Near-infrared Stellar Populations in the Metal-poor, Dwarf Irregular Galaxies Sextans A and Leo A

Ground-based observations of Sextans A and Leo A were obtained through the broadband JHK_s filters using the WIYN High-resolution Infrared Camera (WHIRC; Meixner et al. 2010) mounted on the 3.5 m WIYN telescope at the Kitt Peak National Observatory. The central wavelengths and bandwidths of the three filters can be found in Table 2. The WHIRC camera field of view is 34 × 34 with a plate scale of ≈01 pixel⁻¹. Details of the observations can be found in Tables 3 and 4. The data were collected over four observing runs over a period spanning from 2010 May to 2011 April. Observations were centered around R.A. = 10^h08^m295 and decl. = −0°26^m45^s for Sextans A, and R.A. = 09^h59^m261 and decl. = +30°44^m555 for Leo A. Dedicated background control fields were also observed.

Table 2.  WHIRC Filter Characteristics

Filter	${\lambda }_{c}$(μm)	${\rm{\Delta }}\lambda$(μm)
J	1.250	0.1619
H	1.651	0.3009
${K}_{{\rm{s}}}$	2.150	0.3430

Download table as:  ASCIITypeset image

Table 3.  Observation Log for Sextans A

Band	Observation Date	Interval (days)a	Exp. Time (s)	Seeing FWHM ('')	Airmass
J	2010 May 02	0	120	1.8	1.35–1.45
J	2011 Jan 23	266	180	1	1.70–1.95
J	2011 Jan 24	267	180	1.1	1.72–2.57
J	2011 Apr 15	348	180	⋯	1.26–1.41
H	2011 Jan 24	⋯	180	1.6	1.30–1.43
${K}_{{\rm{s}}}$	2011 Jan 23	0	60	0.7	1.25–1.33
${K}_{{\rm{s}}}$	2011 Jan 23	0	80	0.7	1.48–1.25

Note.

^aInterval is the separation in days between the first observational epoch and latter epochs.

Download table as:  ASCIITypeset image

Table 4.  Observation Log for Leo A

Band	Observation Date	Interval (days)	Exp. Time (s)	Seeing FWHM ('')	Airmass
J	2010 Jan 22	0	180	0.8	1.19–1.03
J	2011 Apr 30	463	120	1.4	1.04–1.12
H	2011 Jan 24	⋯	180	1.6	1.01–1.00
${K}_{{\rm{s}}}$	2011 Jan 22	⋯	90	0.9	1.01–1.38

Download table as:  ASCIITypeset image

The data were reduced in IRAF following the methods detailed in Bruursema et al. (2014) and the steps outlined in the WHIRC Reduction Manual (Joyce 2009).⁷ Briefly, an array linearity correction was performed using the WHIRC task wprep.cl, and the mscred package was used to remove the pupil ghost from the bias-subtracted flat field. The pupil ghost template and pupil removal mask were downloaded from the WIYN-WHIRC NOAO website.⁸ Each individual science image was then sky-subtracted and flat-fielded using a median-filtered sky frame. Finally, bad pixels were removed using standard bad pixel masks and the IRAF fixpix routine.

To determine the astrometry for Sextans A and Leo A, the PyRAF script starfind was used to find the positions of the brightest stars in each field. These stars were then compared to the positions of stars detected in the Two Micron All-sky Survey (2MASS; Cutri & 2MASS Team 2004; Skrutskie et al. 2006) to find the common stars in the image. Using these 2MASS stars as reference stars, the PyRAF packages xyxymatch and geomap were used to find and correct for position offsets in the frames so that the astrometry of the reduced images to match that of the 2MASS images. Field distortions in the images were corrected using the geomap and geotrans tasks in IRAF. Final mosaicked images in each filter for the galaxies were produced using Montage.⁹ The WHIRC maps of Sextans A and Leo A are shown in Figure 1.

Zoom In Zoom Out Reset image size

Source extraction, aperture photometry, and related functions were performed with astropy and its recently developed subsidiary package photutils¹⁰ (Astropy Collaboration et al. 2013).

