X-ray and Near-IR Variability of the Anomalous X-ray Pulsar

a r X i v :0707.2093v 2 [a s t r o -p h ] 17 D e c 2007Accepted in ApJ,2007December 2

Preprint typeset using L A T E X style emulateapj v.10/09/06

X-RAY AND NEAR-IR VARIABILITY OF THE ANOMALOUS X-RAY PULSAR 1E 1048.1?5937:FROM QUIESCENCE BACK TO ACTIVITY

Cindy R.Tam,1Fotis P.Gavriil,2,3Rim Dib,1Victoria M.Kaspi,1,4Peter M.Woods,

5,6Cees Bassa,1

Accepted in ApJ,2007December 2ABSTRACT We report on new and archival X-ray and near-infrared (near-IR)observations of the anomalous X-ray pulsar 1E 1048.1?5937performed between 2001-2007with the Rossi X-ray Timing Explorer (RXTE ),the Chandra X-ray Observatory (CXO ),the Swift Gamma-ray Burst Explorer ,the Hub-ble Space Telescope (HST ),and the Very Large Telescope .During its ～2001-2004active period,1E 1048.1?5937exhibited two large,long-term X-ray pulsed-?ux ?ares as well as short bursts,and large (>10×)torque changes.Monitoring with RXTE revealed that the source entered a phase of timing stability in 2004;at the same time,a series of four simultaneous observations with CXO and HST in 2006showed that its X-ray ?ux and spectrum and near-IR ?ux,all variable prior to 2005,stabilized.Speci?cally,we ?nd the 2006X-ray spectrum to be consistent with a two-component blackbody plus power law,with average kT =0.52keV and power-law index Γ=2.8at a mean ?ux level in the 2?10keV range of ～6.5×10?12erg cm ?2s ?1.The near-IR ?ux,when detected by HST (H ～22.7mag)and VLT (K S ～21.0mag),was considerably fainter than previously measured.Recently,in 2007March,this newfound quiescence was interrupted by a sudden ?ux enhancement,X-ray spectral changes and a pulse morphology change,simultaneous with a large spin-up glitch and near-IR enhancement.Our RXTE observations revealed a sudden pulsed ?ux increase by a factor of ～3in the 2?10keV band.In observations with CXO and Swift ,we found that the total X-ray ?ux increased much more than the pulsed ?ux,reaching a peak value of >7times the quiescent value (2?10keV).With these recent data,we ?nd a strong anti-correlation between X-ray ?ux and pulsed fraction.In addition,we ?nd a correlation between X-ray spectral hardness and ?ux.Simultaneously with the radiative and timing changes,we observed a signi?cant X-ray pulse morphology change such that the pro?le went from nearly sinusoidal to having multiple peaks.We compare these remarkable events with other AXP outbursts and discuss implications in the context of the magnetar model and other models of AXP emission.Subject headings:pulsars:general —pulsars:inpidual (1E 1048.1?5937)—stars:neutron —stars:pulsars 1.INTRODUCTION The small handful of unusual objects known as anoma-lous X-ray pulsars (AXPs)are characterized by their slow spin periods (5–12s),large inferred magnetic ?elds (～1014G),and persistent X-ray luminosities in great ex-cess of available spin-down power.These young,isolated pulsars have properties common to another class of ob-jects,the soft gamma repeaters (SGRs).For recent re-views,see Woods &Thompson (2006)and Kaspi (2007).Potentially the most intriguing and most revealing prop-erty shared by AXPs and SGRs is their highly volatile nature:both are known to exhibit occasional and sudden dramatic ?ux and spin variability in the form of X-ray bursts,?ares,and glitches.The magnetar model (Duncan &Thompson 1992;Thompson &Duncan 1995;Thompson &Duncan 1996)

