NEBULAR METALLICITIES IN TWO ISOLATED LOCAL VOID DWARF GALAXIES

KK246 is a particularly interesting object in the SIGRID sample as it has been identified as the most isolated dwarf galaxy known in the Local Volume (D ≲ 10 Mpc; Karachentsev et al. 2013). It is situated a few Mpc within the Local Void (Tully et al. 2008; Kreckel et al. 2011; Nasonova & Karachentsev 2011). This void is large, at least 23 Mpc in radius (Tully et al. 2008), and KK246 has no apparent neighbors within ∼3 Mpc (Karachentsev et al. 2004). Assuming a spatial peculiar velocity of 100–200 km s⁻¹ this translates into a time of 15–30 Gyr since the last possible galaxy–galaxy encounter. As such, the object is an ideal laboratory in which to study galaxy evolution and self-enrichment in isolation. It is of particular interest in understanding the intrinsic evolutionary processes in the formation of this small galaxy, as almost certainly it has never been affected by tidal disruption by nearby large galaxies, nor by enrichment of the intergalactic medium (IGM) from which it formed by outflows from such galaxies. It has no detectable adjacent dwarf galaxies, unlike the low metallicity low surface brightness galaxies in the Lynx-Cancer void at 18 Mpc investigated by Pustilnik et al. (2011a).

Previous observations of KK246 have shown that the stellar component is embedded in a large surrounding H i cloud (Kreckel et al. 2011) and contains an old stellar population (Karachentsev et al. 2006). Although it currently has only one small, low-luminosity star-forming H ii region, we have been able to measure the nebular metallicity of that region using the WiFeS IFU spectrograph on the ANU 2.3 m telescope at Siding Spring Observatory. This paper reports the first measurement of the nebular spectrum and resultant nebular metallicities.

Using Hubble Space Telescope (HST) imagery and photometry, Karachentsev et al. (2006) measured the TRGB distance to KK246 as 7.83 Mpc. Tully et al. (2008) use a more recent calibration of the T_rgb scale from Rizzi et al. (2007) and give a distance of 6.4 Mpc. With either distance value, KK246 is within the Void (Tully et al. 2008). In the calculations in this paper, we have used the value from Tully et al., but our conclusions do not depend on which value is adopted.

The object's isolation is indicated by the low value of the Tidal Index Θ = −2.2 (Karachentsev et al. 2004) and the SIGRID Δ index (not measurable: no potential tidally effective objects within 10° or ∼1.3 Mpc radius, Nicholls et al. 2011), implying no tidal interference from any nearest neighbor galaxy within a Hubble time.³ Karachentsev et al. (2004) have concluded that there are no potential interacting galaxies within 3 Mpc. KK246 is also listed in the "Local Orphan Galaxies" catalog (Karachentsev et al. 2011) as having the highest recorded isolation index (log₁₀ (ii) = 3.22) using an alternative isolation measure.

KK246 is one of over 4500 H i objects detected in the HIPASS survey (HIPASS J2003-31, Meyer et al. 2004) and was identified with the optical counterpart galaxy ESO 461-G036 (Doyle et al. 2005). It was subsequently observed using the Very Large Array at 21 cm at higher resolution by Kreckel et al. (2011), who found that KK246 is surrounded by a very extended H i region, at least ten times the diameter of the stellar disk.

Kreckel et al. (2011) find a large dynamical mass (4.1 × 10⁹ M_☉) and consequently one of the highest measured dynamic-mass-to-light ratios, M_dyn/L_B  =  89, further emphasizing the unusual nature of this galaxy. Other noteworthy high dynamic-mass-to-light ratio objects include NGC 2915 (M_dyn/L_B  =  76, Meurer et al. 1996), ESO 215-G?009 (M_dyn/L_B  =  22, Warren et al. 2004), and NGC 3741 (M_dyn/L_B  =  107, Begum et al. 2005). However these objects have H ii regions (using GALEX FUV flux as a diagnostic) that are either much brighter (NGC 2915, 3741) or more spatially extended (ESO 215-G?009), than KK246.

As part of a survey of Local Volume galaxies aimed at measuring TRGB distances, Karachentsev et al. (2006) observed KK246 using the HST. The left panel of Figure 1 shows the composite F606W-band image, obtained from the HST archive, from those observations. The location of the H ii region observed in this work is marked.