The DAOstarfinder function was used to identify point sources in the images. Sharp and round limits were implemented in order to obtain a reliable list of point sources; this cut effectively removes cosmic-ray hits and image artifacts, and minimizes the number of extended sources and resolved background galaxies from the source catalog. Aperture photometry was performed on the resulting source list for each image. The radius of apertures was determined using the predicted full width half maximum of the image resolution, and then adjusted, based on the photometric quality. An outer annulus was used to determine the frame background level, and varied with each image proportionally to the area of the inner aperture.

Preliminary catalog cuts were implemented before flux calibration. Sources with error $\geqslant 65$% or with flux <0 were eliminated, as they (a) have a poor signal-to-noise ratio, (b) are background subtraction artifacts, or (c) are non-detections. The remaining sources are considered to be high-confidence point sources with reliable flux measurements.

Detector counts (ADU) of our good sources were calibrated into janskys (Jy) via known 2MASS sources (Cutri & 2MASS Team 2004; Skrutskie et al. 2006) in the images. With many of our images relying on 4–10 calibration-stars, a systematic conversion error of 2.5% was added to the calibrated flux to account for scatter in the resulting corrections. However, the relative photometry accuracy within the images is better, as it only depends on the S/N of the source and the Poisson noise.

2.4.1. The WHIRC Point-source Catalog

The point-source catalog detection statistics are summarized in Table 5. Table 6 lists the 779 sources included in the final point-source catalog for both of the galaxies in our sample. For a source to be included in the full catalog, we require it to be detected in at least two bands or at two epochs with high confidence. We consider a source to be a match if their centroids are within a 1'' radius. If a source was detected over multiple epochs, we present the mean flux and uncertainty for the combined epochs. The WHIRC high-reliability catalog only includes sources detected in two or more WHIRC bands. This is less complete than the full catalog but has a more stringent acceptance criteria.

Table 5.  Detection Statistics

Wavebands	Leo A	Sextans A
Total Detected	287	492
$J,H\,\&\,{K}_{s}$	38	176
J & H	98	88
J & ${K}_{{\rm{s}}}$	1	24
H & ${K}_{{\rm{s}}}$	3	11
High-reliability Catalog	140	299
WHIRC Variables	50	30

Download table as:  ASCIITypeset image

Table 6.  Catalog of Point Sources for Leo A and Sextans A

Galaxy	ID	${\rm{R}}.{\rm{A}}.$	$\mathrm{Decl}.$	U	B	V	I	J	$\sigma J$	H	$\sigma H$	K	$\sigma K$	[3.6]	[4.5]	[5.8]	[8.0]	Variable
LeoA	0	149.8928	30.7250	−99.99	23.72	22.48	21.19	19.19	0.04	19.10	0.07	−99.99	−99.99	19.73	19.64	−99.99	−99.99	0
LeoA	1	149.8593	30.7257	−99.99	23.44	22.34	21.06	19.31	0.04	−99.99	−99.99	−99.99	−99.99	20.12	19.04	−99.99	−99.99	0
LeoA	2	149.8522	30.7259	−99.99	24.57	23.98	−99.99	17.07	0.08	16.87	0.09	15.89	0.10	18.06	17.76	16.30	−99.99	0

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Figure 2 shows a histogram of the number of sources detected in the JHK_s photometric bands as a function of magnitude. The faint source detection limit for the ${K}_{s}$ data of Leo A is almost two magnitudes brighter than for Sextans A due to the high sky background. Although less deep, this data is useful in constraining the SEDs of the brightest sources in the galaxy.

Zoom In Zoom Out Reset image size

2.4.2. Ancillary Photometric Data

In order to fully investigate the nature of the sources in our near-IR catalog; we match our source catalog to all available photometric catalogs for Leo A and Sextans A on a purely positional basis. A matching distance of 3'' was adopted, and if multiple matches occured within a single catalog, only the closest match was kept. The ancillary data was obtained from the Hubble Space Telescope (HST; Bianchi et al. 2012), the Large Binocular Telescope (LBT; Bellazzini et al. 2014), the Subaru Prime Focus Camera (Suprime; Stonkutė et al. 2014), and the Spitzer Space Telescope (Boyer et al. 2009, 2015b, 2015c). These catalogs are not contemporary with the WHIRC data. However, they can aid in source classification and place constraints on the SED from the optical to the mid-IR. Table 7 lists each ancillary catalog, the photometric bands and the number of sources in common with our master catalog.

