Mon.Not.R.Astron.Soc.000,1–??(2010) Printed28January2011 (MNLATEXstylefilev2.2) Hysteresis in the spectral states of the neutron star low-mass X-ray binary EXO 1745–248 1 Arunava Mukherjee1⋆ and Sudip Bhattacharyya1† 1 0 1Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research, Mumbai 400005, India 2 n a J 7 ABSTRACT 2 We study the low-frequency timing properties and the spectral state evolution of the transient neutron star low-mass X-ray binary EXO 1745–248 using the entire Rossi ] E X-ray Timing Explorer Proportional Counter Array data. We tentatively conclude that EXO 1745–248 is an atoll source, and report the discovery of a ≈ 0.45 Hz low- H frequency quasi-periodic oscillation and ∼ 10 Hz peaked noises. If it is an atoll, this . h source is unusual because (1) instead of a ‘C’-like curve, it traced a clear overall p clockwise hysteresis curve in each of the colour-colour diagram and the hardness- - intensity diagram; and (2) the source took at least 2.5 months to trace the softer o banana state, as opposed to a few hours to a day, which is typical for an atoll source. r t The shape of the hysteresis track was intermediate between the characteristic ‘q’-like s a curves of several black hole systems and ‘C’-like curves of atolls, implying that EXO [ 1745–248is animportantsourceforthe unificationofthe blackhole andneutronstar accretion processes. 1 v Keywords: accretion,accretiondiscs—methods:dataanalysis—stars:neutron— 7 techniques: miscellaneous — X-rays: binaries — X-rays: individual (EXO 1745–248) 5 3 5 . 1 1 INTRODUCTION based on spectral and timing properties. The BS is traced 0 1 out on time scales of hours to a day without any hystere- The spectral states and the correlated timing properties 1 sis (van der Klis 2006). On the other hand, the higher HC of neutron star and black hole low-mass X-ray binaries : extreme island state (EIS) is traced out in days to weeks, v (LMXBs)canbeveryusefultounderstandtheextremeenvi- and secular motions in the form of parallel tracks are seen i ronmentsofthesesources(van derKlis 2006).Anexcellent X in EIS.Anatoll source movesfrom EIS toBS viaan island waytostudythesepropertiesistotrackthesesourcesinthe state(IS).Probablytheonlytransientatollsourceshowinga r colour-colourdiagram(CD;hardcolour(HC)vs.softcolour a ‘q’-likehysteresisHIDcurveisAqlX-1(Maitra and Bailyn (SC)) and in the hardness-intensity diagram (HID; hard (2004);Reig et al. (2004);seealsoBelloni et al. (2007)for colour vs.intensity;see§2).From thebeginningof an out- 4U 1636–53 tracks). Such neutron star LMXBs, and more burst,theintensityofatransientblackholesourceincreases, importantly sources showing intermediate tracks between typicallykeepingtheHCatanear-constantvalue.Nearthe ‘q’ and ‘C’, can be very useful (1) to unify the black hole highest intensity the HC value quickly decreases, followed and neutron star accretion processes, and (2) to sort out byanintensity decreaseat alower HCvalue,andasoft-to- the mismatch between the standard EIS-IS-BS framework hardtransitionatalowerintensityvalue.Thusablackhole and the general hysteresis phenomena. In this Letter, we LMXBtypicallytracesa‘q’-likehysteresiscurveintheHID show that the bursting neutron star LMXB EXO 1745– (van derKlis 2006; Belloni 2009). It is usually believed 248 (Markwardt and Swank 2000; Wijnands et al. 2002; that neutron star LMXBs do not trace hysteresis curves Heinkeet al. 2003) is such an intermediate source with in CD/HIDs (van der Klis 2006). For