Zoom In Zoom Out Reset image size

Karachentsev et al. (2006) resolved the brightest two magnitudes of the color–magnitude diagram (CMD) for KK246 from HST WFC3 images, which shows clear evidence of an old stellar population. The large distance to KK246 makes star formation modeling difficult, which rules out calculating the star formation history in detail, but by comparing its CMD with the CMD data for 60 nearby dwarf galaxies in the ANGST survey from Dalcanton et al. (2009), and their star formation histories analyzed by Weisz et al. (2011), it appears likely that KK246 has had at least three epochs of star formation, including the present one involving the H ii region. The right panel of Figure 1 shows our interpretation of the CMD from Karachentsev et al. (2006).

KK246 is detected in both the NUV and FUV images from Galaxy Evolution Explorer (GALEX; see Table 1). The low FUV flux [luminosity] of 28.2 μJy implies that there is only a small population of hot young stars currently present in the galaxy. Using Equations (17) and (18) from Karachentsev et al. (2013) we infer a star formation rate of log (SFR) = −2.43. This is consistent with our H_α observations (see below). Figure 2 (bottom, left) shows the composite GALEX image. Although the resolution and sensitivity are marginal, it is apparent that the brightest UV object coincides with the H ii region.

Zoom In Zoom Out Reset image size

Table 1. Compendium of UV, Optical, Near-IR, and H i Parameters for KK246

Parameter	Value	Reference
R.A. (J2000)	20 03 57.4	Lauberts (1982)
Decl. (J2000)	−31 40 54	Lauberts (1982)
l (deg)	9.72864	Derived from Lauberts (1982)
b (deg)	−28.36894	Derived from Lauberts (1982)
H ii region R.A. (J2000)	20 03 57.7	This work
H ii region Decl. (J2000)	−31 40 48	This work
Dist (Mpc)	6.4a	Tully et al. (2008)
mB (mag)	17.22	Tully et al. (2008)
AB (mag)	1.10	Schlafly & Finkbeiner (2011)
EB − V (mag)	0.154	Schlafly & Finkbeiner (2011)
MB, 0 (mag)	−12.91	This work
log10(LB/L☉)	7.524	This work
mH (mag)	13.9 ± 0.2	Kirby et al. (2008)
MH, 0 (mag)	−15.9 ± 0.3	Kirby et al. (2008)
FUV flux (μJy)	28.18 ± 3.89	GALEX Data Release 6/7
NUV flux (μJy)	26.05 ± 3.72	GALEX Data Release 6/7
Total H i flux (Jy km s−1)	7.3	Kreckel et al. (2011)
	7.5	Meyer et al. (2004)
W20 (km s−1)	93	Kreckel et al. (2011)
	104.5	Meyer et al. (2004)
W50 (km s−1)	71.6	Meyer et al. (2004)
$\log _{10}(\mathcal {M}_{{\rm H\,\scriptsize{I}}}/{\mathcal {M}}_{\odot }$)	8.02 ± 0.03	Kreckel et al. (2011)
	8.03	Derived from Meyer et al. (2004)
$\log _{10}(\mathcal {M}_{\ast }/{\mathcal {M}}_{\odot }$)	7.7 ± 0.2	Kirby et al. (2008)
$M_{\rm H\, \scriptsize {I}}/L_B$	4.656	This work
Θ tidal index	−2.2	Karachentsev et al. (2004)
Δ tidal index	(nil)	Nicholls et al. (2011)
log10(ii) isolation index	3.32	Karachentsev et al. (2011)

Note. ^aTully et al. (2008) use a more recent calibration of the T_rgb scale from Rizzi et al. (2007) and give a distance of 6.4 Mpc, compared to the value of 7.83 Mpc from Karachentsev et al. (2006). This does not affect the conclusions reached here.

Download table as:  ASCIITypeset image

The same figure also shows the visible color image from DSS2 (top left), the single H ii region from a Hα + R-band WiFeS data cube (top right), and an H-band image from Kirby et al. (2008) (bottom right). The small H ii region corresponds to the point marked in the HST image (Figure 1) above. The physical extent, spectrum (see below), and Hα flux confirm that this is an H ii region and not a planetary nebula, unlike the object discussed by Makarov et al. (2012) in the isolated dwarf spheroidal galaxy KKR25.

Table 1 summarizes information available on KK246.

Another interesting galaxy from the SIGRID sample in this context is HIPASS J1609-04. It is located on the edge of the Local Void, 7.65 Mpc from its center (Nasonova & Karachentsev 2011). It is one of the more isolated galaxies in the SIGRID sample (Nicholls et al. 2011), with a Δ tidal index of −2.9. It has an isolation index of 2.38 (Karachentsev et al. 2011), compared to 3.22 for KK246. Its distance of 14.82 Mpc is estimated from local flow data for the Virgo, Shapley, and GA fields. Its baryonic mass is about four times larger than KK246, and it is substantially more actively star forming.