Table 7.  Number of Matches between the WHIRC Catalogs and Other Photometric Catalogs

Survey	Filters	Telescope	Leo A	Sextans A
WHIRC—High Reliability	${{JHK}}_{{\rm{s}}}$	WIYN	140	299
DUSTiNGS; Boyer et al. (2015b)	[3.6], [4.5]	Spitzer	125	272
DUSTiNGS Var; Boyer et al. (2015c)	[3.6], [4.5]	Spitzer	2	9
Boyer et al. (2009)	[5.8], [8.0]	Spitzer	58	120
Bianchi et al. (2012)	$U,B,V,I$	HST	⋯	177
Bellazzini et al. (2014)	$g,r$	LBT	⋯	292
Stonkutė et al. (2014)	$B,V,I$	Subaru	138	⋯

Download table as:  ASCIITypeset image

Most sources (>90%) are matched to at least one of these ancillary catalogs. At longer wavelengths, the number of matches decreases, as dust-enshrouded objects are rarer than stellar photospheres. Of the matched sources, nine stars in Sextans A and two stars in Leo A were identified as AGB variable star candidates by Boyer et al. (2015b, 2015c). Our catalog also includes the two known planetary nebulae (Magrini et al. 2003) in these galaxies. WHIRC sources without optical counterparts are rare, however, the large differences in angular resolution and survey depth increases the likelihood for mismatches compared to the WHIRC-IRAC matches. We discuss these objects further in Section 4.

Evolved stars pulsate, and are variable on timescales of hundreds of days. These pulsations levitate atmospheric material, leading to the formation of dust grains. Thus stars with long-period variations can show a significant IR excess. Our data examines source variability for three epochs, separated by 82, 267, and 363 days for Sextans A, and two epochs separated by 463 days for Leo A. We identify large-amplitude variables from our J-band data using the variability index defined by Vijh et al. (2009). This variability index takes into account the error-weighted flux difference between epochs. Figure 3 shows the variability index distribution for Sextans A and Leo A. Following Vijh et al. (2009), sources with $| \mathrm{Var}| \gt 3$ are classified as variable. This conservative threshold (adopted from the LMC's IR stellar populations) corresponds to a J-band amplitude of a least 0.25 mag for our sample. We find that 30 variables in Sextans A and 50 variables in Leo A meet this criteria (Table 6).

Zoom In Zoom Out Reset image size

Our total sample has 11 sources in common with the DUSTiNGS variable catalog. Of these, we identify three as variable using our J-band criteria. The majority of these candidate variable stars are bright, with well-characterized errors. Variability indices for fainter stars are less constrained, as small changes in amplitude will be masked by the photometric errors, this effects the identification of the lowest amplitude variable stars in our catalog and those stars significantly obscured by dust (Whitelock et al. 2017). This bias is more pronounced for Sextans A due to its greater distance.

In order to classify the variables, we use colors and magnitudes to separate AGB stars from other sources. Near-IR colors are particularly useful for roughly discriminating between O- and C-rich AGB stars, and other stellar populations due to their distinctive spectral characteristics. O-rich stars have spectra dominated by VO, TiO, and H₂O molecules, while stars with an excess of carbon in the atmosphere are characterized by CN and C₂. The CMDs constructed from the dereddened photometric measurements (see Section 3.4) are shown in Figure 4.