example, the near- uniqueproperties. Eddington Z sources trace out roughly ‘Z’ shaped tracks on time scales of hours to a day, while the less luminous atoll sources have ‘C’ shaped tracks (van Straaten et al. 2 DATA ANALYSIS AND RESULTS 2003; vander Klis 2006). The lower HC banana-like por- tion (BS) of the ‘C’ track can be divided into upper ba- The neutron star transient LMXB EXO 1745–248 was ob- nana(UB),lowerbanana(LB)andlowerleftbanana(LLB) served with Rossi X-ray Timing Explorer (RXTE) in 2000 and2002:(1)betweenJul13,2000(starttime:04:43:28)and ⋆ E-mail:[email protected] Nov3, 2000 (endtime: 00:17:52; proposal nos.: P50054 and † E-mail:[email protected] P50138); and (2) between Jul 2, 2002 (start time:20:38:24) 2 Mukherjee and Bhattacharyya andJul22, 2002(endtime:11:04:00; proposal no.:P70412) power spectrum has a significant peaked noise at ∼ 10 Hz for a total observation time of 144 ks. We have produced (seeTable1andpaneleofFig.3).Forexample,thepeaked CD and HID using the entire standard-2 mode data from noise in the data set of Nov 17, 2000 (03:46:10–04:45:07) the top layers of Proportional Counter Unit (PCU) 2. We hasasignificanceof1−7.9×10−175,acentroid-frequencyof have defined HC and SC as the ratio of the background- 11.01±0.25 Hz, an RMS-amplitude of (5.8±0.14)% and a subtracteddetectorcountsinthe(9.2−18.9)/(5.7−9.2) and Q-factor of 0.73±0.046. A typical phase 9 power spectrum (3.9−5.7)/(2.6−3.9) keV energy bands, respectively. We hasarednoise(RMS∼4−12%)below∼0.02Hzwiththe havebeenabletodividethe2000datainninetemporalseg- Leahy-powerreachingafewtimes10at0.004Hz(panelfof ments, i.e., nine phases (see Table 1 for time ranges). Each Fig. 3). phase traces a distiguishable portion of the HID track (see In Fig. 4, we have displayed the locations of the ther- Fig. 1). This figure shows that the source starts from a low monuclearbursts,theLFQPOandthekilohertz(kHz)QPO intensityandahighHCvalue,andinthehardstate(phase (Mukherjeeand Bhattacharyya 2011) in the CD and the 1–4) it traces a few adjacent curved parallel tracks below HID. While the non-photospheric-radius-expansion (non- the intensity ∼ 450 counts/s/PCU and in the HC range of PRE)burstsoccuredinthehardstate,thePREburstsand ∼0.6−1.3. UnlikeAql X–1 and several black-hole sources, thekHzQPO appeared in phase8 (Table 1). theintensityofEXO1745–248doesnotincreasemuchinthe highest HC value. Rather, Fig. 1 shows that the HC value decreasessubstantiallyandtheintensityincreasesatalower 3 DISCUSSION AND CONCLUSIONS HC value of ∼ 0.6 in the hard state. During the transition fromphase4tophase5,thesourcegoesthroughalargein- In this Letter, we have studied the evolution of spectral tensity(in 2.6−18.9 keV)jumpfrom ∼361 counts/s/PCU states of the neutron star LMXB EXO 1745–248. We ten- to ∼ 1135 counts/s/PCU while having a relatively small tatively conclude that it is an atoll source because of the change in the HC value (∼0.60 to ∼0.36; see Table 1 and followingreasons.(1)Fromthespectralfitting,wefindthat Fig. 1). In the high intensity state (phase 5–7) the source theobserved2−30keVunabsorbedsourcefluxvariedinthe shows a clear anti-clockwise loop (hysteresis; Fig. 1). EXO range(0.05−2.12)×10−8 ergscm−2s−1.Suchalargeinten- 1745–248undergoesamoderateintensityjumpfromphase7 sity variation does not happen in a source, which shows an tophase8,andtheintensitysteadilydecreasesuptophase9 exclusive ‘Z’ behaviour (vander Klis 2006). (2) The hard whilekeepingtheHCvaluenearlysame(Fig.1).Thelower colours (Fig. 1) of EXO 1745–248 are consistent with those intensity portion of phase 9 shows a significant increase in of atoll sources, but different from Z sources (Muno et al. HC value. Since the ASM data confirm that the source in- 2002). (3) The source shows parallel tracks for the higher tensity further decreases into the quiescence, Fig. 1 implies hard colour values in HID (Fig. 1), which are typical of aclearoverallclockwise loop(hysteresis)ofthesource.The atollsources(van derKlis 2006).