THE FIRST OBSERVATIONS OF LOW-REDSHIFT DAMPED Lyα SYSTEMS WITH THE COSMIC ORIGINS SPECTROGRAPH^*

These COS data were acquired on 2010 March 29–30 in three orbits as Visit 13 in program GO11598. Two exposures were obtained with the G130M grating at central wavelengths of 1291 and 1309 Å with exposure times of 1497 and 2191 s. Similarly, two exposures were taken with the G160M grating at central wavelengths of 1577 and 1600 Å with exposure times of 2002 and 2007 s.

The DLA line of this system was fit with a single component centered at z_abs = 0.1140 with a best solution of log N(H i) = 20.68 ± 0.10. The fit to the DLA line is shown in Figure 2. Profiles with column densities 1σ above and below the best-fit value are also shown in Figure 2.

Absorption lines from multiple species are shown in the left panel of Figure 3. We detect lines from N i λλ1134, 1199, 1200, 1201; O i λ1302; Si ii λλ1190, 1304, 1526; S ii λλ1250, 1253, 1259; Fe ii λλ1142, 1143, 1144; Ni ii λ1317. An LLS at z_abs = 0.3556 blocks the continuum blueward of ∼1240 Å (see Tumlinson et al. 2011 for an analysis of the LLS) or ∼1115 Å in the rest frame of the DLA.

Zoom In Zoom Out Reset image size

There are only two H₂ lines covered in the COS spectrum due to the presence of the LLS. The Lyman band (X¹Σ⁺_g − B¹Σ_u⁺) transitions from the (1–0) band which have rest-frame wavelengths λ < 1109 Å are all blocked by the LLS. Only the H₂ λλ1120, 1125 lines resulting from the (0–0) transitions are covered in the COS data. These lines are however blended with strong high Lyman series absorption lines from the LLS; consequently, no direct indicator of molecular gas exists for this DLA.

The COS spectra cover the lines of C iv λλ1548,1550 as well as the Si iv λλ1393, 1402 transitions. The C iv lines lie at the edge of the detector where the sensitivity is low and only the C iv λ1548 line is minimally detected at S/N ∼ 3. The Si iv λ1393 line is blended with the Lyα absorption line from a z_abs = 0.2774 system, and there is no detection of the Si iv λ1402 with S/N ∼ 5 in the region.

Optical spectra of SDSS J1009+0713 were obtained with the Keck-II/HIRESb spectrograph (Vogt et al. 1994) on the night of 2010 March 26. The S/N of the data ranges from ≲5 per 1.8 km s⁻¹ pixel at 3200 Å to ∼20 at 4500 Å. These data provide coverage from 2750–4000 Å in the rest frame of the z_abs = 0.1140 DLA, and hence include the lines of Mg ii λλ2796, 2803, Mg i λ2852, Ti ii λλ3242, 3384, and Ca ii λλ3934, 3969. The Mg i λ2852 line is partially blended with the Fe ii λ2344 line from the z_abs = 0.3558 LLS, as can be seen in Figure 3. In addition, with resolution higher (R = 44,000) than the COS spectra (R ∼ 20,000), these data provide valuable constraints on the kinematics and Doppler parameters of the lines that are blended at the resolution of COS.

These data were processed with the Keck/HIRES reduction code hiresredux in IDL.⁶ The right panel in Figure 3 shows the HIRES spectra of several near-UV (NUV) and optical absorption lines that were observed in this DLA.

The profile was fit with only two components having Doppler widths of b_eff = 10.6 and 12.9 km s⁻¹ and centroid velocities of −6.6 and +21.5 km s⁻¹, respectively. The majority of the low ionization gas appears to be in the v = −6.6 km s⁻¹ component, as this component is strongest in all of the non-saturated absorption lines detected here.

The SDSS images show no obvious candidates for the z_abs = 0.1140 DLA. In order to determine the nature of the host galaxy of the DLA system, we acquired imaging in the F390W and F625W filters of this field using the WFC3 on board the HST. Six exposures were taken in both filters in a dither pattern to eliminate the effects of bad pixels and cosmic rays. Total exposure time was 2370 and 2256 s in the F390W and F625W filters, respectively.

