Leveraging High Resolution Spectra to Understand Black Hole Spectra
AAstron. Nachr. / AN , No. 88, 789 – 795 (2017) /
DOI
This.is/not.aDOI
Leveraging High Resolution Spectra to Understand Black Hole Spectra
Michael A. Nowak ,(cid:63) Massachusetts Institute of Technology, Kavli Institute for Astrophysics, Cambridge, MA 02139, USAReceived 9 October 2016, accepted 1 December 2016Published online later
Key words accretion, accretion disks — stars: black holes — X-rays: binaries — X-rays: individual (Cygnus X-1)For the past 17 years, both
XMM-Newton and
Chandra have brought the powerful combination of high spatial and spectralresolution to the study of black hole systems. Each of these attributes requires special consideration— in comparisonto lower spatial resolution, CCD quality spectra— when modeling observations obtained by these spacecraft. A goodunderstanding of the high resolution spectra is in fact required to model properly lower resolution CCD spectra, with theReflection Grating Spectrometer (
RGS ) instrument on
XMM-Newton maintaining the highest “figure of merit” at soft X-ray energies for all missions flying or currently planned for the next decade. Thanks to its even higher spectral resolution,the use of
Chandra -High Energy Transmission Gratings (
HETG ), albeit with longer integration times, allows for one tobring further clarity to
RGS studies. A further promising route for continued studies is the combination of high spectralresolution at soft X-rays, via
RGS and/or
HETG , with contemporaneous broadband coverage extending to hard X-rays(e.g.,
NuSTAR or INTEGRAL spectra). Such studies offer special promise for answering fundamental questions aboutaccretion in black hole systems; however, they have received only moderate consideration to date. This may be due in partto the difficulty of analyzing high resolution spectra. In response, we must continue to develop software tools that makethe analysis of high resolution X-ray spectra more accessible to the wider astrophysics community. c (cid:13) Fundamental questions remain in the study of astrophysicalblack holes. What are the most robust techniques for mea-suring black hole spin? (See the reviews of Middleton 2016;Reynolds & Nowak 2003.) How does this spin relate to theformation of jets in these systems? (See Fender & Mu˜noz-Darias 2016.) What are the relationships between jets andoutflowing winds, and how much mass and accretion en-ergy is transported by the latter? (See Ponti et al. 2012.)What fraction of the hard X-rays can be explained by jetemission, and what fraction can be attributed to a corona?(See Markoff et al. 2015.) Does the disk recede as blackholes transit from high/soft states to low/hard states? (Seethe review by Done et al. 2007.) Answers to these questionshave been sought via multi-wavelength observations rang-ing from radio wavelengths through hard X-ray energies, asmany of the above cited phenomena manifest themselvesover broad energy bands.The soft X-ray, however, holds special importance for anumber of components. It is where the disk spectrum peaks.It is crucial for studies of wind properties. It dominates thebolometric luminosity in Galactic black hole “high states”.Arguably, the two most important instruments for study ofthe soft X-ray spectra of black hole systems have been the
XMM-Newton and
Chandra satellites. Both of these instru-ments share the properties of having unparalleled spatialand spectral resolution. The latter is achieved via the Reflec- (cid:63)
Corresponding author: e-mail: [email protected] tion Gratings Spectrometer (
RGS ; den Herder et al. 2001)of
XMM-Newton , and by the Low Energy TransmissionGratings (
LETG ; Brinkman et al. 2000) and High EnergyTransmission Gratings (
HETG ; Canizares et al. 2005) of
Chandra . The
HETG is comprised of the High Energy Grat-ings (
HEG ) and Medium Energy Gratings (
MEG ). Both thehigh spatial and spectral resolution properties of these in-struments pose unique challenges that are often overlookedin analyses of black hole systems. The high spectral resolu-tion properties especially have not been utilized as often asthey should or could be in multi-wavelength campaigns.We present examples below from both stellar mass andsupermassive black holes. We briefly discuss the synergybetween
RGS and
HETG high spectroscopic resolutionstudies. We then discuss the state of software for high reso-lution spectroscopic analysis. We end with considerations ofthe current use of high resolution spectra in multi-satellite,broadband X-ray campaigns.