(4)ShapeoftheCDtrack IntensityandHCvaluesofphase10(2002data) areconsis- for lower hard colour values looks like a banana (Fig. 2). tent with those of phases 8 and 9. The phases 5–10 show a (5) Hard state to soft state transition of the source was clearbanana-liketrackintheCD(Fig.2).Inthehardstate, plausibly quick (van der Klis 2006). (6) PRE bursts were two phases (1,3), which display substantial changes in HC found in the softer state (Fig. 4), as usually observed for values, show large variation in SC values. fast spinning neutron star LMXBs (Muno et al. 2004). (7) In order to identify the spectral states of EXO 1745– ThekHzQPOwasobservedinthetransitionalstate(plausi- 248, we have analyzed the low-frequency power spectra of blyLB/LLB), which is usualfor atolls (Maitra and Bailyn eachphaseusingallthePCAevent-modedatawithastan- 2004;van derKlis 2006).(8)VLFNat<1HZwasobserved dardtechnique(van derKlis 1989).EachLeahy-normalized inBS,and∼10Hzpeakednoisewasdetectedinthetransi- power spectrumhas aNyquistfrequencyof 128 Hzand the tional state (plausibly LB/LLB), which are usual for atolls best resolution of 0.004 Hz. We have fitted the continuum (Maitra and Bailyn 2004;vander Klis 2006).(9)Rednoise component ofa powerspectrum with aconstant+powerlaw RMS is higher in the hard state (vander Klis 2006). How- model (describing white and red noises, respectively) and ever,althoughwecannotconfirm,thereissomechancethat anynarrowfeaturewithaLorentzian.Thehardstate(Phase at the most intense state, the source transformed into a Z 1–4) power spectra are typically well described with a con- source (e.g., Homan et al. (2010)). This is because, the es- stant+powerlaw, having a strong red noise (typical RMS timated source luminosity in this state was ∼0.5 times the ∼ 25−45%) below ∼ 10 Hz with the Leahy-power reach- Eddingtonluminosity(van der Klis 2006),fora2−30keV ing > 1000 at 0.004 Hz (see panel a of Fig. 3; Table 1). flux of 2.12×10−8 ergs cm−2 s−1, and assuming a 5.5 kpc Only one power spectrum in hard state shows a significant sourcedistance,1.4M⊙neutronstarmass,6.0stellarradius- (1−3.28×10−10) low-frequency quasi-periodic oscillation to-mass ratio and ionized hydrogenic accreted matter. (LFQPO) at 0.452±0.0049 Hz with a quality (Q) factor EXO 1745–248 is very interesting, unusual and impor- of 6.3±1.93 (see Table 1 and panel b of Fig. 3). A high tant for the following reasons. (1) The source exhibited a intensity state (phase 5–7) power spectrum typically shows clear overall clockwise hysteresis in HID and CD (Fig. 1 a very-low-frequency-noise (VLFN) below ∼ 1 Hz with an and 2). A local anti-clockwise hysteresis is also observed RMS ∼ 6−10% having the Leahy-power reaching a few in the high intensity state (plausibly UB). (2) In the hard times100 at 0.004 Hz.Suchapowerspectrum also shows a state(plausibly EIS),unlikea typicalatoll, thehardcolour weakbroadhumpnear0.3Hz(panelcofFig.3;Table1).A changed largely, and no horizontal track is present at the phase8powerspectrumshowsarednoise(RMS∼5−25%) highest hard colour in HID (van derKlis 2006). Moreover, roughly below 0.1 Hzwith theLeahy-powerreachingabout the hard-to-soft transition involved a large change in in- 100 at 0.004 Hz (panel d of Fig. 3). In most cases, such a tensity, unlike several black hole sources. These caused an Spectral states of EXO 1745–248 3 HID-track-shape intermediate between atoll ‘C’ tracks and blackhole‘q’tracks.