Zoom In Zoom Out Reset image size

Sources are identified as TP-AGB stars or RSGs candidates if they are brighter than the RGB tip and redder than foreground objects. The empirical relationship determined by Valenti et al. (2004) was used to estimate the near-IR RGB tip in Leo A and Sextans A. Due to the low mass of these systems, identifying other features in the CMDs is challenging. Instead, we adopt the cuts to the ${K}_{s}$ versus $J-{K}_{s}$ CMD used by Cioni et al. (2006) and Boyer et al. (2011) to separate the C-AGB and O-AGB stars, and identify dusty AGB stars (which experience molecular blanketing) as those with $J-[8.0]\gtrsim 3.4$ mag or $J-{K}_{s}\gtrsim 2.2$ mag (Blum et al. 2006; Boyer et al. 2011). The location of these cuts, determined from the molecular content and opacity in the stellar photospheres, are shown in Figure 4. In near-IR CMDs, O-AGB stars populate a narrow sequence extending to brighter magnitudes and redder colors, above the tip of the red giant branch (TRGB); carbon stars occupy a broad region to the right of the O-AGB sequence in the form of a red branch (McDonald et al. 2012; Boyer et al. 2015a).

The brightest sources in the CMDs are foreground stars and they have relatively blue colors. Using the trilegal Galactic population synthesis code (Girardi et al. 2005), we expect about 10 foreground stars brighter than J = 21 mag in our Leo A field, and 30 foreground stars in Sextans A, all with $J-H\lt 0.6$ mag.

It should be noted that the variation in metallicity between the galaxies may result in changes to the IR color. The metallicity of the population affects the separation between the parallel lines shown in the top panel of Figure 4. At higher metallicity, O-rich AGB stars occupy a broader range of J − K colors, at low-metallicity their color distribution is narrower. Thus caution is necessary in adopting these color classifications, as the location of the cuts are dependent on the metallicity of the galaxy and the extinction correction, limiting the classification accuracy. As both Leo A and Sextans A have populations formed across a range in metalicities, this may affect the photometric chemical classification of approximately 20% of the AGB stars in our sample.

For sources without K_s-band detections, we use the H-band data to determine an initial class following the cuts proposed by Sibbons et al. (2012). Stars with $J-H\lt 0.6$ mag are most likely to be main-sequence stars or foreground dwarfs, and stars with $J-H\gt 1.0$ mag are classified as carbon stars. For completeness, we also use the $J-[3.6]$ versus [3.6] CMD (i.e., $J-[3.6]\gtrsim 3.1$ mag) suggested by Blum et al. (2006) to separate the O-rich, C-rich, and dusty AGB populations. This is necessary as molecular blanketing due to the fact that dust strongly affects the near-IR spectral region.

For red sources, contamination from unresolved background galaxies may be significant. Focusing on the variable star population eliminates the majority of this contamination. Conversely, given the SFHs of Sextans A and Leo A, contamination from massive young stellar objects ($M\gt 5\,{M}_{\odot }$) is likely very small. The vertical sequence seen in the $J-[3.6]$ CMD is populated by objects with a range in stellar effective temperatures (see Section 3.3), and little IR-excess. In both galaxies, the number of very red objects is modest. The results of the photometric classifications of evolved star candidates are listed in Table 8.

Table 8.  Photometric Classification: Evolved Star Candidates

Population	Leo A	Sextans A
$J\lt {\mathrm{TRGB}}_{[J]}$	57	295
O-rich AGB	10	57
C-rich AGB	18	62
Dusty AGB stars	3	18
RSGs	1	8
Foreground	2	3

Download table as:  ASCIITypeset image

In order to calculate the effective temperature (${T}_{\mathrm{eff}}$) and bolometric luminosities (${M}_{\mathrm{bol}}$) of the sources in Sextans A and Leo A, we used the SED-fitting code of McDonald et al. (2012, 2017). This code performs a ${\chi }^{2}$-minimization between the observed SED (corrected for interstellar reddening) and a grid of bt-settl stellar atmosphere models (Allard et al. 2011), which are scaled in flux to derive a bolometric luminosity.

The bt-settl spectral models are based on atomic and molecular data; for each galaxy, we used literature values (given in Table 1) for the metallicity, distance, and interstellar reddening. A ${\chi }^{2}$-minimization is performed, comparing the observed SED to a blackbody function. The derived ${T}_{\mathrm{eff}}$ and ${M}_{\mathrm{bol}}$ is then used as a starting point for an iterative ${\chi }^{2}$-minimization against the interpolated stellar model atmosphere grid. In this way, a precise (±200 K) fit to the data is achieved. For cases where the SED is not well-sampled, uncertainties are higher. Furthermore, in crowded fields, or areas with background nebulosity, luminosities may be more uncertain.