1Department of Physics,Rutherford Physics Building,McGill University,3600University Street,Montreal,QC,H3A 2T8,Canada 2NASA Goddard Space Flight Center,Astrophysics Science Di-vision,Code 662,Greenbelt,MD,20771,USA 3NPP Fellow;Oak Ridge Associated Universities,Building SC ?200,1299Bethel Valley Road,Oak Ridge,TN,37830,USA 4Canada Research Chair;Lorne Trottier Chair;R.Howard Web-ster Fellow of CIF AR 5Dynetics,Inc.,1000Explorer Boulevard,Huntsville,AL,35806,USA 6NSSTC,320Sparkman Drive,Huntsville,AL,35805,USA was instrumental in explaining both the “anomalous”X-ray emission,as well the episodes of burst activ-ity.It identi?ed the sources as highly magnetized iso-lated neutron stars,with ?eld strengths on the order of 1014?1015G.The magnetar model uniquely,and correctly,predicted that SGR-like bursts would be ob-served from AXPs.A major proposed e?ect of this enormous magnetization is the “twisting”of the magne-tosphere,with associated magnetospheric currents and heating of the crust at the twisted ?eld-line footpoints (Thompson et al.2002).This gives rise to the unusual X-ray spectrum,which has generally been empirically well characterized by a two-component model consisting of a power law plus blackbody.Bursts of high-energy emission occur presumably when the crust succumbs to magnetic stresses and deforms,leading to a rapid re-arrangement of the external magnetic ?eld.Recently,Beloborodov &Thompson (2007)proposed the existence of a plasma corona contained within the closed magneto-sphere to explain the broad band spectrum of magnetars that extends from the infrared (IR)to hard X-rays be-yond 100keV (Kuiper et al.2006).1E 1048.1?5937has had an unusual history,even by AXP standards 7.Prior to 2002,monitoring of 7A summary of its properties can be found in the online SGR/AXP Catalog

2

the pulsed ?ux with the Rossi X-ray Timing Explorer (RXTE )showed that its spin down was so unstable that phase coherence could

be maintained for peri-ods of only months at a time (Kaspi et al.2001),be-haviour which echoed earlier reports of ˙P

,?ux and spec-tral variability (Oosterbroek et al.1998).Two small bursts,the ?rst ever found in an AXP,were dis-covered in 2001from 1E 1048.1?5937(Gavriil et al.2002),clinching the suspected association between AXPs and SGRs.Since 2002,tri-weekly RXTE observa-tions of 1E 1048.1?5937allowed the changing pulsed ?ux and spin-down rate to be monitored on rela-tively short timescales.Gavriil &Kaspi (2004,here-after GK04)reported order-of-magnitude torque vari-ability on timescales of weeks to months that only marginally correlated with large luminosity variations.The unusually slow and long-lived pulsed ?ux “?ares”(not to be confused with the SGR giant ?ares;see Woods &Thompson 2006)that began in late 2001and lasted into 2004are in some ways unlike activity seen in any other AXP or SGR thus far.The 2?10keV spectrum and pulsed fraction are also variable on long timescales,the latter being anti-correlated with luminos-ity (Mereghetti et al.2004;Tiengo et al.2005).At near-infrared (near-IR)wavelengths,the ?ux varied dramat-ically during 2002?2003(Wang &Chakrabarty 2002;Israel et al.2002);unfortunately,the sparsity of near-IR observations prohibits measurements of the variability timescale,although a near-IR/X-ray anti-correlation has been suggested (Durant &van Kerkwijk 2005).Further activity in the form of another burst and small pulsed-?ux increase was seen in 2004(Gavriil et al.2006).

Here,we report on a series of simultaneous CXO and HST observations obtained during the course of 2006,as well as on archival CXO ,XMM and VLT observations,in §2.We also present RXTE timing results from the past three years.In 2007March,1E 1048.1?5937unexpect-edly entered yet another new phase of activity,and we report on Target-of-Opportunity observations with CXO and Swift taken after the 2007March event.In §3.1we discuss the 2004-2006period of radiative and spin quiescence in the context of both magentar and accre-tion models,and in §3.2,we describe and interpret the events following the 2007March ?are.