(3)InCD/HID,thesourcemovedfrom EIS to UB, while usually an atoll moves to LB/LLB from EIS(van Straaten et al. 2003;van derKlis 2006).(4)The CD/HID tracks of EXO 1745–248 could be segmented in time (Fig. 1 and 2). Several segments can be distinguished by timing properties (Fig. 3), which shows that these seg- ments are actually in different states, i.e., not in the same stateshiftedbysecularmotions.Thesourcetypicallydwells inasegmentforafewdaystoaboutamonth(Table1).(5) EXO1745–248 took at least 2.5 monthstotrace theBS, as opposedtoafewhourstoaday,whichistypicalforanatoll source (van derKlis 2006). The CD/HIDhysteresis tracksof EXO 1745–248 could be very useful to relate the accretion processes in neutron star systems and black hole systems (§1). Finally, the HID hysteresistrackofEXO1745–248,whichisintermediatebe- tween the ‘q’-like hysteresis track of Aql X-1 and ‘C’-like non-hysteresistracksofmostatollsources,suggeststhatthe popular EIS-IS-BS framework of ‘C’-like tracks might be a specialcaseofamoregeneralhysteresisbehaviour.However, observationsofmoresuchintermediatesourcesarerequired to verify this. REFERENCES Belloni T., 2009, (astro-ph/0909.2474v1) Belloni T., Homan J., Motta S., Ratti E., 2007, MNRAS, 379, 247 Heinke C.O., Edmonds P.D., Grindlay J.E., Lloyd D.A., Cohn H.N.,Lugger P.M., 2003, ApJ, 590, 809 Homan J., van der Klis M., Fridriksson J.K., et al., 2010, ApJ,719, 201 Maitra D., Bailyn C.D., 2004, ApJ,608, 444 Markwardt C.B., Swank J.H., 2000, IAUC, 7454, 1 Mukherjee A., Bhattacharyya S., 2011, (astro- ph/1101.0358) Muno M.P., Galloway D.K., Chakrabarty D., 2004, ApJ, 608, 930 Muno M.P., Remillard R.A., Chakrabarty D., 2002, ApJ, 568, L35 Reig P., van Straaten S., van der Klis M., 2004, ApJ, 602, 918 vanderKlisM.,1989,Eds.H.O¨gelmanandE.P.J.vanden Heuvel,(Kluwer Academic Publishers: Boston), 27 van der Klis M., 2006, in Compact Stellar X-ray Sources, eds. Lewin W.H.G., van der Klis M., Cambridge Univ. Press, 39, 39 van Straaten S., van der Klis M., M`endez M., 2003, ApJ, 596, 1155 Wijnands R., Homan J., Remillard R., 2002, ATel, 101, 1 4 Mukherjee and Bhattacharyya Table 1.Variouspropertiesofthesourceinthe10phases (see§2;seealsoFigs.1and2). Phase Starttime1 Endtime2 Soft-colour3 Hard-colour4 Intensity5 Remarks6 Phase1 13/07/2000 04:43:28 13/07/2000 05:13:04 2.2418623 0.68810917 137.27226 Nonarrowfeature. 21/07/2000 10:23:28 21/07/2000 11:44:00 2.2798728 0.65725367 133.26667 TheLeahy-powersstartrising Phase2 24/07/2000 15:15:28 24/07/2000 16:16:00 2.1178232 0.66027898 213.43736 significantlyabove2in∼2−12Hz, 06/08/2000 12:55:28 06/08/2000 13:28:00 2.0948016 0.65078129 240.75461 andreachabove1000at0.004Hz Phase3 06/08/2000 14:00:32 06/08/2000 15:10:08 2.1486752 0.63243319 252.97392 (seeFig.3-a). 