D   UV-   1.0 < z < 2.5
Stefan Noll1, Daniele Pierini1, & Maurilio Pannella1
1
Introduction
Nearby starburst galaxies, showing vigorous star-formation, probably contain dust with an extinction curve lacking a broad bump
centred at 2175 Å (Calzetti et al. 1994, ApJ 429). This is typical of
the Small Magellanic Cloud (SMC), whose harsh environments,
i.e. strong radiation fields and shocks, and/or low metallicity do
not allow the presence of a large amount of the grains that are carriers of the UV bump (Gordon et al. 2003, ApJ 594). However,
the supershell region surrounding 30 Dor in the Large Magellanic
Cloud (LMC) contains such carriers, which are ubiquitous in the
diffuse interstellar medium of the Milky Way.
Are there star-forming galaxies at high redshift exhibiting the dust
absorption feature at 2175 Å in their spectra? While for most object types the answer to this question is no or at least doubtful
(e.g., Vijh et al. 2003, ApJ 587; Wang et al. 2004, ApJ 609;
York et al. 2006, astro-ph/0601279), Noll & Pierini (2005, A&A
444, NP05) find that the majority of red UV-luminous galaxies at
2 < z < 2.5 show clear evidence of the 2175 Å bump. These
galaxies indicate strong dust reddening (Noll et al. 2004, A&A
418) and seem to be massive and metal rich (Mehlert et al. 2002,
A&A 393; Daddi et al. 2004, ApJ 600; Shapley et al. 2004, ApJ
612). In order to investigate whether the carriers of the 2175 Å feature also exist in other galaxy populations, we extend our study of
UV-continuum properties to additional UV-luminous and massive
galaxies at 1 < z < 2.5.
MPE Garching
z < 1.2 show very strong 2175 Å features indicated by γ34 < −5.
At 1.5 < z < 2.2 there are only two candidates for galaxies with
γ34 < −2. In one case the redshift is uncertain and in the other
case R = 24.4, which does not fulfil the R < 24 criterion. Hence,
objects with γ34 < −2 like those found at z ∼ 2.4 seem to become
very rare at 1.5 < z < 2.2. On the hand, due to selection effects
our 1.5 < z < 2.2 galaxies are too luminous (see Fig. 4) to prove a
possible decrease of objects typically found at z ∼ 1.1.
F 1: The UV-bump indicator γ34 versus redshift for 108 actively
also seems to be true for galaxies at 1 < z < 1.5. The average γ34
decreases with increasing reddening parameter βb as expected for
extinction curves showing a significant UV bump. For β b < −1.5
the Ks-weighted average hγ34 i = −0.44 ± 0.18, while for βb > −0.5
hγ34i = −3.48 ± 0.51.
Fig. 3 shows a comparison of the Ks-weighted composite spectra
of 1 < z < 1.5 galaxies with R < 24 having a weak (γ34 > −2)
and a strong (γ34 < −2) 2175 Å feature, respectively. The presence of the feature in the more reddened composite with γ34 < −2
is evident. Moreover, the redder spectrum seems to have a higher
relative contribution of an older stellar population at > 3800 Å.
This is consistent with the observed lower average Ks magnitude
of 19.7 compared to 20.7 for the γ34 > −2 composite. The equivalent widths (EW) of the star-formation indicator [O II] of both
composites do not show a significant difference.
The luminosities at 1500 Å of the 1 < z < 1.5 galaxies are generally lower than those of the sample objects at 2 < z < 2.5 (see
Fig. 4). For γ34 < −2 the high-redshift galaxies show a 4 − 5×
higher average L1500 than the objects at lower redshift. Since the
UV continuum reddening is similar for both subsamples, the ratio
of the intrinsic UV luminosites (Leitherer et al. 2002) should be
of the same order of magnitude, if the relation between reddening
and attenuation does not significantly change with redshift. Hence,
the 1 < z < 1.5 galaxies with γ34 < −2 have obviously lower starformation rates (SFRs) than their counterparts at 2 < z < 2.5.
star-forming galaxies at 1 < z < 2.5 selected from the FDF Spectroscopic Survey (blue), the K20 Survey (red), and the GDDS (green).
Crosses mark galaxies without a UV continuum slope derivation. They
are no more considered in Figs. 2 to 4.
The Sample
In addition to the 34 galaxies at 2 < z < 2.5 with RVega < 24.8 already studied by NP05, we have also selected 32 1 < z < 2 objects
with R < 24 from the I-limited FORS Deep Field (FDF) Spectroscopic Survey (Noll et al. 2004). This sample is complemented
by 34 star-forming galaxies at 1 < z < 2.3 selected from the spectroscopic catalogue of the K20 Survey in the Chandra Deep Field
South and a field around the quasar 0055-2659 (Cimatti et al. 2002,
A&A 392; Mignoli et al. 2005, A&A 437). Besides Ks < 20 all objects but two have R < 24. Finally, we consider eight star-forming
galaxies at 1.5 < z < 2.3 from the I and K-limited Gemini Deep
Deep Survey (GDDS, Abraham et al. 2004; AJ 127), all showing
R < 24.7 and Ks . 21.0.
