The soft X-ray spectrum of the luminous narrow line Seyfert galaxy PG1211+143 - evidence for a second high velocity outflow component
An XMM-Newton observation of the luminous Seyfert galaxy PG1211+143 in 2001 revealed the first clear evidence for a highly ionised high speed wind (in a non-BAL AGN), with a velocity of v0.09c based on the identification of blue-shifted absorption lines in both EPIC and RGS spectra. A subsequent analysis of EPIC spectra, including additional absorption lines, led to an upward revision of the wind speed to 0.14c, while broad band modelling indicated the need for a second, partial covering absorber to account for continuum curvature and spectral variability. We show here, in a new analysis of the XMM-Newton RGS data, that this additional absorber is detected in the soft X-ray spectra, with the higher spectral resolution providing a much improved velocity constraint, with v0.07c. Similar variability of the 0.07c and 0.14c outflow components suggest they are physically linked, and we speculate that occurs by the fast (primary) wind impacting on small clumps of higher density, slow moving matter close to the disc. We show that strong, velocity broadened soft X-ray emission features, located at the redshift of PG1211+143 indicate the extended scale of the ionised outflow.
keywords:galaxies: active – galaxies: Seyfert: quasars: general – galaxies: individual: PG1211+143 – X-ray: galaxies
X-ray spectra from an XMM-Newton observation of the narrow-line Seyfert galaxy PG1211+143 in 2001 provided the first detection in a non-BAL AGN of blue-shifted resonance line absorption corresponding to a sub-relativistic velocity of 0.09c (Pounds et al. 2003). Although a much lower velocity was claimed from a separate analysis, principally based on the RGS soft X-ray data (Kaspi and Behar 2006), the high velocity was confirmed - and adjusted to 0.140.01c - in a re-analysis where the higher energy resolution of the MOS cameras provided revised line identification (Pounds and Page 2006). Repeated observations with XMM-Newton, Chandra and Suzaku have since shown the high velocity outflow to be persistent, but of variable strength (eg Reeves et al. 2008). Confirmation that the outflow in PG1211+143 was both massive and energetic - with potential importance for galaxy feedback - was obtained from the detection of PCygni and other broad emission features in stacking of the 2001, 2004 and 2007 XMM-Newton EPIC spectra (Pounds and Reeves 2007, 2009).
More recently the examination of archival data from XMM-Newton has shown high velocity ionised winds (UFOs) to be relatively common in nearby, bright type 1 AGN (Tombesi 2010, 2011, 2012), a finding supported by a similar analysis of the Suzaku archive (Gofford 2013). The frequency of these detections confirms a typically large covering factor and hence significant mass and kinetic energy in such winds. Indeed, their integrated mechanical energy may be an order of magnitude greater than required to disrupt the bulge gas in the host galaxy, suggesting much of the energy in a persistent wind must be lost before reaching the star forming region.
The first evidence of a fast AGN wind shocking with the ISM or previous ejecta, at a radial distance where strong Compton cooling will cause much of the flow energy to be lost, has recently been found in an extended XMM-Newton study of the narrow line Seyfert 1 NGC4051 (Pounds and Vaughan 2011, Pounds and King 2013). That outcome may be expected for many of the low redshift AGN in the Tombesi et al. sample, with the resulting momentum-driven flow being the effective agent of galaxy feedback (King 2003, 2005).
The evidence for a cooling post-shock flow in NGC4051 was provided by the higher resolution RGS grating spectra, and it is relevant to note that all existing X-ray detections of high velocity outflows (v 10000 km s; 0.03c) have been based on blue-shifted absorption lines identified with highly ionised Fe in lower resolution CCD spectra. In the present paper we re-examine the RGS soft X-ray spectra of PG1211+143 to seek further definition of this prototype energetic outflow.
We assume a redshift for PG1211+143 of (Marziani et al. 1996). Spectral fitting is based on the XSPEC package (Arnaud 1996) and includes absorption due to the line-of-sight Galactic column of (Murphy et al. 1996). Estimates for the black hole mass in PG1211+143 range from (Kaspi et al. 2000) to (Bentz et al. 2009), with the lower value making the luminosity close to Eddington. It is interesting to note that historical X-ray fluxes of PG1211+143 are generally higher than in the XMM-Newton era (Bachev et al. 2009) further strengthening the case for continuum driving of the fast wind in this source (King and Pounds 2003).
