a r X i v :a s t r o -p h /0407025v 1 1 J u l 2004
The X-ray Outburst of H1743–322:High-Frequency QPOs with a
3:2Frequency Ratio
Ronald A.Remillard
Center for Space Research,MIT,Cambridge,MA 02139-4307
rr@https://www.wendangku.net/doc/5b9331519.html,
and
Je?rey E.McClintock
Harvard-Smithsonian Center for Astrophysics,60Garden St.MS-3,
Cambridge,MA 02138jem@https://www.wendangku.net/doc/5b9331519.html,
and
Jerome A.Orosz
Dept.of Astronomy,San Diego State University,5500Campanile Drive,
San Diego,CA 82182-1221orosz@https://www.wendangku.net/doc/5b9331519.html,
and
Alan M.Levine
Center for Space Research,MIT,Cambridge,MA 02139-4307
aml@https://www.wendangku.net/doc/5b9331519.html,
ABSTRACT
The 2003X-ray outburst of the candidate black-hole binary,H1743–322,was investigated in frequent pointed observations (2–250keV)with the Rossi X-ray Timing Explorer.We consider one particular program of 130observations.We organized these data into 111time intervals and conducted a search for the presence of high-frequency quasiperiodic oscillations (HFQPOs)in the range 50–2000Hz.Only a single observation (2003June 13)yielded a detection above 4σ;the central frequency of 239±4Hz is consistent with the 240Hz QPO reported for this source on 2003May 28(Homan et al.2003).We next grouped the observations in several di?erent ways and computed the average power-density spectra (PDS)in a search for further evidence of HFQPOs.This e?ort yielded two signi?cant results for those observations de?ned by the presence of low-frequency QPOs (0.1-20Hz)and an absence of “band-limited”power continua:(1)The 9time intervals with the highest X-ray ?ux yielded an average PDS with a QPO at 166±5Hz.(4.1σ;3–35keV);and (2)a second group with lower X-ray ?ux (24time intervals)produced an average PDS with a QPO at 242±3Hz (6.0σ;7–35keV).The ratio of these two frequencies is 1.46±0.05.This ?nding is consistent with results obtained for three other black hole systems that exhibit commensurate HFQPOs in a 3:2ratio.Furthermore,the occurrence of H1743-322’s slower HFQPO at times of higher X-ray luminosity closely resembles the behavior of XTE J1550-564and GRO J1655-40.We discuss our results in terms of a resonance model that invokes frequencies set by general relativity for orbital motions near a black-hole event horizon.
1.Introduction
A key question in black hole binary(BHB) research is whether high-frequency QPOs(HFQ-POs),observed in the range of40–450Hz,repre-sent a unique timing signature that may constrain a black hole’s mass and spin via a model rooted in general relativity(GR;see McClintock&Remil-lard2003).The HFQPO signals are weak,and an attack on this problem requires observations of bright X-ray transients during their rare and short-lived(~months)outbursts.
HFQPOs have been detected in7con?rmed or candidate BHB systems,including H1743–322 (see below).In each of three sources(GRO J1655–40,XTE J1550–564,and GRS1915+105),tran-sient HFQPOs have appeared at a pair of fre-quencies that are commensurate with a3:2ratio (Remillard et al.2002a;Remillard et al.2003). Furthermore,when the HFQPOs were compared among the three sources,it was found that the frequencies scale inversely with black hole mass. This result is consistent with expectations for oscillations produced at some characteristic ra-dius in a strong-gravity environment described by GR theory,but only in the special case where the values of the dimensionless black hole spin parameter are similar for the three sources (McClintock&Remillard2003).