F 4: Luminosity at 1500 Å versus redshift for 88 galaxies of the
FDF Spectroscopic Survey (blue circles) and the K20 survey (red circles). Galaxies with γ34 < −2 are marked by filled symbols.
F 2: The UV-bump indicator γ34 versus the reddening measure βb
Analysis
for our FDF (blue circles), and K20 galaxies (red circles), and a comparison sample of 27 local starburst galaxies (crosses) observed with
IUE (see NP05). The diagram also shows different dust attenuation
models (see legend) for a Maraston (2005, MNRAS 362) model with
In order to study the UV SEDs of our sample galaxies we use parameters based on power-law fits of the form f (λ) ∝ λγ to subregions of the UV continuum. The UV-bump indicator γ 34 =
γ3 − γ4 (NP05) is based on continuum slope measurements in the
wavelength ranges 1900 − 2175 Å (γ3) and 2175 − 2500 Å (γ4),
respectively. The “classical” reddening measure β is taken for
1250 − 1750 Å (Calzetti et al. 1994; Leitherer et al. 2002, ApJS
140). Due to a different accessible wavelength range we take βb as
a reddening measure for galaxies at z < 2. This parameter is derived at 1750 − 2600 Å excluding 1950 − 2400 Å, i.e. the range of
the 2175 Å feature. Wavelengths affected by strong spectral lines
are excluded from the fitting procedure for all parameters (Calzetti
et al. 1994; NP05).
Salpeter IMF, continuous SF, an age of 200 Myr, and solar metallicity.
The symbols are plotted in intervals of ∆E B−V = 0.1.
Conclusions
Results
Fig. 1 shows the UV-bump indicator γ34 versus redshift for our
spectroscopic sample of 108 actively star-forming galaxies at 1 <
z < 2.5. Galaxies with strong 2175 Å features, i.e. γ34 < −2, are
mainly found at 2.3 < z < 2.5 (38% of the FDF galaxies in this
redshift range) and 1 < z < 1.5. Considering that about 28% of
the photometric FDF sample at 1 < z < 1.5 with R < 24 show
Ks < 20 (Heidt et al. 2003, A&A 398; Gabasch et al. 2004,
A&A 421) we can combine the FDF and K20 samples and estimate that about 26% of the 1 < z < 1.5 galaxies with R < 24
show γ34 < −2. Among the 1 < z < 1.5 galaxies only objects with
In accordance with the “downsizing” scenario (Cowie et al. 1996,
AJ 112) the total stellar masses (Drory et al. 2005, ApJ 619;
Pannella et al. 2006, astro-ph/0601338) also tend to be lower at
1 < z < 1.5. For γ34 < −2 hMstarsi ∼ 2 × 1010 M at 1 < z < 1.5,
while hMstarsi ∼ 5×1010 M at 2 < z < 2.5. The metallicities of the
γ34 < −2 objects do not seem to show significant evolution in the
redshift range studied. Typical metallicities are close to the solar
value (Mehlert et al. 2002; Savaglio et al. 2004, ApJ 602). NP05
find that for constant UV-continuum reddening β mass, size, and
metallicity of the 2 < z < 2.5 galaxies do not obviously correlate
with the strength of the UV bump. However, 1 < z < 1.5 galaxies
with −1.5 < βb < 0.5 show on average higher masses, larger sizes,
and lower Sersic indices (Pannella et al. 2006), if the strength of
the 2175 Å feature increases.
F 3: Comparison of Ks -weighted composite spectra of 1 < z < 1.5
galaxies with R < 24 having γ34 > −2 (blue) and γ34 < −2 (red),
respectively. The lower panel gives the ratio of both composites, normalised at 2400 − 2570 Å.
NP05 find that the UV continua of UV-luminous galaxies at 2 <
z < 2.5 suggest effective extinction curves ranging from those typical of the SMC to those typical of the LMC. Fig. 2 shows that this
We have investigated the dust properties of 108 UV-luminous
galaxies at 1 < z < 2.5. In our sample the strongest 2175 Å
dust features (which are as strong as those found in the LMC)
are apparently clustered around z ∼ 1.1 and z ∼ 2.4. At higher
redshift galaxies with γ34 < −2 are highly dust-enshrouded and
indicate a high neutral clouds’ covering fraction in direction towards the observer (as suggested by a study of strong interstellar
UV absorption lines; NP05). This implies that the carriers of the
2175 Å feature are protected from the strong and hard radiation
fields and/or shocks in these ultraluminous galaxies by dust selfshielding due to high dust column densities. At lower redshift the
galaxies with γ34 < −2 are less massive, show lower SFRs, and
are larger. This suggests that the radiation fields in these objects
are weaker. Hence, the carriers of the 2175 Å feature can better
survive although the amount of dust is lower.