Major Contributor to AGN Feedback: VLT X-shooter Observations of S IV BALQSO Outflows

TL;DR: The most energetic BALQSO outflow measured to date, with a kinetic luminosity of at least 10{sup 46} erg s{sup -1}, was reported in this paper.

Abstract: We present the most energetic BALQSO outflow measured to date, with a kinetic luminosity of at least 10{sup 46} erg s{sup -1}, which is 5% of the bolometric luminosity of this high Eddington ratio quasar. The associated mass-flow rate is 400 solar masses per year. Such kinetic luminosity and mass-flow rate should provide strong active galactic nucleus feedback effects. The outflow is located at about 300 pc from the quasar and has a velocity of roughly 8000 km s{sup -1}. Our distance and energetic measurements are based in large part on the identification and measurement of S IV and S IV* broad absorption lines (BALs). The use of this high-ionization species allows us to generalize the result to the majority of high-ionization BALQSOs that are identified by their C IV absorption. We also report the energetics of two other outflows seen in another object using the same technique. The distances of all three outflows from the central source (100-2000 pc) suggest that we observe BAL troughs much farther away from the central source than the assumed acceleration region of these outflows (0.01-0.1 pc).

Summary

1. INTRODUCTION

• Broad absorption line (BAL) outflows are observed as blueshifted troughs in the rest-frame spectrum of ∼20% of quasars (Hewett & Foltz 2003; Ganguly & Brotherton 2008; Knigge et al. 2008).
• The research program developed by their team has led to the determination of Ṁ and Ėk in several quasar outflows (e.g., ∗ Based on observations collected at the European Southern Observatory, Chile, PID: 87.B-0229.
• One way to alleviate this uncertainty is to target objects which possess absorption troughs from excited states of highionization species, where S iv/S iv* λλ1062.66, 1072.97 are especially promising.
• The ionization similarity of C iv and S iv makes S iv/S iv* outflows a much better agent than the usual low-ionization species for the determination of the feedback from the high-ionization outflows (see discussion in Section 5.2).

2. OBSERVATIONS AND DATA REDUCTION

• X-shooter is the second-generation, medium spectral resolution (R ∼ 6000–9000) spectrograph installed at the Cassegrain focus of VLT/UT2 (Vernet et al. 2011).
• The unique design of the instrument, in which the incoming light is split into three independent arms (UVB, VIS, and NIR) each composed of a prism-cross-dispersed echelle spectrograph, allows the simultaneous covering of a wide spectral band (3000–24,000 Å) in a single exposure.

3. SPECTRAL FITTING

• The radial velocity values across the absorption troughs are determined with respect to the systemic redshift of the quasar.
• Hewett & Wild (2010) report an improved redshift value3 for these two SDSS DR7 quasars by cross-correlating a quasar emission line template to the observed SDSS spectrum.
• Note that while the NIR range of the X-shooter observations of J1106+1939 covers the Hβ+ [O iii] λ5007 and also the Mg ii rest-frame regions, these portions of the spectrum are located within dense and strong H2O and CO2 atmospheric bands preventing us from determining a more accurate redshift from the fit of these emission features.
• The authors also examined the expected position of [O ii] λ3727 and found no emission feature in its spectral vicinity.

3.1. Unabsorbed Emission Model

• Deriving ionic column densities from absorption troughs requires knowledge of the underlying unabsorbed emission F0(λ).
• This approach could strongly underestimate the true underlying O vi emission.
• The resulting emission model for the UVB+VIS region is shown in Figure 1.

4. PHOTOIONIZATION ANALYSIS

• The authors use photoionization models in order to determine the ionization equilibrium of the outflow, its total hydrogen column density (NH), and to constrain its metallicity.
• Given the lack of observational constraints in wavebands outside the X-shooter spectral range for both objects, the authors choose the UV-soft spectral energy distribution (SED) model for high-luminosity radio-quiet quasars described in Dunn et al. (2010).
• The use of this SED model, lacking the so-called big blue bump from the classical MF87 SED (Mathews & Ferland 1987), is motivated by the rather soft FUV slopes observed by Telfer et al. (2002) over a large sample of Hubble Space Telescope spectra of typical radio-quiet quasars (see Dunn et al. 2010 for a detailed discussion).
• Ionic column densities predicted by the models are tabulated and compared with the measured values in order to determine the models that best reproduce the data.

