Characterizing the quiescent X-ray variability of the black hole low mass x-ray binary V404 Cyg

0.3 MB·

by  Federico Bernardini

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed17January2014 (MNLATEXstylefilev2.2)
Characterizing the quiescent X-ray variability of the black
hole low mass x-ray binary V404 Cyg
F. Bernardini1,2⋆, E. M. Cackett1
1 Department of Physics & Astronomy, Wayne State University, 666 W. Hancock St., Detroit, MI 48201, USA
2 INAF, Osservatorio Astronomico di Capodimonte, Salita Moiariello 16, 80131 Napoli, Italia
4
1
0
2

n
a
J ABSTRACT

6
1 We conducted the first long-term (75 days) X-ray monitoring of the black hole
low mass X-ray binary V404 Cyg, with the goal of understanding and characterizing

] itsvariabilityduringquiescence.TheX-raylightcurveofV404showsseveralflareson
E
timescales of hours with a count rate change of a factor of about 5-8. The root mean
H squarevariabilityisFvar =57.0±3.2%.Thefirstorderstructurefunctionisconsistent
. with both a power spectrum of index -1 (flicker noise), or with a power spectrum of
h
index 0 (white noise), implying that the light curve is variable on timescales from
p
daysto months. The X-rayspectrum is wellfitted by a powerlawwith spectralindex
-
o Γ = 2.10−2.35, and we found that the spectral shape remains roughly constant as
r the flux changes. A constant spectral shape with respect to a change in the X-ray
t
s flux may favour a scenario in which the X-ray emission is dominated by synchrotron
a radiation produced in a jet.
[

Key words: stars: black holes – X-rays: binaries – X-rays: individual: V404 Cyg
1
v
3
1
1
1 INTRODUCTION escent state of low level of emission, like in the case of the
4
family of models known as advection dominated accretion
. In low mass X-ray binaries (LMXBs) a compact object, a
1 flow(ADAF)solutions, mainly developed byNarayan & Yi
black hole (BH) or a neutron star (NS), is coupled with a
0 (1994, 1995a,b), but see also Ichimaru (1977) for some ear-
4 late-typecompanion star, with mass lower than that of the lieridea.Inthisscenario,ifthemassaccretionratefromthe
1 Sun.SomeLMXBsarepersistent,constantly accreting mat-
companion drops below a critical value, and consequently
v: ter from the companion through a standard accretion disk the X-ray luminosity becomes very low, . 10−3LEdd, the
(Shakura& Sunyaev1973).Consequently,theyproducelu-
i accretiondiskbecomesunstableandlikelyundergoesadras-
X minous X-ray emission (up to 1037−38 erg/s) close to the
tic structural change. The inner part of the disk, which in
r Eddingtonlimit,LEdd.OtherLMXBsareinsteadtransient. outburst extends close to the compact object, can evapo-
a Theyalternatelongperiodsoflowluminosityemission(qui-
rate, resulting in a disk truncated further out where the
escence), that can last years or decades, to short and spo-
structureremainsthatofastandarddisk(e.g.geometrically
radic periods of intense activity (outbursts), lasting weeks-
thin and optically thick). The evaporated gas instead as-
months. During quiescence the X-ray luminosity is of the
sumesaspherical shapearound thecompact object andro-
order of 1031−33 erg/s. However, during outburst theX-ray
tatesmuchslowerthantheKeplerianvelocity.Theaccretion
luminositydramaticallyincreasesbyseveralordersofmagni-
timescale becomes more rapid than the cooling timescale,
tude,andthesources becomesimilar topersistent LMXBs.
and thus the accretion flow is radiatively inefficient. The
The quiescence to outburst cycle is broadly explained by
energy stored in the form of heat (entropy) in the gas is
thediskinstabilitymodel(DIM,e.g.Cannizzo1993;Lasota
not radiated away (unlike what happens in a standard ac-
2001),wheretheaccretingmatterisaccumulatedintheac-
cretion disk) but is transported inward (advection). In the
cretion disk during quiescence and then is suddenly trans-
case of a BH the energy is finally lost in the event hori-
ferred to thecompact object duringoutburst.On theother
zon, and in the case of a NS it is transferred to the stel-
hand,the origin of theX-ray emission from LMXBs during
lar surface. However, in order to explain the extremely low
quiescence is still debated.
level of X-ray emission from some faint LMXBs, the ac-
Severalmodels havebeen proposed toexplain thisqui-
tion of an extra component in carrying away matter and
radiation from the compact object is mandatory. The na-
⋆ E-mail:[email protected] ture of this extra component is still a matter of investiga-
(cid:13)c 0000RAS
2 Bernardini & Cackett
tionanddebate,andlikelydependsonthecharacteristicsof were made using a single X-ray observations lasting 16 and
thespecific accreting object. In thecase of NSs,theaccret- 60ksonlyrespectively).However,uptonow,regularX-ray
ing matter could be pushed away by the centrifugal force monitoringofaquiescentBHoveralongbaseline(months)
of the rotating magnetosphere (see, e.g. Asai et al. 1998; hasnotbeenperformed.But,meaningfulvariabilityonsuch
Menou et al. 1999), or, alternatively, a ‘dead disk’ could timescales is seen in persistent sources and those in out-
form, inhibiting but not completely preventing matter ac- burst.Forinstance,duetospectralstatetransitions(see,e.g.
cretingontotheNSsurface(D’Angelo & Spruit2010,2012). Remillard & McClintock 2006), or other mechanisms such
Otherpossibilitiesinvoketheformationofgasoutflows(that as precessing or warped disks (Charles et al. 2008). Moti-
the ADAF itself is inclined to produce, as first suggested vated by this, and with the goal to shed some light on the
byNarayan & Yi1994,1995a),orastrongoutflowingwind quiescentstateofBHLMXBs,weundertookauniquestudy
(ADIOS;Blandford & Begelman 1999) or a (radiatively in- monitoring the BH LMXB V404 Cyg with the Swift satel-
efficient) jet (e.g. Fender, Gallo & Jonker 2003), where the liteforapproximatelytwomonths,onalternatingdays.This
majorityoftheaccretionpowerisreleasedwiththeoutgoing allowed us, for the first time ever for a BH LMXB in qui-
matter. Moreover, hybrid jet/ADAFmodels havebeen also escence, to characterize the properties of this variability on
suggested (see, e.g. Yuan,Cui & Narayan 2005). See Sect. timescales of daystomonths.Wepresent thesourcein Sec.
3.6 of Narayan & McClintock (2008) for a clarifying review 2, the Swift observations and data reduction in Sec. 3, the
aboutADAFplusoutflowandjetmodels.Allthelattersce- timingandspectralresultsinSec.4thatwelaterdiscussin
narios apply to both BH and NS LMXBs. Sec. 5 (making also a comparison with the case of Cen X-4
The X-ray spectrum of quiescent LMXBs shows no inSec.5.1).Finally,inSec.6,wereportabriefsummaryof
sign of the accretion disk (though the disk is X-ray bright themain conclusions.
during outburst and produces a bright thermal component
then). For BHs, in particular, only a non thermal compo-
nent is present. Its shape, between 0.5 and 10 keV, is
∼ ∼
apowerlaw anditsphysicalorigin isstill strongly debated.
