High-resolution observations of two O vi absorbers at z ≈ 2 toward PKS 1448−232
Universität Potsdam, Institut für Physik und Astronomie,
Karl-Liebknecht-Strasse 24-25, Haus
Received: 15 February 2011
Accepted: 14 October 2011
To explore the ionization conditions in highly-ionized absorbers at high redshift, we study in detail two intervening O vi absorbers at z ≈ 2 toward the quasar PKS 1448−232, based on high (R ≈ 75 000) and intermediate (R ≈ 45 000) resolution optical VLT/UVES spectra. We find that both absorption systems are composed of several narrow subcomponents with typical C iv/O vi Doppler-parameters of b < 10 km s-1. This implies that the gas temperatures are T < 105 K and that the absorbers are photoionized by the UV background. The system at z = 2.1098 represents a simple, isolated O vi absorber that has only two absorption components and is relatively metal-rich (Z ~ 0.6 solar). Ioinization modeling implies that the system is photoionized with O vi, C iv, and H i coexisting in the same gas phase. The second system at z = 2.1660 represents a complicated, multi-component absorption system with eight O vi components spanning almost 300 km s-1 in radial velocity. The photoionization modeling implies that the metallicity is non-uniform and relatively low (≤0.1 solar) and that the O vi absorption must arise in a gas phase that differs from that traced by C iv, C iii, and H i. Our detailed study of the two O vi systems towards PKS 1448−232 shows that multi-phase, multi-component high-ion absorbers similar to the one at z = 2.1660 can be described by applying a detailed ionization modeling of the various subcomponents to obtain reliable measurements of the physical conditions and the metal abundances in the gas.
Key words: intergalactic medium / quasars: absorption lines / cosmology: observations
© ESO, 2012
Highly ionized species such as O vi and C iv, evidence of which is observed in the spectra of distant quasars, are excellent tracers of metal-enriched ionized gas in the filamentary intergalactic medium (IGM) and the circumgalactic environment of galaxies. Hence, the analysis of intervening O vi and C iv absorbers towards low- and high-redshift QSOs is crucial to improving our understanding of the physical nature, distribution, evolution, and baryon and metal content of the IGM in the context of galaxy evolution. Owing to the high cosmic abundance of oxygen, the large oscillator strength of the O vi doublet (located in the far-ultraviolet at λλ1031.9,1037.6 Å), and the high ionization energies of the ionization states O+4 (113.9 eV) and O+5 (138.1 eV), the O vi ion is a particularly powerful tracer of the metal-enriched IGM and the gaseous environment of galaxies. Using QSO absorption spectroscopy, O vi absorption now is commonly detected in various different galactic and intergalactic environments in the redshift range z ≈ 0−3.
In the local Universe, O vi absorption in interstellar and intergalactic gas can be observed in the FUV spectra of stars and extragalactic background sources. For instance, O vi absorption is known to arise in the thick disk of the Milky Way (e.g. Savage et al. 2003), in the extended, multi-phase gas halos of the Milky Way and other galaxies e.g. (e.g. Sembach et al. 2003; Wakker & Savage 2009; Prochaska et al. 2011), and in intervening O vi absorption-line systems that trace metal-enriched gas in the IGM (e.g., Tripp et al. 2000; Savage et al. 2002; Richter et al. 2004; Sembach et al. 2004; Tripp et al. 2008; Thom & Chen 2008a,b; Danforth & Shull 2008; for a review see Richter et al. 2008). Intervening O vi absorbers at low redshift were considered as major baryon reservoirs in the IGM, possibly tracing the shock-heated and collisionally ionized intergalactic gas that results from large-scale structure formation (Cen & Ostriker 1999; Davé et al. 2001). This so-called warm-hot intergalactic medium (WHIM) has gas temperatures in the range 105 < T < 107 K and is believed to host 30−40% of the baryons at z = 0 (Cen & Ostriker 1999). Observational and theoretical studies indicate, however, that some O vi absorbers at low z may trace low-density, photoionized gas or conductive, turbulent, or shocked boundary layers between cold/warm (~103−104 K) gas clouds and an ambient hot (~106−107 K) plasma rather than the shock-heated WHIM (Fox 2011, see discussion in). Thus, a simple estimate of the ionization state of the gas in the absorbers from the observed O vi/H i ratios may lead to erroneous results because of the complex multi-phase character of the gas (Tepper-García et al. 2011).
For redshifts z > 2, O vi absorption is detectable from the ground, where it can be observed in optical QSO absorption spectra of relatively high signal-to-noise ratio (S/N). One very problematic aspect of the analysis of O vi absorbers at high redshift is the often severe blending of the O vi absorption with the Ly α forest. As at low redshifts, the origin and nature of O vi absorbers at high z is expected to be manifold. It has been shown by simulations (e.g. Theuns et al. 2002; Oppenheimer & Davé 2008) that shock-heating by collapsing large-scale structures is inefficient at high redshift in providing a widespread warm-hot intergalactic phase in the early Universe. Galactic winds instead probably contribute substantially to the population of photoionized and collisionally ionized O vi absorbers at high redshifts, enriching the surrounding circumgalactic and intergalactic gas with heavy elements at relatively high gas temperatures (Fangano et al. 2007; Kawata & Rauch 2007). Many of the strong O vi absorbers at high z indeed exhibit the complex absorption patterns that would be expected for a circumgalactic multi-phase gas environment (e.g. Bergeron & Herbert-Fort 2005). As for circumgalactic absorbers in the local Universe, a considerable fraction of the O vi absorbers at high z thus may arise in conductive, turbulent, or shocked boundary layers.