Figure 3 shows the GALEX NUV+FUV, visible, H_α and R-band, and H-band images. There are several active star forming regions. The H-band image is from the VISTA Phase III survey⁴ and the Hα + R-band is from the SINGG survey (Meurer et al. 2006).

Zoom In Zoom Out Reset image size

Table 2 summarizes information available on J1609-04. Stellar masses were calculated using the methods discussed by Kirby et al. (2008), based on the stellar mass-to-light ratio models of Bell & de Jong (2001). Bell & de Jong (2001) found that the major contributor to uncertainties in estimating total stellar mass from H- or other IR band magnitudes was the initial mass function (IMF). As this is not well known, we have adopted the same approach as Kirby et al. (2008). Drawing on their more extensive observational base, we assume the same mass-to-light ratio, $\Upsilon _*^H=1.0\pm 0.4$.

Table 2. Compendium of UV, Optical, Near-IR, and H i Parameters for J1609

Parameter	Value	Reference
R.A. (J2000)	16 09 36.8	NED
Decl. (J2000)	−04 37 12.6	NED
l (deg)	7.134450	Derived from Lauberts (1982)
b (deg)	32.607155	Derived from Lauberts (1982)
Dist (Mpc)	14.82	NED
mB (mag)	15.35	Doyle et al. (2005)
AB (mag)	0.862	Schlafly & Finkbeiner (2011)
EB − V (mag)	0.201	Schlafly & Finkbeiner (2011)
MB, 0 (mag)	−15.50	Derived here
log10(LB/L☉)	8.39	From NED
mH (mag)	13.26 ± 0.06	From VISTA Phase III imagerya
MH, 0 (mag)	−17.59 ± 0.06	From VISTA Phase III imagery
FUV flux (μJy)	361.26 ± 29.68	GALEX Data Release 6/7
NUV flux (μJy)	345.87 ± 17.19	GALEX Data Release 6/7
Total H i flux (Jy km s−1)	7.2	Meyer et al. (2004)
W20 (km s−1)	111.2	Meyer et al. (2004)
W50 (km s−1)	71.9	Meyer et al. (2004)
$\log _{10}(\mathcal {M}_{{\rm H\,\scriptsize{I}}}/\mathcal {M}_{\odot }$)	8.55 ± 0.03	Meurer et al. (2006)b
$\log _{10}(\mathcal {M}_{\ast }/{\mathcal {M}}_{\odot }$)	8.37 ± 0.22	From VISTA Phase III imageryc
$M_{\rm H\, \scriptsize {I}}/L_B$	1.45 ± 0.10	Derived here
Δ tidal index	−2.9	Nicholls et al. (2011)
log10(ii) isolation index	2.38	Karachentsev et al. (2011)

Notes. ^aThe H-band magnitude was calculated from the VISTA image using the standard equation m_H = −2.5log₁₀(flux) + photozp. ^bRe-calculated using the flow-corrected distance taking into account the Virgo Cluster, the Great Attractor, and the Shapley concentration. ^cThe major part or the error (±0.2) arises from uncertainties in determining the mass-to-light ratio from the H-band luminosity. See above.

Download table as:  ASCIITypeset image

For the stellar mass and error estimates, photometry was performed on the VISTA H-band FITS image, using an elliptical aperture large enough to cover the complete galaxy to measure the total H-band flux. The same ellipse was used to measure the noise characteristics in five adjacent, signal free areas (no apparent stars or background galaxies). The errors quoted in Table 2 combine the effect of the (large) mass-to-light ratio uncertainties and the (small) average of the luminosity noise measurements.

The spectra of both KK246 and J1609-04 show, at most, only faint stellar continua, and as a result measurement of equivalent widths of the spectral lines is unreliable. The brighter H ii regions 2 and 5 in J1609-04 exhibit faint broad stellar absorption underlying the Hβ, γ, and δ nebular emission lines, but the stellar continuum is faint. Measurement noise makes it difficult to estimate the equivalent widths of the absorption features, but they have been taken into account in estimating emission line fluxes. In addition, our reduction pipeline does not currently allow accurate rendering of the absolute continuum for faint continua at the short wavelength end of the blue spectrum, below 4200 Å.

Equivalent widths for emission lines are similarly problematic. The low stellar continuum creates large EW values for the bright lines, e.g., EW [O iii 5007] = 135 for J1609-04(5), with reliable errors difficult to estimate. For the UV [O ii] lines, the measurement noise is sufficiently large that EW estimates are meaningless. For this reason we have not included EW measurements in this paper.