2 Absorption velocity profile in the soft X-ray spectrum of PG1211+143
PG1211+143 was observed by XMM-Newton on 2001 June 15, 2004 June 21 and 2007 December 21 and 23. Here we concentrate on data from the Reflection Grating Spectrometer (RGS, Den Herder et al.2001), with effective exposures of 54 ks, 47 ks, 49 ks and 38 ks, respectively.
Although individual soft X-ray spectra are rather noisy from these relatively short exposures, several absorption lines were claimed in the original analysis of the 2001 observation (Pounds et al. 2003), identification with H- and He-like ions of C, N, O and Ne indicating an outflow velocity in the range 22000-24000 km s. However, the individual absorption lines are weak and - as noted in the Introduction - ion by ion modelling by Kaspi and Behar (2006) failed to confirm this high velocity in the RGS data. In the present analysis we use the velocity transformation method described in Pounds and Vaughan (2011a) which allows stacking of spectra for multiple resonance lines in order to enhance the visibility of absorption features.
Figure 1 (top panel) shows the composite velocity profile for the 5 principal resonance absorption lines of OVIII, OVII, NVII, NVI and CVI in the 2001 RGS spectra. The bin width of 400 km s matches the mid-band RGS spectral resolution. The blue-shifted soft X-ray absorption reported in Pounds et al. (2003) is clearly detected, with a broad absorption trough near -22000 km s contributing to a poor fit to a level continuum ( of 123 for 85 degrees of freedom). Addition of a negative Gaussian to model that absorption (figure 1, mid panel) finds an outflow velocity of 22230240 km s and width =1020250 km s, with a revised of 94/82. The absorption feature is highly significant, with an f-test random probability of , or allowing for 35 possible (1 line width) values across the velocity band. A Gaussian search across the velocity plot finds a weaker absorption velocity at 5590230 km s, with width =570240 km s (figure 1, mid panel), yielding an overall of 83/79. The latter absorption is intriguing, being close to the factor 4 reduction in velocity expected across a strong shock. However, the f-test shows this second feature to be only marginally significant (in the 2001 data) if no prior velocity is assumed.
The width of the high velocity absorption feature might be due to velocity variations over the data integration period or to physically separate velocity components. We find compelling evidence for the latter in Section 3, in terms of different ionisation levels, and meanwhile quantify this separation with an additional, narrow (=400 km s) Gaussian. The second absorption component lies close to the systemic velocity of PG1211+143, indicated in the lower panel of figure 1 by the symbol G. Allowing for a mid-range uncertainty in the RGS wavelength calibration of 8 mÅ (120 km s), local absorption - perhaps in the Galactic halo - appears an interesting possibility.
A repeat of the above exercise for the 2004 and 2007 RGS data (see Appendix) finds weaker absorption near -22000 km s, consistent with weaker Fe K resonance line absorption in the 2004 and 2007 EPIC data (Pounds and Reeves 2007, 2009). An indication of the near-systemic velocity component in both 2004 and 2007 profiles strengthens the case for a constant component in the summed RGS data from all 4 XMM-Newton observations.
|Comp||velocity||width ()||flux ( counts|
3 Comparison with an ionised outflow
To quantify the properties of the absorption features indicated in the velocity plots (figure 1) the full 2001 RGS spectral data were then modelled with absorption in a photoionised gas, using grids 18 and 21 from the XSTAR code (Kallman et al 1996). Grids 18 and 21 differ in having fixed turbulence velocities of 100 km s and 1000 km s, respectively, with free parameters being column density and ionisation parameter, with outflow (or inflow) velocities in the AGN rest frame obtained from the apparent redshift of the absorbing gas. Element abundances were fixed at solar values. In the final fit, described below, grid 21 was preferred for component a, giving a better constrained column density, with grid 18 being retained for components b and c.
The soft X-ray continuum was first fitted over the full 8-36 Å waveband by a power law plus black body, both attenuated by the Galactic column of . Positive and negative spectral features in the residuals contributed to a fit statistic of 960/858. Positive Gaussians were then added to represent broad line emission of OVII and OVIII and four REDGE components (to replicate recombination continua), with parameters determined from the stacked data in described in Section 4. Re-fitting the 8-36 Å spectrum with adjusted continuum components gave a revised baseline fit statistic of 920/858.