2.OBSERVATIONS,ANALYSIS AND RESULTS

As part of a long-term project,we regularly monitor AXPs using the Rossi X-ray Timing Explorer (RXTE ).1E 1048.1?5937is by far the most frequently observed AXP due to its relatively poor rotational stability.The RXTE observations (see §2.1)are crucial in measur-ing the spin evolution of the source and its pulsed ?ux.However,the high background and large unim-aged ?eld-of-view of RXTE make spectral measurements of 1E 1048.1?5937di?cult and total ?ux (and con-sequentially pulsed fraction)measurements impossible.To make these measurements,we obtained ?ve Chan-dra X-ray Observatory observations (CXO ;see §2.2)approximately equispaced throughout 2006.In order to probe the origin of the near-IR emission of this source,we observed with the Hubble Space Telescope (HST ;see §2.3)simultaneously with CXO .The motivation for si-http://www.physics.mcgill.ca/～pulsar/magnetar/main .

Fig. 1.—The long-term evolution of 1E 1048.1?5937’s pulsed properties.The three vertical dashed lines indicate the approxi-mate beginnings of the ?ux ?ares on (from left to right)2001Oc-tober 26,2002April 6,and 2007March 21,respectively.Horizon-tal dotted lines represent the value of the plotted parameter prior to all observed ?ares.(a )Spin frequency as observed by RXTE .Small points show the measured frequencies,while intervals over which phase coherence has been maintained are shown as thick lines (GK04;Dib et al.,in preparation).(b )Frequency derivative as a function of time.While the frequency was varying in 2002-2004and a phase-coherent timing solution could not be found,˙νwas de-termined in short intervals by calculating the slope of three consec-utive νmeasurements (after GK04).Since 2004,phase coherence has been maintained and ˙νre?ects the standard technique (Dib et al.,in preparation).(c )2?10keV rms pulsed ?ux as observed by RXTE .(d )Simulated total 2?10keV unabsorbed ?ux,shown as green points.Total ?ux is estimated from the RXTE pulsed ?ux and the power-law correlation between pulsed fraction and measured total ?ux,shown as red points,as described in §3.2.2.The simulated ?uxes (green )have been scaled to match the ac-tual measured total ?uxes (red ).(e )2?10keV rms pulsed fraction calculated using the method described in Woods et al.(2004).

multaneous CXO and HST observations was to search for correlated variability in the di?erent bands,as that provides insight into the physical mechanisms generat-ing the radiation.We also examined archival K S -band observations with the Very Large Telescope (VLT ;see §2.4),which contribute to our monitoring of its near-IR brightness.Following the 2007March glitch detected in our RXTE monitoring observations of 1E 1048.1?5937(Dib et al.2007b),we initiated three CXO and two Swift Gamma-Ray Burst Explorer (Swift ;see §2.2)Target-of-Opportunity (ToO)observations to follow the X-ray ?ux and spectrum of the source.In order to characterize the long-term evolution of 1E 1048.1?5937we also incor-porate and reanalyze archival XMM-Newton and CXO observations (see Gavriil et al.2006,for details on how these data were processed).Separate post-outburst op-tical and near-IR observations of 1E 1048.1?5937are reported in detail elsewhere (Wang et al.2007).

2.1.RXTE

We have observed 1E 1048.1?5937regularly since 1997with RXTE (Kaspi et al.2001;GK04).Our data were obtained using the Proportional Counter Array (PCA)on board RXTE which consists of ?ve identical

3 and independent Xenon/Methane Proportional Counter

Units(PCUs).We use our RXTE observations of

1E1048.1?5937to look for the presence of bursts(see

Gavriil et al.2004for details),to look for pulse pro?le

changes,to monitor its pulsed?ux,and to monitor its

frequency evolution using phase-coherent timing when

possible.

For the timing analysis,we created barycentered

lightcurves in the2?5.5keV band with31.25ms time res-

olution.As past monitoring has shown that it can be dif-

?cult to maintain pulse phase coherence over timescales

longer than a few weeks,observations of1E1048.1?5937

since2002are done three times per week,with the three

observations carefully spaced so as to allow a phase-

coherent analysis and a precise frequency measurement.