As a first example, Fig. 1 shows a spectrum of the blackhole candidate (BHC) and X-ray binary, Cyg X-1. The ob-servation occurred at orbital phase 0 (superior conjunction),when the optical companion was directly between our lineof sight to the black hole. This particular observation wasquasi-simultaneous in all X-ray satellites flying at that time (April 2006), with the
Chandra - HETG and
Suzaku spectrapreviously being discussed in Nowak et al. (2011). There c (cid:13) a r X i v : . [ a s t r o - ph . H E ] D ec
90 Michael A. Nowak: Leveraging High Resolution Spectra − × − × − × − ν F ν ( e r g s c m − s − ) − χ Energy (keV)
XIS HETG − − ν F ν ( e r g s c m − s − ) − χ Energy (keV)
EPIC-PN
Fig. 1
Left: Comparison of quasi-simultaneous flux-corrected
Suzaku -XIS (blue diamonds) and
Chandra - HETG (greyhistogram) spectra of a hard state of Cyg X-1. (Times of dips, see below, have been excluded.) The model (red and orangehistograms, respectively) consists of a disk, powerlaw, and Fe K α emission line absorbed by both neutral and ionizedemission. Scattering of the soft X-ray photons out of the line of sight by a foreground dust halo is included in the HETG ,but not
Suzaku , fit. Residuals (histogram colors match their respective data colors) show the fit absent both the Fe K α emission and ionized absorption (Nowak et al. 2011). Right: the same spectra as on the left, now with the addition of quasi-simultaneous EPIC-PN spectra (orange stars), fit with an absorbed and dust-scattered (
Chandra - HETG and
EPIC-PN only)powerlaw (model not shown), with cross normalizations set so that the spectra match at 7 keV.are a number of issues that arose in analyzing these spec-tra. First is the effect of spatial resolution in
XMM-Newton and
Chandra . To a first approximation, the presence of fore-ground dust scatters soft X-rays out of the arcsec scale fieldof view of these instruments, but X-rays scatter back in (al-beit time-delayed) on the arcminute scale field of view ofinstruments such as
Suzaku . This effect is noticed on theright side of Fig. 1, where the “flux-corrected” spectra from
EPIC-PN and
HETG lie below that of
Suzaku . This is not (or at least not predominantly) a calibration effect, but is in-stead due to dust scattering that had to be included in themodeling on the left portion of Fig. 1. This effect is de-scribed in more detail for various X-ray instruments by Cor-rales et al. (2016).A recent
Chandra / Swift study of the black hole candi-date V404 Cyg by Heinz et al. (2016) shows that obser-vations of black holes with long-term time variations canprobe the location and composition of dust in the interstellarmedium (ISM).
XMM-Newton is perhaps the current bestinstrument to perform such observations going forward, asit has the proper combination of high (enough) spatial reso-lution with large effective area in the soft X-rays, yet is rela-tively free of pileup (see Davis 2001) compared to
Chandra .The second effect noticeable in Fig. 1 is the pres-ence of ionized absorption, clearly visible in the
Chandra - HETG spectra, yet completely unresolved but extremelystatistically significant in the
Suzaku and
EPIC-PN spec-tra (Nowak et al. 2011). It is important to note that boththe ionized absorption, as well as prominent dipping eventsthat occur as a function of orbital phase in Cyg X-1, areassociated with absorption by a powerful wind from the Ostar secondary in this system (see the discussions and refer-ences in Hanke et al. 2009; Miˇskoviˇcov´a et al. 2016). The spectra shown here occur outside of all detectable dippingevents at phase 0 . Such ionized absorption, only spectro-scopically resolvable by
RGS , LETG , and
HETG spectra,is a ubiquitous feature in spectra of Cyg X-1 (Miˇskoviˇcov´aet al. 2016). This is demonstrated in the right most figureof Fig. 2, where the relative amplitude of this componentis shown as a function of orbital phase in non-dip
Suzaku spectra of Cyg X-1.Fig. 2 shows color-color diagrams from the full
Suzaku and
Chandra observations associated with Fig. 1. The ex-tension to the lower left (excluded in the shown spectra) isdue to the interspersal of dense clumps passing in front ofour line of sight during the observations. The “tail” fromthe lower left to the lower right, however, is due to the “par-tial covering” nature of these clumps. In