2 V404 CYG
Bildsten & Rutledge(2000) proposed that thequiescent X-
rayemissionofBHLMXBscouldbegeneratedbythestellar V404 Cyg, also known as GS 2023+338, was discovered
corona of the rapid rotating companion. However, Lasota in 1989 by the Ginga X-ray satellite during an outburst
(2000) argued that the expected coronal emission would (Makino 1989). While its distance was first believed to
have likely been far below that observed in quiescent sys- be about 3.5 kpc, more recent parallax observations, ob-
tems, as later unambiguously demonstrated by Kong et al. tainedwithverylongbaselineinterferometryatradiowave-
(2002). Consequently, the X-ray emission of BH LMXBs, lengths, unambiguously put V404 Cyg at 2.40 0.14 kpc
±
and its power law spectral shape, must be intrinsic to the (Miller-Jones et al.2009).Itisthemostluminousofthequi-
accretion flow. escentBHLMXBs,withaX-rayluminosityofabout7 1032
ThemajorityofLMXBscontainingBHs,spendthebulk (d/2.4 kpc)2 erg/s (Miller-Jones et al. 2009). It sh×ows a
of their time in the quiescent state, this is also true for transient behaviour with three confirmed outbursts: 1938,
the nearby population of active galactic nuclei (AGN, see 1956,and1989(see,e.g.Richter1989;Z˙ycki,Done& Smith
e.g. Soria et al. 2006; Ho 2009; Gallo et al. 2010; Pellegrini 1999,thelatterforadetailedanalysisofthe1989outburst).
2010), or for the super massive BH in the center of our Thecompanionisa 1M⊙,K0( 1)III–Vstar,almostfill-
galaxy, Sgr A∗. However, also due to the low level of X- ingitsRochelobe(W∼agner et al.±1992;Casares et al.1993).
ray emission during quiescence, our knowledge of this state Casares, Charles & Naylor (1992) and Casares & Charles
is still relatively poor. The quiescence of several LMXBs, (1994), using spectroscopic optical data, measured an or-
on the contrary to what ”quiescence” might suggest, have bital modulation of 6.473 0.001 d, and unambiguously
±
been proven to be highly variable, and the origin of this derived the lower limit of the mass of the compact object
variability is still unclear. For example, theNS LMXB Cen 6.08 0.06M⊙, that implies the presence of a black hole.
±
X-4hasshownX-rayvariabilityonawiderangeoftimescale By usinginfrared (IR)photometry,combined with spectro-
fromhundredsofsecondsuptoyears(Campana et al.1997; scopic data, Shahbazet al. (1994) constrained the value of
Rutledgeet al. 2001; Campana et al. 2004; Cackett et al. the mass ratio q = 16.7 (Mx/Mc) and the inclination i =
2010,2013;Bernardini et al.2013).ThequiescentX-rayand 56 4degrees.Thatledtoanaccuratemeasureoftheblack
±
theoptical/UV emission of Cen X-4are strongly correlated hole mass which is 12 2M⊙. Later, Shahbazet al. (1996)
±
down to a timescale of a few hundreds seconds. A residual by using IR spectroscopy derived a 15% upper limit to the
low level of accretion, down to the NS surface, that trig- contamination to the IR light by the accretion disk. This
gers optical/UV reprocessing on the companion star, must reduced the mass derived with IR photometry to 10M⊙,
be the origin of this variability (Bernardini et al. 2013). still confirming that the compact object is a stellar mass
Aql X-1, another NS LMXB, also showed X-ray variabil- black hole. During quiescence the source 0.5–10 keV X-ray
ity on timescales from weeks to year (Rutledge et al. 2002; spectrum is well fitted, as usual for quiescent BH LMXB,
Cackett et al. 2011;Coti Zelati et al. 2013). by a power law with spectral index Γ 2 multiplied by
LMXBs containing BHs also seem to show some X- an interstellar hydrogen column density∼of 0.8 1.2 1022
ray variability in quiescence on timescales of years (e.g. cm−2 (Bradley et al. 2007; Plotkin, Gallo & J−onker×2013;
Reynoldset al. 2013, and more references therein). More- Reynoldset al. 2013). As it is the most luminous quiescent
over,V404Cygalsoshowedafactorof10–20X-rayvariabil- BH, and most of its main characteristics have been con-
ityinlessthanaday,andafactorof 2variabilityin 30 strained,V404Cygisthebestcandidatetoperformastudy
∼ ∼
minutes (Wagner et al. 1994; Hyneset al. 2004, both work focused on X-rayvariability during quiescence.
(cid:13)c 0000RAS,MNRAS000,000–000
Quiescent X-ray variability in V404 Cyg 3
3 OBSERVATIONS AND DATA REDUCTION 3.2 UVOT data reduction
The Swift satellite is a multi-wavelength instrument, with We performed the UVOT data reduction for each point-
three telescopes on board that allow simultaneous cover- ingbyapplyingthefourstepprocedure:1)uvotbadpix,2)
ageoftheoptical/ultravioletband(UltravioletOpticalTele- uvotexpmap,3)uvotimsum,4)uvotdetect.Wedescribe
scope,UVOTRoming et al.2005;Breeveld et al.2010),the itinmoredetailinthefollowing.Fromtherawimagefile,we
soft X-ray band (X-ray Telescope, XRT), and the hard X- generatedthebadpixelmapwithuvotbadpix.Then,from
rayband(Burst Alert Telescope, BATGehrels et al. 2004). the level II image fits file, we produced the exposure maps
BecauseofthelowX-rayfluxofV404Cyg,thesourceisun- by using the previously generated bad pixel map and the
detectedbyBATand,consequently,we onlyanalysed XRT uat.fits file for the attitude file. With the uvotimsum tool,
data, all taken in photon counting (PC) mode. Concerning we summed all the multiple image extensions contained in
theoptical/UVanalysisweexploreddataintheUVW1filter thelevelIIimagefitsfileofeachpointing(theUVOTpoint-
(2600 ˚A), the only one used. Swift has observed the source ingsarealwayscomposedofanumberofmultiplesnapshots
for a total of 34 times with an exposure time of 5 ks each. of variable duration). This procedure generates the total
33 observations were performed, approximately every two pointing image file. With the uvotimsum tool we did the
days, as part of the main observational campaign, between samefortheexposuremap,constructingthetotalexposure
03 July 2012 and 16 September 2012 (obsid: 00031403002– mapofeachpointing.Finally,weruntheuvotdetecttool
00031403034). Moreover, an archival observation was per- and we ascertained that V404 Cyg was not detected in any
formed on 26 April 2009 (obsid: 00031403001). pointing in the UVW1band.