5.1. Determining R: The Distance of the Outflow from the Central Source

• Measuring R is crucial for estimating of Ṁ and Ėk , and is also essential for understanding the relationship of the outflows to the host galaxy and its surroundings.
• In Section 4, the authors derived the ionization parameter (UH) for each outflow.

5.1.1. Determining nH

• In highly ionized plasma the hydrogen number density is related to the electron number density through ne 1.2nH.
• In this case the systematic errors mentioned above should dominate.
• The different distribution of the absorbing material for the F1+F2 model causes a much larger deviation in the derived S iv* and S iv column densities than possible errors due to their continuum placement.

5.2. Constraining Ω

• For a full discussion about constraining Ω that is needed for Equations (2) and (3), the authors refer the reader to Section 5.2 in Dunn et al. (2010).
• The authors only compare BAL outflows for two reasons.
• The spectral coverage of the SDSS can show these S iv/S iv* troughs only for redshifts z > 2.8, where the forest is very thick and greatly complicates the identification of narrow S iv troughs.
• For their purposes here, the authors need to account for both pure S iv outflow (since low-density outflows may not show S iv* even when S iv absorption is unambiguous) and also cases where the S iv and S iv* are so wide that they blend into one trough.
• There are also three plausible cases of matching narrower S iv and/or S iv* absorption features, but the authors do not include them as the possibility of a false positive identification of unrelated Lyα absorption features is quite significant.

5.3. Results

• Inserting the parameters derived in Sections 4 and 5 into Equations (1)–(3), the authors calculate the distance and energetics for the three S iv outflows.
• These quantities, as well as most of the parameters they are derived from, are listed in Table 5.
• In Table 5 the authors present three different models for the outflow of SDSS J1106+1939.
• Using solar abundances and the UV-soft SED (see Section 4 for details).
• The advantage of this model is its simplicity and the ease of comparison with any other outflows that are modeled with solar abundances.

6. RELIABILITY OF THE MAIN STEPS IN

• Due to the potential importance of this result to AGN feedback processes, the authors review in this section the steps that were taken in order to arrive at this result.
• The authors aim is to address possible caveats or systematic issues that might affect this result, especially whether Ėk can differ significantly from the most plausible value the authors report in Table 5 (Z = 4 Z model).

6.1. Ionic Column Density Extraction and Its Implications

• Reliable measurements of the absorption ionic column densities (Nion) in the troughs are crucial for determining almost every physical aspect of the outflows: ionization equilibrium and abundances, number density, distance, mass flux, and kinetic luminosity.
• A firm lower limit on Nion of a given trough is produced by integrating the AOD (τAOD) of the trough across its width (see Borguet et al. 2012).
• As shown in Paper II, the actual Nion can be 1000 times larger than the value inferred from τAOD.
• (2) τAOD of the blue doublet component is significantly larger than the τAOD of the red doublet component within the measurement errors.
• The authors note that in such a case the estimated Ėk will be larger for two compounding reasons.

6.2. Photoionization Modeling: Sensitivity to Different SEDs and Abundances

• The lack of observational constraints on the incident SED, especially in the critical region between 13.6 eV and the soft X-ray region, motivates us to check the sensitivity of the above results to other AGN SEDs.
• The authors note that, as shown in Table 5, the derived mass-flow rate and kinetic luminosity using the MF87-determined NH and UH are consistent with those derived using the UV-soft SED to better than 25%.
• The authors conclude that the derived results are only mildly sensitive to other physically plausible quasar SEDs.
• In contrast, the derived Ṁ and Ėk are more sensitive to departure from solar metallicity.
• A comparison of the results in the first and third models given in Table 5 shows the following.

6.3. Reliability of the Distance Estimate

• The authors distance estimate is derived from measuring the ratio of S iv*/S iv column densities.
• The converse is true for cases where the S iv*/S iv column density ratio is smaller than unity, which is the case for their SDSS J1106+1939 measurements.
• Therefore, the distance the authors derive for this outflow is technically a lower limit.
• Thus, again their reported Ėk is conservative.

6.4. Global Covering Factor (Ω)

• As noted in the Introduction, this aspect is one of the major strengths of the current analysis.
• For the energetics calculation, the authors chose the more conservative Ω = 0.08, which minimizes Ėk .

