EDP Sciences
Free Access
Issue
A&A
Volume 610, February 2018
Article Number L13
Number of page(s) 4
Section Letters to the Editor
DOI https://doi.org/10.1051/0004-6361/201732406
Published online 23 February 2018

© ESO 2018

1 Introduction

Since the discovery of the relation between the mass of a super-massive black hole (SMBH) and the bulges of their host galaxies (i.e., the “M–σ relation”, Kormendy & Richstone 1995; Magorrian et al. 1998), we know that SMBHs likely play a role in the formation and growth of the galaxies (Fabian 2012). Active galactic nuclei (AGN)-driven ultra-fast outflows (UFOs; Tombesi et al. 2010a) have been recently proposed as a major feedback process whereby sweeping out and/or compressing the interstellar gas may influence the formation and growth of the galaxies (Fabian 2012; King & Pounds 2015).

Resonant absorption lines detected in the ~7–10 keV energy range due to highly ionized iron are the UFO signatures. They are measured in both radio-quiet and radio-loud objects and both in type-I and -II AGNs (Tombesi et al. 2010a,b; Gofford et al. 2013). While the average properties of UFOs are known at low z, we have only a few UFO detections at z ≥ 1.5, that is, where they may have acted to shape the M–σ relation seen today (Hasinger et al. 2002; Chartas et al. 2002, 2003, 2007, 2016; Lanzuisi et al. 2012; Vignali et al. 2015).

Here we present the XMM-Newton spectrum of MG J0414+0534, a lensed (magnification factor μ ~ 30–60, Trotter et al. 2000; Minezaki et al. 2009) and radio-loud type-I QSO at z = 2.64 (Lawrence et al. 1995). The target is also a hyper-luminous infrared and red QSO (Lawrence et al. 1995; McLeod et al. 1998). These objects are thought to represent a dust-enshrouded phase in AGN evolution during which nuclear winds are expected to be present and expel/heat the cold gas in the hosting galaxy (Georgakakis et al. 2009; Urrutia et al. 2009) thus enabling feedback processes between the SMBH and the galaxy bulges (Fabian 2012; King & Pounds 2015). In X-ray, the source was previously pointed by Chandra and the spectrum was described by an absorbed power-law (Γ = 1.7 ± 0.1, NH = 4.7 ± 0.7 × 1022 cm−2, errors are at 90% confidence level for one parameter of interest here and throughout the paper, Avni 1976) plus an iron line in emission (EFeKα = 6.4 ± 0.1 keV, EWFeKα = 200 ± 100 eV; Chartas et al. 2002).

2 Data reduction and analysis

XMM-Newton pointed to MG J0414+0534 on March 11, 2017. SAS-15 and the latest available software and response matrices were used to reduce and analyze the data. The observation lasted ≈78 and ≈76 ks for EPIC-pn and EPIC-MOS instruments, respectively. Since it was affected by soft-p+ flares, high-background intervals were removed through an iterative sigma-clipping procedure applied to the 10–15 keV band data; we were left with cleaned exposure times of 48.5, 66.5, and 69.1 ks of exposure for pn, MOS1, and MOS2, respectively.

The images of MG J0414+0534 are within ≈ 3′′ (Chartas et al. 2002), thus they form a single “point-like” source in XMM-Newton. Source counts (≈9000 in total in the 0.3–10 keV band) have been extracted from circular regions with radii of 25′′ for pn and 20′′ for the MOS. Background was extracted from larger, source-free circular regions in the same chip of the target; it contributes ≈5–10% of the flux in the 2–5 keV band.

Since no significant variability was observed, we performed a time-averaged spectral analysis. Data were grouped so as to obtain 20 source plus background counts per bin. Spectral features are clearly visible once the spectrum is modeled with a power-law and, among them, a drop of counts at E ≈ 2.5 keV (Fig. 1). The X-ray flux of MG J04141+0534 is F0.5 8 keV ≈ 4 × 10−13 erg s−1 cm−2 (see model #1 in Table 1) which is within the range of values previously recorded (F0.5–8 keV ≈ 2.5–8 × 10−13 erg s−1 cm−2, Chartas et al. 2002; Pooley et al. 2012). Following Chartas et al. (2002) we modeled the XMM-Newton spectrum with an absorbed power law plus an iron line in emission finding consistent results, that is, Γ = 1.75 ± 0.05, NH = 3.9 ± 0.5 × 1022 cm−2, EFeKα = 6.37 ± 0.10 keV with an equivalent width EWFeKα = 95 ± 53 eV.