In order to maximize the signal-to-noise ratio we
summedtogetherall availableUVOTobservationswith the
UVW1 filter. Then, we followed the above procedure, with
theonlydifferencethatatstepnumber3weaddedtogether
3.1 XRT data reduction
images and exposure maps of different pointings. We de-
We used the web tool developed at the UK Swift Science finedthesourceregionwithacircleof5arcsecsize,centered
DataCentertoproducethe0.3 10 keVX-raylight curve, at the best position for V404 Cyg, and the background re-
−
with the binning time corresponding to each pointing and gion with acircle of 10 arcsec in aclose byregion freefrom
eachsnapshotduration(theXRTpointingsarealwayscom- othersourcecontamination.Asalaststep,usingtheuvot-
posed of a number of multiple snapshots of variable dura- sourcetool,weget a3σ upperlimit fortheaveragesource
tion). This tool is available on the University of Leicester flux density in the UVW1 band of F < 1.62 10−18 erg
λ
website1 (see Evans et al. 2009, for a detailed description s−1 cm−2 ˚A−1. This is consistent with the flu×x measured
of the light curve extraction tool). The 0.3 10 keV aver- by Hyneset al. (2009) in the F250 HubbleSpace Telescope
−
agesourcecountrateis0.014 0.008c/s,whiletheaverage band (0.52 0.20 10−18 ergcm−2s−1) which is close in
± ± ×
backgroundcountrateis0.002 0.001 c/s.Alluncertainties wavelength tothe SwiftUVW1band.
±
are hereafter at the1σ confidence level.
Due to the source faintness, in order to create spec-
tra with the maximum level of signal to noise (S/N), we
4 ANALYSIS AND RESULTS
summedtogetherspectrafrom differentpointingswithsim-
ilar count rates (see Sect. 4.3 for details about the selected 4.1 X-ray light curve
pointings),usingthefollowingprocedure.First,usingHEA-
soft 6.12, we generated the cleaned event file for each of We show the XRT 0.3–10 keV background subtracted light
the 34 pointings with the xrtpipeline command, together curve on the pointing timescale in Fig. 1. For plotting pur-
poses we do not include the first (archival) pointing which
with therespective exposure maps, manually providing the
does not belong to the main observational campaign (it is
bestsourceposition.Then,wesummedtheindividualevent
files of interest by using XSelect and, one by one, the in- at t = −1164.25 d, and has a count rate of 0.015±0.002
c/s). The light curve is highly variable and shows dra-
dividualexposure maps by using thecommand sumima in
theXImagepackage.Weusedasourceextractionregion of matic count rate changes. The largest variability that is
significant at greater than the 3σ level are changes of a
20 pixels size, centered at the best source position, and a
factor of about 4–5. There are some larger changes that
background extraction region of 60 pixels size, in a nearby
are significant at approximately 2σ level, like between 26.0
clean region of the sky. Finally, we extracted the summed
spectrafromthesummedeventfileswithXSelectandgen- d (0.039 ± 0.008 c/s) and 34.0 d (0.0021 ± 0.0009 c/s),
where we measure a decrease of a factor of 18.4 8.8.
erated the summed ancillary response file (ARF) using the ±
In order to properly quantify the intensity of this vari-
summed exposuremaps. Before fitting,all thespectra have
ability we also calculated the root mean square variability
beenrebinnedinordertohaveaminimumnumberofcounts
greater or equal to 20. This allows a correct use of the χ2 (see Vaughan et al. 2005, for more details), finding Fvar =
statistic.AllspectralfitswereperformedwithXSpec12.7.1 57±3%.ForcomparisonofthequiescentX-rayvariabilityof
(Arnaud 1996) and the latest calibration files available in V404Cygondifferenttimescales,wealsocalculatedFvar for
an archival XMM-Newton observation (obsid 0304000201)
October 2012.
presented in Bradley et al. (2007). Using the background-
subtracted EPIC-pn lightcurve that is 34 ks long, we mea-
sure FXMM = 94 2% when using a binning time of 1000
var ±
s.
1 http://www.swift.ac.uk/ We note that many of the Swift light curve points lie
(cid:13)c 0000RAS,MNRAS000,000–000
4 Bernardini & Cackett
wellaboveorbelow3σfromtheaveragecountrateof0.014 the measurement noise, and it has an amplitude of twice
0.008 (see Fig. 1). Moreover, significant rapid changes, b±e- the variance of the measurement noise (2σ2 ). The sec-
noise
tween approximately the average level and higher or lower ond plateau usually occurs at a timescale greater than the
count rate, frequently happen on a two days timescale. See inspected physical process, and it has an amplitude twice
e.g. t 18 d, t 20 d, and t 22 d where the count rates is thevarianceofthefluctuations(2σ2 ).Thetimein-
∼ ∼ ∼ fluctuation
0.009 0.002 c/s,0.032 0.003 c/s,and0.011 0.002 c/sre- tervalwhere thecurveconnecting thetwo plateaus extends
± ± ±
spectively.Thesefindingssuggestthatthesourceisvariable correspondstothetimescalewherecorrelated variability is
bothonatwo-daytimescale(theminimumdistancebetween detected. The slope of this curve is linked to the slope of
two pointings) and up to a seventy five days timescale (the the power spectrum. For example, the first order structure
maximum distance between two pointings). function of a power spectrum with P(f) f−2 (red noise)
We also produced the XRT 0.3 10 keV background will be V(τ) τ1 (see, e.g. Hughes, Aller∝& Aller 1992, for
− ∝
subtracted light curve on the snapshot timescale (average moredetailsandexamples)intheregionconnectingthetwo
exposure time 560 s), which we show in the bottom plateaus.WeshowthefirstorderstructurefunctionofV404
∼
panel of Fig. 1. Once again the light curve is highly vari- Cyg in Fig. 2.