Suzaku , this is alsoan artifact of the dust scattering halo, as we are seeing thetime-delayed scattering from foreground halos on arcminutescales, with the fitted covering fraction being consistentwith the fraction of flux scattered by dust (Nowak et al.2011). Such scales are resolved out by the
Chandra pointspread function (PSF); however, there remains an ≈ un-covered fraction. Although this may be due to the inner coreof the dust halo, another possibility is that it is due to partialcovering in time. That is, we might be seeing large dippingevents passing through our line of sight on time scales fasterthan the ≈ . s integration times of these Chandra observa-tions. Exploring this possibility, however, would require thelarger effective area and faster timing capabilities of
XMM .This latter experiment potentially has been conductedwith the recent (Summer 2016) observations of Cyg X-1 conducted by
XMM-Newton , which were designed (PI:P. Uttley) to conduct spectral-timing studies of the contin-uum, Fe band, and reflection spectra of Cyg X-1, as well as c (cid:13) stron. Nachr. / AN (2017) 791 . . (0.5 − − ( . − k e V ) / ( − k e V ) . . a r d C h a nd r a c o l o r : c oun t s ( . − k e V ) / c oun t s ( − k e V ) soft Chandra color: counts(0.5 − − . . ( . − k e V ) / ( − k e V ) . . − − . . . . Orbital Phase R e l a ti v e I on i ze d A b s o r p ti on Fig. 2
Left: Color-color diagram of the full
Suzaku observation of Cyg X-1, from Fig. 1, now including dipping events.The red line is a simple partial covering model that describes these events with a varying column, but fixed partial coveringfraction (Nowak et al. 2011). Middle: Color-color diagram derived from
Chandra - HETG data of Cyg X-1, includingdipping events. Lines show simple partial covering models as on the left, with covering fractions ranging from 20–98%and columns ranging over (0.6–40) × cm − . (See the discussions in Hanke et al. 2009.) Right: Relative degree ofionized absorption in Suzaku observations of Cyg X-1 hard states, compared to that found in the orbital phase 0 (superiorconjunction) observation of Fig. 1. All of these fits are for times without any dipping events .to perform high-resolution absorption spectroscopy of thesecondary wind. It is important to note here that
XMM-Newton - RGS is ideally suited for such studies as it has thehighest “figure of merit” for soft X-ray band spectroscopy,as shown in Fig. 3.
This would still be true below ≈ . keV,even if the Hitomi mission had not been lost.
The power ofthis
RGS capability can be seen in Fig. 3 where two (non-simultaneous, but very comparable in terms of flux, spectralshape, and integration time) observations of the same BHCare shown, highlighting
MEG and
RGS spectra of the oxy-gen absorption edge region. Among potential future studiesto be considered with
RGS are detailed spectral-temporalobservations of galactic black hole systems, especially overbinary orbits, to separate out absorption components localto the system from those due to the ISM.The advantage of galactic binary BHC studies, asidefrom spectral signal-to-noise, are the time scales involved.In Cyg X-1, thanks to high resolution spectroscopy from
RGS and
HETG , we know that both ionized absorptionand “partial covering” models are applicable. We can tracktheir changes over binary orbits, and thus have the strongpromise of separating out their effects from those associ-ated with the inner accretion flow, e.g., the relativisticallybroadened Fe K α line. Such effects are more difficult, butno less important, to disentangle with high resolution spec-troscopic observations of Active Galactic Nuclei (AGN).This is directly relevant to studies of relativistic features inAGN, where some researchers have suggested that what ismodeled as a broadened line is in fact due to features domi-nated by partial covering (Miller et al. 2009). For the case ofNGC 3783, Brenneman et al. (2011) showed the necessityof simultaneously fitting the warm absorption and broad linefeatures. Further work by Reynolds et al. (2012) showedhow the model parameter regions of Miller et al. (2009) andBrenneman et al. (2011) are related, with the latter being fa- vored in Markov Chain Monte Carlo (MCMC) analyses ofthe spectra. Also important in the work of Brenneman et al.(2011) and Reynolds et al. (2012) was the inclusion of hardX-ray spectra, which we discuss further below. XMM-Newton / Chandra
Synergies
