surface brightness that corresponds to these faint regions is ∼33.7 mag arcsec⁻² (in g), which is calculated at the background surface density level of ∼0.62 RGB stars per square arc-minute, by assuming a Salpeter initial mass function (Salpeter 1955) and a 10 Gyr population with [Fe/H] = −1.5.

The gray area in Figure 9 marks the boundary (∼12') beyond which crowding no longer significantly affects our data (see Section 2.2 and Figure 6). Hence, for further analysis, we use only the data points right of the gray area (r > 12').

Next we fit functions to the background subtracted radial profiles. The background for each of the stellar tracers is computed as the completeness corrected surface density between 25' and 38'. We focus on single exponential and broken exponential fits to the data, because they have provided good descriptions of dwarf galaxies in the past (Hidalgo et al. 2003; Vansevičius et al. 2004; Hunter & Elmegreen 2006; Zhang et al. 2012; Herrmann et al. 2013), and experimenting with such fits suggested they would also provide reasonable fits to the IC 1613 data. The functional form for the exponential fit is:

$$\begin{eqnarray*}&&\rho (r)={\rho }_{0}\exp \left(-\tfrac{r}{{r}_{0}}\right),\end{eqnarray*}$$

where ρ₀ is the central surface density and r₀ is the scale length, which is defined as the radius at which the density falls by e. Sérsic profiles also fit the broken exponential profiles (e.g., Herrmann et al. 2013), but to facilitate comparison with past studies of IC 1613's inner profile, we focus on (broken) exponentials for our analysis rather than using Sérsic fits.

For all the profiles, we used a nonlinear least-squares method for the fits to the data from 12' to 25'. For the MS stars, a broken exponential did not represent an improvement over the single exponential fit, while the more complex models had a significantly lower reduced χ² for the RGB and HB stars.

The best quality fits for all three stellar tracers are listed in Table 1 and shown in Figure 10. The bottom panel of the figure shows the surface density profiles after the background is added back. The profiles have been extrapolated in the inner (r < 12') regions and shown as dotted lines. The lower panels in all six plots show the fractional residuals left over after the fits.

Zoom In Zoom Out Reset image size

Table 1.  Scale Lengths of Exponentials for the Three Stellar Tracers

Stellar Tracer	Inner r0 (r < 12')	Intermediate r0	Outer r0	rb	χred2
	(Bernard et al. 2007)	(12' < r < rb)	(r > rb)
MS	119 ± 004	195 ± 013	195 ± 013	⋯	13.4
RGB	38 ± 01	204 ± 005	134 ± 007	158	4.0
HB	⋯	263 ± 023	070 ± 033	172	29.9

Note. The inner scale length is from Bernard et al. (2007). The rest of the columns are from the fits done in this work. r_b denotes the radius of the break between the two exponentials.

Download table as:  ASCIITypeset image

3.4.1. Opposite Structures for Younger and Older Stellar Populations

As discussed in Section 1, previous studies of IC 1613 mostly concentrated on the inner regions of the galaxy. Here we focus on comparisons to the results of Bernard et al. (2007), who also studied the structure of different stellar populations.

Bernard et al. (2007) constructed radial profiles up to ∼15' from the center of IC 1613, and fit exponential functions to populations including MS and RGB stars. For comparison, the scale lengths for these populations are also listed in Table 1.

On comparing these scale lengths with the values that we obtained from our fits (see Table 1), it is clear that our outer (r > 12') profiles do not match with their inner (r < 12') profiles for both RGB and MS stars. This points to a break for both the populations within 12'. Combined with the break that we found for the RGB and HB profiles at ∼165, we conclude that the old population of IC 1613 is made of at least three components, with the profile becoming steeper as we move outwards from the center of the galaxy. The young MS population has at least two components, with the outer component shallower than the inner component.

We list three takeaway points from our investigation of the structure of resolved stellar populations of the dwarf irregular galaxy IC 1613. First, the young MS stars are almost as extended as the old RGB and HB stars, even though the density of young stars is comparatively low in the outer regions of the galaxy. Second, the radial surface density distribution of old stars is steeper in the outer regions compared to the inner regions. Third, the radial surface density distribution of young stars is steeper in the inner regions compared to the outer regions: this behavior is opposite to that of the old stars.

We discuss the interpretation of these findings further in Section 4.

The presence of an outer break in the structure of the old stellar populations in IC 1613 suggests the possibility of an "extra" component, which could be identified with an accreted stellar halo.

Another way to search for evidence of accreted material is to search for evidence of asymmetry in the outer surface density profile. As an initial exploration of this idea, we show a map of the stellar tracers (RGB, HB, and MS) divided into different quadrants is shown in Figure 11.

Zoom In Zoom Out Reset image size

In each quadrant, the surface density of different stellar tracers is calculated using the method described in Section 3.4. The resulting surface density profiles of MS, RGB, and HB stars are given in the left, middle, and right panels of Figure 12, respectively. We do not observe any signs of obvious asymmetry in the four profiles for any of the stellar tracers. Hence we see no evidence, at least from this simple test, for a large-scale asymmetry that could reflect the presence of substantial accretion event(s) in the halo.

Zoom In Zoom Out Reset image size

There are a number of ideas that have been proposed to explain the extended stellar structures in dwarf galaxies. Akin to the classification for more massive galaxies, it is reasonable to separate these explanations into in situ scenarios in which the extended structures are formed through internal galactic processes, and scenarios in which the halos are built through the hierarchical accretion of less massive galaxies.