Thus,for each observation,we fold at the pulse period

determined via periodogram,cross-correlate the folded

pro?les with a high signal-to-noise template,and?t the

resulting phases with a linear function whose slope pro-

vides the average frequency.Frequencies determined in

this way are shown in Figure1.As can be seen in this?g-

ure,the2004-2006frequencies were actually much more

stable than in the past;we therefore attempted a fully

phase-coherent analysis as well.This will be described

elsewhere(Dib et al.,in preparation).Frequency deriva-

tives,which are also displayed in Figure1and are from

GK04and Dib et al.,(in preparation),are clearly much

more stable in the2004-2006interval than previously.

This period of rotational stability was accompanied

by X-ray pulsed?ux stability and relative quiescence:

the pulsed?ux time series of1E1048.1?5937in the

2?10keV band is presented in Figure1.The pulsed

?ux was calculated using a method similar to that de-

scribed in Woods et al.(2004)and is based on the rms

of the folded pro?le,but without variance subtraction.

The long period of rotational stability ended in2007

March,when a large glitch was observed(Dib et al.

2007b).The glitch also signaled the end of the period

of pulsed?ux quiescence.The pulsed?ux suddenly in-

creased by a factor of～3in the energy range2?10keV.

The upper limit on the rise time of the pulsed?ux for

this event is approximately one week.The peak pulsed

?ux reached by the source was～10%larger than the

peak?ux reached during the largest of the two previ-

ously observed?ares,and the rise time was at least4

times smaller.On2007May17,the date of the last

X-ray imaging observation included in this paper,the

pulsed?ux had decreased by～10%.It is as yet di?cult

to compare the decay timescale of the pulsed?ux of this

new event to that of the previous?ares however it will

be possible in the near future.

2.2.CXO and Swift

X-ray imaging observations were carried out with the

CXO and Swift telescopes.The date,total exposure

time and resulting count rate for each region-?ltered

background-subtracted observation are listed in Table1

and a detailed description of our spectral?tting is given

in§2.2.1.

1E1048.1?5937was observed?ve times8in2006with

8Although four simultaneous observations were originally

planned,HST experienced technical di?culties during one,result-

ing in a?fth CXO observation.

Fig.2.—The evolution of1E1048.1?5937’s X-ray spectral prop-

erties.For consistency,all the spectra from CXO,Swift,and XMM

were?t jointly with a two-component absorbed blackbody plus

power-law model,that produced a best-?t N H=(0.97±0.01)×

1022cm?2(see§2.2.1).The uncertainties shown re?ect statistical

errors only.Vertical dashed lines indicate the beginnings of the

?ux?ares(see Fig.1caption),while horizontal dotted lines indi-

cate quiescent values.(a)Blackbody temperature kT.(b)Photon

indexΓ.(c)Ratio of the2?10keV?ux contribution from the

blackbody and power law components.(d)Total unabsorbed?ux

in2?10keV.(e)2?10keV rms pulsed fraction calculated using

the method described in Woods et al.(2004).(f)For reference,we

show the2?10keV rms pulsed?ux as observed by RXTE.

CXO in quasi-equispaced intervals,and three times in

2007April.The observations were made with the Ad-

vanced CCD Imaging Spectrometer(ACIS)camera us-

ing the S3chip in continuous clocking(CC)mode.This

mode generates a1×1024pixel image,which is read out

every2.8ms;pile-up in this mode is not an issue.The

subsequent analysis was based on the“Level2”events

?les for which the event times are photon arrival times

(as opposed to readout times for the“Level1”events).