able, and thanks to the better time resolution, we clearly We fit the structure function with a power law of the
seethepresenceofmultipleflares.TheX-raycountratein- formy(t)=cxγ,findingγ = 0.1 0.1,c=1.6 0.4 10−4
creasesanddecreases, multipletimes,byafactorof 5 8 with χ2 1.1 for 12 dof. H−ence±, the slope o±f the×power
∼ − ν ∼
on timescale of hours. See e.g. around 19 days where the law is consistent to be zero within 1σ, and thus V(τ)
count rate goes from 0.075 0.0011 c/s to 0.0140 0.004 τ0. By looking at Fig. 2, we see tha∼t V(τ) has an intensit∝y
± ±
c/s, implying a change of a factor 5.4 1.5, or around 54 close to the value expected for the long time-lag plateau
days where the count rate increases fro±m 0.006 0.002 c/s (2σ2 1.3 10−4). Such a kind of plateau can be
± fluctuation∼ ×
to0.051 0.006c/s,implyingachangeofafactor8.5 3.0). produced in different ways. A flat structure function could
However±, we note that the lowest count rate bins f±or the beproducedbyaflickernoisepowerspectrum,P(f) f−1,
∝
snapshot timescale could correspond to statistical fluctua- atatimelagwherethefluctuationsarestillcorrelated,orit
tions below the mean flux at that time, and consequently couldalsoarisefromawhitenoisepowerspectrum,P(f)
thereis thepossibility thattheamplitudeof real variations f0, at a time lag where there is no more correlation. ∝
isover-estimated.Wecannotmeasuretherootmeansquare We confirm this by performing simulations of
variability for this snapshot lightcurve because the source lightcurves with different power spectral slopes and deter-
is not detected in several of the snapshots due to the short mining their structure function. We simulated the limit of
exposures. adataset with nonoise and muchbettersampling than the
real case. Ifthetwo structurefunctions look thesame for a
highS/Ndatasetwithbettersampling,wecannotexpectto
4.2 Structure Function detect a difference in the real data. We generated 104 light
The first order structure function V(τ) (see Do et al. 2009, curveswith mean and standard deviation equal to that ob-
and references therein for more details) is a useful tool to served(0.140c/sand0.008c/srespectively),forthecasesof
studythevariability inthetimedomain whentheavailable both a f−1 and a f0 power spectrum.The lightcurves were
datasetisunevenlysampled.V(τ)isrelatedtotheautocor- generated using the Timmer & Koenig (1995) algorithm to
relation function and the power spectrum and represents produce the light curves that are 750 days long (10 times
another, straightforward, way to define the power of the more than our data set), and sampled 10 times per day.
variability at a given timescale. We computed V(τ) only Then,for each simulated light curve,we computedV(τ)by
for the light curve in Fig. 1 corresponding to the pointing usingabinningthatisthreetimessmallerthanthatofFig.
timescale(itcannotbecomputedonthesnapshottimescale 2.WefittedeachsimulatedV(τ)withapowerlawfunction.
lightcurvebecauseofthemultiplenondetections).Thefirst Wemeasuretheslope γ andthecoefficient (c)ofthepower
order structurefunction is defined as: lawandwederivedthemean andthestandarddeviationof
thedistributionsofthetwovalues.Weget γ =0.11 0.11,
V(τ)=h[s(t+τ)−s(t)]2i (1) c =6.7 2.3 10−5 for f−1, and γ =h i0.004 ±0.052,
where the angled bracket implies an average quantity, s(t) hci = 1.1± 0.2× 10−4 for f0. Herhe tihe a−ngled b±rackets
h i ± ×
isaset of measures (thecountratein theX-rayband)per- imply the mean value of the distribution. Weget fully con-
formed at the time t. Wemeasured [s(t+τ) s(t)]2 for all sistent values for the two structure functions within uncer-
thepossiblepairsoftimelagssinceourdatas−etisunevenly tainty. We therefore conclude that both power spectra are
sampled.Werebinnedthetimelags withavariablebinsize producingtwofullyconsistentstructurefunctionswithzero
to have a minimum numberof pairs equal to 20 and a con- slope.Summarizing,wealsoconcludethatV404Cygisvari-
sistentnumberofpairswithineachbin.Wetakecentralbin able over the inspected timescale (2-75 days), but from the
value as the bin time lag, while the average of V(τ) is the structurefunction we cannot determine if this variability is
valueofthestructurefunctionforeachbinatthattimelag. generated by a correlated (flicker noise) or an uncorrelated
Thestandarddeviation of V(τ),dividedbythesquareroot process (white) noise.
ofthenumberofpointsinthebin(σ /√N ),represents
bin bin
theuncertainty in V(τ).
4.3 X-ray Spectral analysis
The expected shape of an ideal structure function con-
sists of two plateaus connected by a curve whose slope is Due to the source faintness, in order to search for spectral
linked to the nature of the source variability (red noise, variability asafunction oftheX-rayflux,weselected three
flicker noise etc.). The first plateau is connected to size of count rate ranges: low <0.01 c/s, medium 0.011–0.020 c/s,
(cid:13)c 0000RAS,MNRAS000,000–000
Quiescent X-ray variability in V404 Cyg 5
4
0
0.
2
0
0.
e
at
r
nt
u]
coc/s 0
pweFxahipngioceuhlsr,ueibrseu1nt.toioTtmnopeptlo:h∼teXte5X-draRk 0.3−10 keV hysT)eb.rs [eaTnccakhopgers00.050.1rrhroeeousfptneord0tenimnsducseebsttctoraiamltcehtee(ead(av2rle0icrg1Thah2igiv-tme0ac7lee2u- xo00rspvb3ineosics2doue0f r:020Ve00040t:1i203m2341−)4eC00c∼37yo0g−r50r0i62en30s4 (pt20tsho0=)en.:-0d010s.13:6t2−6o3. 21t[5h0deakdyeabsVye]sg,rina∼nn6gi00ne.g1o5nocft/hosbe).sXiBdRoT0tt0o0pm3o1i:n4Tt0i3hn0eg0st2ai.mmTeeshcaeasloetnh(leayvuepproapigneert
0−4
×1 counts for each spectrum. The high spectrum has 465
3 ∼
counts for 23.7 ks of exposure, the medium spectrum
∼
has 550 counts for 41.9 ks of exposure, and the low
∼ ∼
spectrum has 490 counts for 74.1 ks of exposure. The
0−4 spectracovert∼he0.3–10keVran∼ge,andwearemainlyinter-
1
×
2 ested in verifying the presence of spectral changes with the
τV() fltruaxl.mWoedefilrustsuraelsltyriacptepdlieodurtoanBaHlysLisMtXoBthiensqimuipeslecsetncsep.ecIt-
consistsofapowerlawmultipliedbythephabscomponent,
10−4 which accounts for the photoelectric absorption due to the
interstellar medium.Westarted fittingthethreespectrasi-
multaneouslywithallthecomponentsfreetovary.Weveri-
fiedthat thevalueofNH was constant within uncertainties
0 amongthethree.Thus,we imposed NH tobethesame be-
0 20 40 60 80
tween different spectra, but free to vary to minimize the
Time lag τ [days]
χ2, and we determined a new fit. We report the value of
Figure2.FirstorderstructurefunctionV(τ)fortheX-raylight
all model components, together with the 0.3–10 keV ab-
curve (pointing timescale) of V404 Cyg in Fig. 1 (upper panel).
sorbed/unabsorbedflux,the0.3–10 keVluminosity andthe
The dashed line corresponds to the intensity of the measure-
ment errors (2σ2 ∼ 2×10−6). The dot-dashed line corre- bolometricluminosityinTab.1.Thelatterwascalculatedby
noise
sponds to the expected intensity of the long time lag plateau extrapolating our model to the 0.01–100 keV energy range,
(2σ2 ). The solid line represents the fit made with a with the caveat that the true spectral shape is unknown
fluctuation
powerlaw. both above 10 keV and much below 1 keV.