Even with the superb image quality and resolution that WFC3 enables, galaxies at small impact parameter to the central bright QSO may only appear after subtracting the QSO with a suitably chosen point-spread function (PSF). The PSFs used here were acquired from the Tiny Tim online interface⁷ at the pixel coordinates appropriate for the position of the QSO in the images. This PSF was then interactively subtracted from each individual dithered image with idp3⁸ and the subtracted images were then drizzle combined with multidrizzle. The final drizzled images are shown in Figure 4. A color combined image was also created from the PSF-subtracted images, and is shown in Figure 5. The F625W, F390W, and an average of the two were used in the red, blue, and green channels of the color image and the images were scaled using the asinh  scaling described in Lupton et al. (2004).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Multiple galaxies are seen in these WFC3 images within 10'' of the QSO, or ∼20 kpc at z = 0.1140. A large spiral galaxy is clearly seen in Figures 4 and 5. Spectra of objects $80\_5$, $86\_4$, $92\_7$, and $170\_9$ as well as the H ii region labeled in Figure 5 were obtained with Keck/LRIS.⁹ A 1'' slit was used and aligned such that multiple galaxies and the H ii region marked in Figure 5 were observed simultaneously. Emission lines from O iii] λλ4960, 5008 and Hβ λ4862 are clearly seen in galaxies $86\_4$ and $170\_9$ placing them at z = 0.35458 and z = 0.35557, respectively. These two galaxies are likely associated with a LLS at z_abs = 0.3556 (see Tumlinson et al. 2011 for an analysis of the LLS).

Two emission features from galaxy $80\_5$ are seen at 7004 and 8158 Å, which we interpret as O ii] λ3727 and Hγ 4341 at z = 0.87899. A single strong emission line at 8508 Å is observed in the spectrum of galaxy $92\_7$. No prominent emission lines expected from a galaxy at a redshift z = 0.1140 correspond to this observed wavelength. Assuming this emission line is from O ii] λλ3727, 3729 places this galaxy at z = 1.283.

Narrow emission lines from Hβ λ4862 and O iii] λλ4960, 5008 from the H ii region in the spiral galaxy labeled in Figure 5 give a redshift of z = 0.45699. Some flux from the nucleus of the QSO is observed in the spectra; however, no emission lines from another redshift were seen in the spectra increasing our confidence in the determination of the redshift of the H ii region. Combined with the close alignment of the nucleus of the QSO and disk of the galaxy, we interpret the spiral galaxy to be the host of the QSO itself.

Emission lines from intervening galaxies are sometimes seen superimposed on the continuum of the QSO in SDSS spectra (Noterdaeme et al. 2010; Borthakur et al. 2010), although most QSOs do not show such emission features in their spectrum even when strong absorption lines are seen. No narrow emission lines are seen in the SDSS spectrum of this QSO at the absorption redshift of the DLA or any other redshift.

We performed aperture photometry on the galaxies in Figure 5 with apphot in iraf. AB magnitudes are given in Table 2. Magnitudes have been corrected for Galactic dust extinction¹⁰ from Schlegel et al. (1998). As galaxies 170$\_$9, 80$\_$5, and 92$\_$7 are spectroscopically confirmed to be at significantly higher redshift than the absorption redshift of the DLA, an obvious question arises: where is the host galaxy of the DLA?

Table 2. AB Magnitudes and Position Offsets for the Galaxies Near SDSS J1009+0713

ID	Δα	Δδ	m625W	m390W	zgal
	('')	('')	AB	AB
170$\_$9	+1.58	−9.27	21.33 ± 0.08	22.16 ± 0.12	0.35569
86$\_$4	+2.87	+0.16	23.21 ± 0.21	24.10 ± 0.20	0.35455
80$\_$5	+4.00	+0.81	22.50 ± 0.12	23.09 ± 0.18	0.87899
92$\_$7	+6.26	−0.44	23.67 ± 0.37	24.45 ± 0.36	1.283?

Note. Identifiers are based on the position angle and distance from the target QSO.

Download table as:  ASCIITypeset image

We have used the galfit software package (Peng et al. 2010) to fit a multicomponent profile to the F625W image of the QSO. A PSF for the nucleus, Sersic profile for the bulge, and exponential disk for the spiral structure was used in the fit. The best-fit solution from galfit using these components gives m_B = 18.68, m_D = 18.71, and m_N = 17.99 where m_B, m_D, and m_N are the magnitudes of the bulge, disk, and nucleus, respectively. The combined magnitude of the bulge and disk is m_tot = 17.91, comparable to that of the nucleus. We show the results of the galfit decomposition in Figure 6.