Following the standard threads we then corrected the

“Level2”events for certain caveats related to ACIS CC-

mode data9.We extracted a rectangular region centered

on the source with a width typically of～30pixels.We

estimated the background using rectangular regions on

either side of the source which extended from～10pixels

beyond the edge of the source region to～10pixels from

the edge of the image.Following the standard threads,

the source and background spectra were then extracted

using CIAO10v3.4.We grouped the source spectrum such

that there were no fewer than20counts per bin after

background subtraction.The response matrix function

(rmf)and area response function(arf)for each observa-

tion were also generated using CIAO and CALDB.We used

Timed Exposure(TE)mode response matrices because

9e1ff7b4d767f5acfa1c7cde8/ciao3.4/why/ccmode

10e1ff7b4d767f5acfa1c7cde8/ciao

4

of the absence of available spectral calibration for CC mode.1E 1048.1?5937was observed three times 11in 2007with Swift using the X-ray Telescope (XRT).XRT has two observing modes:photon counting (PC)and windowed-timing (WT).PC mode provides a ～600×600pixel 2image with low (～2.5s)time resolution,whereas WT mode provides a one-dimensional ～200-pixel wide image with high (～0.74ms)time resolution.XRT au-tomatically switches from WT to PC-mode when the count rate exceeds ～2counts s ?1.In our analysis we ignored the WT mode data as they were too short to be of use.We used the PC-mode event ?les from the standard pipeline;these were cleaned and e1ff7b4d767f5acfa1c7cde8ing FTOOL xrtcentroid ,we determined the centroid of the image,and input the cleaned,barycentered events into the command line interface xselect 12.We selected a circular source region with a 25pixel radius centered on the centroid of the image.Since the source was slightly piled up,we excluded a 4pixel radius circular region centered on the centroid of the image.We used 4pixels because that is the recommended exclusion radius for a source with 1E 1048.1?5937’s spectrum (L.Angelini,pri-vate communication).For the background,we selected events from an annulus with an inner radius of 50pixels and outer radius of 130pixels.We further ?ltered our events by selecting only those that had grades 0to 12.With xselect we then created source and background PHA ?les.The FTOOL xrtmkarf was used to create the arfs,and we input the source PHA ?le with the excluded center so that it would be corrected for pile-up.We used the ?le provided by CALDB appropriate for PC mode data events of grades 0to 12to make the rmf.

2.2.1.Global Spectral Fit We extracted spectra of 1E 1048.1?5937from all archival,monitoring,and ToO observations collected from CXO ,XMM and Swift .A description of the re-cent data analysis is in §2.2,and the archival anal-ysis was described in Gavriil et al.(2006).Using the ?tting package XSPEC 13v12.

3.1we modeled the X-ray spectra with a photoelectrically absorbed black-body plus power law.We used the phabs photoelec-tric absorption XSPEC model,which assumes the so-lar abundance table of Anders &Grevesse (1989)and uses the bcmc photoionization cross-section table from Balucinska-Church &McCammon (1992)with the new He cross-section from Balucinska-Church &McCammon (1998).We ?t this model to all the observations si-multaneously,allowing the column density N H to vary but with the sole constraint that it be the same for all observations.We restricted our ?t to the 0.7–5keV band.The total global ?t had χ2ν=1.07for ν=4200degrees of freedom (dof).Inpidually,the ob-servations were equally well modeled (except for a rel-atively high χ2value in one observation,see §2.2.2).Our best-?t column density obtained from the global ?t is N H =(0.97±0.01)×1022cm ?2,in agreement with Durant &van Kerkwijk (2006b).The results of our 11Two additional Swift observations taken contemporaneously were omitted from our analysis because their short exposures yielded prohibitively large uncertainties.12e1ff7b4d767f5acfa1c7cde8/docs/software/lheasoft/ftools/xselect/xselect 13e1ff7b4d767f5acfa1c7cde8 spectral ?tting are listed in Table 1and plotted in Fig-ure 2. 2.2.2.Possible Spectral Feature