In addition to O vi absorbers that trace highly-ionized gas in the immediate environment of galaxies, intergalactic O vi absorbers (i.e., absorbers that are not gravitationally bound to individual galaxies) may arise in regions that are sufficiently enriched with heavy elements. Previous surveys of high-redshift O vi absorbers (Bergeron et al. 2002; Simcoe et al. 2002, 2004, 2006; Carswell et al. 2002; Bergeron & Herbert-Fort 2005) have shown that there are many narrow O vi absorbers with Doppler-parameters b ≤ 10 km s-1. Such narrow lines cannot arise from collisionally ionized gas but must be related to photoionized (possibly intergalactic) gas with temperatures T < 105 K. Many of these narrow O vi absorbers at both low and high redshift display velocity-centroid offsets between O vi, C iv, and H i, suggesting that these ions do not arise in the same gas phase. Unfortunately, this crucial aspect is only partially considered in previous O vi surveys.
To explore the multi-phase character of high-ion absorbers and improve our understanding of the ionization conditions in O vi systems, it is important to investigate in detail the absorption characteristics and ionization conditions of carefully selected absorption-line systems. It is particularly important to study absorbers that can be observed at high S/N for which the O vi absorption is not blended by Ly α forest lines. Owing to the complexity of many high-ion absorbers that often are composed of several velocity subcomponents a spectral resolution of R ≈ 45 000 and higher is desired. We note that while the analysis of individual high-ion absorbers is a common strategy for exploring the nature of O vi absorbers at low redshift (e.g. Tumlinson et al. 2011; Savage et al. 2011), detailed studies of individual O vi absorption systems at high redshift are rare (e.g., Fox et al. 2011).
In this paper, we present VLT/UVES observations at intermediate (R ≈ 45 000) and high (R ≈ 75 000) spectral resolution of two particularly interesting O vi systems at z ≈ 2 along the line of sight toward the quasar PKS 1448−232. This sightline was selected by us for a detailed study because it contains two unsaturated O vi systems at zabs = 2.1098 and zabs = 2.1660, that both appear to have a well-defined subcomponent structure with narrow O vi/C iv absorption components and no major blending with Ly α forest lines (Bergeron et al. 2002; Fox et al. 2008). These two absorption systems are therefore ideal targets to study in detail the physical conditions in photoionized, multi-phase high-ion absorbers at high redshift.
Absorption profiles for the O vi absorber at z = 2.1098 in the high-resolution data (left panel) and the intermediate-resolution data (right panel).
|Open with DEXTER|
Absorption profiles for the O vi absorber at z = 2.1660 in the high-resolution data (left panel) and the intermediate-resolution data (right panel). The strong absorption observed in O vi λ1037.62 plot is a Si iii line at z = 1.7236.
|Open with DEXTER|
Our data set consists of intermediate- and high-resolution spectra of the quasar PKS 1448−232 (zem = 2.208; V = 16.9), observed at the VLT with the UVES spectrograph. The intermediate-resolution data have a spectral resolution of R ≈ 45 000, corresponding to a velocity resolution of Δv ≈ 6.7 km s-1 FWHM. These data were obtained and reduced as part of the ESO Large Programme “The Cosmic Evolution of the IGM” (Bergeron et al. 2002). The wavelength coverage of the intermediate resolution data is 3050−10 400 Å. The S/N in the data varies between 15 and 90 per spectral resolution element.
The high-resolution data have R ≈ 75 000, corresponding to Δv ≈ 4.0 km s-1 FWHM velocity resolution and were obtained with VLT/UVES in 2007 in an independent observing run (program ID 079.A−0303(A)). The wavelength coverage of the high resolution data is 3000−6687 Å. The raw data were reduced using the UVES pipeline implemented in the ESO-MIDAS software package. The pipeline reduction includes flat-fielding, bias- and sky-subtraction and a relative wavelength calibration. The individual spectra were then corrected to vacuum wavelengths and coadded. The S/N in the high-resolution data is 20−70 per resolution element.
Fit parameters for the system at z = 2.1098.
The detected absorption features that are associated with the two absorbers at zabs = 2.1098 and zabs = 2.1660 were fitted independently in both spectra (at intermediate resolution and high resolution) with Gaussian profiles using the CANDALF fitting routine1, which uses a standard Levenberg-Marquard minimization algorithm. The program simultaneously fits the continuum and the absorption lines, delivering ion column densities, N, and Doppler parameters, b, for each absorption component. The continuum is modeled as a Legendre polynomial with an order of up to 4. The one-σ fitting errors in N and b (as listed in Tables 1−3) are estimated using the diagonals of the Hesse matrix.