Commensurate HFQPO frequencies can be in-terpreted as a signature of an oscillation driven by some type of resonance condition.Abramowicz& Kluzniak(2001)had earlier proposed that QPOs could represent enhanced emission from a particu-lar radius where there is a resonance in the GR co-ordinate frequencies for orbital/epicyclic motions in strong gravity(see Merloni et al.1999).Reso-nances in some form may be applicable to both BH and NS systems(Abramowicz et al.2003).We note that GR coordinate frequencies and associ-ated beat frequencies in the inner accretion disk were invoked in earlier work on variable-frequency QPOs in both neutron-star and some BHB sys-tems(Stella,Vietri,&Morsink1999).
Herein,we report on a general search for HFQ-POs in the source H1743–322.The2003out-burst of this source was?rst detected on March 21in hard X-rays(15–200keV)by INTEGRAL (IGR J17464–3213;Revnivtsev et al.2003). Follow-up observations with RXTE led to the recognition that the source is a recurrent X-ray nova?rst observed with HEAO1in1977–1978 (Markwardt&Swank2003;Gursky et al.1978). The X-ray spectral and temporal properties ex-hibited during the?rst outburst had established H1743–322as a black-hole candidate;its X-ray spectrum contained both a soft component(1–10 keV)and a hard X-ray tail(10–100keV;Cooke et al.1984),and there were no pulsations or X-ray bursts that would identify the source as a neutron-star system(Tanaka&Lewin1995).
An X-ray light curve(1.5–12keV)for the recent outburst is displayed in the top panel of Fig.1,us-ing data from the RXTE All-Sky Monitor(ASM; Levine et al.1996).The bottom panel shows mea-surements of the spectral hardness ratio,de?ned as the ratio of the ASM count rate at5–12keV to that at3–5keV.These data show that the initially hard spectrum rapidly evolves to a softer one.A number of bright?ares are seen before the source settles into a relatively steady and soft X-ray state. Several bright?ares reach intensities near1.5Crab (i.e.3.6×10?8erg cm?2s?1at2–10keV)during 2003April18-24.
Radio detections of H1743–322were obtained during the?rst few weeks of the2003outburst, yielding a precise celestial position and evidence for a transient radio jet(Rupen,Mioduszewski, &Dhawan2003a;2003b;2003c).Soon there-after,a faint optical counterpart was detected (R=21.9),despite~13mag of dust extinction (Steeghs et al.2003).
RXTE made227pointed observations of H1743–322(=XTE J1746–326)during the2003–2004out-burst,in the course of carrying out?ve di?erent observing programs.In this paper we report on the results of a search for HFQPOs(50–2000Hz) from this source using the130observations from program ID80146.The times of these observa-tions are indicated in the row of vertical lines in the top panel of Fig. 1.An HFQPO was reported for H1743–322after a25ks observa-tion with RXTE on2003May28for program ID80135(Homan et al.2003).This observations yielded an HFQPO detection(4.5σ)at240Hz, with a possible second feature(2.5σ)at160Hz (Homan et al.2003).
In the following sections we describe our anal-ysis techniques and methods for data grouping that lead to signi?cant detections of HFQPOs at
242and166Hz.Assuming H1743–322is a gen-uine BH,our discovery establishes this source as the fourth BH where a pair of HFQPOs appear to have commensurate frequencies in a3:2ratio. It is further be shown that additional properties of these HFQPOs bear a striking resemblance to those seen in GRO J1655–40and XTE J1550–564 (Remillard et al.2002a).
2.Observations and Data Analysis
We consider all of the130observations of H1743–322conducted by RXTE under program ID80146.This data set contains occasional se-quences of short observations,and we therefore chose to collect and organize the observations into the111time intervals that are listed in Table1. The number of observations combined for each time interval is given in column4.The maximum time span for any interval is0.48days,and the total exposure time is498ks.
Our X-ray timing analyses utilize data from the RXTE Proportional Counter Array(PCA).The data telemetry modes chosen for this program al-low us to construct PCA light curves with125μs time resolution in3energy bands:3-7,7-14,and 14-35keV.The PCA consists of5detector units (PCUs).However,the observations of H1743–322 utilized a variable number of PCUs(usually3)be-cause some of the PCUs are cycled o?periodically to avoid problems with detector breakdown.For each observation interval,the telemetry modes for high-speed data merged all of the good events from the PCUs that were in operation.