Figure 5 shows the Hertzsprung–Russell diagram of Leo A and Sextans A we have produced using the derived effective temperatures and bolometric luminosities for all the stars in our field of view. In Leo A, a concentration of stars on a well defined RGB is clearly visible, with the RGB tip at approximately 3800 K, 2500 ${L}_{\odot }$. A small AGB population appears to continue onward from this, ascending to the top right of the diagram. To the left of the RGB is a population of hotter stars, these are likely to be foreground stars, main-sequence stars, or point sources with insufficient IR data constraining the SED. However, some post-AGB stars, OB stars, Herbig Ae/Be stars, and certain kinds of binary stars, could be present in this region at higher luminosities. In Leo A, sources cooler than 3500–3000 K, with luminosities less than 3000 ${L}_{\odot }$ are likely to be background galaxies, although some galaxies may also reside among the RGB and AGB stars.

Zoom In Zoom Out Reset image size

In Sextans A, there is a large amount of scatter in the H–R diagram, and the precise location of the RGB is difficult to define. The scatter is due to a more populated main-sequence and giant branches, combined with worse photometric accuracy. Foreground stars are visible toward the top left of the H–R diagram, while background galaxies lie at cooler temperatures and low luminosities.

In an attempt to separate the RGB and AGB stars from the rest of the population in Sextans A, we overplot our variable star candidates and compare their location to that of the RGB in Leo A. When comparing the two galaxies, it is clear that in Sextans A we are sensitive to only the brightest AGB stars and our H–R diagram is therefore incomplete, even at the tip of the RGB, while in Leo A we reach a fainter giant branch population. There may also be a difference between the bright AGB populations. In Sextans A, the scatter in the AGB population may be a result of a higher concentration of carbon stars due to the lower bulk metallicity of the galaxy.

Mass-loss rates for every evolved star in Leo A and Sextans A can be computed by fitting their broadband spectral energy distributions (SEDs) with radiative-transfer models. This allows us to characterize the cumulative dust production in these galaxies, and constrain the evolutionary status and chemical type of the individual stars.

To fit the SEDs of all the stars in our catalog, we use the grid of dusty RSG and AGB models (grams; Sargent et al. 2011; Srinivasan et al. 2011). These models were computed using the 2Dust radiative-transfer code (Ueta & Meixner 2003), and stellar photosphere models for O-rich (Kučinskas et al. 2005, 2006) and C-rich (Aringer et al. 2009) AGB stars. The O-rich dust is modeled using astronomical silicates (Ossenkopf et al. 1992), and for carbon stars, a mixture of amorphous carbon (amC) and 10% silicon carbide (SiC) was used.

To model the stars, we use SEDs constructed from the mean flux at each wavelength and we correct for interstellar reddening using the Cardelli et al. (1989) extinction law with ${R}_{{\rm{v}}}=3.1$ mag. Uncertainties in the optical and near-IR photometric data were increased in quadrature using representative LPV amplitudes, to account for stellar pulsations (see Srinivasan et al. 2016). This allows us to fit SEDs for which a precise phase correction could not be determined.

As the grams models are computed for stars in the LMC, the model fluxes must be scaled before performing the fitting. After correcting for distance and extinction, an optimal scaling factor (η) was determined by minimizing the difference between the observed and model fluxes. We then perform a ${\chi }^{2}$ minimization between the models and the data to obtain the best-fit parameters, which include the DPR, and chemical type. The ${\chi }^{2}$ value is defined by

$$\begin{eqnarray}&&{\chi }_{\nu }^{2}=\displaystyle \frac{1}{N}\sum _{i}\displaystyle \frac{{({f}_{{\mathrm{obs}}_{i}}-{f}_{{m}_{i}})}^{2}}{{\sigma }_{i}^{2}},\end{eqnarray} \tag{ 1 }$$

where ${f}_{{\mathrm{obs}}_{i}}$ and ${f}_{{m}_{i}}$ are the observed and model flux in the ith band with error bar ${\sigma }_{i}$, and N is the total number of data points. As a scaling factor (η) has been applied to the models, the model luminosity and wind parameters must also be adjusted. The luminosity scales by η and the mass-loss rate by the factor $\sqrt{\eta }$, following Equation (2) of Groenewegen (2006).