The addition of 3 photoionised absorbers successively improved the spectral fit to of 854/849. The fit parameters and significance of each XSTAR absorption component are listed in Table 2. For the first two components the fit was started with default values of the variable parameters (log N=21, log=0, z = 0), and repeated trials showed the fitted parameters to be robust. For component c, the initial redshift parameter was set at 0.06 in order to explore the intermediate velocity indicated in figure 1.
|Comp||log||N||velocity (km s)|
The two most significant absorption components in the photoionised modelling (a and b) are both of high velocity (relative to the AGN), with values close to the strongest absorption feature in the velocity plot of figure 1. However, those XSTAR absorbers have very different ionisation parameters and column densities, which might indicate that the fast outflow has entrained cooler gas, albeit with a substantially lower column density. An interesting alternative, given the proximity to the systemic velocity of PG1211+143, is that the lower ionisation component (b) is not associated with the AGN, but arises from low redshift matter in the same line of sight.
While the 3rd photoionised absorption component is of lower statistical significance, it is again consistent with a velocity absorption feature in Figure 1. Although the ionisation parameter of component 3 is typical of a ‘warm absorber’ the velocity would be unusually high. We return to this point in Section 7.
4 Evidence for velocity broadened emission
In an attempt to quantify the covering factor/collimation of the fast outflow in PG1211+143 Pounds and Reeves (2007) modelled the broad band EPIC spectrum with photoionised absorption and emission, assuming both to arise from the same ionised flow. An interesting factor in achieving a good fit was a requirement to broaden the emission lines by inclusion of a Gaussian smoothing parameter (GSMOOTH in XSPEC). In addition, the XSTAR emission spectrum indicated a mean velocity (in the AGN rest frame) of 30003000 km s, locating the ionised gas observed in absorption close to the redshift of the AGN. Here we examine individual emission features obtained by stacking the soft X-ray spectra from all four RGS observations to clarify the above indications. As reported previously, the only strong emission features are broad.
Figure 3 (top) shows coarsely binned RGS 1 spectra summed over all four XMM-Newton observations, plotted as a ratio to a power law plus black body continuum. Five broad emission features are modelled with positive Gaussians to determine the approximate wavelength and width of each feature, being found - for all 5 - to lie close to the respective wavelength (adjusted for the AGN redshift) of OVIII Lyman-, the OVII triplet and the radiative recombination continua (RRC) of OVIII, OVII and CVI. This is an important confirmation that the ionised gas seen in absorption is indeed associated with the AGN, while the width of the broad OVIII Lyman- emission line would correspond to velocity broadening of 20000 km s (FWHM), a value similar to that found for the PCygni line observed in Fe K (Pounds and Reeves 2009), and consistent with a wide angle outflow.
To better quantify the individual emission features, with the RRC widths potentially containing important information on the related electron temperature, the stacked XMM-Newton soft X-ray data were then modelled in XSPEC, where each RRC (including NeIX) was more appropriately fitted with a REDGE component.
In the lower panel of figure 3 the XSPEC fit is reproduced, including four significant RRC and broad and narrow emission components of OVIII Lyman- and the OVII triplet. The component parameters of the main broad emission features and corresponding statistical improvement over the baseline continuum fit are listed in Table 3.
Component energies are adjusted for the redshift of PG1211+143 in the table to allow direct comparison with theoretical values. The agreement is good, with the mean energy of the broad Lyman- line again suggesting a small blueshift. Assuming the RRC are affected by similar velocity broadening to the OVIII Lyman- line, yields an electron temperature of the relevant gas component of kT10–20 eV, consistent with the fast outflow being close to the Compton temperature in the radiation field of the AGN. (The derived value is consistent with an unpublished OVII RRC electron temperature of 145 eV from the Chandra LETG observation in 2004; Reeves private communication).
5 Scattering from the fast ionised wind
The emission features described above provide an important constraint on the scale of the outflow at v0.07c.
The XSPEC fit to the summed RGS data finds a flux for the strong OVII RRC of photons cm s. Assuming a temperature of the soft X-ray emitting gas of kT10 eV, the relevant recombination rate is of order cm s (Verner and Feldman 1996). For a solar abundance of oxygen and parent ion fraction of 0.3, we find an emission measure of cm, for a redshift-distance to PG1211+143 of 350 Mpc.