All observations had spectra that were well modeled by a photoelectrically absorbed blackbody plus power law.However,the CXO observation on 2007April 6(obser-vation ID 7647),the ?rst after the 2007March event,had a relatively high χ2ν=1.33(ν=353dof).There is possible evidence for an absorption line at ～2.7keV (see Fig.3).Adding a Gaussian line improved the ?t (χ2ν=1.20for ν=351dof)with ?χ2=48.3.With the line width ?xed at 0.1keV,we measure a line energy of 2.73±0.03keV.Allowing the line width σto be a free parameter,we obtain σ=0.16±0.03keV,a line energy of 2.74±0.03keV and χ2ν=1.19for 351dof.We ?nd no evidence of a phase dependence for this possible line.A detailed phase-resolved spectroscopic analysis will be presented in a forthcoming paper.To test the signi?cance of adding such a line,we per-formed the following simulation.We generated 10000simulated spectra by adding Poisson noise to a black-body plus power-law model spectrum having the same parameters as our best-?t model.We then determined the maximum change in χ2after adding a Gaussian line.To avoid local minima and ensure that we found the true minimum χ2for each simulation iteration,instead of ?t-ting for the peak energy,we stepped through di?erent line energies between 0.6and 7.0keV with a step size of 0.025keV.The width of each line was held ?xed at 0.1keV and its normalization was allowed to vary.In all the simulation iterations,none had a change in χ2greater than 48.3.The probability that this feature is due to random chance is<0.13%,accounting for the number of trials.We note however that with the addition of the line,the overall ?t is still unacceptable;we speculate that other features may be present as well,though at marginal signi?cance.For example,there is a small but intriguing feature at ～1.3keV,which is approximately half the en-ergy of the line discussed above.The evidence for a line must be considered tentative because CC mode,which uses the TE mode response matrices,is not spectrally calibrated,and there exist calibration lines between 1.7and 3keV in the TE response.However,this is true for the other CXO observations,including those in which we ?nd no such features.2.2.3.Pulse Morphology and Pulsed Fraction Study Using our CXO and Swift observations we were also able to study the source’s pulse morphology,?ux and pulsed fraction.For the CXO observations,we barycentered our cor-rected Level 2event lists and then extracted background-subtracted light curves in the di?erent bands using the same source and background regions as for the spectral analysis.We folded these lightcurves at the optimal fre-quency as determined by a periodogram.The frequencies we obtained all agreed with the ephemerides determined with our contemporaneous RXTE data.Figure 4(left )displays the CXO pulse pro?les in the 1?3and 3?10keV bands.Notice that after the ?are (last 3panels in Fig-ure 4left ),the pulse pro?le changed from single to at least triple peaked.To study the pulse morphology evo-

lution quantitatively we decomposed the pulse pro?les

5

TABLE1

CXO,XMM and Swift observing parameters and results

Date MJD Obs.ID Exposure Count rateΓa kT a Unabs.Flux a,b F PL/F BB a,c Pulsed

(s)(counts s?1)(keV)(10?12erg cm?2s?1)Fraction d

Archival XMM Observations

Archival CXO Observations

Monitoring CXO Observations

ToO CXO Observations

ToO Swift Observations

√B,and the peak-to-peak

pulsed fraction is given by P F pp=A

6

0.000.020.040.06

MJD 53196

MJD 53201

0.000.02

0.040.06

MJD 53792

MJD 53845

0.000.020.040.06

MJD 53900

MJD 53946

0.000.020.040.06

MJD 54001

MJD 54196

0.00

0.020.040.06

0.0

0.5

1.0

1.5

MJD 54206

0.0

0.5

1.0

1.5

Phase (Cycles)N o r m a l i z e d P u l s e d A m p l i t u d e

MJD 54218

0.00

0.05

MJD 53196

MJD 53201

0.00

0.05MJD 53792

MJD 53845

0.00

0.05

MJD 53900

MJD 53946

0.00

0.05

MJD 54001

MJD 54196

0.00

0.05

5

10

MJD 54206

5

10

Harmonic (n)

P o w e r (P n /P 1)

MJD 54218

Fig.4.—Left :1?3keV (blue )and 3?10keV (red )normalized pulsed pro?les as observed by CXO .The pro?les are phase aligned,have had their minimum bins subtracted,and are normalized such that the area underneath them is unity.Note that the 2007event occured around MJD 54183.Right :1?3keV (blue )and 3?10keV (red )Fourier decomposition of the pulse pro?les.The powers P n are plotted in terms of their ratio to the fundamental P 1.The n =0and the n =1(fundamental)components are excluded.Notice how the pulse pro?le has additional structure after the glitch/?ux enhancement.

count rate,what its true pulsed fraction would be as-suming a pulse pro?le with as much harmonic content as the nearest CXO observation.On average we found a reduction of no more than ～73%.Note in Figure 2the observed tight correlation between total ?ux and pulsed fraction (see also Tiengo et al.2005)which we discuss in detail in §3.2.2.