To avoid degeneracies in the model fits, we compare the best-fit parameters from the carbon- and oxygen-rich models separately, adopting an approach similar to that of Srinivasan et al. (2016). Here we consider all models that satisfy ${\chi }^{2}-{\chi }_{\mathrm{best}}^{2}\lt 2{N}_{\mathrm{data}}$, to be acceptable, where ${\chi }_{\mathrm{best}}^{2}$ is the reduced ${\chi }^{2}$ value of the best-fitting model and ${N}_{\mathrm{data}}$ is the number of photometric data points. The uncertainty in each best-fitting parameter is determined from the median absolute deviation of these models. The chemical classification (carbon or oxygen rich) is established from the model with the smallest overall ${\chi }^{2}$, with a chemical uncertainty determined from the ${\chi }^{2}$ ratio of the best-fit carbon- and oxygen-rich models.

Figures 6 and 7 show the SEDs and best-fitting models for the six dustiest sources in each galaxy. In both figures, the range of acceptable fits for each source is shown in grey, and, where appropriate, both the best-fit carbon and oxygen-rich model. In instances where only one chemistry provides an acceptable fit to the data, only that chemical type is shown. For the reddest sources, and SEDs with high ${\chi }^{2}$ values, we carefully inspect the SED and model output to verify the quality of the fitting procedure. In most instances, objects with a high ${\chi }^{2}$ have SEDs that are inconsistent with a stellar object; they are disjointed, with a poorly defined shape (produced from a mismatch between catalogs) or have too few valid data points. We consider these SEDs to be invalid and discarded their fits from the dust-budget estimates. The results of the model fits for 162 stars in Leo A and 281 stars in Sextans A are listed in Table 9.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 9.  Luminosities and Dust-production Rates of the 443 Evolved Stars with Valid Fits in Leo A and Sextans A

Galaxy	WHIRC_ID	${T}_{\mathrm{eff}}$	${M}_{\mathrm{bol}}$	DPR	σDPR	GRAMS Chem.
		(k)		(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)	(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)
LeoA	3	3800	−4.1	5.0E−11	1.1E−11	C
LeoA	4	4100	−3.7	2.3E−13	4.8E−12	O
LeoA	7	2500	−6.1	1.3E−09	4.3E−10	C
LeoA	9	4200	−3.2	1.4E−13	2.9E−12	O

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

3.4.1. Uncertainties in DPRs

In Figure 8, we compare the relative uncertainty in the DPR as a function of the DPR for our sample. Sources with DPRs of $\mathrm{log}(\mathrm{DPR})\lt -11$ have large uncertainties and are consistent with stellar photospheres with no apparent circumstellar excess due to dust. In both Sextans A and Leo A, the majority of the stars detected (∼70%) are low-DPR sources hence they do not contribute to the dust budget of the galaxy.

Zoom In Zoom Out Reset image size

When comparing the DPR between galaxies, we have assumed that there is no change in the dust properties with metallicity, and that the wind expansion velocity is constant. A fixed expansion velocity is unlikely if radiation pressure on dust is the primary accelerant of the wind. This has been observed by Marshall et al. (2004), Groenewegen et al. (2016), and Goldman et al. (2017) who found a decrease in expansion velocity for stars with comparable properties at lower metallicity. Furthermore, Jones et al. (2012, 2014) and McDonald et al. (2010) have shown that the composition of oxygen-rich dust species has a dependence on metallicity. These systematically increase the absolute uncertainty in both the individual DPRs and the global dust injection rates (see Table 2 of McDonald et al. 2011 for a detailed analysis of the inherent uncertainties in DPRs). However, the star-to-star variations are much smaller.