We assume the wind extends from launch at the escape velocity radius and coasts at near-constant velocity before colliding with the ISM. King and Pounds (2013) suggest that will typically be where the initial ISM has been swept up by radiation pressure into an optically thin shell. For PG1211+143 the predicted transparency radius 8 pc, yielding an emission volume () of cm. For a flow collimation b (=) of 0.5 (Pounds and Reeves 2009), comparison with the above emission measure then implies a mean electron density of cm. While that would give a column density over 8pc a factor 5 larger than component 1 in table 2, the latter is a crude estimate for a coasting wind where the particle density will fall as and the absorbing column is dominated by the inner radii.
The strength of the broad soft X-ray emission features, and their lack of obvious variability over several years, supports the view the corresponding ionised outflow continues unchecked to a relatively large distance.
6 Comparison with previous EPIC analyses
As noted in the Introduction, previous XMM-Newton studies of the outflow in PG1211+143 have been primarily based on the EPIC spectra. Although exhibiting a strong ‘soft excess’ over a ‘primary’ power law continuum (2.2), RGS spectra show only weak narrow spectral features. Detailed modelling of pn and MOS spectra (Pounds and Reeves 2007) suggested an explanation, with velocity-broadened emission lines and soft X-ray absorption lines being ‘diluted’ by a second, softer and less absorbed, continuum component (3.2).
Broad-band spectral modelling required two ionised absorbers in the partial covering fit, with log1.5 and 2.9. Outflow velocities from the XSTAR modelling were, respectively, 0.070.02c and 0.140.01c, with the higher velocity, higher ionisation absorption evidently driven by fitting blue-shifted resonance lines of Fe K and other heavy metal ions (as in Pounds and Page 2006), while the lower ionisation, lower velocity absorption was primarily responsible for the continuum curvature at lower energies, with the velocity determination consequently less secure.
Finding an outflow velocity of 0.07c in the 2001 RGS data now greatly strengthens the case for two high velocities in the PG1211+143 outflow, with a physical association being indicated by Fe K resonance lines and broad band absorption also being more pronounced in 2001 (eg Figure 2 in Pounds and Reeves 2009). The absence of absorption in the RGS data at the higher velocity of 0.14c can be explained by that component being more highly ionised.
We discuss the implications that both 0.07c and 0.14c flow components were present during the 2001 observation in Section 8.
7 Absorption in matter at lower redshift
Component b in the Xstar modelling of the 2001 RGS data, evidently linked with the Gaussian component 1a in Table 1, provides the first evidence of absorption in matter perhaps not associated with PG1211+143. The persistence of a narrow absorption feature close to the systemic velocity in the 2004 and 2007 RGS data (figure 4) lends support to a local origin. However, the derived velocity (Tables 2 and 3) is 500-1000 km s (2 and 3 ) less than the systemic velocity, arguing against an origin in the Galaxy, although Herenz et al.(2013) report significant CIV absorption in high velocity Galactic Halo clouds receding at 169 and 184 km s in the line of sight to PG1211+143.
The reality of component c in the XSTAR modelling is also backed by a similar velocity (relative to the AGN) of component 3 in figure 1, and might be the first evidence for a warm absorber in PG1211+143. The factor 4 difference in velocity compared to component ‘a’ would be consistent with the recent suggestion that the warm absorber arises from gas accumulating after the fast wind impacts the ISM or previous ejecta (Pounds and King 2013).
An interesting alternative explanation for component c is absorption in dwarf galaxies along the line of sight to PG1211+143. Prochaska et al. (2011) report OVI absorbing column densities of 2 cm z0.0510 and z0.0645. Component 3 in table 2 would give an OVI column only a factor of a few greater, while a small range in ionisation parameter would provide consistency.
Establishing a second high velocity outflow component in PG1211+143 is the most important result of the present analysis, being the first example, to our knowledge, of simultaneous multiple high velocities in a UFO. Although it is relatively common for broadband X-ray spectra of AGN to be fitted with a partial covering model, no velocity information has been reported for the partial covering absorber.
Detection of multiple high velocities will, of course, affect the overall energy and momentum budget, with important relevance to feedback. We restrict discusion here to considering how these velocity components may be physically linked.