Table1also provides,for each observation in-terval,the mean count rate(c/s/PCU)at2-35 keV(col.6)and at7–35keV(col.7).The lat-ter quantity is useful in evaluating the strength of nonthermal radiation in H1743–322and was used to group observations in the search for HFQ-POs.The PCA count rates are derived from background-subtracted energy spectra processed from the‘Standard2’data mode using PCU#2, which was in use(along with PCU#0)during ev-ery observation of program80146.All of the spec-tral reductions were performed via the‘ftools’rou-tines in the HEAsoft software package distributed by NASA’s HEASARC.
The procedures used to compute power den-sity spectra(PDS)and to search for QPOs are described in Remillard et al.(2002).Brie?y,for each of the111time intervals(Table1),we com-pute Fourier transforms for each256-s segment in the PCA light curve,and we output the av-erage PDS for the available number of segments. We then subtract the deadtime-corrected Poisson noise and normalize the PDS to units of(rms de-viation/mean)2Hz?1.In each frequency bin, we compute the uncertainty as the larger of ei-ther the statistical error(2/N0.5prior to rms nor-malization)or the empirical standard deviation of the mean power,where N is the number of trans-forms(of256s duration)in a given observation interval.We note that this conservative approach can yield error bars that appear large,compared to bin-by-bin?uctuations,when the broad power continuum varies signi?cantly during an observa-tion.Finally,the power densities and uncertain-ties are re-binned in logarithmic intervals of fre-quency(ν),maintaining(for this study)a mini-mum?ν/ν=0.04.When we average the PDS for groups of observations,we compute the weighted mean(usingσ?2)and the net uncertainty for each frequency bin.
We search for HFQPOs using statistical tests to detect a Lorentzian peak rising above the local power continuum.The continuum,in terms of log Pν,is modeled with a second order polynomial in logν,which presumes a power-law function with allowances for broad curvature.QPOs are distin-guished from broad power peaks using a coherence parameter,Q=ν/F W HM 2.For each PDS, we useχ2minimization to obtain the best?t for the QPO pro?le and the local power continuum. Finally,in our general QPO searches,we seek re-sults that have acceptable values ofχ2νwhile sur-passing a signi?cance threshold for the integrated QPO power(P),relative to the uncertainty(σP): P/σP 4.This elevated threshold compensates for the large number of PDSs and trial frequencies considered in the analysis of data sets obtained for typical bright X-ray transients.
3.Results
We searched for QPOs in the range50–2000 Hz for each of the111observing intervals(Table 1).For each interval,we considered separately the average PDS for the energy ranges3-35,7-35, and14-35keV.Only one interval(2003June13=
MJD52803)yields a detection above4σ.In this case the energy range is7–35keV,the central fre-quency is239±4Hz,the integrated rms amplitude
is r=2.0±0.2%,Q~9,and the detection signif-icance is4.3σ.The central frequency is consistent with the240Hz QPO detected for this source on 2003May28(MJD52787)in observations from
a di?erent RXTE program(Homan et al.2003). There is one additional noteworthy?nding from this initial search for HFQPOs.For the interval with highest PCA count rate(1.5Cra
b on MJD 52765),an HFQPO candidate(3.8σ)appears at 162±7Hz with Q~4and r=1.1±0.2%(3-35 keV).
Most of the HFQPOs detections in BHB out-bursts have been found in PDS averaged over a number of RXTE observations(Remillard et al.1999; Cui et al.2000;Remillard et al.2002c)because high statistical precision is required in order to detect a faint signal spread over a bandwidth
of20to50Hz.Furthermore,judicious data selection is required,since HFQPOs are usu-ally detected only in the“steep power-law state”(McClintock&Remillard2003),and within this state further considerations may be required to deal with amplitude variability,frequency switch-ing,and choice of PDS energy range.We therefore investigated several strategies for grouping the ob-servations of H1743–322prior to further searches for HFQPO detections.