The figure of merit presented in Fig 3 tells only part ofthe story; absolute spectral resolution is still an importantmetric. For the absorption edge of 4U 1957 +
11 shown inFig. 3, the higher effective area of the
RGS as compared tothe
MEG clearly greatly benefits our ability to model thesefeatures. Here, however, the edge and its associated absorp-tion lines (i.e., transitions of atomic oxygen; see Juett et al.2004), as well as the possible ionized features from eitherthe interstellar medium (see Yao & Wang 2005) or the localsystem (Nowak et al. 2008), are well-isolated. This is notalways the case, especially with absorption features in thesoft X-ray band covered by the
RGS .As an example in Fig. 4 we show a portion of an
RGS spectrum of the Seyfert galaxy PG1211 + EPIC-PN studies(Pounds et al. 2003) and these more recent
RGS studies,it has been argued that PG1211 +
143 exhibits blueshifted,ionized outflows moving at speeds in the galaxy rest frameof O (10%) the speed of light. The best evidence, however,is that there are multiple ionized components at a varietyof outflow velocities that are blended in the RGS spectra(Pounds et al. 2016a,b). Furthermore, basic line properties,such as line widths, are not resolved by the
RGS spectra.Cleanly separating ionized outflow components and resolv-ing their individual line widths requires the resolution of the
Chandra - HETG ; however, such observations require largeinvestments of observing time. c (cid:13)(cid:13)
Chandra - HETG ; however, such observations require largeinvestments of observing time. c (cid:13)(cid:13)
92 Michael A. Nowak: Leveraging High Resolution Spectra . . Energy (keV) R e l a ti v e ( A e ff · R ) / H ito m i LET G EPI C − P N HEG M E G RG S1+ RG S2
20 22 24 × − × − × − − Wavelength (Å) λ F λ ( e r g s c m − s − )
20 22 24 × − × − × − − Wavelength (Å) λ F λ ( e r g s c m − s − ) Fig. 3
Left: Figure of merit for emission and absorption line studies (square root of effective area times spectral resolu-tion) for various instruments, normalized to that for
EPIC-PN . The vertical dashed lines show the energies below which
Chandra - HEG (right) and
XMM-Newton - RGS (left) spectral resolution exceeds that achieved by the
Hitomi calorimeter.Middle/Right: Comparison of two comparable (in terms of flux, spectral shape, and exposure time)
HETG (middle) and
RGS (right) observations of the oxygen-edge region of the black hole candidate 4U 1957 +
11 (see Nowak et al. 2012). Thespectrum is a multi-color disk with peak temperature ≈ . keV absorbed by a neutral column of . × cm − . Thereis also evidence for ionized absorption by the warms phase of the interstellar medium, as well as by material local to thesystem. F e XXV N V II − − × − × − ν F ν ( e r g s c m − s − ) − χ Energy (keV) F e K ? S i X I V S i X III M g X II ? M g X I ? N e X N e I X ? O V III F e XV II ?? O V II O V III F e XV III O V II ?
15 20 25 30 − × − × − c oun t s s − Å − c m − observed wavelength (Å)rgs 2014 Fig. 4
Top:
XMM-Newton - RGS spectra of the Seyfertgalaxy PG1211 +
143 (figure from Pounds et al. 2016b; seealso Pounds et al. 2016a), fit with an absorbed continuumand several ionized outflow components. Bottom:
Chandra - HETG spectrum of PG1211 +
143 (Danehkar et al. 2016,in prep.). The figure is labeled with possible lines, foundfrom a blind search, consistent with an outflowing wind at asingle (galaxy restframe) blueshift. Question marks are po-tential lines also found in the blind search that have not yetbeen identified with absorption or emission components. To this end, we have obtained an
HETG observation(PI: Julia Lee) of PG1211 +
143 for a total integration timeof 450 ks. Fig. 4 shows the binned, combined
HETG spec-tra, with “blind search” line fits (Danehkar et al. 2016, inprep.) derived from the unbinned spectra. This figure fur-ther shows possible line identifications for a set of featuresat a common blueshift of . c , consistent with one of theoutflows described by Pounds et al. (2016a). The full anal-ysis will be described elsewhere by Danehkar et al. (2016);however, we note that the analysis heavily draws upon thehigher effective area, albeit lower resolution, RGS analysis.It likely would have been impossible to obtain the
HETG observations for this integration time, or have found thesefeatures in even a blind search in shorter
HETG obser-vations, without first having the preliminary studies with