2.3.HST

We observed 1E 1048.1?5937using HST simultane-ously with CXO on four occasions in 2006:February 26,April 20,July 30,and September 23(Program ID 10761).Observations were made with the Near Infrared Camera and Multi-Object Spectrometer (NICMOS)instrument,a 256×256square pixel HgCdTe e1ff7b4d767f5acfa1c7cde8bined with camera 3,it provided a focus ratio of f /17,a ?eld of view of 51′×51′,and a plate scale of 0.′′2pixel ?1.The detector was read out in MULTIACCUM mode,and a spiral dither pattern with 5′′spacings was applied.We observed using the ?lter F110W (similar to the ground based J -band ?lter)on two occasions and F160W (sim-ilar to H -band)on four occasions.The average FWHM of the point spread function (PSF)was ～0.′′4for all NICMOS observations.Observing parameters are sum-marized in Table 2.

The data underwent “On-The-Fly-Reprocessing”in-volving standard pipeline routines before being retrieved from the HST Archive.The calibration routine calnica performed basic data reduction steps,and calnicb pro-duced a single ?nal image for each observation,com-

Fig.5.—The evolution of 1E 1048.1?5937’s near-IR brightness.VLT and HST results from this work fall after 2004,while data prior to that are from previous literature (Wang &Chakrabarty 2002;Israel et al.2002;Durant &van Kerkwijk 2005).We also show the corresponding magnitude on the right axis;however the translation is not exact due to HST ’s unique ?lters.Vertical dashed lines indicate the beginnings of the ?ux ?ares (see Fig.1caption),while horizontal dotted lines indicate quiescent values.(a )“J -band”?ux νF ν.(b )“H -band”?ux νF ν.The detection made from combining three HST observations (see §2.3)is shown in grey.(c )“K S -band”?ux νF ν.Results from 2007are described in detail elsewhere (Wang et al.2007).(d )For reference,we show the 2?10keV rms pulsed ?ux as observed by RXTE .

7 bining the dithered mosaic input images.DAOPHOT

(Stetson1987)for IRAF v2.12.2was used to perform

PSF photometry.Absolute calibration of our measure-

ments was done using the NICMOS Photometric Key-

words15for which the uncertainties are believed to be

less than5%;we found the aperture correction to corre-

spond with the calibration keywords by simulating PSFs

with the Tiny Tim16software package.

In order to determine limiting instrumental magni-

tudes in each of the observations,we performed PSF

photometry on a set of a hundred trial images to which

a single arti?cial star was added.This arti?cial star was

placed in a0.′′3box in a blank region near the nominal

position of the IR counterpart.By varying the bright-

ness of the star,we de?ne the3σdetection limit at the

instrumental magnitude for which the PSF photometry

recovers the arti?cial star with an uncertainy of0.3mag

(see Hulleman et al.2000).

At the position of1E1048.1?5937,as measured by

Wang&Chakrabarty(2002)in K S-band with the Mag-

ellan telescope,we?nd one point source in the F160W

image from2006February26only.To con?rm the po-

sitional coincidence of this source with the Magellan ob-

ject,we astrometrically tied N=31nearby?eld stars

in our HST image to those from the Magellan image.

A～0.′′05o?set was found for the1E1048.1?5937

candidate.The Magellan source’s relative positional

uncertainty,given its0.′′4FWHM radius,is roughly

√

FWHM/

8

TABLE2

HST and VLT observing parameters and results

Date MJD Exposure Filter a Limiting DetectionνFνb

(s)Magnitude(10?15ergs s?1cm?2)

VLT Observations

2005Apr1152324900K S>20.3···<7.0

2005Apr2953471900K S>20.6···<5.3

2005May2753489900K S>21.221.0(3)<3.1;3.7±1.0

2005Jun653527900K S>21.4···<2.6

HST Observations

2006Feb26537922637F110W>25.0···<0.37

2006Feb26537921196F160W>23.722.70(14)<0.49;1.3±0.2

2006Apr20538452637F110W>24.8···<0.45

2006Apr20538451196F160W>23.1···<0.89

2006Jul30539461037F160W>23.3···<0.74

2006Sep23540011037F160W>22.9···<1.0

Note.—Magnitude uncertainties re?ect errors determined by DAOPHOT;upper limits are