The spectral fits are shown in Fig. 3. We measure
NH =8.9 0.9 1021 cm−2,consistentwiththatpreviously
± ×
and high > 0.020 c/s (see Fig. 1), and we produced three reported for the source in quiescence (e.g. Bradley et al.
summed spectra2. This gives roughly the same number of 2007; Plotkin, Gallo & Jonker 2013), and slightly higher
than the total in the direction of the source that lies be-
tween 6.6 1021 cm−2 and 8.1 21 cm−2 (Kalberla et al.
× ×
2 Thethreespectraaremadewiththesumofthefollowingobsid 2005; Dickey & Lockman 1990, respectively). This could
0003140300+: 2,9,11,12,13,24 (high); 1,5,15,17,22,23,26,28,29,30 imply the presence of localized absorbing material intrin-
(medium),3,4,5,7,8,10,14,16,18,20,21,25,27,31,32,33,34(low) sic to the source (as also suggested by e.g. Wagner et al.
(cid:13)c 0000RAS,MNRAS000,000–000
6 Bernardini & Cackett
%
0.01
V−1
ke
unts s −1 10−3 $"#
o
d c
malize & (
or10−4
n
$
2
χ 0
−2
0.5 1 2 5 !"#!"# $ $"# %
Energy (keV)
’
Figure 3. Summed spectra of V404 Cyg for three count rate
ranges:low<0.011c/s(bluetriangles),medium0.011–0.020c/s Figure 4. Contour plot showing the allowed region in the Γh-
(redcircles),andhigh>0.020c/s(blacksquares).Thesolidlines Γl plane for our best fit power law model on the three count
represent the model made by a power law multiplied by phabs rate spectra. The three contours illustrate the 68%, 90%, and
(photoelectricabsorption).Residualareshowninthelowerpanel. 99% confidence level (black, red, green). The solid straight line
representsΓh=Γl.
1994). We get a roughly constant power law index, Γ =
l
2.10 0.15, Γ = 2.20 0.14, Γ = 2.35 0.15, consistent
l l
± ± ±
with that for the source in quiescence (e.g. Bradley et al.
2007;Plotkin, Gallo & Jonker 2013, who analysed Chandra 15
0
and XMM-Newton data respectively). To further verify the 0.
consistence betweenΓ and Γ ,weproducedacontourplot
of the two componentls thathwe show in Fig. 4. The two c/s]
pthoewe1rσlcaownfiinddenicceeslaerveel.coHnoswisetevnetr,waitshminallaclhitatnlegem, loerses tthhaann e keV [ 0.01
at
11%couldbepresent,butnotdetectablewithpresentdata. nt r
u
We defined here the maximum percentage change, as the co
absolute valueof the differencebetween Γ of thehigh spec- 10 0−3
− 1
0 ×
trum and that of the low spectrum, divided by its average 2. 5
value, e.g.: 3( Γh-Γl )/(Γh+Γm+Γl). We conclude that,
| |
within statistical uncertainty, only the power law normal-
ization is clearly changing. This suggests that, likely, the 0
spectralshaperemainsalmostthesameasthefluxchanges. 0 5×10−3 0.01 0.015
0.3−2.0 count rate keV [c/s]
To investigate how the spectrum changes on shorter
timescales,wecomparedthe0.3–2keVcountratewiththat Figure 5. 0.3–2.0 keV count rate vs 2–10 keV count rate. The
solidlinerepresentsthebestfitmadewithalinearcomponent.
in the 2–10 keV range for each XRT pointing (see Fig. 5).
This allows us to explore the spectral variability between
every observation (with two-day spacing) rather than just
comparing three average spectra. We fit a straight line to ingfrom bothaspherical Comptonizing corona orfrom the
the 0.3–2 keV versus 2–10 keV count rate relation, which inner part of an ADAF region, and consists of a compTT
gives a best-fitting slope of 0.86 0.04 with χ2 = 3.6. Al- component (Titarchuk 1994). Both models are multiplied
though some scatter is present, m±aking the χ2νvalue high, by phabs which we imposed to be the same among differ-
ν
there is no significant deviation from a linear fit, confirm- ent spectra, but free to vary. We report all fitting results
ing that the overall spectral shape remains roughly con- in Tab 2. Concerning the bremsstrahlung model, kT it is
stant as the flux changes. The fact that the slope of the constrained to lie in the range 3.3–4.7 keV, and only the
straight line is 0.83 simply implies that the flux in the 0.2– normalization clearly varies. We get χ2 = 1.17 for 70 dof.
ν
10 keV band is lower than that in the 0.3–2 keV band (i.e. The inspection of the contour plot of kT vs kT shows
h l
F2−10keV = 0.86F0.3−2keV). However, the relative flux in that the two are consistent within 1σ (see Fig. 6). For
∼
thetwo bands stay thesame as thesource varies. the compTT model, which has a higher number of free pa-
As a second step we fitted the three spectra with two rameters, we first fixed the temperature of the Comptoniz-
different more physical models that are normally used for ing corona to a reasonable value, kT=60 keV (this com-
BH LMXBS. The first one is appropriate for the X-ray ponent is otherwise unconstrained due to the fact that we
emission from hot electrons and ions in the outer part of are covering herethe0.3–10 keV energy range only). Then,
an ADAF region, and consists of a bremsstrahlung compo- we verified that the temperature of the seed photons (T0)
nent. The second one is appropriate X-ray emission aris- wasconstantwithinuncertaintiesbetweendifferentspectra.
(cid:13)c 0000RAS,MNRAS000,000–000
Quiescent X-ray variability in V404 Cyg 7
Table1.Spectralresultsasafunctionofthecountrate:low<0.011c/s,medium0.011–0.020c/s,andhigh>0.020c/s(χ2 =1.14,70
ν
dof).Thethreespectraarefittedtogetherimposingacommoncolumndensity(freetovary)NH =8.9±0.8×1021cm−2.Theresultsof
fitonthespectrummadesummingtogetheralltheexistingpointing(total)arealsoreported.ForthelatterwegetNH =9.1±0.9×1021
cm−2 (χ2 = 1.22, 68 dof). The 0.3–10 keV absorbed flux, the 0.3–10 keV unabsorbed flux, and the 0.3-10 keV and 0.01–100 keV
ν
luminosityat2.4kpcarealsoreported.Uncertaintyareat1σ cl.