The chemical classification determined from the best-fit grams models also has a large influence on the uncertainty in the DPRs. Low-confidence chemical classifications, where both the O- and C-rich models produce almost an equally good fit to the data, can have DPR estimates that differ by over an order of magnitude. Figure 9 shows the uncertainty in the chemical classification. The majority of the sources in our sample with $\mathrm{log}(\mathrm{DPR})\gt -11$ were deemed to have a large chemical uncertainty, with a value close to unity. This is a consequence of a lack of data at mid-IR wavelengths, necessary for breaking the degeneracy in dust chemistry.

Zoom In Zoom Out Reset image size

The differences in the temperature and luminosity distributions between the stellar populations of Leo A and Sextans A can be clearly seen in Figure 10. In Leo A, there is a significant decline in source density toward brighter magnitudes at ${M}_{\mathrm{bol}}=-3.6$, which corresponds to the TRGB (Ferraro et al. 2000). Stars below the TRGB are of low mass ($M\sim 0.8-2\,{M}_{\odot }$), likely from a population-aged between 1 and 13 Gyr. The depth of the photometry for Sextans A is not sufficient to reach the TRGB; here the luminosity function has a rapid decline toward fainter magnitudes at ${M}_{\mathrm{bol}}=-4.5$. Our reported AGB population should therefore be considered a lower limit. The discontinuity near ${M}_{\mathrm{bol}}=-4.5$, may also occur if there was a significant increase in the SFR a few hundred megayears ago.

Zoom In Zoom Out Reset image size

Leo A has a blue population of giants that have low circumstellar extinction and populate fainter magnitudes, which is consistent with RGB or early-AGB stars. Leo A has only a few red sources that appear to be evolving on the TP-AGB, these stars are bluer than those in Sextans A. The presence of carbon stars in the CMDs provides confirmation of an intermediate-age population in both Leo A and Sextans A. Only intermediate-mass AGB stars (∼2–4 ${M}_{\odot };$ Marigo & Girardi 2007; Ventura & Marigo 2010) can become carbon-rich, although the transition luminosity is lower in metal-poor populations. Sextans A appears to have a higher relative fraction of carbon stars compared to Leo A (Table 9), and its evolved stellar population (above the TRGB) is more luminous.

In both galaxies, there are a few stars with bolometric magnitudes brighter than the traditional AGB luminosity limit of ${M}_{\mathrm{bol}}=-7.1$ (Wood et al. 1983). These bright sources represent a small population of RSGs or hot-bottom-burning AGB stars (with $M\gt 8\,{M}_{\odot }$), whose ages are consistent with a recent increase in the SFR ∼1 Gyr ago (Dolphin et al. 2005; Cole et al. 2007). The discrepancy in the size of the AGB populations unlikely due to the difference in stellar mass of the host galaxy. More likely it is due to differences in both the mean metallicity and SFH.

Stars with $\mathrm{log}(\mathrm{DPR})\gt -11$ have a measurable IR excess due to dust. There are 32 stars in Leo A and 101 stars in Sextans A, which meet this criteria, with a low-moderate relative uncertainty in the DPR. As the uncertainty in the chemical classification is high for most sources (Figure 9), we compute four values for the total dust-production rate: (1) using the best-fit models irrespective of dust chemistry; (2) assuming all stars are O-rich; (3) assuming all stars are C-rich; and (4) only including stars identified as variable. Table 10 lists the total DPRs for each galaxy. These estimates exclude any source with a relative uncertainty in the DPR > 0 (see Figure 8).

Table 10.  The Global Dust-production Rates for Leo A and Sextans A

Population	Leo A		Sextans A
	Total DPR	Median DPR	Total DPR	Median DPR
	(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)	(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)	(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)	(${M}_{\odot }\,{\mathrm{yr}}^{-1}$)
Best-fitting Chemical Type	$(8.2\pm 1.8)\times {10}^{-9}$	$8.1\times {10}^{-11}$	$(6.2\pm 0.2)\times {10}^{-7}$	$1.8\times {10}^{-10}$
If all O-rich	$(6.5\pm 2.8)\times {10}^{-9}$	$6.2\times {10}^{-10}$	$(6.3\pm 1.5)\times {10}^{-7}$	$4.5\times {10}^{-10}$
If all C-rich	$(5.0\pm 1.5)\times {10}^{-9}$	$8.1\times {10}^{-11}$	$(4.4\pm 1.3)\times {10}^{-8}$	$1.1\times {10}^{-10}$
Variables Only	$(2.9\pm 0.8)\times {10}^{-9}$	$2.1\times {10}^{-10}$	$(5.7\pm 0.1)\times {10}^{-7}$	$1.6\times {10}^{-9}$