Figure 1 shows the velocity profiles of O vi (λλ1031.9,1037.6), C iv (λλ1548.2,1550.8), and H i, Ly α, and Ly β (λλ1215.7, 1025.7) for the z = 2.1098 absorber in the high-resolution data (left panel) and the intermediate-resolution data (right panel). From the visual inspection of both panels, we find no significant differences between the two data sets. The S/N is on average higher for the high resolution data, except for the C iv region where this ratio is slightly higher at intermediate resolution. Therefore, the differences in the values for N, b, and z derived for the individual absorption components in the intermediate and high-resolution spectra are a result of the different S/N values for the two data sets.
Two absorption components are detected in each of these ions. The O vi absorption is relatively strong compared to C iv. The H i absorption is weak relative to other O vi absorbers at similar redshift (e.g., Bergeron et al. 2002) with a central absorption depth in the H i Ly α line of less than 70 percent. We note that the second, weaker, component of H i Ly α and Ly β absorption associated with the high-ion absorption is blended, so that the true component structure of the H i and the relative H i column densities and H i b-values remain somewhat uncertain. The blending aspect is not taken into account in the formal error estimate for N and b given in Table 1, which is based on the profile fitting. While for the stronger of these components the absorption of O vi, C iv, and H i is well aligned, there appears to be a small (<10 km s-1) velocity shift between H i and the high ions in the weaker component (see Table 1). If real, this shift may indicate that the H i and the metal ions may not trace the same gas phase in the weaker absorption component. Owing to the blending of the H i absorption, however, the reality of this shift remains unclear.
We derive for the column densities, listed in Table 1, log N(O vi) ≈ 14.3, log N(C iv) ≈ 13.1, and log N(H i) ≈ 13.4 in the stronger of the two components. The resulting ion-to-hydrogen ratios of N(O vi)/N(H i) ~ 8 and N(C iv)/N(H i) ~ 0.5 already indicate that the metallicity of this absorber must be fairly high (Bergeron & Herbert-Fort 2005). We note that because of the blending problem in the Ly α and Ly β lines, the H i column density may be regarded as an upper limit, thus the ratios given above might be even higher.
H i Ly α fit parameters for the absorption system at z = 2.1660.
The O vi system at z = 2.1660 exhibits a significantly more complex absorption pattern than the absorber at z = 2.1098, as can be seen in the velocity profiles presented in Fig. 2. We detect O vi absorption in eight individual absorption components, spanning a velocity range as large as ~300 km s-1. From our visual inspection, it is evident that the absorption pattern of O vi differs than those of the other detected intermediate and high ions (C iii, C iv) and H i, although some of the components appear to be aligned in velocity space. As for the system at z = 2.1098, there are no significant differences between the absorption characteristics of the high-resolution data and the intermediate-resolution data. However, the S/N is somewhat lower in the latter for lines that are located in the blue part of the spectrum, hence the resulting fitting values for N, b, and z for the individual absorption components differ slightly (Tables 2 and 3).
We modeled the H i absorption by simultaneously fitting Ly α and Ly β in four absorption components (components 2−5; see Table 2), obtaining column densities in the range 13.2 < log N(H i) < 15.2. One additional component (component 1) is present in the Ly α absorption, but is blended in Ly β (see Fig. 2), so that N(H i) was derived solely from Ly α. We note that for the H i fit we did not attempt to link the H i component structure to the structure seen in the the metal ions, as this requires knowledge about the physical conditions in the absorber. We discuss this aspect in detail in Sect. 3.2, where we try to reconstruct the H i absorption pattern based on a photoionization model. We fitted the H i absorption with the minimum number of absorption components required to match the observations (Fig. 2, lowest panel) and to obtain an estimate on the total H i column in the absorber.
By summing over the column densities in the individual absorption components, we derive total column densities of log N(O vi) ≈ 14.2, log N(C iii) ≈ 13.7, log N(C iv) ≈ 13.5, and log N(H i) ≈ 15.3. The resulting ion-to-hydrogen ratios of N(O vi)/N(H i) ~ 0.1 and N(C iv)/N(H i) ~ 0.02 (representing the average over all components) are substantially smaller than in the z = 2.1098 system, which is indicative of a lower (mean) absorber metallicity.
The complexity of the absorption patterns for the various species in this system and the large velocity spread suggests that this absorber arises in an extended multi-phase gas structure.
To infer information about the physical properties of the two O vi absorbers towards PKS 1448−232, we modeled in detail the ionization conditions in these systems. Since the two absorbers at z = 2.1098 and z = 2.1660 have redshifts close to the quasar redshift (zQSO = 2.208), it is necessary to check whether the two systems lie in the proximity of the background quasar and are influenced by its ionizing radiation.
With the above given redshifts, the two absorbers have velocity separations from the QSO of δv2.1098 ≈ 9000 km s-1 and δv2.1660 ≈ 4000 km s-1, thus the absorber at z = 2.1660 can be regarded (depending on the definition) as an associated system. With a (monochromatic) luminosity at the Lyman limit of L912 = 3.39 × 1031 erg s-1 Hz-1, the size of the sphere-of-influence of the ionizing radiation from PKS 1448−232 is known to be 6.7 Mpc, corresponding to a velocity separation of ~1400 km s-1 (Fox et al. 2008). Therefore, it is safe to assume that the ionizing radiation coming from PKS 1448−232 itself has no measurable influence on the ionization conditions in the two O vi systems.