XMM-Newton - RGS .A further example of this synergy between
XMM-Newton - RGS and
Chandra - HETG is shown in Fig. 5,where we show a simulation of an upcoming 500 ks
HETG observation of NGC 1313 X-1. Analysis of archival
RGS observations suggests ionized emission at systemic veloc-ities, as well as ionized absorption at both systemic andblueshifted outflow ( v ≈ . c ) velocities (Pinto et al.2016). These components, however, are unresolved by the RGS , but would have taken significant integration timesto discover with
HETG observations alone. Based uponthe initial
RGS studies, a commitment was made to use500 ks (out of an annual allocation of 700 ks to
HETG
PIClaude Canizares) of
HETG
Guaranteed Time Observa-tions (GTO). The goals of these observations will be to re-solve line widths and to obtain more precise outflow ve-locity measurements. Comparing the simulations shown inFig. 5 to Fig. 3, we see that much of the
HETG gains willcome from the 1–2 keV region where its figure of merit ex-ceeds that of
RGS . We see again, however, the utility ofhaving first used the higher effective area
RGS to identifya high resolution spectroscopy target of interest. Further-more, the
RGS observations allowed us to determine a plau- c (cid:13) stron. Nachr. / AN (2017) 793 − × − × − × − k e V P ho t on s c m − s − k e V − − χ Energy (keV)
Fig. 5
Simulated 500 ksec
Chandra - HETG spectrum ofthe ULX NGC 1313 X-1, based upon ionized absorptionand emission line fits of
RGS spectra presented by Pintoet al. (2016) (see their Fig. 1). The model fit correspondsto a “blind line search” implemented in
ISIS (Houck &Denicola 2000) (see § HETG integration time for follow-up study.
A difficulty in the analysis of high resolution X-ray spectrais the complexity of models required to describe the data.Above ≈ keV, e.g., the Fe region, where spectral lines arerelatively few and widely spaced, simple direct modeling ofindividual line features may be feasible. In the soft X-raybands covered by the RGS , LETG , and
MEG , lines canbe numerous and closely spaced, complicating the analysis.Few codes exist for describing emission and absorption bythe photoionized plasmas relevant to the study of black holesystems, among them are
CLOUDY , (Ferland et al. 1998,2013),
XSTAR (Kallman & Bautista 2001), and the mod-els included in the
SPEX analysis package (Kaastra et al.1996). These codes take a global approach to descriptionof the high resolution spectra, modeling multiple individuallines and blends simultaneously. As such, however, search-ing parameter space can be slow and unintuitive, and thecodes and code results can be difficult to interpret. E.g., forthe lines that comprise the best fit, identification of individ-ual components, and obtaining access to the relevant atomicdata that led to those components, can be difficult. Creat-ing custom interfaces to these codes can be cumbersome,especially in cases where the source code is not publiclyavailable (e.g.,
SPEX ). These issues can lead to a “poten-tial barrier”, discouraging the use of high resolution X-rayspectra.To further reduce the barrier to entry to the study ofhigh resolution X-ray spectra, a number of researchers havebeen creating more user-friendly tools for both fitting and interpreting such models. Graphical user interfaces to theatomic databases for piecewise fits to restricted wavelengthregions are being explored. (See the individual talks by T.Kallman and R. Smith at the 2016 meeting of the Interna-tional Astronomical Consortium for High Energy Calibra-tion[IACHEC]). XSTARDB is a code suite for use within
ISIS that provides easier access to the
XSTAR atomicdata (e.g., individual line identifications, line searches basedon transition strengths, etc.), that also manages and runs
XSTAR spectral fits.In the example of PG1211 +
143 above, we have em-ployed an
ISIS script suite that we are developing thatallows for more phenomenological analysis of high reso-lution data. (Preliminary code versions are available as partof the