Conceivably, multiple speeds could arise from different radii of the inner disc, with corresponding escape velocities, or reflect real variability over the duration of an observation. However, for continuum driving that would require (King 2010, equ. 8) a factor two change in accretion rate over a few hours in the 2001 observation, which may seem unlikely.
An interesting alternative, given the importance of obtaining a better understanding of the fate of UFOs as they move out into the surrounding medium, is that the lower velocity, lower ionisation component represents the collision of two parcels of gas with similar masses, where one has the primary wind speed and the other is more or less static until the collision. Once a similar wind mass has shocked, the speed is automatically v/2. An important constraint on this idea is that ‘similar mass’ requires a ‘similar column density’, suggesting the collision must occur at a relatively small radius for a radially expanding wind.
A re-analysis of soft X-ray spectra of the luminouus Seyfert galaxy PG1211+143 has confirmed that the high velocity outflow observed in highly ionised Fe K absorption lines is also evident in soft X-ray lines. In particular, the absorption velocity profile obtained by combining spectra near multiple resonance lines finds apparent outflow velocities (wrt the AGN) of 23730230 and 21910190 km s.
Spectral fitting with the XSTAR photoionisation code (Kallman et al.1986) confirms the existence of two separate absorption components, with similar apparent velocities but very different ionisation levels. The similarity of the low ionisation soft X-ray component velocity to the systemic velocity of PG1211+143 (24270 km s) raises the interesting possibility of absorption in local matter in line of sight to the AGN, with that local interpretation being supported by the narrow absorption feature persisting in the 2004 and 2007 data.
In contrast, the 22000 km s absorption is clearly linked to the high velocity wind in PG1211+143, comparison with previous analyses of EPIC data finding similar variability in both Fe K resonance line and continuum absorption, both being strongest in 2001. Intriguingly, broad band spectral modelling in Pounds and Reeves (2007, 2009) found outflow velocities of 0.07c and 0.14c, with the lower value being driven by the continuum curvature below 2 keV.
While the present soft X-ray analysis provides a more secure measure of the 0.07c flow component, the higher velocity is not detected, presumably due to that flow component being too highly ionised for significant soft X-ray opacity. Establishing a direct physical link between the two high velocity absorbers will require more extended observations, but the velocity ratio does offer the interesting possibility that the lower velocity matter is being entrained when the fast wind impacts on the ISM or slow moving ejecta. While the impact will shock, with likely loss of mechanical energy by Compton cooling, the momentum of the residual flow will be conserved, with a factor 2 velocity difference perhaps most readily detectable.
Although the presence of discrete ‘clumps’ of low ionisation matter has been proposed to explain rapid, spectrally neutral flux changes (eg. Brennenman et al. 2013), the present analysis would provide the first confirmation of a high radial velocity for low ionisation matter, with implications for both the wind acceleration and outflow energetics.
Coarse binning of the summed soft X-ray spectra together with XSPEC spectral fitting confirms the velocity-broadened soft X-ray emission features indicated in the earlier EPIC analyses, with resonance line and RRC fluxes being interpreted as scattering of the AGN X-ray continuum from a wide angle fast outflow, with an emission measure consistent with the wind extending to a relatively large radius.
Finally, a weak absorption feature at 5600-6100 km s may represent the first detection of a Warm Absorber in PG1211+143. The weakness of such soft X-ray features and the absence of UV absorption (Ganguly et al. 2013) may suggest the outflow in PG1211+143 remains relatively highly ionised even after the anticipated shocking with the ISM or previous ejecta (Zubovas and King 2012).
Substantially deeper soft X-ray observations of ultra fast outflows, such as that in PG1211+143, will have considerable potential in clarifying the ionisation and dynamical structure in powerful ionised winds in AGN, and their subsequent interaction with the ISM or slower moving ejecta.
The author is pleased to acknowledge a continuing and stimulating dialogue with Andrew King attempting to clarify and understand the properties of AGN winds. The results reported here are 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).