6.5. Can the Outflows Hide a Significant Amount of Undetected Mass

• The highest ionization species available in their spectrum is O vi, for which the authors can obtain only a lower limit for Nion.
• This does not allow us to probe the higher ionization material that is known to exist in AGN outflows via X-ray observations, the socalled warm absorber (WA) material.
• At least in one case, UV spectra of outflows from luminous quasars show a similar phenomenon where very high-ionization lines (Ne viii and Mg x) show that the bulk of the outflowing material is in this very highionization component (see Muzahid et al.

7. DISCUSSION

• The powerful BAL outflow observed in SDSS J1106+1939 possesses a kinetic luminosity high enough to play a major role in AGN feedback processes, which typically require a mechanical energy input of roughly 0.5%–5% of the Eddington luminosity of the quasar (Hopkins & Elvis 2010; Scannapieco & Oh 2004, respectively).
• 5%LBol, it has enough energy to drive the theoretically invoked AGN feedback processes.
• The investigation described here gives the first reliable estimates of R, Ṁ , and Ėk for a few high-ionization, highluminosity quasar outflows.
• Furthermore, the absorption spectra of these two objects look very similar to the run-of-the-mill BALQSO spectra longward of Lyα.
• The authors conclude that most AGN outflows are observed very far from their initial assumed acceleration region.

Figures

Figure 4. Identification of the S iv and S iv* intrinsic absorption troughs in the spectrum of SDSS J1512+1119. The red wing of the troughs associated with component 2 of S iv* λ1072.97 is mainly affected by a blend due to the S iv* λ1073.51 transition of component 2 that allowed us to pinpoint the S iv* column density for that component (see Paper II).

Table 2 Ionic Column Densities in the Outflow of SDSS J1106+1939 Using Model 2

Figure 3. Fitting the high-ionization P v, S iv+S iv*, and Si iv BAL profiles with the Model 2 optical depth profile. As for Model 1, the optical depth distribution associated with a transition is assumed to be known, leaving only the maximum optical depths of each component as a free parameter in each fit. Model 2 is composed of a sum of two functions: F1 (a Gaussian with a modified red wing) represented by a dotted line and F2 (a Gaussian with identical parameters to G2) represented by a dashed line. The total model for the trough is plotted as a solid line. For P v the fit requires a ratio of optical depths between both components of the doublet of 1:1.1, suggesting the possibility of moderate non-black saturation.

Table 5 Physical Properties of the S iv Quasar Outflows

Table 3 Ionic Column Densities Associated with Component 4 of the SDSS J1512+1119 Outflow

Figure 8. Density diagnostic (and errors) for SDSS J1106+1939 and kinematic component 4 of the SDSS J1512+1119 outflow. We plot the theoretical ratio of the level population of the first excited states of S iv (E = 951 cm−1) to the level population of the ground state versus the electron number density ne for three representative temperatures. The density diagnostic provided is fairly temperature independent for the range of densities represented in this plot. We determine the electron number density associated with the two outflows using the curve for the temperature of 10,000 K found for our best-fit Cloudy models (see the text).

Figure 7. Photoionization modeling of component 4 of the outflow in SDSS J1512+1119 assuming solar abundances. As in Figure 5, a measured column density is represented by a shaded area (showing the uncertainties), an upper limit on a column density is represented by a dashed line, and a lower limit by a dot-dashed line. The region of the phase space in which models are able to reproduce the observed constraints is roughly delimited by the trapezium.

Figure 1. Reduced and fluxed UVB+VIS X-shooter spectra of the quasar SDSS J1106+1939. We indicate the positions of the absorption lines associated with the intrinsic outflow. The VIS part of the spectrum (λobs > 5700 Å) has not been corrected for atmospheric absorption. This does not affect our study since our diagnostic lines are located in regions free of such contamination. The dashed line represents our unabsorbed emission model (see Section 3.1).

Table 4 Results of the Photoionization Modeling for SDSS J1106+1939

Figure 2. Fits to the mildly blended high-ionization absorption troughs (P v, S iv, and Si iv), as well as the non-saturated troughs (He i*, Mg ii, C ii, Al ii, Al iii) in the X-shooter spectrum of SDSS J1106+1939. Each transition is fitted with an identical optical depth profile whose shape is described by a sum of two Gaussians G1+G2 (Model 1, see Section 3.2) leaving only the maximum optical depth as a free parameter in each fit. G1 is represented by a dotted line and G2 by a dashed line. The full model for the trough is plotted as a solid line. This simple model provides a good fit to He i* and the low-ionization species as well as to the unblended blue wing of the P v absorption. The poor fit of the P v, S iv+ S iv*, and Si iv BAL red wings suggests the existence of another component that is not seen in the low-ionization species (see the text and Figure 3).