Note. Best-fitting chemical type: 25 C-rich and 6 O-rich in Leo A, and 68 C-rich and 33 O-rich in Sextans A. The total DPR in Sextans A should be considered a lower limit, as we do not detect AGB stars to the tip of the RGB.

Download table as:  ASCIITypeset image

In both galaxies, the total dust input into the ISM is dominated by one or two stars. In Leo A, the star with the highest DPR is LeoA-143, with a DPR of $(2.1\pm 0.1)\times {10}^{-7}$ ${M}_{\odot }\,{\mathrm{yr}}^{-1}$. This is two orders of magnitude greater than the second-highest DPR in Leo A. This object has a rising SED and was identified as a candidate PNe from Hα observations by Magrini et al. (2003), hence we exclude this source from the global dust-budget estimate.

In Sextans A, the two sources with the highest DPRs are IDs 17 and 448 (SSTDUSTG J101059.43-044249.3 and J101058.05-044304.1, with $(5.5\pm 0.1)\times {10}^{-7}$ ${M}_{\odot }\,{\mathrm{yr}}^{-1}$ and $(8.1\pm 3.9)\times {10}^{-9}$ ${M}_{\odot }\,{\mathrm{yr}}^{-1}$, respectively). These also have rising SEDs; however, both of these sources have been identified as variable x-AGB candidates by DUSTiNGS (Boyer et al. 2015c). Our fitting routine classified SextansA-17 as an oxygen-rich AGB-star; it is responsible for 89% of the global dust production of Sextans A. If this source is instead carbon rich, then its DPR is significantly lower: $(1.2\pm 0.3)\,\times {10}^{-8}$ ${M}_{\odot }\,{\mathrm{yr}}^{-1}$. If SextansA-17 is confirmed to be an oxygen-rich AGB star, then it conclusively shows that individual metal-poor ($[\mathrm{Fe}/{\rm{H}}]\sim -1.85$ to −1.40) evolved stars can produce significant amounts of dust, and contribute substantially to the enrichment of the ISM in their host system.

Figure 11 compares the DPR to a source's bolometric magnitude. For the stars with a circumstellar excess, $\mathrm{log}(\mathrm{DPR})\gt -11$, no trend is seen between mass-loss rate and luminosity. This is surprising, as both luminosity and DPR should increase during the stars' evolution on the AGB, as expected from most mass-loss prescriptions (e.g., Vassiliadis & Wood 1993; Bloecker 1995). A similar effect was seen in the LMC by Riebel et al. (2012). Three luminous carbon stars are observed in Sextans A, which are in excess of the maximum AGB luminosity limit; however, all three have low-confidence chemical classifications. If they are indeed carbon-rich they are unlikely to be AGB stars associated with the galaxy and are presumably foreground stars. All stars brighter than ${M}_{\mathrm{bol}}=-9.0$ are considered foreground sources, and are excluded from global DPR estimates.

Zoom In Zoom Out Reset image size

The most extreme AGB stars, producing the largest amount of dust are near to the end of their AGB evolution. The RGB-tip is at approximately ${M}_{\mathrm{bol}}\sim -3.6$ mag, TP-AGB stars should all be brighter than this limit (e.g., Rosenfield et al. 2014). Consequently, our data represents a near-census of the TP-AGB stars in Leo A and we can separate the dust-producing AGB stars from the larger RGB population. In Sextans A, we do not reach the RGB tip, as we are limited by sensitivity, so our dust-production value for this galaxy should be considered a lower limit.

4.2.1. Comparison of Global Dust-production