The small b-values measured for O vi, C iv, and H i indicate that collisional ionization is not the origin of the O vi existence in the gas. It is common to assume that the observed Doppler parameters (bobs) are composed of a thermal and a turbulent component (bth and bturb, respectively), such that . The thermal component can be expressed by , where T is the gas temperature and m is the mass of the considered ion. The Doppler parameters measured for the O vi components in the two absorbers are all b < 16 km s-1 and many of them are b < 10 km s-1 (see Tables 1 and 3), indicating that T < 105 K. This value is below the peak temperature of O vi in a collisional ionization equilibrium (T ~ 3 × 105 K; Sutherland & Dopita 1993); it is also lower than the temperature range expected for O vi arising in turbulent mixing layers in the interface regions between cold and hot gas (T = 105−106 K; Kwak & Shelton 2010). Consequently, photoionization by the hard UV background remains as the only plausible origin of O vi in the two high-ion absorbers towards PKS 1448−232.
On the basis of these considerations, we modeled the ion column densities in the two O vi systems using the photoionization code CLOUDY (version C08; Ferland et al. 1998). We assumed a solar abundance pattern of O and C and an optically thin plane-parallel geometry in photoionization equilibrium exposed to a Haardt & Madau (2001) UV background spectrum at z = 2.16, which had been normalized to log J912 = −21.15 (Scott et al. 2000) at the Lyman limit.
We assumed that each of the observed velocity components is produced by a “cloud”, which we modeled as an individual entity. As input parameters, we considered the measured column densities of C iii (for only the z = 2.1660 absorber), C iv, O vi, the metallicity Z (in solar units), and the hydrogen particle density nH. The metallicity of each cloud and the hydrogen density were varied across a range appropriate to intergalactic clouds (i.e., −3 ≤ log Z ≤ 0 and −5 ≤ log nH ≤ 0).
We then applied the following iterative modeling procedure. In a first step, we derived models using CLOUDY for a set of values of Z, nH, and N(H i), where N(H i) has been constrained by the observations. In a second step, the corresponding values of N(C iii), N(C iv), and N(O vi) were calculated. The output was compared with the observed column densities and, in the case of a mismatch, the input parameters Z and nH were adjusted before the next iteration step. This process was repeated until the differences between the output column densities and the observed values became negligible and we obtained a unique solution. In addition to the ion column densities, our CLOUDY model provides information about the neutral hydrogen fraction, fHI, the gas temperature, T, and the absorption pathlength, L = N(H i)/(fHI nH).
Fit parameters for metal absorption in the absorber at z = 2.1660.
Modeled column densities for the absorber at z = 2.1098.
As we have mentioned earlier, absorption by O vi and C iv is well-aligned in both components in this system, while the true component structure of the H i is uncertain because of blending effects in the Ly α and Ly β lines. Owing to the alignment of O vi and C iv, we assumed a single-phase model, in which each of the two components (clouds) at v = 0 and +25 km s-1 in the z = 2.1098 rest frame hosts O vi, C iv, and H i of column densities similar to those derived from the profile fitting. Consequently, we assumed log N(H i) = 13.37 and 13.38 as input for the CLOUDY modeling and followed the procedure outlined above. The results of the CLOUDY modeling of the z = 2.1098 absorber are summarized in Table 4. Our model closely reproduces the observed O vi and C iv column densities in both components, if the clouds have a density of log nH ≈ −4.2, a temperature of log T ≈ 4.6, and a neutral hydrogen fraction of log fHI ≈ −5.3. However, to match the observations, the second component (at +25 km s-1) in our initial model (Table 4, first two rows) needs to have a metallicity of log Z = −1.02, which is ~0.8 dex lower than that of the other component (log Z = −0.24). The absorption path lengths are ~20 kpc for the component at 0 km s-1 and ~30 kpc for the component at +25 km s-1.
Owing to the blending problem for the H i Ly α and Ly β absorption, which affects in particular the estimate of N(H i) in the cloud at +25 km s-1 (Fig. 1), we set up a second CLOUDY model in which we tied the metallicity of the +25 km s-1 component to the metallicity of the other component (log Z = −0.24), but left the N(H i) of this component as a free parameter. From this, we derived a value of log N(H i) = 12.57 for the cloud at +25 km s-1 and the absorption path-length decreased to L = 4.7 kpc. In terms of the blending, we regard this model as more plausible than the model with two different metallicities and to have a larger absorption path-length.
In summary, our CLOUDY modeling suggests that the z = 2.1098 absorber towards PKS 1448−232 represents a relatively simple, metal-rich O vi absorber in which the highly ionized O vi and C iv states coexist in a single gas-phase.
Modeled column densities for the C iii/C iv absorbing phase in the z = 2.1660 absorber.