In the case of XTE J1550-564,it was shown that the phase lags(13-30keV vs.2-13keV)
in low-frequency QPOs(LFQPOs;0.1-20Hz) could be used to de?ne three LFQPO types that are well correlated with HFQPO properties (Remillard et al.2002c).These LFQPO types predicted,respectively,the absence of HFQPOs or the appearance of HFQPOs at either185or276 Hz.We investigated this method for the case of
H1743–322,but we found the results to be statis-tically unsatisfactory because the source is fainter than XTE J1550-564and the LFQPO amplitudes are comparatively low.
As an alternative approach,one can capital-ize on correlations found between HFQPOs and the properties of the power continuum and the energy spectrum.The LFQPO type(“C”)that forecasts an absence of HFQPOs in XTE J1550-564can also be recognized for its association with
a“band-limited”power continuum in which the power density is?at at low frequencies and then drops abruptly at frequencies above the LFQPO and its harmonic.Such a PDS can be recog-nized via the integrated rms power derived from the PDS(0.1to10Hz),and those values are given in col.8of Table1.Observation inter-vals for H1743–322with integrated rms>0.11 all show LFQPOs and band-limited power con-tinua,and we label these as PDS type“q-bl”in col.9of Table1.Intervals with PDS that are de-void of LFQPOs are noted as PDS type“0”;these cases also exhibit the lowest values of the PCA hardness ratio(HR=col.6/col.7<0.017, with one exception).These properties suggest that PDS type0designations identify the times when the source is in a thermal-dominant state (McClintock&Remillard2003).The remaining intervals contain LFQPOs without band-limited power continua and are labeled“q”in col.9of Table1.
We computed average PDS for groups“q-bl”(23observation intervals)and“0”(44intervals), excluding the last9intervals where the2-35keV count rate drops below550c/s/PCU(~0.2Crab) and the PDS become statistically weak.We then divided the“q”type PDS into2groups distin-guished by the source brightness above or below 430c/s/PCU at7-35keV(col.8of Table1).This latter criterion attempts to exploit the fact that the strength of the steep power-law component distinguishes the observations that yield each of the QPOs with frequencies that scale in a3:2ratio (hereafter referred to as the2ν0QPO and the3ν0 QPO)in both XTE J1550-564and GRO J1655-40 (Remillard et al.2002a).
The PDS for the4groups,plotted in units of log(ν×Pν)vs.logν,are shown in Fig. 2.The PDS fro the“q-bl”group shows a strong and very broad feature that peaks near3Hz.The PDS for the“0”group appears to show the same feature, although reduced in strength by a factor~30. HFQPO features are apparent only in the PDS for the“q”groups,and they occur at di?erent fre-quencies.We note that all of the four PDS show a residual continuum above100Hz that is an arti-fact of an imperfect deadtime model for the PCA instrument.
In Fig.3we show the HFQPO detections for the two“q”groups at higher frequency resolution, along with the?tted models for the QPO pro?les.
The9“q”intervals with the higher X-ray?uxes produce an average PDS(2-35keV)with a QPO (4.1σ)at166±5Hz,with r=0.60±0.08%and Q=5.7±1.6.The“q”group comprising26in-tervals at lower X-ray?ux produces an average PDS(7-35keV)with a QPO(6.0σ)at242±3 Hz,with r=1.1±0.1%and Q=11±1.Note that the transition in H1743–322from a broader HFQPO at2ν0in a broad bandwidth to a nar-rower feature at3ν0in a harder energy band oc-curs with decreasing7-35keV count rate,which is presumably dominated by a non-thermal spectral component.This is precisely the behavior exhib-ited by the HFQPOs from XTE J1550-564and GRO J1655-40(Remillard et al.2002a).