Zoom In Zoom Out Reset image size

The host galaxies of QSOs are known to be luminous, with L > L^⋆, and come from a variety of morphological types across the entire Hubble sequence (Bahcall et al. 1997; Hamilton et al. 2002). Given a measured apparent magnitude m_λ, the absolute magnitude M_λ is given by

$$\begin{equation} M_{\lambda }=m_{\lambda }-5{\rm {log}}\Big (\frac{D_L}{10\ \rm {pc}}\Big) - K_{\lambda } - A_{\lambda }, \end{equation} \tag{ 1 }$$

where D_L is the luminosity distance, K_λ is the k-correction, and A_λ is the extinction due to the Galaxy. Based on the morphology of the galaxy seen in Figure 5, we assume an Sb type spectrum and find a K-correction of K_λ = 0.25 magnitudes in the F625W filter at z_em = 0.456 and apply an extinction of A_λ = 0.05 mag giving M_F625W = −24.4. The median luminosity of radio loud QSO host galaxies in the sample of Hamilton et al. (2002) is M_V = −23.5, although galaxies classified as having spiral morphologies were typically more luminous than elliptical counterparts.

A small feature at small impact parameter appears more clearly in the PSF-subtracted images shown in Figure 4 and is marked with an arrow in Figure 5. The angular separation from the nucleus of the QSO is 07, or ∼1.5 kpc at z = 0.114. It is possible that this is the host galaxy of the DLA, or perhaps an H ii region in the QSO host galaxy. For example, if either of the galaxies labeled 80$\_$5 or 86$\_$4 in Figure 5 were placed at small impact parameter to the QSO, it would be difficult to disentangle the two even with the resolution of WFC3. The WFC3 grism is likely the best (if not only) way to definitively determine the redshift the spiral galaxy and object at small impact parameter seen in Figure 5.

This QSO does show radio continuum emission, with a central source and two lobes detected in the Faint Images of the Radio Sky at Twenty cm (FIRST) survey with peak emission of ∼6 mJy (Becker et al. 1995). Radio loud QSOs are almost all found in elliptical galaxies; out of the 26 radio loud QSOs observed by Hamilton et al. (2002), only 4 had spiral type morphologies (16%).

Logarithmic abundances (defined as [X/H] = log X − log H − [X/H]_☉) for the DLA system as determined by the profile fits to the COS and HIRES data are presented here and given in Table 3. Solar system reference abundances ([X/H]_☉) are taken from Lodders (2003). Rest-frame wavelengths as well as f values are taken from Morton (2003) for all lines aside from the C i lines, for which we use the revised f values determined in Jenkins & Tripp (2001) and Ni ii λλ1317, 1370 lines for which we use the revised f values from Jenkins & Tripp (2006).

The COS spectra were fit using the component structure determined with the higher resolution ground-based Keck spectra. Both the Doppler parameters and velocity centroids were held fixed in the fits, and multiple transitions were fit simultaneously where possible (i.e., the S ii λλ1250, 1253, and 1259 lines were all fit simultaneously). The synthetic Voigt profiles were convolved with the COS line-spread functions (LSFs)¹¹ as determined by the COS science team at the nearest tabulated wavelength (Ghavamian et al. 2009). The Keck/HIRES spectra were convolved with a Gaussian LSF with FWHM = 2.5 pixels. Profile fitting was performed using the profile-fit code of Fitzpatrick & Spitzer (1994).

In the case of a particular resonance line not being detected in these data, we have used the following procedure to determine a limiting column density for the line. (1) First, we again use the Doppler parameters and velocity centroids as determined in the higher resolution data. (2) We take the column densities for one of the better constrained fits to determine the ratio of the column densities in each of the components. (3) Synthetic Voigt profiles are then produced and multiplied by the normalized flux, and an equivalent width is determined. The integration and statistics are performed over a region from the lowest to highest velocity component, with an additional 30 km s⁻¹ (or ∼2 resolution elements) on each end to ensure that all of the absorption are taken into account. (4) The significance of the line is determined by calculating EW/σ_tot, where $\sigma _{{\rm tot}}=\sqrt{\vphantom{A^A}\smash{\hbox{${\sigma _c^2+\sigma _p^2}$}}}$ and σ_c is the uncertainty in the EW due to the continuum placement and σ_p is the uncertainty This again perhaps indicate that the spiral seen in Figure 5 is not the host of the DLA galaxy due to photon noise. (5) The column densities are then increased until the line has a 3σ significance.

Strong transitions such as O i λ1302, C ii λ1334, and Mg ii λλ2793, 2803 are almost universally saturated, and only lower limits to the column densities can be determined by fits to these lines. Lower limits in these cases were determined by creating synthetic Voigt profiles and increasing the column densities in the components until the profile reached the observed flux at the line core. We have determined a permissible range of 15.35 < N(Mg ii) < 16.0 which is equivalent to −0.9 <[Mg/H]<−0.2 by measuring an upper limit on the weak transition of Mg ii λ1239.