13/08/2000 09:58:24 13/08/2000 11:08:00 3.1275264 0.87214917 84.047352 Non-PREburstsobserved. Phase4 13/08/2000 11:39:28 13/08/2000 12:50:08 2.0437827 0.61713214 195.17507 Similartothephases1−3.AnLFQPO 15/08/2000 17:57:52 15/08/2000 19:26:40 2.054908 0.59863458 360.6038 isdetected inonesegment(Fig.3-b). Phase5 18/08/2000 13:08:32 18/08/2000 13:41:36 1.9404588 0.35567838 1135.4601 TheLeahy-powersstartrising 21/08/2000 19:36:32 21/08/2000 19:55:44 1.8951522 0.35778984 977.661 significantlyabove2below∼1Hz, Phase6 24/08/2000 10:48:48 24/08/2000 11:19:44 2.0420255 0.45687916 1191.8152 andreachabove100at0.004Hz. 27/08/2000 05:51:28 27/08/2000 06:28:00 1.9212794 0.41142752 897.9504 Aplausiblehumpat∼0.3Hzis Phase7 27/08/2000 06:53:36 27/08/2000 08:10:56 1.8786671 0.36700101 806.13514 typicallyseen(seeFig.3-c). 30/08/2000 15:23:28 30/08/2000 15:46:40 1.9175748 0.39400707 777.30473 Phase8 05/09/2000 09:41:20 05/09/2000 11:13:04 1.809879 0.37902471 340.92112 TheLeahy-powersstartrising 06/10/2000 05:00:32 06/10/2000 05:13:36 1.75375 0.38471589 164.56586 significantly above 2 below ∼ 0.1 Hz, andreach≈100at0.004Hz.Apeaked noise around 10 Hz is typically seen (seeFigs.3-dand3-e).PREburstsand kHzQPOfoundinthisphase. Phase9 09/10/2000 05:13:20 09/10/2000 05:50:56 1.732144 0.4048109 120.68564 TheLeahy-powersstartrising 03/11/2000 00:02:24 03/11/2000 00:17:52 1.3439725 0.45464597 4.9097872 significantlyabove2below ∼0.02Hz, andreachabove10at0.004Hz(Fig.3- f). Phase10 02/07/2002 20:38:24 02/07/2002 20:47:44 1.6929414 0.30983275 280.54887 Roughlysimilartophase9. 22/07/2002 08:46:24 22/07/2002 11:04:00 1.4488735 0.56095016 20.41566 1Starttimeofthefirst(upperline)andthelast(lowerline)continuous timesegments ofthephase. 2 Endtimeofthefirst(upper line)andthelast(lowerline)continuous timesegments ofthephase. 3Soft-colours(defined in§2)ofthefirsttimebinofthefirstcontinuous timesegment(upper line),andthelasttimebinofthelast continuous timesegment(lowerline)ofthephase. 4Hard-colours(definedin§2)ofthefirsttimebinofthefirstcontinuous timesegment(upper line),andthelasttimebinofthelast continuous timesegment(lowerline)ofthephase. 5Intensities(defined in§2)ofthefirsttimebinofthefirstcontinuous timesegment(upper line),andthelasttimebinofthelast continuous timesegment(lowerline)ofthephase. 6Primarilyashortdescriptionofatypicalpowerspectrum ofthephase(§2). Spectral states of EXO 1745–248 5 Figure 1.Hardness-intensitydiagram(HID)ofEXO1745–248 usingtheRXTEPCAdata. Hardcolourandintensity(forPCU2)are definedin§2.Varioustemporalsegments(phases;seeTable1and§2)areshownwithdifferentsymbolsandphasenumbers(seeTable 1).TheschematicintheinsetshowsthemotionofthesourcealongtheHIDtrackwithtime.Thisfigureclearlyshowshysteresisinthe spectralstates. 6 Mukherjee and Bhattacharyya Figure 2.Colour-colourdiagramofEXO1745–248usingtheRXTEPCAdata.Hardcolourandsoftcolouraredefinedin§2.Various temporal segments (phases; see Table 1 and § 2) are shown with different symbols and phase numbers (see Table 1). Typical 1σ error barsforsomeofthephasesareshown.Thisfiguresuggestshysteresisinthespectralstates. Spectral states of EXO 1745–248 7 Figure 3.Typicallow-frequencypowerspectraofvariousphasesofEXO1745–248 (seeTable1and§2).Panela:phase1–3.;panel b: the power spectrum of phase 4withLFQPO;panel c:phase 5–7; panel d:phase 8; panel e: sameas panel d, but ina differentscale to showthepeaked noiseclearly;andpanelf:phase9. 8 Mukherjee and Bhattacharyya Figure4.Colour-colourdiagram(leftpanel)andhardness-intensitydiagramofEXO1745–248(sameasFig.2andFig.1).Thetriangles mark the non-PRE thremonuclear X-ray bursts and the two squares mark the PRE bursts (§ 2). The plus signs mark the continuous datasetcontainingtheLFQPOandthediamondsignsmarkthecontinuous datasetduringwhichthekHzQPOappeared(§2).