4.Discussion
The X-ray light curve and variability charac-teristics of H1743–322during its2003outburst fully support the identi?cation of this source as a black hole candidate(see§1).The behavior of H1743–322resembles the BHBs XTE J1550-564 and GRO J1655-40in many ways,although it was not as bright as those sources(by a factor of3–4), comparing the brightest4-month intervals of each outburst.This is the likely reason why we do not detect HFQPOs from H1743–322during individ-ual observations,other than those on MJD52787 and possibly MJD52765.
The e?ort to group the PDS in order to gain statistically signi?cant HFQPO detections for H1743–322was guided by results for XTE J1550-564and GRO J1655-40(Remillard et al.2002a). Our best results were found simply by selecting ob-servations with LFQPOs present,excluding cases with q-bl type power continua,and then grouping the PDS according to the7-35keV count rate, which represents the strength of the nonthermal X-ray?ux.This e?ort yielded two HFQPOs with properties that closely resemble the HFQPOs of XTE J1550-564and GRO J1655-40,i.e.the cen-tral frequencies are consistent with a3:2ratio, the amplitudes have rms~1%,and the lower frequency QPO(166Hz for H1743–322)is seen at times of highest non-thermal?ux.In addition,the higher frequency QPO(242Hz)has a narrower pro?le(i.e.higher Q value)and the detection bandwidth has a relatively higher mean photon energy.We note that the one remaining BH sys-tem with HFQPOs in a3:2ratio(GRS1915+105) was investigated using a di?erent analysis method in which each QPO was extracted from a portion of a particular type of violently variable light curve (Remillard et al.2002b;Remillard et al.2003).
HFQPOs with frequencies at2ν0and3ν0can be considered as expressions of a single frequency system.As with the other BHBs discussed above, the frequency system in H1743–322appears to be invariant through changes in luminosity or inter-ruptions due to the source’s evolution through a thermal-dominant or other states.The observa-tions that correspond with the two groups that yield HFQPO detections(i.e.PDS type“q”in Table1)span a range204–1083c/s/PCU at7–35keV and1234–3775c/s/PCU at2–35keV.The temporal order of these observations is shown with small arrows in the top panel of Fig.1,where the upper/lower row represents the242/166Hz group, respectively.There are three gaps in the series of q-type observations(https://www.wendangku.net/doc/5b9331519.html,bined arrows)that occur when the PDS types are either0or q-bl (see Table1).
In summary,black hole HFQPOs appear at sta-ble pairs of frequencies with3:2ratio,and they exhibit common behavior patterns linked to the properties of the energy spectra.These results support the view that HFQPOs convey a dis-tinct temporal signature for each accreting black hole.The detection bandwidth and high frequen-cies suggest that HFQPOs originate near the black hole event horizon.Furthermore,the stability of HFQPO pairs with changing luminosity suggests that the frequencies may depend only on the in-herent properties of the black hole,viz.its mass and spin.
We have shown that the HFQPOs in H1743–322appear to originate from the same mecha-nism as the HFQPOs with3:2frequency ratio in the other BHBs.An empirical relationship has been derived(only3sources)between the HFQPO frequency systems and black hole mass (McClintock&Remillard2003),and we may ap-ply this to the case of H1743–322.We assume that HFQPO frequencies depend only on the black hole mass(M x)and spin,which is frequently eval-uated in terms of the dimensionless spin param-eter:a?=cJ/GM2x,where J is the angular mo-mentum of the black hole.If the black hole in H1743–322has a similar value of a?as the other
BHBs,then the HFQPOs(withν0~81Hz)sug-gests:M x=931/ν0~11.5M⊙.
Detailed models of the HFQPO properties and behavior patterns must confront the prob-lem that we do not understand the origin of the steep power-law spectrum.This radiation com-ponent is always fairly strong when the3:2oscil-lations appear.Furthermore,the strength of the steep power-law appears to regulate the frequency switching from2ν0to3ν0.