-  Arnaud K.A. 1996, ASP Conf. Series, 101, 17
-  Bachev R. et al. 2009, MNRAS, 399, 750
-  Bentz M.C., Peterson B.M., Pogge R.W., Vestergaard M. 2009, ApJ, 694, L166
-  Brennenman L.W., Risaliti G., Elvis M., Nardini E. 2013, MNRAS, 429, 2662
-  den Herder J.W. et al. 2001, A&A, 365, L7
-  Gofford J., Reeves J.N., Tombesi T., Braito V., Turner T.J., Miller L., Cappi M. 2013, MNRAS, 430, 60
-  Herenz P., Richter P, Charlton J.C., Masiero J.R. 2013, A&A, 550, A87
-  Kallman T., Liedahl D., Osterheld A., Goldstein W., Kahn S. 1996, ApJ, 465, 994
-  Kaspi S. Smith P.S., Netzer H., Maoz D., Jannuzi B.T., Giveon U. et al. 2000, ApJ, 533, 631
-  Kaspi S., Behar E. 2006, ApJ, 636, 674
-  King A.R., Pounds K.A. 2003, MNRAS, 345, 657
-  King A.R. 2003, ApJ, 596, L27
-  King A.R. 2005, ApJ, 635, 121
-  King A.R. 2010, MNRAS, 402, 1516
-  King A.R., Pounds K.A. 2013, MNRAS, accepted (arXiv 1310 3969)
-  Marziani P., Sulentic J.W., Dultzin-Hacyan D., Clavani M., Moles M. 1996, ApJS, 104, 37
-  Murphy E.M., Lockman F.J., Laor A., Elvis M. 1996, ApJS, 105, 369
-  Pounds K.A., Reeves J.N., King A.R., Page K.L., O’Brien P.T., Turner M.J.L. 2003, MNRAS, 345, 705
-  Pounds K.A., Page K.L. 2006, MNRAS, 360, 1123
-  Pounds K. A., Reeves J.N. 2007, MNRAS, 374, 823
-  Pounds K. A., Reeves J.N. 2009, MNRAS, 397, 249
-  Pounds K.A. and Vaughan S. 2011 MNRAS, 413, 1251
-  Pounds K.A. and Vaughan S. 2011a, MNRAS, 415, 2379
-  Pounds K.A. and King A.R. 2013, MNRAS, 433, 1369
-  Prochaska J.X., Weiner B., Chen H.-W., Mulchaey J., Cooksey K. 2011, ApJ, 740, 91
-  Reeves J.N., Done C., Pounds K.A., Tereshima Y., Hayashida K., Anabuki N., Uchino M., Turner M.J.L. 2008, MNRAS, 385, L108
-  Tombesi F., Cappi M., Reeves J.N., Palumbo G.C., Yaqoob T., Braito V., Dadina M. 2010, A&A, 521, A57
-  Tombesi F., Cappi M., Reeves J.N., Nemmen R.S., Braito V., Gaspari M., Reynolds C.S. 2013, MNRAS, 430, 1102
-  Verner D.A. and Ferland G.J. 1996, ApJS, 103, 467
-  Zubovas K., King A.R. 2012, ApJ, 745, L34
Appendix A Appendix
In Section 2 the composite velocity profile for the 5 principal resonance absorption of CVI, NVI, NVII, OVII and OVIII from the 2001 RGS data showed a significant broad high velocity feature which may be a blend of blue-shifted absorption at 22200 and 23700 km s, the latter being close to the systemic velocity for the redshift of PG1211+143. The existence of 2 separate absorption components was supported by the XSTAR modelling in Section 3, which finds a lower ionisation parameter for the near systemic velocity.
Figure 4 compares the same composite velocity profiles for the sum of the 2004 and 2007 RGS data where the the broad absorption near -22000 km s is much less evident, while the narrow component close to the systemic velocity remains, as does the absorption feature near -5600 km s (in the AGN rest frame). The lower panel of figure 4 shows the same profile for the combined 2001, 2004 and 2007 RGS data, again binned at 400 km s.
While it is difficult to assess such weak features, evidence for the -22000 km s feature arising from high velocity soft X-ray absorption in the highly ionised wind of PG1211+143 is strengthened by the its variability, being much weaker in the later observations (as were the continuum and resonance line absorption in the EPIC spectra). The consistent velocities and strengths of both the near-systemic absorption feature and that -5600 km s supports their reality. While the latter feature might be the first evidence for a warm absorber in PG1211+143 the velocity would be unusually high (Tombesi et al. 2013), while an intriguing alternative would be ionised gas associated with a pair of dwarf galaxies showing strong OVI absorption at cz= 0.0510 and 0.0645 Prochaska et al. (2011).
|Comp||velocity||width ()||flux (counts)|