Parameter Unit High Medium Low Total
Γ 2.10±0.15 2.20±0.14 2.35±0.15 2.31±0.11
Norm. 10−4 5.1±0.8 3.5±0.5 2.1±0.4 3.1±0.4
Fabsorb 10−12 1.20±0.10 0.75±0.05 0.37±0.03 0.58±0.03
0.3−10keV
erg/s/cm2
Funabsorb 10−12 2.7±0.7 1.7±0.5 1.0±0.3 1.5±0.3
0.3−10keV
erg/s/cm2
L2.4kpc a 1033 ∼1.9 ∼1.2 ∼0.7 ∼1.0
0.3−10keV
erg/s
L2.4kpc a 1033 ∼5.5 ∼4.1 ∼3.2 ∼4.3
0.01−100keV
erg/s
χ2 (dof) 1.14(70) 1.22(68)
ν
Consequently,weimposed it tobethesamebetween differ-
ent spectra, but free to vary, in order to minimize the χ2,
andwetried toputsome constrain onthevariabilityof the
plasma opticaldepth(τp).Weget T0 =0.33 0.07, andan
±
opticaldepth,τp =0.9 1.5,thatisconstantwithinslightly
−
morethan1σ(seeFig.7),andonlythenormalizationclearly 10
changes. In both cases, bremsstrahlung and compTT mod-
els, the spectral shape remains roughly the same, within
statistical uncertainty, as the flux varies. We note that the V]
derivedopticaldepthishigh,butitdependsonthevalueof ke
kT, meaning that for increasing value of kT τp decreases. kT [l
Moreover, we are interested in measuring the percentage
5
variability range of τp, more than is absolute value. Such
modelingallowsustoconstrainthemaximumchangeinthe
ionandelectrontemperature(kTbrem)andtheopticaldepth +
of the Comptonizing corona (τp) respectively. We measure
∆KTbrem < 34% and ∆τp < 49% respectively. The latter
2 4 6 8 10 12
valuesarelessconstrained thanthatobtainedwith apower
kT [keV]
h
law model (∆Γ < 11%), where no assumptions are made
Figure6.ThesameasFig.4butforthebremsstrahlungmodel,
(whilefortheCompTT modelwearbitrarilyfixedthevalue
ofT0).Consequently,inordertoquantifythemaximumfrac- where kTh and kTl is the bremsstrahlung temperature for the
highandlowcountratespectrum respectively.
tional spectral change between different count rate spectra,
we use thepower law model.
equivalenthydrogencolumntoE(B-V)dereddeningtheflux
4.3.1 Energetics fursoimngGthu¨evgeras&-toO¨-dzuelst(2ra0t0i9o),.NFHin(aclmly−,2t)o=es6ti.m86a×te10t2h1eEe(xBti-nVc)-
We estimated the flux in the UVW1 band, by summing tioncorrectionat2600˚A(UVW1)weusetheinterstellarex-
together all available pointings. The source is undetected tinctioncurveofCardelli, Clayton & Mathis(1989),getting
in the UVW1 band, with an lower limit to its magnitude 3760 as conversion factor (this means that the dereddened
(AB system) > 25. The 3σ upper limit to the source flux flux is 3760 times the observed flux). We therefore obtain
density is Fλ < 1.62 10−18 ergcm−2s−1˚A−1. Comparing a 3σ upper limit to the UVW1 luminosity Lλ < 4.2 1030
with Fig.7from Hyn×eset al.(2009)it canbeseen thatthe erg/s (λL <1.1 1034 erg s−1). ×
λ
×
SwiftUVW1filteriscoveringaregionofthesourcespectral We get an estimate of the X-ray luminosity by fitting
energy distribution ( 1.1 1015 Hz) where the flux dra- thetotalspectrum(summedfromallpointings),onceagain,
∼ ×
matically drops due to Galactic extinction. To convert the withamodelconsistingofapowerlawmultipliedbyphabs.
fluxdensitylimittoaluminositylimitwefirstconvertfrom WereportallbestfitmodelparameterinTab.1.Weinfera
(cid:13)c 0000RAS,MNRAS000,000–000
8 Bernardini & Cackett
Table 2. Spectral results as a function of the count rate: low < 0.011 c/s, medium 0.011–0.020 c/s, and high > 0.020 c/s for the
bremsstrahlungandthecompTT model.Uncertaintyareat1σ cl.
Bremsstrahlung
Parameter Unit High Medium Low
NH 1022 cm−2 0.68±0.06a
KTbrem keV 4.7±1.3 4.2±0.9 3.3±0.6
0.9 0.7 0.5
Norm. 10−4 4.2±0.5 2.8±0.3 1.7±0.2
χ2 (dof) 1.17(70)
ν
CompTT
Parameter Unit High Medium Low
NH 1022 cm−2 0.5±0.2a
T0 keV 0.33±0.07a
τp keV 1.5±0.4 1.3±0.3 0.9±0.2
Norm. 10−6 9±2 6±2 3±1
χ2 (dof) 1.12(69)
ν
a Thevalueisimposedtobethesamebetweendifferentspectra,butisleftfreetovarytominimizetheχ2.
5 DISCUSSION
3 WeshowedthatthequiescentX-raylightcurveofV404Cyg
ishighly variable on timescales ofdaystomonths,with the
fractional root mean square variability Fvar = 57.0 3.2%.
±
The first order structure function, Vτ, has a shape consis-
tent with both a flicker noise power spectrum (index 1)
2 −
and awhite noise powerspectrum (index 0).This implies
−
that variability is detectedat all inspected timescales, from
l
about 2 days up to about 75 days. However, the structure
functionanalysiscannotdetermineifthevariabilityisdueto
1 acorrelatedortoanuncorrelatedprocess.Weobservedmul-
+
tipletimesafactorof4–5variabilityontimescales ofabout
a week. Higher magnitude changes (a factor of 10–20) have
beenpreviouslyobservedon 12hourstimescalefromV404
∼
CygduringquiescencebyWagner et al.(1994);Hyneset al.
1 2 3 (2004). We also showed that multiple flares are presents in
h the X-ray light curve on timescale of about one hour, with
Figure7.ThesameasFig.4butfortheCompTT model,where thecountratechangingbyafactorof 5 8.Afactorofa
τh andτl istheplasmaopticaldepthforthehighandlowcount few variability has been already seen o∼n bo−th 30 minutes
ratespectrumrespectively. and years timescales (Wagner et al. 1994; R∼eynolds et al.
2013).Consequently,V404Cygisvariableduringquiescence
onawiderangeoftimescales fromtensofminutestoyears.
The power law observed in the energy spectrum of BH
LMXBs is likely generated in the inner region of the ac-
cretion flow, closer to the BH, where a population of high-
energy electrons is present. Moreover, there is no sign of
0.3-10keVunabsorbedfluxof1.5 0.3 10−12 ergcm−2s−1. the presence of the accretion disk in the X-ray spectrum of
± ×
Assuming a distance of 2.4 kpc and a spectral power law V404Cygduringquiescence(norfromquiescentBHLMXBs
shape(thatisexpectedtoextendabove10keV)weextrap- in general) which would instead produce thermal emission.
olate this to a 0.01–100 keV luminosity Lx 4.3 1033 Consequently, the observed variability is not originating in
∼ ×
erg/s. The latter and the following values should be taken theaccretiondisk,butislikelydrivenbymassaccretionrate
as indicative only, since the real spectral shape both above fluctuations in the inner region of the accretion flow, close
10keVandmuchbelow1keVisnotknown.Sincethespec- to the BH. We can only speculate about the real nature of
tral change is almost constant as the flux varies, we can the matter in the inner accretion flow and about how it is
get an estimateof the0.01–100 keVluminosity correspond- triggering thevariability.
ing to the two Swift snapshots with the highest and lowest If the matter around the BH is in an ADAF state,
count rate respectively, by simple proportionality. We get hot electrons in the outer region of the ADAF (at a dis-
Lmax 2.8 1034 erg/s (obsid 00031403012, 0.1 c/s) tance & 104 Schwarzschild radii from the central BH)
which∼implies× 2 10−5 LEdd, and Lmin 1 ∼1033 erg/s shouldcooldownthroughthermalbremsstrahlung(see,e.g.