We started to model this system with CLOUDY, again under the assumption of a single gas-phase hosting the observed intermediately and highly ionized C iii, C iv, and O vi states in the various subcomponents. However, during the modeling process it quickly turned out that it is impossible to match the observed column densities of C iii and O vi by assuming a single gas-phase in the components, when these two ions are aligned in velocity space. Our modeling indicates that the C iii absorption must instead arise in an environment that has a relatively high gas density and is spatially distinct from the O vi phase. In a second step, we tried to tie the highly ionized C iv and O vi states in a single gas phase (as for the z = 2.1098 system) in the relevant absorption components, ignoring the C iii phase. However, this approach did not deliver satisfying results, as we obtained for some components, for which C iv/O vi was constrained by observations, very low gas densities and very large absorption path-lengths on Mpc scales, which are highly unrealistic. Given that the overall component structures of O vi and C iv differ substantially from each other in this system (Fig. 2), this result is not really surprising.
The only modeling approach for which we obtain realistic results for gas densities, temperatures, and absorption path-lengths in this system and its subcomponents is a two-phase model, in which C iii coexists with C iv and part of the H i in one (spatially relatively confined) phase, and O vi and the remainder of the H i in a second (spatially relatively extended) phase. The coexistence of C iii and C iv in one phase is also suggested by the C iii and C iv absorption, which is well- aligned in velocity space (see Fig. 2). The results of this two-phase model are presented in Tables 5 and 6. A critical issue for the modeling of this complex multi-phase absorber with its many absorption components is the assumption of a neutral gas column density in each subcomponent (and phase). Since in the H i Ly α and Ly β absorption, most subcomponents are smeared together to one large absorption trough, the observational data provide little information about the distribution of the H i column densities among the individual components. Nevertheless, the data provide a solid estimate of the total H i column density in the absorber (log N ≈ 15.3; see Sect. 2.4), which must match the sum of N(H i) over all subcomponents considered in our model. Consequently, we included in our iteration procedure the constraints on N(H i)tot and the shape of the (total) H i absorption profile. The latter aspect also concerns the choice of the gas temperature in the model, as T regulates the thermal Doppler-broadening and thus the width of the modeled H i lines. We modeled the H i width following the approach of Ding et al. (2003).
With these various constraints, we first modeled the C iii/C iv phase in the absorber. However, owing to the extremely complex parameter space, we did not find a unique solution for (T,nH,Z) among the individual components, but had to make additional constraints. Since the individual components observed in C iii/C iv are very close together in velocity space, we assumed they all have the same metallicity and, based on the Z range allowed in the model, we set log Z = −1.5 for all subcomponents. This model was able to match the observed column densities of these two ions in the individual subcomponents, but did not match closely the overall shape of the overall H i absorption, implying that the metallicity in this absorber is non-uniform among the individual absorption components. Therefore, we refined our model by using two different metallicities, log Z = −1.7 for the saturated H i components and log Z = −1.0 for the weaker H i components (see Tables 5 and 6 for details). Although imperfect, this model delivers a satisfying match between the modeled spectrum and the UVES data.
Adopting this model, we found that the C iii/C iv absorbing components have temperatures between log T = 4.3 and 4.6, densities between log nH = −3.7 and −2.7, and neutral gas fractions between log fHI = −4.8 and −3.6 (see Table 5). The absorption path lengths were found to vary between 0.3 and 16.3 kpc for the components with log Z = −1.7, and between 0.01 and 0.04 kpc for the components with log Z = −1.0. These numbers suggest that the C iii/C iv absorbing phase resides in relatively small and confined gas clumps. This scenario is consistent with the small turbulent b-values of <6 km s-1 for the subcomponents that we derive in our model. We note that in Table 5, we also list the predicted column densities for O vi, which are typically 1−2 orders of magnitude below the observed ones in this absorber. This, again, indicates that C iii/C iv and O vi must reside in different gas phases with different physical conditions to explain the observed column densities.
Finally, we modeled the O vi absorbing phase in the z = 2.1660 absorber, based on the observed O vi column densities. Since we had information for no ions other than H i and O vi that could provide information about the physical conditions in this phase, we fixed the metallicity of the gas to log Z = −1.7 and log Z = −1.0 (equal to the C iii/C iv phase) and constrained the temperature range [Tmin,Tmax] in the CLOUDY models based on the observed line widths of O vi (giving Tmax) and the modeling results of the C iii/C iv phase (giving Tmin for all components except the first two). The results of this model are shown in Table 6. We derived gas densities in the range log nH = −4.6 to −3.2 and neutral gas fractions in the range log fHI = −5.8 to −4.6. The absorption path length was found to vary between 19.8 and 83.3 kpc for the components with log Z = −1.7, and between 1.3 and 38.3 kpc for those with log Z = −1.0. The mismatch between N(O vi) of the model and the data for components one and nine (see Table 6) implies that the metallicity distribution among the individual absorption components is even more complex than the one assumed in our model. Despite this (minor) concern, our CLOUDY modeling for O vi provides clear evidence that the O vi absorbing phase has substantially lower gas densities than the C iii/C iv absorbing phase and is spatially more extended.