ISIS scripts maintained and developed at RemeisObservatory .) The scripts allow for modeling of individ-ual lines using a choice of standard profiles (e.g., gaussianor Voigt profiles, with parameters input in energy or wave-length, with or with explicit redshift parameters). Lines areadded to the model parameter file either in wavelength orenergy order, with users being able to name the line profile(e.g., a redshifted gaussian added as a description of FeK α could be named zg FeKa ). For Fig. 4, the lines wereplaced in a multiplicative model (to allow them to describeeither emission or absorption lines, with the latter never for-mally describing negative counts), with their names enu-merating the statistical order in which they were added ina blind line search. (E.g., the most significant line, likelyfrom Ne X , was called zg 0 , while the seventh most signifi-cant line, likely from Ne IX , was called zg 6 .)We plan to expand this code to allow for the easy addi-tion of multiple levels of functional ties, such as tying red-shifts across profiles expected to come from similar tem-perature/ionization states, while simultaneously tying linestrengths within a line series. The scripts will include proce-dures for applying blind line searches to the data. The goalis to allow a preliminary, straightforward phenomenologicaldescription of the spectra as a prior step before embarkingupon more physical descriptions with, for example, com-plex and slow running photoionization codes. As discussed above, high resolution spectra provide uniqueinformation about black hole systems. Looking forward tofuture observations with
XMM-Newton - RGS and
Chan-dra - HETG , one important area that we must consider isbroadband, multi-satellite observations. For the exampleof NGC 3783, as shown by Brenneman et al. (2011) andReynolds et al. (2012), both high resolution, soft X-rayspectra and broadband (in that case,
Suzaku ; going forwardin the future, most likely utilizing
NuSTAR ) are neces-sary to understand relativistic features. Indeed, Garc´ıa et al. web.mit.edu/iachec/meetings/2016/ c (cid:13)(cid:13)
NuSTAR ) are neces-sary to understand relativistic features. Indeed, Garc´ıa et al. web.mit.edu/iachec/meetings/2016/ c (cid:13)(cid:13)
94 Michael A. Nowak: Leveraging High Resolution Spectra (2015) has recently shown that under ideal circumstance,given knowledge of the underlying continuum and qual-ity observations, the soft X-ray reflection spectrum mea-sures the cutoff at high energies. In practice, however, con-cerns about ionized absorption, as discussed in Nowak et al.(2011) for the case of Cyg X-1, must first be addressed with,e.g, high spectral resolution observations. Clearly, however,just as their is a synergy between CCD and gratings qual-ity spectra ( § RGS spectra and higher resolution
HETG spectra ( § XMM-Newton and
NuSTAR ob-servations, an examination of the literature to date showsthat of the approximately 100 papers describing
XMM-Newton / NuSTAR observations, only 11 discuss
RGS spec-tra in any capacity. Of the dozen accepted joint
Chan-dra / NuSTAR proposals, only three utilize the
HETG . (Oneof these campaigns is a joint observation of NGC 3783; PIL. Brenneman.)When sufficient telemetry exists, and instrumental con-straints aren’t violated,
RGS spectra will always be ob-tained with an
XMM-Newton observation. This does notnecessarily guarantee that the
RGS will have sufficientsignal-to-noise for use, but in many cases it does. How dowe encourage users to analyze these data? We note that thework of Pinto et al. (2016) was derived from archival obser-vations from proposals that were initially designed to obtainCCD-quality spectra. How do we encourage users to explic-itly design programs around the use of
RGS spectra?For
Chandra proposals, gratings observations (typicallydominated by the use of
HETG ) comprise only ≈ ofthe accepted program. This is not because gratings propos-als are accepted at a lower rate than non-gratings proposals,rather it is because they represent only 15% of the submit-ted proposals. How do we encourage users to apply for moregratings observations?Part of the issue is undoubtedly the complexity of thespectra, and the lack of an “easy entry” to the high resolu-tion spectroscopy software ( § RGS below ≈ . keV and HETG between ≈ . –2 keV are the premiere instruments forspectroscopic studies. A relaunch of a satellite comparableto Hitomi would not alter the situation for
RGS . High res-olution spectroscopy has the capability of providing mea-sures of density, temperatures, and velocities, and is in factthe prime scientific mission of the proposed
Athena and
X-ray Surveyor missions. Continued studies with
RGS , LETG , and
HETG must pave the way for these future mis-sions.
Acknowledgements.