As the component structure in the more highly ionized gas traced by the C iv λλ1548, 1550 and Si iv λλ1393, 1402 lines is often different than that of the low ions and the S/N in the region of the C iv lines is too low to discern the component structure, we have measured the column density of C iv by direct integration of the line using the apparent optical depth (AOD) method. We determine log N(C iv) = 13.86 ± 0.20. We also place an upper limit on the column density of Si iv based on the non-detection of the Si iv λ1402 line of log N(Si iv) < 13.3.

The depletion in this system is mild with [S/Fe] = +0.24 ± 0.22, [S/Ti] = +0.28 ± 0.15, and [S/Ni] = +0.35 ± 0.22, even less than what is seen in the halo of the Milky Way (Welty et al. 1997). Depletion levels in the ISM of the Milky Way are seen to vary over two orders of magnitude, from [Zn/Fe] >+2 in cold clouds to [Zn/Fe] ∼ + 0.5 in the halo of the Milky Way (Savage & Sembach 1996; Jenkins 2009). We note, however, that similar mild depletion was seen at very small impact parameter toward HS1543+593 (Bowen et al. 2005), a case in which the connection between the absorber and host galaxy is unambiguous.

We have used the model of the depletion patterns in the ISM of the Milky Way from Jenkins (2009) to analyze the depletion in this DLA system. In this model, a single value (F_*) indicates the overall level of depletion in a sightline, with larger values of F_* indicating higher levels of dust depletion. We find that the line-of-sight depletion factor F_* = −0.48 ± 0.24 for this DLA, implying an intrinsic metallicity of [Z/H] = −0.77 ± 0.44, consistent with the metallicity determined from the S ii lines. Such negative values of F_* are typically seen along sightlines with extremely low average densities in the halo of our Galaxy (Jenkins 2009) implying that the sightline passes through the outer edge of a dwarf galaxy.

Kinematically, the component structure of this DLA is extremely simple (see Figure 3). Only two components are required to fit the profiles of the absorption lines. The kinematic width Δv₉₀ (defined as the region where 90% of the AOD in the line occurs) has often been used as an indicator of the kinematics of QSO absorption line systems (Prochaska & Wolfe 1997; Ledoux et al. 2006). We find a kinematic width for this system of Δv₉₀ = 52 km s⁻¹ based on the Ti ii λ3384 line.

The nucleosynthetic origins of N seem to include both primary (production of N from newly synthesized C via the CNO cycle) and secondary production (N produced again in the CNO cycle, but from previously created C and O from previous star formation) processes in intermediate mass stars (Marigo 2001). Of the ∼30 determinations of N abundances in DLA systems, all are at high z where the N i lines are redshifted into wavelengths accessible with ground-based telescopes (Pettini et al. 2008; Henry & Prochaska 2007). Some constraints have been placed on nitrogen in low-z sub-DLAs where, like the high-z DLAs, nitrogen is found to be underabundant (e.g., Tripp et al. (2005)).

Abundances from N can be readily obtained from optical spectra of H ii regions in nearby star-forming galaxies of a range of luminosities. The low-luminosity galaxies from Van Zee & Haynes (2006) show a flat plateau in the [N/O] ratio with respect to metallicity, while the higher metallicity H ii regions in Van Zee et al. (1998) show a sharp increase in [N/O] with increasing metallicity, indicative of secondary production of N.

Observations of N in extremely low metallicity DLAs are important for the study of N production and galactic chemical evolution in general (Pettini et al. 2008; Centurión et al. 2003). It is still yet to be seen if the [N/O] ratio in DLAs displays the same upturn as metallicity increases as most DLAs are metal poor and at high z the N i lines are often blended with the Lyα forest. The DLA presented here is the highest metallicity DLA with both N and α abundance measurements.