In summary, our CLOUDY modeling of the z = 2.1660 absorber suggests that this system contains a complex multi-phase gas structure, in which a number of cooler, C iii/C iv absorbing cloudlets are embedded in a spatially more extended, O vi absorbing gas phase spanning a total velocity range of ~300 km s-1. Although the metallicity is not well-constrained in our model, it appears that log Z ≤ −1 in the absorber, which is ~0.8 dex below the value obtained for the system at z = 2.1098.
Modeled column densities for the O vi absorbing phase in the z = 2.1660 absorber
Our detailed analysis of the two O vi absorbers at z = 2.1098 and z = 2.1660 towards the quasar PKS 1448−232 has clearly illustrated the large diversity and complexity of high-ion absorbers at high redshift.
A number of studies based on both optical observations (e.g., Bergeron et al. 2002; Carswell et al. 2002; Simcoe et al. 2002, 2004, 2006; Bergeron & Herbert-Fort 2005; Aguirre et al. 2008) and numerical simulations (e.g., Fangano et al. 2007; Kawata & Rauch 2007) have investigated the properties of high-redshift O vi systems and their relation to galaxies.
As a result of their survey of O vi absorbers at redshifts z = 2.0−2.6, Bergeron & Herbert-Fort (2005) suggested that O vi systems may be classified into two different types: metal-rich absorbers (“type 1”) that have large N(O vi)/N(H i) ratios and that appear to be linked to both galaxies and galactic winds, and metal-poor absorbers (“type 0”) with small N(O vi)/N(H i) ratios, that are embedded in the intergalactic medium. The two absorbers observed towards PKS 1448−232 that we have discussed in this paper do not match the classification scheme of Bergeron & Herbert-Fort (2005). The absorber at z = 2.1098 has a very large N(O vi)/N(H i) ratio of ~8 (i.e., it is of type 1): it is a simple, single-phase, metal-rich system with a metallicity slightly below the solar value. Nevertheless, this system is completely isolated with no strong H i Ly α absorption within 1000 km s-1. In contrast, the absorber at z = 2.1660 has a N(O vi)/N(H i) ratio of only ~0.1 and a metallicity of 0.1 solar or lower (i.e., it is of type 0 according to Bergeron & Herbert-Fort 2005). This absorber is a complex multi-phase system with a non-uniform metallicity, suggesting that it originates in a circumgalactic environment. While this mismatch with the Bergeron & Herbert-Fort (2005) classification scheme certainly has no statistical relevance for the general interpretation of O vi absorbers at high redshift, the results suggest that for a thorough understanding of highly-ionized gas at high redshift the absorption characteristics of O vi systems may be too diverse to permit a simple classification scheme based solely on the observed (and partly averaged) column density ratios of O vi, H i, and other ions.
One critical drawback of many previous O vi surveys at high z is that they often considered only simplified models for the ionization conditions in their sample of highly ionized absorbers, so that the multi-phase character of the gas and the possible ionization conditions far from a photoionization equilibrium are insufficiently taken into account. As pointed out by Fox (2011), single-phase, single-component ionization models, if applied, will provide physically irrelevant results for most of the O vi systems at high z. This implies that previous estimates of the baryon- and metal-content of O vi absorbers at low and high z are possibly affected by large systematic uncertainties.
One firm conclusion of many previous observational and theoretical studies of highly ionized absorbers is that a considerable fraction of the O vi systems at low and high z must arise in the metal-enriched circumgalactic environment of (star-forming) galaxies (e.g., Wakker & Savage 2009; Prochaska et al. 2011; Fox et al. 2011; Tepper-García et al. 2011; Fangano et al. 2007). Thus, the complex absorption pattern observed in the z = 2.1660 system towards PKS 1448−232 and many other O vi absorbers at high z may reflect the complex gas distribution of enriched gaseous material that was ejected from galaxies into the IGM during their wind-blowing phase (e.g., Kawata & Rauch 2007). In this context, Schaye et al. (2007) suggested that the intergalactic metals were transported from galaxies by means of galactic winds and reside in the form of dense, low- and high-metallicity patches within large hydrogen clouds. These authors point out that much of the scatter in the metallicities derived for high-redshift absorbers could be explained by the spatially varying number of the metal-rich patches and the different absorption path lengths through the surrounding metal-poor intergalactic filament instead of an overall (large-scale) metallicity scatter in the IGM. In this scenario, the substantial differences in the metallicities of the two O vi systems towards PKS 1448−232, and even the intrinsic metallicity variations within the z = 2.1660 system, could be explained by the different geometries of the absorbing structures, suggesting that much of the H i that is associated with the metal absorption in velocity space, arises in a spatially distinct region. A similar conclusion was drawn by Tepper-García et al. (2011), who studied the nature of O vi absorbers at low z using a set of cosmological simulations. We note that absorbers with larger H i column densities, such as Lyman-limit systems (LLS) and damped-Lyman α systems (DLAs), sometimes exhibit abundance variations among the different velocity subcomponents (e.g., Richter et al. 2005; Prochter et al. 2010). This indicates that the metals in the gas surrounding high z galaxies have not been well-mixed.