For the two brightest regions in J1609-04 (2 and 5), the [O iii] auroral line λ4363 is evident. For these we can measure the metallicity using the direct T_e method. To improve the signal-to-noise ratio, we have combined the spectra of the two regions. Using the models presented in Dopita et al. (2013); Nicholls et al. (2013), we calculate the [O iii] electron temperature from the combined spectra, as: T_e = 12,806 ± 556 K. Using the methods described in Izotov et al. (2006), the oxygen ([O iii]+[O ii]) metallicity calculated from the electron temperature of the combination of regions 2 and 5 of J1609-04 is 7.96 ± 0.06. As is characteristic of metallicities derived conventionally from the T_e method (López-Sánchez et al. 2012), this value is lower by ∼0.2 dex than the value determined using the latest strong line methods (Dopita et al. 2013).

Table 6 shows the complete line ratios for κ = ∞, with two sets of averages. The first set takes the means over all strong line diagnostics. The second, significantly more precise, uses only the mean values of diagnostic ratios using [N ii]/[S ii]. The latter diagnostic ratios accurately return input values from test spectra generated by the Mappings IV code, whereas the diagnostic ratios involving [N ii]/[O ii] return significantly higher values, in some cases. It is not clear yet whether this indicates a problem with the diagnostic itself, or, more likely, whether the interpolation process (Dopita et al. 2013) is at fault. Further work is necessary. In any case, as shown in Figures 7 and 8, the errors in observed flux ratios are significantly greater for ratios involving [O ii]. For this reason we adopt values for Z and log(q) using only diagnostics involving [N ii]/[S ii]. Other ratios including [N ii]/[Ha] tend to confirm the [N ii]/[S ii] values.

As shown in Table 6, the metallicity for KK246, assuming an equilibrium electron energy distribution, is 8.17 ± 0.01, and for the H ii regions in J1609-04, the metallicity values span a range between 8.02 and 8.20. For a non-equilibrium κ electron energy distributions with κ = 50 and 20, the metallicities are virtually identical, a consequence of choosing strong line diagnostics that are not especially sensitive to κ, and which therefore result in very consistent metallicities.

The interpretation of the new diagnostic ratios used here is explored in depth by Dopita et al. (2013), but there is scope for further work. In the meantime, especially for low metallicity H ii regions, diagnostics involving [N ii]/[S ii] appear to offer very consistent results.

One of the more important parameters in understanding galactic evolution is the nitrogen metallicity, and in particular, the ratio of nitrogen to oxygen. The observations reported here include good measurements of both [N ii] and [O ii], allowing us to explore the values of log(N/O) for each H ii region. To calculate the value of N/O from [N ii] and [O ii] line fluxes, we use empirical formulae from Izotov et al. (2006), Equations (3) and (6). These reduce to:

$$\begin{eqnarray} \log \left(\frac{{\rm N}}{{\rm O}}\right) &=& \log \left(\frac{{\rm N\,\scriptsize{II}}\,6584+6548}{{\rm O\,\scriptsize{II}}\, 3726+3729}\right) + 0.273 - 0.726/T_{e4}\nonumber\\ && +0.007*T_{e4} - 0.02*\log (T_{e4}), \end{eqnarray} \tag{ 1 }$$

where T_e4 is the electron temperature in units of 10,000 K. This equation differs only by a small constant offset (0.033) from that quoted by Pagel et al. (1992, Equation (9)). We assume the same electron temperature for O ii and N ii (reasonable, as they both arise primarily from the outer parts of the H ii region), and further, that N⁺/O⁺ = N/O, following Pilyugin et al. (2010) and others. The errors from these assumptions are likely to be of the same order as the measurement uncertainties. The results are shown in Table 7.

The values of log(N/O) for these regions are uniformly low, and similar to the value adopted by Lebouteiller et al. (2013) for 1Zw18, −1.61 ± 0.10.

In these data, as noted above, the value of Z (see table 6) was calculated using only the averaged values derived from diagnostics involving the ratio [N ii]/[S ii], as these diagnostics (at present) return the most accurate values (±0.01) of metallicity for test spectra using Mappings IV.

Figure 9 shows log(N/O) versus Z, calculated as above, plotted with the extensive data from van Zee et al. (1998). A linear fit to the distribution is usually taken as indicating objects with secondary nitrogen only (i.e., produced locally by previous generations of stars) (van Zee et al. 1998; Vila Costas & Edmunds 1993). The current observations smoothly extend to lower values the relationship between log(N/O) and Z in the van Zee data, and are consistent with there being mainly secondary nitrogen present in the H ii regions observed, and low primary nitrogen, as defined by Vila Costas & Edmunds (1993). There is no evidence of the transition "knee" from secondary (diagonal slope) to primary nitrogen (horizontal slope) at low (oxygen) metallicity. This behavior appears reasonable for isolated dwarf galaxies with low or occasional star formation, and away from any possible enrichment of the IGM by large galaxies and galaxy mergers.

