5 Candidates for very low metallicities

There has been a long debate on whether the BCDs are truly young dwarf galaxies undergoing their first burst of star formation or whether the present burst occurs within an older galaxy. The low abundances found in these galaxies make them good candidates for the least chemically evolved galaxies. But the detection of an extended faint stellar underlying component in the majority of BCDs (Loose & Thuan 1986; Kunth et al. 1988; Telles et al. 1997; Vennik et al. 1999) supports the idea that they are not truly primordial galaxies, but older LSB dwarf galaxies undergoing transient periods of star formation. A clear demonstration of this two - component structure was recently presented by Schulte-Ladbeck et al. (1998, 1999) for the very nearby BCD VII Zw 403 (UGC 6456), based on HST photometry of its individual stars down to the red giant branch stars. Even for I Zw 18, the lowest metallicity known galaxy, Garnett et al. (1997) found C/O to be much larger than the mean value of other metal-deficient galaxies. This was interpreted as a proof that I Zw 18 has experienced carbon-enriching episodes of star formation in the past, and is therefore not a young galaxy. Recently Aloisi et al. (1999) used synthetic colour-magnitude diagrams to investigate deep HST data for I Zw 18, and they found again that the present burst is not the first one to occur in this galaxy.

However, Izotov & Thuan (1999) argued that the extreme low metallicity BCD SBS 0335-052 is a good example for a young galaxy; the HST V and I imaging of this galaxy (Thuan et al. 1997) showed extremely blue colours, not only in the region of current star formation but also in the extended low surface brightness underlying component. Thuan & Izotov (1997) also argued that the large HI cloud associated with this BCG (Pustilnik et al. 1999) is made of pristine gas, unpolluted by metals. Based on abundance measurements, Izotov & Thuan (1999) suggested that in fact all galaxies with 12+log(O/H)$\leq7.6$ are young, with ages not exceeding 40 Myr, while those with $7.6 < 12+{\rm log}({\rm O}/{\rm H}) < 8.2$ are not older than 1-2 Gyr. Furthermore, Lynds et al. (1998) discussed the possibility that VII Zw 403 (UGC 6456) is not young (as also found by Schulte-Ladbeck et al. 1998, 1999), but could still be considerably younger than a Hubble time.

While the controversy on the age of the BCDs reflects our current knowledge of galaxy formation and evolution (Schulte-Ladbeck et al. 1999), more work on both statistical samples of very low-metallicity BCDs as well as HST imaging of nearby BCDs could give clues relevant to this debate. In this section we give a list of candidates for very-low metallicity BCDs, which were found in the spectroscopic analyses of our sample. Nevertheless, high-resolution spectroscopy is needed to confirm the metallicity that we assign for each object.

 

 

Table 7:

Galaxy	F(H $\beta)_{{\rm obs}}$ [ergs-1cm-2]
	HII1	HII2	HII3	HII4
HS 1232+3846	4.25e-16	2.33e-15	4.19e-16	3.62e-16
HS 1236+3821	5.01e-15	1.90e-16
HS 1244+3648	1.25e-15	1.39e-14
HS 1304+3529	2.57e-15	4.42e-14	1.45e-15
HS 1312+3508	2.12e-15	1.00e-13
HS 1340+3307	6.20e-16	9.11e-15
HS 1614+4709	5.92e-14	1.17e-15	5.81e-15
HS 1641+5053	1.37e-15	1.73e-14

 

 

Table 8:

Galaxy		Intensitya
HS 0915+5540	D
HS 1304+3529	D	5
HS 1312+3847	C
HS 1312+3508	D	4
HS 1424+3836	D	9
HS 1442+4250	C
HS 1507+3743	D	6
HS 1657+5735	D	4
HS 1728+5655	C

^a Dereddened line fluxes relative to I(H${\beta }$) = 100.

Figure 1: The distribution of the oxygen abundances 12+log(O/H)

 

 

Table 9: Candidates for very low metallicity galaxies

Galaxy	12+log(O/H)	MB
HS 1222+3741	7.64	-17.67
HS 1236+3937	7.47	-15.67
HS 1304+3529	7.66	-17.38
HS 1318+3239	7.81	-17.22
HS 1319+3224	7.59	-15.3
HS 1330+3651	7.66	-17.15
HS 1347+3811	7.94	-15.30
HS 1424+3836	7.97	-16.06
HS 1440+3120	7.91	-17.4
HS 1442+4250	7.89	-14.73
HS 1507+3743	7.74	-17.58

Figure 2: The oxygen abundances (12+log(O/H)) - absolute magnitude relation as derived by Skillman et al. (1989) for the local dwarf irregular galaxies (open squares). Some selected BCDs from Thuan et al. (1997) are also plotted (open triangles) together with our low-abundance candidates (filled dots)

 

 

Table 10: Frequency distribution of the ELG types in the University of Michigan survey (UM) and in our sample

Type	this paper	UM
	[%]	[%]
SS	4.8 $\pm$ 2.4	9.9
DHIIH	39.3 $\pm$ 6.8	29.7
HIIH	26.2 $\pm$ 5.6	24.8
DANS	14.3 $\pm$ 4.1	12.4
SBN	8.3 $\pm$ 3.1	13.2
IP	7.1 $\pm$ 2.9	9.9

We derived the metallicities for our sample using both the models of Dopita & Evans (1986) as well as a five level atom program and the ionisation correction method of Mathis & Rosa (1991), the latter for the few objects for which the [OIII]${\lambda}$4363 line was detected in the spectrum. The distribution of 12+log(O/H) (Fig. 1) has a maximum around 8.5, with a long tail towards low and very low metallicities. All the galaxies with metallicities less than 8.0 were considered candidates for very low metallicity objects and are listed in Table 9. The value of the metallicity we assign should be taken with caution, and only as a preliminary result. As a simple consistency check,

we used the data from Table 9 to plot the metallicity - luminosity relation of star forming dwarf galaxies, as given by Skillman et al. (1989). Since the dwarf galaxies used by Skilmann et al. are all dwarf irregular galaxies, we further added some BCD's from Thuan et al. (1997). Our data distribute in the same region as the Thuan et al. BCDs, with a similar amount of scatter and are - within the errors - in good agreement to the relation proposed by Skillman et al. (1989) (see Fig. 2). We conclude that the errors in our preliminary 12+log(O/H) determinations are small enough to indicate the existence in our sample of some interesting low-abundance BCDs. However, we cannot exclude a small offset in our O/H scale.

Figure 3: The distribution of the H${\beta }$ luminosities (corrected for internal absorption for: a) the whole sample; b) SS and DHIIH galaxies; c) HIIH and DANS galaxies; d) the sample of Salzer et al. (1989); e) the sample of Terlevich et al. (1991); f) The logarithm of the H${\beta }$ luminosities as a function of blue absolute magnitude. The different symbols indicate the ELG type: ${\circ }$ - Searle-Sargent Galaxy, $\times$ - Dwarf HII Hotspot Galaxy, ${\Box }$ - HII Hotspot Galaxy, $\triangle$ - Dwarf Amor. Nuclear Starburst, $\Diamond$ - Starburst Nucleus, $\ast$ - Interacting Pair. The solid line represent a linear least-square fit to the data