The observed velocity differences between O vi and other ions and the multi-phase nature of the gas provide further evidence of an inhomogeneous metallicity and density distribution in intervening highly ionized absorbers. It is an interesting that the velocity misalignment appears to concern only the O vi absorbing phase in highly ionized absorbers at high redshift, while other highly ionized states such as N v and C iv generally appear to be well-aligned with H i, even in systems that exhibit a complex velocity-component structure (Fechner & Richter 2009). This puzzling aspect underlines that additional detailed studies of individual O vi absorption systems could be very important to our understanding of intergalactic and circumgalactic gas at high redshift, as this ion traces a metal-enriched gas phase that cannot be observed by other means.
We have investigated two O vi absorbers at z = 2.1098 and z = 2.1660 towards the quasar PKS 1448−232. For this, we have used high- (R ≈ 75 000) and intermediate-resolution (R ≈ 45 000) optical spectra obtained with the VLT/UVES instrument and CLOUDY photoionization models.
The O vi system at z = 2.1098 is characterized by strong O vi absorption and weak H i absorption in a relatively simple, two-component absorption pattern. The absorption by O vi, C iv, and H i are well aligned in velocity space, indicating that they trace the same gas phase. From a detailed photoionization modeling of this system, we derived a metallicity of ~0.6 solar, a characteristic density of log nH ≈ −4.2, a temperature of log T ≈ 4.6, and a total absorption path length of ~30 kpc. The absorber is isolated with no strong H i Ly α absorption within 1000 km s-1.
The O vi absorber at z = 2.1660 represents a complicated, multi-component absorption system with eight relatively weak and narrow O vi absorption components spanning almost 300 km s-1 in radial velocity. The O vi components are accompanied by strong H i absorption and C iii, C iv absorption. The O vi component structure differs from that of H i and C iv, indicating the absorber contains a multi-phase IGM. Our photoionization modeling with CLOUDY suggests that there are (at least) two distinct gas phases in this system. The first consists of C iii, C iv, and most of the H i, which appear to coexist in several relatively compact cloudlets at gas densities of from log nH ≈ −3.7 to −2.7, temperatures of log T ≈ 4.3−4.6, and absorption path lengths of <16 kpc. We have found that O vi appears to reside in a highly ionized, more spatially extended gas phase at densities in the range log nH ≈ −4.6 to −3.2, temperatures between log T ≈ 4.3 and 5.3, and absorption path-lengths up to 83 kpc. While the precise metallicity of the absorber is not well-constrained, our modeling is most consistent with a non-uniform metal abundance among the individual absorption components with (at least) two different metallicities of log Z = −1.7 and log Z = −1.0.
Our study has illustrated the large diversity and complexity of O vi systems at high redshift. We speculate that some of the observed differences between the two highly ionized absorbers towards PKS 1448−232 could be the result of a inhomogeneous metallicity and density distribution in the photoionized IGM. Our study indicates that multi-phase, multi-component highly ionized absorbers similar to the one at z = 2.1660 can be described by a detailed ionization modeling of the various subcomponents to obtain reliable information about the physical conditions and metal-abundances of the gas. We conclude that much effort will be required to achieve a more complete view of the nature of O vi absorbers at high redshift.
In the future, we plan to continue our investigation of these systems by studying a larger sample of O vi absorbers
in high-quality UVES archival data and compare their absorption characteristics with artificial spectra generated from numerical simulations of star-forming galaxies and their intergalactic environment.
N.D. and P.R. acknowledge financial support by the German Deutsche Forschungsgemeinschaft, DFG, through grant Ri 1124/5-1.
- Aguirre, A., Dow-Hygelund, C., Schaye, J., & Theuns, T. 2008, ApJ, 689, 851 [NASA ADS] [CrossRef] (In the text)
- Bergeron, J., & Herbert-Fort, S. 2005 [arXiv:astro-ph/0506700] (In the text)
- Bergeron, J., Aracil, B., Petitjean, P., & Pichon, C. 2002, A&A, 396, L11 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Carswell, B., Schaye, J., & Kim, T. 2002, ApJ, 578, 43 [NASA ADS] [CrossRef] (In the text)