In Figure 7, we show [N/α] versus [α/H] for H ii regions in nearby galaxies from Van Zee et al. (1998) and Van Zee & Haynes (2006), as well as the DLA sample from Pettini et al. (2008) and this DLA. The DLA from this work is shown with the red circular point, and lies below the "knee" of the diagram where N production transitions from primary to secondary production. The mean [N/α] ratio from the samples of Van Zee et al. (1998) and Van Zee & Haynes (2006) with [α/H]<−0.3 is 〈[N/α]〉 = −0.55 ± 0.15, 3σ higher than seen in this DLA of [N/S] = −0.78 ± 0.16. Pettini et al. (2008) note this discrepancy in the sample of DLAs shown in Figure 7 and propose that the lower values seen in high z DLAs could be due to the delayed enrichment of N with respect to O as the delay of ∼250 Myr is a greater fraction of the time available for star formation at high redshift than in local galaxies, so at high z this effect should be more pronounced. However, in this low-z DLA we see the same decrement in [N/α] compared to the values from local H ii regions. It would be interesting to see if other high-metallicity DLAs are also below the plateau, perhaps indicating a systematic offset in abundances determined via absorption and emission line diagnostics.

Zoom In Zoom Out Reset image size

Both C i and C ii* are important diagnostic resonance lines in the study of the ISM. Emission from the 158 μm emission line of singly ionized carbon in the ISM of the Milky Way is the dominant coolant, and is expected also to be so in high-z DLAs as well. This line originates from the transition between the ²P_3/2 and ²P_1/2 levels in the ground state of C⁺. In the cold neutral medium (CNM; T ∼ 10² K), the excited state is populated largely by collisions with hydrogen atoms, where in the more diffuse warm neutral medium (WNM; T ∼ 10⁴ K) both collisions with free electrons as well as photon pumping can populate the excited state (Wolfe et al. 2003; Bowen et al. 2005).

In the WNM, the electron density can be determined from observations of the C ii and C ii* lines as

$$\begin{equation} n(e)=18.3\sqrt{T_4}e^{0.091/T_4} \left[ \frac{N(\rm C\,\mathsc{ii}^*)}{N(\rm C\,\mathsc{ii})} \right] \ \rm cm^{-3}, \end{equation} \tag{ 2 }$$

where T₄ is T/10⁴ K (Lehner et al. 2004). In the ISM of the Galaxy, the CNM is characterized by having large depletions of the refractory elements such as Fe and Ti relative to the volatile elements of S and Zn which do not condense onto dust grains. As the depletions levels in this DLA are modest, with [Ti/S] = −0.37 ± 0.12, we assume that the sightline is passing through mainly a WNM. Depletion levels seen in the CNM of the Milky Way are typically 10 times higher than what is observed in this DLA (Savage & Sembach 1996; Jenkins 2009).

The C ii* λ1335 line was not detected in these data, with S/N ∼ 11 per resolution element in the region around the line. We have used the procedure discussed above to determine a limiting column density of log N(C ii*) < 13.7. We show a synthetic Voigt profile with this total column density overplotted on the data in Figure 8.

Zoom In Zoom Out Reset image size

As was mentioned above, the C ii λ1334 is strongly saturated and only a limiting column density of log N(C ii)>15.90 can be determined from this transition. However, in sightlines with low depletion such as this Carbon is mildly depleted and we can use the S abundance to estimate a total C ii column density assuming [C/S] ∼ 0.0. The above analysis yields log N(C ii) = 16.45. Using the above values in Equation (2) with a temperature of 6000 K characteristic of the WNM, the electron density is determined to be n(e) < 0.08 cm⁻³.

The cooling rate per H atom due to C ii] λ158 μm emission can be estimated by

$$\begin{equation} I_{c}=2.89\times 10^{-20}\left[\frac{ N({\rm C}\,\mathsc{ii}^*)}{N(\rm H\,\mathsc{i})}\right] \ \rm {erg\ s^{-1}\ (H\ atom)^{-1}} \end{equation} \tag{ 3 }$$

(Lehner et al. 2004). Using the values above, we determine log I_c < −26.5. This is typical of what has been observed in higher z DLA systems (Wolfe et al. 2003; Lehner et al. 2004).

C i is a tracer of cold gas in the ISM, and is usually seen in DLA absorbers bearing H₂ (Srianand et al. 2005). As such, C i absorption lines are indicative of cold, dense material in DLAs. The C i λ1277 line provides the strongest constraint on the C⁰ column density in the COS spectrum as the S/N in the region around the stronger transition of C i λ1560 was much lower. This line was not detected in the spectrum and we have determined a limiting column density of log N(C i) < 13.6 using the prescription above. A synthetic Voigt profile with this column density is shown is Figure 9. In the two-phase ISM model of Liszt (2002), the ratio of N(C ii)/N(C i) is indicative of the gas density. The observed ratio of N(C ii)/N(C i) > 700 in this system implies that n(H) < 10 cm⁻³.

Zoom In Zoom Out Reset image size