Michael Nowak gratefully acknowledgesfunding support from the National Aeronautics and Space Admin-istration through the Smithsonian Astrophysical Observatory con-tract SV3-73016 to MIT for support of the
Chandra
X-ray Center,which is operated by the Smithsonian Astrophysical Observatoryfor and on behalf of the National Aeronautics Space Administra-tion under contract NAS8-03060. He further acknowledges sup-port by NASA Grant NNX12AE37G. He would like to thank V.Grinberg, D. Huenemoerder, M. Middleton, and J. Wilms for use-ful conversations. c (cid:13) stron. Nachr. / AN (2017) 795 References
Brenneman, L. W., Reynolds, C. S., Nowak, M. A., et al. 2011,ApJ, 736, 103Brinkman, B. C., Gunsing, T., Kaastra, J. S., et al. 2000, in Societyof Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeries, ed. J. E. Truemper & B. Aschenbach, Vol. 4012, Societyof Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeries, 81Canizares, C. R., Davis, J. E., Dewey, D., et al. 2005, PASP, 117,1144Corrales, L. R., Garc´ıa, J., Wilms, J., & Baganoff, F. 2016, MN-RAS, 458, 1345Davis, J. E., 2001, ApJ, 562, 575den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001, A&A,365, L7Done, C., Gierli´nski, M., & Kubota, A. 2007, ARA&A, 15, 1Fender, R., & Mu˜noz-Darias, T. 2016, in Lecture Notes in Physics,Berlin Springer Verlag, ed. F. Haardt, V. Gorini, U. Moschella,A. Treves, M. Colpi, Vol. 905, Lecture Notes in Physics, BerlinSpringer Verlag, 65Ferland, G. J., Korista, K. T., Verner, D. A., et al. 1998, PASP, 110,761Ferland, G. J., Porter, R. L., van Hoof, P. A. M., et al. 2013, RevistaMexicana de Astronom´ıa y Astrof´ısica, 49, 137Garc´ıa, J. A., Dauser, T., Steiner, J. F., et al. 2015, ApJ, 808, L37Hanke, M., Wilms, J., Nowak, M. A., et al. 2009, ApJ, 690, 330Heinz, S., Corrales, L., Smith, R., et al. 2016, ApJ, 825, 15Houck, J. C., & Denicola, L. A. 2000, in ASP Conf. Ser. 216:Astronomical Data Analysis Software and Systems IX, Vol. 9,591Juett, A. M., Schulz, N. S., & Chakrabarty, D. 2004, ApJ, 612, 308Kaastra, J. S., Mewe, R., & Nieuwenhuijzen, H. 1996, in UV andX-ray Spectroscopy of Astrophysical and Laboratory Plasmas,ed. K. Yamashita, T. Watanabe, 411Kallman, T., & Bautista, M. 2001, ApJS, 133, 221Markoff, S., Nowak, M. A., Gallo, E., et al. 2015, ApJ, 812, L25Middleton, M., 2016, in Astrophysics of Black Holes: From Fun-damental Aspects to Latest Developments, ed. C. Bambi, Vol.440, Astrophysics and Space Science Library, 99Miller, L., Turner, T. J., & Reeves, J. N. 2009, MNRAS, 399, L69Miˇskoviˇcov´a, I., Hell, N., Hanke, M., et al. 2016, A&A, 590, A114Nowak, M. A., Hanke, M., Trowbridge, S. N., et al. 2011, ApJ,728, 13Nowak, M. A., Juett, A., Homan, J., et al. 2008, ApJ, 689, 1199Nowak, M. A., Wilms, J., Pottschmidt, K., et al. 2012, ApJ, 744,107Pinto, C., Middleton, M. J., & Fabian, A. C. 2016, Nature, 533, 64Ponti, G., Fender, R. P., Begelman, M. C., et al. 2012, MNRAS,L417Pounds, K., Lobban, A., Reeves, J., & Vaughan, S. 2016a, MN-RAS, 457, 2951Pounds, K. A., Lobban, A., Reeves, J. N., Vaughan, S., & Costa,M. 2016b, MNRAS, 459, 4389Pounds, K. A., Reeves, J. N., King, A. R., O’Brien, P. T., & Turner,M. J. L. 2003, MNRAS, 345, 705Reynolds, C. S., Brenneman, L. W., Lohfink, A. M., et al. 2012,ApJ, 755, 88Reynolds, C. S., & Nowak, M. A. 2003, Physics Reports, 377, 389Yao, Y., & Wang, Q. D. 2005, ApJ, 624, 751 c (cid:13)(cid:13)