The addition of a Gaussian in absorption at E ≈ 2.5 keV (E ≈9.2 keV rest frame) is required by the data (Δ χ2 ≈ 27 for two parameters of interest corresponding, using the F-test, to a 5σ detection; model #2 in Table 1 and Fig. 3). Its EW is ≈ 235± 70 eV (rest frame)and it is consistent to be narrow (if the line width is left free to vary we obtain Δ χ2 ≈ 0.1 and a 90% confidence upper limit of σ ≤ 250 eV, rest frame). This makes an edge origin implausible for at least part of the feature as proposed for APM 08279+5255 (Hagino et al. 2017): indeed if we substitute the Gaussian with an edge we obtain a worst fit by Δ χ2 ≈ 5 for the same number of parameters. To test the absorption feature, we searched for its presence in each single EPIC detector dataset. We used the model #1 in Table 1 as a baseline. All the parameters of the model (except for the width of the lines which was fixed to σ = 0) were free to vary. The result is plotted in Fig. 4. The absorption line is detected at more than 99% confidence level in both MOS1 (Δχ2 = 14.2 for two parameters of interest) and pn datasets (Δ χ2 = 13.0 for two parameters of interest), while there are hints of its presence in the same energy range in the MOS2 (Δ χ2 = 3.5 for two parameters of interest). A similar combination of independent detections is highly improbable. We performed 1000 Monte-Carlo simulations for each EPIC detector using model #1 of Table 1 as baseline. We searched for detections of spurious absorption lines between (rest frame) 7 and 14 keV (corresponding to outflow velocities of ~0.01–0.6c) in the simulated spectra. We found that none of the 1000 simulations allowed us to obtain detections for which the line centroids are within 1 keV range for the three detectors (rest frame, see Fig. 4) and with a Δ χ2 of at least 10 for two instruments and 3 for the other. Thus, considering the conservative approach that we used, we can assess that the probability of measuring an absorption feature as seen in MG J0414+0534 by chance is well below 0.1% and fully consistent with the combined probability obtained with the F-test (see above). Since the line is close to some instrumental edges (E ≈ 2.35 and E ≈ 2.8 keV), we also tried, without success, to account for the E ≈ 2.5 keV feature allowing the detector gain to change (using “gain fit” within Xspec). We finally searched for a similar line in the longest Chandra exposure, finding that, fixing the line at E = 9.2 keV, the EW is ≤130 eV (90% confidence level); that is, if present, the line has varied since then(EW = 235 ± 70 eV today).

The detection of the FeKα emission line may indicate the presence of a reflection component. This feature is commonly observed in nearby Seyfert galaxies (e.g., Perola et al. 2002), and recently it has been detected also in some high-z QSO (Dadina et al. 2016; Lanzuisi et al. 2016). To test this hypothesisand to further probe the robustness of the detection of the absorption feature against a more complex underlying continuum, we tried the Pexmon reflection model (Nandra et al. 2007) fixing the inclination angle (Θ =30) and the high-energy cut-off (Ecut-off = 100 keV). The data do not require this component and we obtained an upper limit (90% confidence level) on the relative reflected-to-direct emission normalization of r ≤ 0.1. The detection of the absorption line is highly significant (more than 99%) also in this case.

We finally tried a UFO scenario in which the absorption feature at 2.5 keV is due to ionized and outflowing gas. To this end, we used the warmabs model based on Xstar (Kallman & Bautista 2001). We fixed the abundances at all elements to the solar value and the turbulence velocity to vturb = 3000 km s−1, in agreement with what is measured in local AGN (Tombesi et al. 2012, 2014). The free parameters of the fit are the column of the ionized absorbing gas, its ionization parameter, and the redshift at which the absorber is detected. This last value allows us to infer the outflow velocity of the absorber. As presented in Table 1 (model #3) and plotted in Fig. 5, we obtained a good fit and the absorber is found to be at redshift zobs ≈ 1.72, which corresponds, considering the relativistic effects along the line of sight, to an outflow velocity of vout = (0.28 ± 0.05)c. The ionization state (Log(ξ) ≈ 4) strongly indicates that the absorption line is due to FeXXVI (see also Tombesi et al. 2011).

thumbnail Fig. 1

Data-to-model ratio expressed in terms of standard deviations with respect to a power-lawabsorbed by Galactic column density (1.02 × 1021 cm−2, Kalberla et al. 2005). A deep and narrow drop of counts at E ≈ 2.5 keV (observed frame) is clearly present. Black data points indicate EPIC-pn, while green and red data points represent EPIC-MOS1 and MOS2, respectively.