Figure 5. Photoionization modeling of the outflow in SDSS J1106+1939 assuming solar abundances. The shaded regions represent the locus of points (UH, NH) able to reproduce the observed ionic column density for a given species (see Table 1); dashed lines represent an upper limit on the column density of the ion; and dot-dashed lines represent a lower limit on the column density. The best model is marked by a diamond and the systemic error on the solution is represented by the cross. For clarity we do not represent the Al ii line since it is cospatial with the C ii line.

Figure 6. Photoionization modeling of the outflow in SDSS J1106+1939 assuming a metallicity Z/Z = 4. The description of the figure is identical to Figure 5.

Table 1 Ionic Column Densities in the Outflow of SDSS J1106+1939 Using Model 1

Frequently asked questions

Q1. What are the contributions in "C: " ?

The authors present the most energetic BALQSO outflow measured to date, with a kinetic luminosity of at least 1046 erg s−1, which is 5 % of the bolometric luminosity of this high Eddington ratio quasar. The authors also report the energetics of two other outflows seen in another object using the same technique. The distances of all three outflows from the central source ( 100–2000 pc ) suggest that the authors observe BAL troughs much farther away from the central source than the assumed acceleration region of these outflows ( 0. 01–0. 1 pc ). 

Q2. What are the parameters that determine the kinetic luminosity of the outflows?

Reliable measurements of the absorption ionic column densities (Nion) in the troughs are crucial for determining almost every physical aspect of the outflows: ionization equilibrium and abundances, number density, distance, mass flux, and kinetic luminosity. 

Q3. What is the ionization parameter for the outflow?

The ionization parameterUH ≡ QH 4πR2cnH , (1)(where QH is the source emission rate of hydrogen ionizing photons, R is the distance to the absorber from the source, c is the speed of light, and nH is the hydrogen number density) and NH of the outflow are determined by self-consistently solving the ionization and thermal balance equations with version c08.00 of the spectral synthesis code Cloudy, last described in Ferland et al. (1998). 

Q4. What is the role of the BAL outflows in shaping the early universe?

The energy, mass, and momentum carried by these outflows are thought to play a crucial role in shaping the early universe and dictating its evolution (e.g., Scannapieco & Oh 2004; Levine & Gnedin 2005; Hopkins et al. 

Q5. What is the way to alleviate this uncertainty?

One way to alleviate this uncertainty is to target objects which possess absorption troughs from excited states of highionization species, where S iv/S iv* λλ1062.66, 1072.97 are especially promising. 

Q6. What is the simplest method for estimating the column density of an ionic species?

The column density of an ionic species i associated with a given kinematic component is estimated by modeling the residual intensity Ii(λ) ≡ Fobs(λ)/F0(λ) as a function of the radial velocity. 

Q7. Why is modeling of the unabsorbed emission in the region of the spectrum important?

modeling of the unabsorbed emission in that region is not important for their BAL analysis due to the fact that it contains only heavily blended diagnostic lines coupled with a limited signal-to-noise ratio (S/N). 

Q8. What is the best-fit photoionization model for MF87?

The best-fit photoionization models are parameterized by log UH = −0.2 and log NH = 22.6 cm−2 for MF87, within 0.2 dex of the values obtained with the UV-soft SED. 

Q9. What is the spectral energy distribution for the quasar?

Given the lack of observational constraints in wavebands outside the X-shooter spectral range for both objects, the authors choose the UV-soft spectral energy distribution (SED) model for high-luminosity radio-quiet quasars described in Dunn et al. (2010). 

Q10. How did the authors find the ionization parameter and column density?

Using the SED developed in Mathews & Ferland (1987), the authors found the ionization parameter and column density dropped by ≈0.2 and 0.3 dex, respectively. 

Q11. What is the main assumption when using this technique?

The main assumption made when using this technique is that the physical properties of the absorbing gas do not significantly change as a function of the radial velocity for a given kinematic component (e.g., Moe et al. 2009; Dunn et al. 2010). 

Q12. How are the Si iv lines compared to other ionic species?

the Si iv lines are narrow enough and unblended (Δv ∼ 2000 km s−1) to be adopted as a template to identify the kinematic components in other ionic species.