(obsid 0003140∼3027×, 0.002 c/s) which im∼plies× 8 10−7 Narayan & Yi 1994, 1995a). However, this emission should
LEdd, where LEdd =1∼.26 1039 erg/s, for a 10 M∼⊙ B×H. beconsistentwithapowerlawspectralshapewithindex1,
×
(cid:13)c 0000RAS,MNRAS000,000–000
Quiescent X-ray variability in V404 Cyg 9
while here we observe a softer power law with index 2.10 likely observe here, then the synchrotron dominating sce-
−
2.35.Weconclude,asalsosuggestedbyQuataert & Narayan nario would be favoured. This is in agreement with the
(1999), that the soft X-ray emission we observe is not results of Hyneset al. (2009), who found evidence of the
thermal bremsstrahlung arising from the outer region of presence of a jet in quiescent radio, and IR data. We stress
an ADAF, but it could be instead inverse Compton emis- that we are only proposing a possible scenario, and that
sion produced by hot electron close to the BH. However, this must not be taken as a firm conclusion, since there is
Hyneset al.(2009)analysedthespectralenergydistribution stillalotofuncertaintyaboutthecorona/jet geometry and
ofV404Cyginquiescenceandarguedthatthecontribution its microphysics. However, we emphasize that future simul-
totheaccretion lightintheUVbandisminimal,disfavour- taneous multiwavelength observations, including the whole
inganADAFscenario alone,thatinstead shouldproducea sourceSED,andinparticularcoveringthehardX-rayrange,
moreabundantUVemission. Ontheotherhand,anADAF like in the case of the recently launched NuSTAR satellite,
plusanoutflowingwind(see,e.g.Quataert & Narayan1999) could certainly help in solving themystery about thephys-
couldbeconsistentwiththeSEDofV404Cyg(thewindhas ical mechanism generating hard X-ray photons: Compton,
the effect to suppress the UV emission from the ADAF it- synchrotron, and bremsstrahlung emission are expected to
self). We emphasize that a formal fit of this model against show significant differences in thespectrum above10 keV.
real data has neverbeen performed so far.
Hot electrons surrounding the black hole, and emit-
5.1 A comparison with Cen X-4
ting a Compton like spectrum, are also part of the so-
called jet models (see, e.g. Gallo et al. 2007, and more ref- In this section we compare V404 Cyg with Cen X-4, likely
erence therein). In this scenario, inverse Compton emission thebeststudiedquiescentNSLMXB,thatwerecentlymon-
is produced at the base of the jet, in a region linked to itoredwithasimilarcampaign(Bernardini et al.2013).De-
a compact and magnetic corona, where there is a popu- spite their intrinsic difference, due to the nature and the
lation of quasi-thermal particles. Attached to the corona, massofthecompactobject,andtothelengthofthebinary
and departing from it, there is a jet, where non-thermal orbital period ( 15 hours for Cen X-4, and 6.4 days for
∼ ∼
particles are strongly accelerated and emit predominantly V404Cyg)thetwosourcesshareseveralcommonproperties.
through optically thin synchrotron.The dominant emission They both show a strongly variable X-ray light curve, with
process in the jet scenario can vary depending on the na- amulti-peaked,flare-likestructure,withvariabilityonshort
ture of the accreting object and its specific spectral state. timescales( days).Theyarebothvariableoverawiderange
∼
Markoff, Falcke & Fender (2001), for example, found that oftimescales: fromhundredsofsecondsandtensofminutes
jet synchrotron emission could account for the whole SED for Cen X-4 and V404 Cyg respectively to years. However,
of the BH XTE J1118+480 in the hard state. This means in the case of Cen X-4 the X-ray light curve is very likely
that synchrotron is dominating from the radio through the generated by a red noise power spectrum, while the X-ray
X-ray.Gallo et al. (2007) found instead that theX-rayqui- lightcurveofV404Cygcouldbegeneratedbyeitheraflicker
escent emission of the BH A0620-00 is primarily produced noise power spectrum or by a white noise power spectrum.
by inverse Compton processes at the base of the jet. We Consequently, we can not determine if it is due to a corre-
know that a compact jet (see also Miller-Jones et al. 2008) latedprocessornot.Bothsystemsshowedaclearcorrelation
is likely driven from the inner region of the accretion disk between their X-ray and optical emission (see Hyneset al.
producingaflatradiospectrumandanexcessinthemid-IR. 2004; Bernardini et al. 2013, for the case of V404 Cyg and
Incorona/Compton models, that can or cannot includethe for the case of Cen X-4 respectively). The correlation was
presenceofajet,variabilityinthemassaccretionratewould interpreted in either case as X-ray reprocessing (from the
leadtoachangeindensityinthecorona,andhenceachange companion star and the accretion disk in the case of Cen
in the probability of Compton scattering in the corona. In X-4 and V404 Cyg respectively). We note that the X-ray
thetypical unsaturated Comptonization scenario, theslope emission of Cen X-4is also strongly correlated with that in
of the power law in the energy spectrum will depend on theUVband,however,nosimultaneousX-ray/UVcoverage
the optical depth, and hence the mass accretion rate. We wasavailableforV404Cyg.ForbothsourcestheX-rayspec-
constrained themaximumfractional variability of thespec- tral shape remained approximately constant as the X-ray
trum to be lower than 11%, and we demonstrated that the fluxvariedduringquiescence.Allthesesimilaritieslikelyim-
spectral shape remains constant within uncertainties as the ply, as suggested by Cackett et al. (2013); Bernardini et al.
fluxvaries.ThiswouldfavourascenarioinwhichtheX-ray (2013),thatalowlevelofaccretionisstilloccurringforCen
emissionisdominatedbysynchrotronradiationproducedin X-4also in quiescence, as it occurs for V404 Cyg.
a jet rather than Comptonization. In the synchrotron sce- Comparing the power-law index between the two
narioonemaynaivelyexpectlessdependenceofthespectral sources, however, we see a difference. The power law spec-
slope on the mass accretion rate and more likely a change tral index of Cen X-4 is slightly harder 1.4–2.0 compared
in the normalization. This is because the spectral slope is to2.10–2.35 ofV404Cyg.Thetwodifferentspectralshapes
mostly set byplasma physics(e.g. acceleration probability) could indicate two distinct emission mechanisms at work.
and while there may be some dependence on the mass ac- BHLMXBscould beinajet dominated statealso in quies-
cretion rate, it is likely weak. cence,likeinthecaseofA0620-00(Gallo et al.2007),where
Summarizing, a significant change in the slope of the the jet is dominating from the radio (through synchrotron
power law (and so a spectral shape change with the flux) emission) through the soft X-ray (through Compton emis-
wouldnaivelyfavoraComptonization dominatingscenario, sion) and also in the case of V404 Cyg itself (Hyneset al.