Open with DEXTER
Table 1

Spectral models.

thumbnail Fig. 2

Confidence contours of the photon index vs. the absorbing column (upper panel) and of the Gaussian emission line normalization vs. its energy centroid (lower panel; rest frame) adopting model #1 in Table 1. Contours are at 99%, 90% and 68%.

Open with DEXTER
thumbnail Fig. 3

Confidence contours plot of the Gaussian absorption line normalization vs. its rest-frame energy centroid (see model #2 in Table 1). Contours confidence levels are as in Fig. 2.

Open with DEXTER
thumbnail Fig. 4

Confidence contours plot of the Gaussian absorption line normalization vs. its rest-frame energy for each single EPIC instrument. Contours confidence levels are as in Fig. 2.

Open with DEXTER
thumbnail Fig. 5

Unfolded X-ray observed frame energy spectrum of MG J0414+0534 (upper panel) obtained using model #3 in Table 1 and displayed in the middle panel. This model fits well the data and no strong residuals are left (lower panel). Color-code is as in Fig. 1.

Open with DEXTER

3 Discussion and results

We present the results obtained analyzing the XMM-Newton data of the radio-loud quasar MG J0414+0534 taken on March 11, 2017. We probed its radio-loudness using the parameter R = f5 GHz/f4400 (R ≥ 10 for radio-loud sources, Kellermann et al. 1989). To obtain the rest frame value of R we used the observed fluxes in H band (mH ≈ 13.95) (Skrutskie et al. 2006) and at 1.4 GHz (f1.4 GHz = 2.1 ± 0.1 Jy, Condon et al. 1998). The result is that R ≈ 780. The average shape of the X-ray continuum is very much in agreement with what was previously found by Chartas et al. (2002) for the brightest image (Image A) of the source. The photon index is Γ ≈ 1.7 and there is a cold absorbing column of NH ≈ 4 × 1022 cm−2. We also detected a cold iron line (EFeKα = 6.4 ± 0.1 keV) in emission with EWFeKα = 95 ± 53 eV. According to the present analysis, the iron emission line may be due to the same matter responsible for the cold absorption assuming an almost spherical distribution of such a component (e.g., Leahy & Creighton 1993).

The observed luminosity of MG J0414+534, once corrected only for absorption, is L2 10 keV ≈ 1.5 × 1046 erg s−1 adopting a standard Λ CDM cosmology with H0 = 70 km s−1 Mpc−1 and Ω λ = 0.73. If we assume the magnification factor μ = 45 between the estimated values of 30 and 60 (Trotter et al. 2000; Minezaki et al. 2009), we can infer an intrinsic X-ray luminosity L210 keV ≈ 3 × 1044 erg s−1 that corresponds to an intrinsic Lbol ≈ 1 × 1046 erg s−1 assuming the bolometric correction factor (kbol ≈ 30) by Lusso et al. (2012). Based on the Hβ broadening, the SMBH mass has been estimated to be (Peng et al. 2006) and this implies that the source is emitting at ≈5% of its Eddington limit (LEdd ≈ 2 × 1047 erg s−1).