Zoom In Zoom Out Reset image size

The low (N/O) values are not fully consistent with the "no winds" models proposed by Gavilán et al. (2013), which strongly suggest a secondary to primary transition at metallicities found in the objects reported here. This may indicate that the models need to take into account recent star formation, and different IMFs than previously considered. It is conceivable that relatively low stellar masses in these H ii regions may be subject to "small number statistics," which make IMF modeling difficult. The previously reported scatter of log(N/O) at low metallicity may be another feature of small number statistics, as proposed by Carigi & Hernandez (2008).

Mass versus metallicity behavior is one of the important evolutionary diagnostics for galaxies. It has been extensively mapped for larger galaxies (e.g., Tremonti et al. 2004), but it is less well known for dwarf galaxies (Lee et al. 2006). Exploring it was one of the motivations for the SIGRID sample (Nicholls et al. 2011). Figure 10 shows the stellar masses versus metallicities for KK246 (red) and J1609-04 (yellow, summed over all regions) plotted with data on similar low mass objects from Lee et al. (2006) for comparison. Metallicity for the Lee objects was taken from several sources of good quality spectra, estimated using the T_e method (and older atomic data). The values for KK246 and J1609-04 have been reduced by 0.2 dex to compensate for the typical difference between strong line and T_e method metallicities. Figure 10 shows that KK246 and J1609-04 follow the same trend as Lee et al.'s non-void dwarf objects, although, as has been found for void galaxies (Pustilnik et al. 2011b), J1609-04 falls a little below the trend.

Zoom In Zoom Out Reset image size

Another question of interest is how the mass to light ratio versus absolute magnitudes of these two galaxies compare with other similar galaxies. Are they intrinsically low luminosity objects? Values for $M_{\rm H\, \scriptsize {I}}/L_B$ are shown in Tables 1 and 2: for KK246 the value is 4.66 and for J1609-04, 1.45. In Figure 11 we plot KK246 and J1609-04 on data from Figure 2 from Warren et al. (2007). The shaded area shows 752 more luminous galaxies from the HIPASS Bright Galaxy Catalog. The dashed line shows the locus of an upper limit for the H i mass-to-light ratio, as a function of luminosity. The individual galaxies marked with error bars are those discussed by Warren et al. (2007) and comprise late-type galaxies with a wide range of physical parameters. It is clear that there is nothing extreme about the two objects presented here—their $M_{\rm H\, \scriptsize {I}}/L_B$ ratios are typical of the 38 late-type galaxies with a range of mass-to-light ratios discussed by Warren et al. (2007).

Zoom In Zoom Out Reset image size

The subject of galaxies in voids in the Local Volume has received increasing attention recently (Pustilnik et al. 2011a, 2011b; Kreckel et al. 2012). Pustilnik et al. (2011a, 2011b) have investigated galaxies in the Lynx-Cancer Void using their own and Sloan Digital Sky Survey (SDSS) data, and find that in general, the metallicities are lower by ∼30% than for higher density regions. In a study using SDSS data for galaxies in Local Volume voids, van de Weygaert et al. (2011) and Kreckel et al. (2012) found that, despite predictions from simulations, there was no significant population of H i-rich low-luminosity galaxies filling the voids. They also found evidence of cold-mode accretion of H i in some objects. Both studies included late-type spiral galaxies as well as dwarf irregulars. For 48 Lynx-Cancer Void galaxies from Pustilnik et al. (2011b), the listed metallicity varies between 7.14 and 8.36. These values are measured using the T_e method, or using a semi-empirical method based on it, using older atomic data, and are likely underestimated by ∼0.2 dex compared to strong line metallicities (López-Sánchez et al. 2012; Nicholls et al. 2013).

Figure 12 shows metallicity versus M_{B, 0} for KK246 and J1609-04 plotted in the data for 42 Lynx-Cancer Void galaxies from Pustilnik et al. (2011b). The metallicities for KK246 and J1609-04 have been reduced by 0.2 dex to reflect the discrepancy between T_e metallicities and strong line metallicities. The dashed line is for dI galaxies from Lee et al. (2003). J1609-04 is typical of the other galaxies plotted, and lies near the trend defined by the dIrrs. KK246 is somewhat more metal rich, or conversely, less luminous in the B-band, than the others. This may be explained if the galaxy is nearing a "post-starburst" phase, with fewer blue stars present than would be expected from the current metallicity.

Zoom In Zoom Out Reset image size