Focusing first on the in situ scenarios, there is some evidence from numerical simulations of dwarfs for "outside-in" star formation. At early times when the gas supply was high, star formation could occur at larger distances from the center, but as gas is consumed, pressure support decreases and star formation is only viable at smaller radii (Stinson et al. 2009). Scattering in the orbits of old stars and shocks driven by bursts of star formation can also produce more extended distributions of old stars. (Mashchenko et al. 2008; Stinson et al. 2009; Maxwell et al. 2012). Using the FIRE simulations, El-Badry et al. (2016) showed that stellar feedback and bursty star formation can lead to net stellar migration on timescales of a ∼few ×10⁸ yr, as well as pushing star-forming gas to larger radii.

Hybrid scenarios where inner old stars are formed in situ, but redistributed to large radii through galactic interactions, might also be plausible. For example, Zolotov et al. (2009) use simulations to show that a portion of the inner stellar halos of Milky Way mass galaxies were indeed formed near the center of the galaxy and displaced to large radii by major mergers. An alternative model is if, akin to the major mergers thought to have formed some early-type galaxies, a low-mass galaxy was formed through the merger of two lower-mass gas-rich dwarfs. This scenario can lead to a remnant with an extended outer spheroid and an inner star-forming disk (Bekki 2008).

Finally, we consider the standard scenario in which extended structures are built through the accretion of less massive galaxies, as discussed in detail in the Introduction. Perhaps the most obvious evidence that this process has occurred would be the detection of clear streams or tidal features in the halo. More subtle evidence could be the presence of structural features in the radial distribution of stars that could be unambiguously identified with accretion. For example, the stellar density profile of the Milky Way follows a broken power law, with a break at ∼25 kpc (Jurić et al. 2008). This break radius has been interpreted as a pile-up of stars at their orbital apocenters associated with a single massive accretion event (Deason et al. 2013, 2018).

First considering the young MS stars: the presence of such stars at large projected radii (>4.2 kpc; ∼4 r_h) in IC 1613 is puzzling considering that the H i column density falls to ∼10¹⁶ atoms cm⁻² by ∼3.2 kpc (Hunter et al. 2012), and at such low densities star formation is unlikely to occur. A recent burst of star formation that pushed gas to larger radii than presently observed could hence explain the presence of young stars at those distances (El-Badry et al. 2016).

Extending this hypothesis to the older stars, a series of such bursts could also have led to the presence of intermediate-age and old stars at large radii. A possible issue with this scenario is that IC 1613 has an almost constant star formation history (Cole et al. 1999; Skillman et al. 2003, 2014) with little or no evidence of distinct bursts. The necessary averaging of the star formation history into bins could smooth out bursts that occurred on timescales ≲500 Myr. Nonetheless, it cannot be said that there is specific evidence for the sort of bursts that could have led to large-scale rearrangement of the stellar distribution in IC 1613. Hence this hypothesis seems better-supported for the young stars than for the old stars.

Indeed, the lack of bursts of star formation also presents a problem for any scenario positing a major gas-rich merger, since such a merger is expected to lead to a noticeable elevation in the star formation range (Bekki 2008).

In principle, the breaks in the radial profiles of intermediate-age and old stars could be consistent with either in situ outside-in formation (e.g., Stinson et al. 2009) or with accretion of smaller galaxies. The occurrence of breaks in both the RGB and HB stars at a common radius of ∼165 (∼3.5 kpc = 3.4 r_h) might be more easily explained in an accretion scenario: accreted galaxies might well have both intermediate-age and old stars that would end up on similar orbits, while similar break radii for a wide range of ages would not seem to be a straightforward prediction of a gradual outside-in formation scenario.

Not all evidence favors the accretion scenario: the lack of evidence for streams or other obvious tidal material is notable. In addition, IC 1613 has no known globular clusters, long considered as luminous tracers of stellar halos (Brodie & Strader 2006), even at large radii (see, e.g., the Local Group search of di Tullio Zinn & Zinn 2015). We did not find any new globular cluster candidates in our data, though without formal artificial cluster tests (planned for a future paper) we cannot quantify these limits on the presence of clusters.

Overall, we conclude that the extended intermediate-age and old components in IC 1613 have observed properties consistent with those expected for accreted stellar halos, but that in situ scenarios, such as outside-in star formation, could also explain some of the observed properties of these components, and may separately be required to explain the unusually extended young stars in the galaxy.

Detailed numerical simulations could help with the interpretation of this and future, similar data sets. In particular, it would be useful to know whether simulated dwarf galaxies with M_⋆ ∼ 10⁷–10⁹ M_⊙ show evidence for complex radial profiles among their intermediate and older stellar populations, even without a clearly bursty star formation history. This is one example of the need to better understand the relationship between the outer structure of the galaxy and its in situ star formation history. This would better allow one to infer information about the accretion history of the dwarf from the structure of its stellar populations.

On the observational side, spectroscopy of large samples of luminous red giants in the outer regions of IC 1613 would be feasible, and could allow more sensitive searches for evidence of accretion events in phase space than the relatively simple searches in projected physical space that we conducted in this paper.

Looking to future facilities, LSST will extend wide-field observations of dwarf galaxies to comparably faint magnitudes over the entire southern sky. From space, the James Webb Space Telescope and Wide Field Infrared Survey Telescope will also provide complementary deep and wide-field images that can resolve stars to large radii and offer superior star/galaxy separation to faint magnitudes, allowing one to construct the full star formation history of IC 1613 and similar dwarfs over a wider range of radii.