- Cen, R., & Ostriker, J. P. 1999, ApJ, 519, L109 [NASA ADS] [CrossRef] (In the text)
- Danforth, C. W., & Shull, J. M. 2008, ApJ, 679, 194 [NASA ADS] [CrossRef] (In the text)
- Davé, R., Cen, R., Ostriker, J. P., et al. 2001, ApJ, 552, 473 [NASA ADS] [CrossRef] (In the text)
- Ding, J., Charlton, J. C., Bond, N. A., Zonak, S. G., & Churchill, C. W. 2003, ApJ, 587, 551 [NASA ADS] [CrossRef] (In the text)
- Fangano, A. P. M., Ferrara, A., & Richter, P. 2007, MNRAS, 381, 469 [NASA ADS] [CrossRef] (In the text)
- Fechner, C., & Richter, P. 2009, A&A, 496, 31 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Ferland, G. J., Korista, K. T., Verner, D. A., et al. 1998, PASP, 110, 761 [NASA ADS] [CrossRef] (In the text)
- Fox, A. J. 2011, ApJ, 730, 58 [NASA ADS] [CrossRef] (In the text)
- Fox, A. J., Bergeron, J., & Petitjean, P. 2008, MNRAS, 388, 1557 [NASA ADS] [CrossRef] (In the text)
- Fox, A. J., Ledoux, C., Petitjean, P., Srianand, R., & Guimarães, R. 2011, A&A, 534, A82 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Haardt, F., & Madau, P. 2001, in Clusters of Galaxies and the High Redshift Universe Observed in X-rays, ed. D. M. Neumann & J. T. V. Tran (In the text)
- Kawata, D., & Rauch, M. 2007, ApJ, 663, 38 [NASA ADS] [CrossRef] (In the text)
- Kwak, K., & Shelton, R. L. 2010, ApJ, 719, 523 [NASA ADS] [CrossRef] (In the text)
- Oppenheimer, B. D., & Davé, R. 2008, MNRAS, 387, 577 [NASA ADS] [CrossRef] (In the text)
- Prochaska, J. X., Weiner, B., Chen, H., Mulchaey, J. S., & Cooksey, K. L. 2011, ApJ, 740, 91 [NASA ADS] [CrossRef] (In the text)
- Prochter, G. E., Prochaska, J. X., O’Meara, J. M., Burles, S., & Bernstein, R. A. 2010, ApJ, 708, 1221 [NASA ADS] [CrossRef] (In the text)
- Richter, P., Savage, B. D., Tripp, T. M., & Sembach, K. R. 2004, ApJS, 153, 165 [NASA ADS] [CrossRef] (In the text)
- Richter, P., Ledoux, C., Petitjean, P., & Bergeron, J. 2005, A&A, 440, 819 [NASA ADS] [CrossRef] [EDP Sciences] (In the text)
- Richter, P., Paerels, F. B. S., & Kaastra, J. S. 2008, Space Sci. Rev., 134, 25 [NASA ADS] [CrossRef] (In the text)
- Savage, B. D., Sembach, K. R., Tripp, T. M., & Richter, P. 2002, ApJ, 564, 631 [NASA ADS] [CrossRef] (In the text)
- Savage, B. D., Sembach, K. R., Wakker, B. P., et al. 2003, ApJS, 146, 125 [NASA ADS] [CrossRef] (In the text)
- Savage, B. D., Narayanan, A., Lehner, N., & Wakker, B. P. 2011, ApJ, 731, 14 [NASA ADS] [CrossRef] (In the text)
- Schaye, J., Carswell, R. F., & Kim, T. 2007, MNRAS, 379, 1169 [NASA ADS] [CrossRef] (In the text)
- Scott, J., Bechtold, J., Dobrzycki, A., & Kulkarni, V. P. 2000, ApJS, 130, 67 [NASA ADS] [CrossRef] (In the text)
- Sembach, K. R., Wakker, B. P., Savage, B. D., et al. 2003, ApJS, 146, 165 [NASA ADS] [CrossRef] (In the text)
- Sembach, K. R., Tripp, T. M., Savage, B. D., & Richter, P. 2004, ApJS, 155, 351 [NASA ADS] [CrossRef] (In the text)
- Simcoe, R. A., Sargent, W. L. W., & Rauch, M. 2002, ApJ, 578, 737 [NASA ADS] [CrossRef] (In the text)
- Simcoe, R. A., Sargent, W. L. W., & Rauch, M. 2004, in Origin and Evolution of the Elements, Carnegie Observatories Astrophysics Series, ed A. McWilliam, & M. Rauch (Pasadena: Carnegie Observatories) (In the text)
- Simcoe, R. A., Sargent, W. L. W., Rauch, M., & Becker, G. 2006, ApJ, 637, 648 [NASA ADS] [CrossRef] (In the text)
- Sutherland, R. S., & Dopita, M. A. 1993, ApJS, 88, 253 [NASA ADS] [CrossRef] (In the text)
- Tepper-García, T., Richter, P., Schaye, J., et al. 2011, MNRAS, 413, 190 [NASA ADS] [CrossRef] (In the text)
- Theuns, T., Viel, M., Kay, S., Schaye, J., Carswell, R. F., & Tzanavaris, P. 2002, ApJ, 578, L5 [NASA ADS] [CrossRef] (In the text)
- Thom, C., & Chen, H.-W. 2008a, ApJS, 179, 37 [NASA ADS] [CrossRef] (In the text)
- Thom, C., & Chen, H.-W. 2008b, ApJ, 683, 22 [NASA ADS] [CrossRef] (In the text)
- Tripp, T. M., Savage, B. D., & Jenkins, E. B. 2000, ApJ, 534, L1 [NASA ADS] [CrossRef] [PubMed] (In the text)
- Tripp, T. M., Sembach, K. R., Bowen, D. V., Savage, et al. 2008, ApJS, 177, 39 [NASA ADS] [CrossRef] (In the text)
- Tumlinson, J., Werk, J. K., Thom, C., et al. 2011, ApJ, 733, 111 [NASA ADS] [CrossRef] (In the text)
- Wakker, B. P., & Savage, B. D. 2009, ApJS, 182, 378 [NASA ADS] [CrossRef] (In the text)
Absorption profiles for the O vi absorber at z = 2.1098 in the high-resolution data (left panel) and the intermediate-resolution data (right panel).
|Open with DEXTER|
|In the text|