The main result of our analysis is the first detection, to our knowledge, of an UFO in a radio-loud object at z ≥ 1.5. The absorption feature is due to iron resonant absorption (essentially FeXXVI) in ionized and outflowing gas (Log(ξ) ≈ 3.9, NH ≈ 8 × 1023 cm−2) with a velocity of vout = (0.28 ± 0.05)c. It is worth noting here that the UFO characteristics of MG J0414+0534 are consistent with what observed in nearby radio-loud AGNs (vout ~ 0.04–0.43c, NH ≥ 1022 cm−2, Log(ξ) ~ = 1.4–5.6, Tombesi et al. 2014). Assuming standard “recipes” and considering the large uncertainities on parameters such as M, μ and kbol, we can try to infer a very rough and purely indicative estimate of the outflow mechanical output. Following Tombesi et al. (2012), we assume that the outflowing gas has been detected at a radius at which the observed velocity corresponds to the escape velocity, that is, . Using M and vout reported above, we obtain that r ≈ 6.8 × 1015 cm, that is, r ≈ 25rg. Following Crenshaw et al. (2003), we can estimate the mass-outflow rate as: (1) where mp is the proton mass, μ the mean atomic mass per proton (1.4 for solar abundances), NH the column of the ionized gas, vout the line of sight outflow velocity, r the absorber’s radial location, and C the global covering factor (C ≈ 0.5 here, i.e., similar at the median value obtained by Tombesi et al. 2010a). The obtained mass-outflow rate is of the order of  yr−1 that corresponds to Ė kin ≈ 2.5 × 1046 erg s−1, that is, ≈2.5Lbol, and to an outflow momentum rate of out ≈ 6 × 1036 g cm s−2 , that is approximately 17 times the radiation force rad = Lbolc. The ĖLbol ≈ 2.5 ratio is much larger than what is observed in the local Universe (Ė kinLbol ≈ 0.01, Tombesi et al. 2013; Gofford et al. 2015) and well above the limit to switch-on/off feedback mechanisms by AGN-driven winds (Ė kinLbol ≥ 0.01, Di Matteo et al. 2005; Hopkins & Elvis 2010) but it is consistent with what is observed in other distant quasars (see for example thecase of the radio-quiet quasar HS 0810+2554 where Ė kin ≈ 9Lbol, Chartas et al. (2016). Moreover, together with the large ratio between the wind and radiation forces ( outrad ≈ 17), this may indicate that the magnetic field is probably dominating in driving the outflow in accordance with the radio loudness of the source.

Finally, it is worth noting here that MG J0414+0534 is the seventh QSO at z ≥ 1.5 in which UFOs have been detected. Excluding HS 1700+6416 (Lanzuisi et al. 2012) and PID352 (Vignali et al. 2015) which are non-lensed, the remaining objects, namely APM 08279+5255, PG1115+080, H1413+117, HS 0810+2554 (Hasinger et al. 2002; Chartas et al. 2003, 2007, 2009, 2016) and MG J0414+0534 are lensed. The flux enhancement due to the lensing does certainly help to collect good quality X-ray spectra and this may help in detecting such features. Alternatively, one can speculate that the flux enhancement makes it easier to probe weaker fluxes and, if the anti-correlation between the absorption line EW and the source flux observed in IRAS 13224−3809 (Parker et al. 2017) holds also at high-z, gravitational lensing helps in getting stronger features. However, the current absence of large enough samples of good-quality X-ray spectra of either lensed or non-lensed high-z QSO has hampered the study of these or other possible effects which must be accounted for if we want to understand how the feedback mechanism worked along cosmic time to shape the observed Mσ relation.

Acknowledgments

We thank the referee, James Reeves, for his helpful comments. This work is based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and the USA (NASA). M.D. acknowledges financial under contract ASI – INAF grant I/037/12/0. G.P. acknowledges the Bundesministerium für Wirtschaft und Technologie/Deutsches Zentrum für Luft- und Raumfahrt (BMWI/DLR, FKZ 50 OR 1408) and the Max Planck Society. B.D.M. acknowledges support from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 665778 via the Polish National Science Center grant Polonez UMO-2016/21/P/ST9/04025.

References

All Tables

Table 1

Spectral models.

All Figures

thumbnail Fig. 1

Data-to-model ratio expressed in terms of standard deviations with respect to a power-lawabsorbed by Galactic column density (1.02 × 1021 cm−2, Kalberla et al. 2005). A deep and narrow drop of counts at E ≈ 2.5 keV (observed frame) is clearly present. Black data points indicate EPIC-pn, while green and red data points represent EPIC-MOS1 and MOS2, respectively.

Open with DEXTER
In the text
thumbnail Fig. 2

Confidence contours of the photon index vs. the absorbing column (upper panel) and of the Gaussian emission line normalization vs. its energy centroid (lower panel; rest frame) adopting model #1 in Table 1. Contours are at 99%, 90% and 68%.

Open with DEXTER
In the text
thumbnail Fig. 3

Confidence contours plot of the Gaussian absorption line normalization vs. its rest-frame energy centroid (see model #2 in Table 1). Contours confidence levels are as in Fig. 2.

Open with DEXTER
In the text
thumbnail Fig. 4

Confidence contours plot of the Gaussian absorption line normalization vs. its rest-frame energy for each single EPIC instrument. Contours confidence levels are as in Fig. 2.

Open with DEXTER
In the text
thumbnail Fig. 5

Unfolded X-ray observed frame energy spectrum of MG J0414+0534 (upper panel) obtained using model #3 in Table 1 and displayed in the middle panel. This model fits well the data and no strong residuals are left (lower panel). Color-code is as in Fig. 1.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.