while if the slope is instead steady, but the normalization 2009), where a steady jet is producing synchrotron emis-
is clearly changing with mass accretion rate, as we more sionintheradioandlikelymid-IRbands(anditmightalso
(cid:13)c 0000RAS,MNRAS000,000–000
10 Bernardini & Cackett
somehow influence the X-ray emission). On the contrary, ThisresearchhasmadeuseoftheXRTDataAnalysisSoft-
radio emission has neverbeen detected from NS LMXBs in ware (XRTDAS) developed under the responsibility of the
quiescence. Recent studies suggest instead that NSs could ASIScience Data Center (ASDC),Italy.
never reach a jet dominated state, always remaining X-ray
dominatedobjects(Migliari & Fender2006).Morerecently,
Bernardini et al. (2013) suggested that the power law of
REFERENCES
Cen X-4 could be produced by residual accretion, since it
isstrictlylinkedtothethermalcomponentarisingfrom the ArnaudK.A.,1996,inAstronomicalSocietyofthePacific
NS surface, they are both changing in tandem as the flux Conference Series, Vol. 101, Astronomical Data Analysis
vary. Moreover, for NSs in general, when their spectrum is Software and Systems V, Jacoby G. H., Barnes J., eds.,
dominated by the non thermal power law component (that p.17
isnotthecaseofCenX-4)thepresenceoftheNSmagnetic Arons J., Tavani M., 1993, ApJ, 403, 249
field could also provide further mechanisms, different from Asai K., Dotani T., Hoshi R., Tanaka Y.,Robinson C. R.,
a jet, for creating a power-law like spectrum. For example Terada K., 1998, PASJ, 50, 611
residual accretion on the NS magnetosphere could occur, Bernardini F., Cackett E. M., Brown E. F., D’Angelo C.,
or alternatively a shock could form due to the interaction Degenaar N., Miller J. M., Reynolds M., Wijnands R.,
of the wind of a quenched pulsar and the accreting matter 2013, MNRAS
(Arons& Tavani1993; Campana et al. 1998, respectively). Bildsten L., RutledgeR.E., 2000, ApJ,541, 908
Blandford R.D.,Begelman M. C.,1999, MNRAS,303, L1
Bradley C.K.,HynesR.I.,KongA.K.H.,HaswellC.A.,
Casares J., Gallo E., 2007, ApJ, 667, 427
6 SUMMARY Breeveld A. A.et al., 2010, MNRAS,406, 1687
Cackett E. M., Brown E. F., Degenaar N., Miller J. M.,
We analysed 34 Swift X-ray observations of the BH LMXB
ReynoldsM., Wijnands R., 2013, MNRAS,433, 1362
V404 Cyginquiescence, with thegoal ofcharacterizing the
Cackett E. M., Brown E. F., Miller J. M., Wijnands R.,
source variability. The main results of this work are briefly
2010, ApJ, 720, 1325
summarized in thefollowing.
Cackett E. M., Fridriksson J. K., Homan J., Miller J. M.,
TheX-raylightcurveisstronglyvariable,showingsev-
Wijnands R.,2011, MNRAS,414, 3006
eral flares with count rate changes of a factor of 5 8
∼ − CampanaS.,ColpiM.,MereghettiS.,StellaL.,TavaniM.,
on timescale of hours. We measure the root mean square
1998, AA&AR,8, 279
variability Fvar =57.0 3.2%.
± Campana S., Israel G. L., Stella L., Gastaldello F.,
The first order structure function is consistent with
Mereghetti S., 2004, ApJ,601, 474
both a flicker noise power spectrum (of index -1) and a
CampanaS.,MereghettiS.,StellaL.,ColpiM.,1997,A&A,
white noise power spectrum (of index 0). This implies that
324, 941
the source is variable at all the timescales inspected here
Cannizzo J. K., 1993, ApJ,419, 318
(days-months).Thiscompletesthewiderangeofvariability
CardelliJ.A.,ClaytonG.C.,MathisJ.S.,1989,ApJ,345,
timescalesalreadyknowforthesource:fromtensofminutes
245
to years.
Casares J., Charles P. A., 1994, MNRAS,271, L5
WeproducedthreeaveragedX-rayspectraasafunction
Casares J., Charles P. A., Naylor T., 1992, Nature, 355,
of the source count rate, low < 0.01 c/s, medium 0.011–
614
0.020 c/s, and high >0.020 c/s. We found the spectrum to
Casares J.,CharlesP.A.,NaylorT.,PavlenkoE.P.,1993,
be well fitted by an absorbed power-law model. We mea-
MNRAS,265, 834
surea powerlaw indexΓ=2.10 2.35 and found that it is
− CharlesP.,ClarksonW.,CornelisseR.,ShihC.,2008,New
constant within 1σ,as thefluxchanges (also between sin-
∼ Astronomy Reviews, 51, 768
gle two-day spaced observations). We found that the max-
Coti Zelati F., Campana S., D’Avanzo P., Melandri A.,
imum fractional variability of the energy spectrum slope is
2013, ArXiv e-prints
∆Γ<11%.
D’Angelo C. R.,Spruit H. C., 2010, MNRAS,406, 1208
Concerningthenatureofthematterintheinneraccre-
D’Angelo C. R.,Spruit H. C., 2012, MNRAS,420, 416
tion flow, our results could naively favour the presence of a
Dickey J. M., Lockman F. J., 1990, ARA&A,28, 215
jet (synchrotron dominated) as in such a scenario changes
Do T., Ghez A. M., Morris M. R., Yelda S., Meyer L., Lu
in the X-rayluminosity would likely produce only a change
J.R.,HornsteinS.D.,MatthewsK.,2009,ApJ,691,1021
inoverallfluxandnotasignificantchangeinspectralshape
Evans P.A. et al., 2009, MNRAS,397, 1177
in theX-rayband.
Fender R. P., Gallo E., Jonker P. G., 2003, MNRAS, 343,
L99
Gallo E., Migliari S., Markoff S., Tomsick J. A., Bailyn
C. D., Berta S., Fender R., Miller-Jones J. C. A., 2007,
ACKNOWLEDGMENTS
ApJ,670, 600
FB thanks Michele Donato and Jean-Charles Tropato for Gallo E., Treu T., Marshall P. J., Woo J.-H., Leipski C.,
their useful help about coding. We thank Sera Markoff and AntonucciR., 2010, ApJ,714, 25
Elena Gallo for useful discussions regarding jet models in Gehrels N.et al., 2004, ApJ, 611, 1005
quiescence.ThisworkmadeuseofdatasuppliedbytheUK Gu¨ver T., O¨zel F., 2009, MNRAS,400, 2050
Swift Science Data Center at the University of Leicester. Ho L. C., 2009, ApJ,699, 626
(cid:13)c 0000RAS,MNRAS000,000–000