3σlimits,as de?ned in the text.Measured quantities are“observed”,ie.have not been corrected

for reddening/extinction e?ects.

a Although the?lter system of HST does not precisely match standard ground-based near-IR

?lters,note that F110W≈J-band and F160W≈H-band.b To convert standard IR magnitudes

to?ux,we take as the K S=0mag zero pointνFν=9.28×10?7ergs s?1cm?2,derived from

Cox(2000).HST?uxes are determined from NICMOS Photometric Keywords.

tosphere(Eichler et al.2002;Beloborodov&Thompson 2007),but so far this has not be con?rmed.Are the low-and high-energy emission mechanisms intimately connected(Heyl&Hernquist2005)?Correlated post-burst?ux decay in the X-ray and near-IR regimes has been seen in at least one other AXP(Tam et al.2004). Given the many observational properties that have been characterized for1E1048.1?5937in this work and for other AXPs in similar studies,which property will emerge as being the most constraining of physics is hard to know.The most promising behaviors are those which show clear correlations with others,or those that are common to many or all AXPs and SGRs.In this section, we discuss the behavior of1E1048.1?5937,and focus on the phenomena that are potentially the most useful for testing the magnetar or other competing models.We ?rst consider the quiescent phase we have observed in 2004-2006,and its implications for AXP models,then subsequently discuss the source’s return to activity in 2007March.

3.1.The2004-2006Quiescent Phase

3.1.1.X-ray Flux and Spectrum

The X-ray?aring observed pre-2004contrasts strongly with the stable pulsed?uxes we observed in2004-2006 (Fig.1c).Clearly the source’s stable state is also its faintest.The X-ray spectrum during this quiescent phase is also fairly constant,with kT evolving post-?are,on a time scale of several years.Interestingly,kT does not return to its pre-?are value as measured in2000Decem-ber(see Fig.2a).Similarly,the photon index(Fig.2b) decreased in2004-2006,away from its softer pre-?are value.Meanwhile the pulsed fraction(Figs.1e and2e) was slowly rising in2004-2006,as the source grew fainter, as if,unlike kT andΓ,it was slowly recovering to the pre-?are value.Thus,overall,the quiescent period is char-acterized by slow evolution,on timescales of years,in which the source?ux and pulsed fraction slowly relaxed back to their pre-?are values,while the source spectrum varied signi?cantly(Fig.2c)and did not relax back to its pre-?are state.This is suggestive of a?xed energy loss rate in quiescence,though perhaps with a di?erent magnetospheric current con?guration,which impacts the surface emission via return currents(Thompson et al. 2002).We note,on the other hand,that the2006and pre-2001pulse pro?le were very similar if not identical (Kaspi et al.2001),and that the Thompson et al.(2002) simple prediction for the decay time of the sort of magne-tospheric twist required for1E1048.1?5937is an order of magnitude too large for standard parameters.The evolution could also be purely due to the thermal com-ponent(¨Ozel&Guver2007).

3.1.2.Timing Stability

Simultaneous with the slow evolution of the X-ray?ux, pulsed fraction,and spectrum,the pulsar’s rotational be-havior clearly stabilized,with the source spinning down relatively smoothly at a value close to the long-term aver-age in2004-2006(Fig.1a).Whereas previously this AXP distinguished itself from others by defying attempts at phase-coherent timing and exhibiting large torque vari-ations(Fig.1b),in2004-2006,1E1048.1?5937resem-bled,from a timing point of view,other AXPs which are relatively stable rotators,at least when not glitching (e.g.Kaspi et al.1999;Dib et al.2007a).This behavior demonstrates a clear relationship between timing and ra-diative properties,and that the“noise”seen in the fre-quency derivative during2001-2004is likely physically di?erent from the“timing noise”observed ubiquitously in radio pulsars and in otherwise radiatively stable AXPs. In the context of the magnetar model,magnetospheric activity can account for both torque and X-ray?ux vari-ability,although the former is most sensitive to currents anchored closest to the magnetic poles,so that only a broad correlation between?ux and torque is expected (Thompson et al.2002)and indeed is seen when consid-ering many AXPs and SGRs(Marsden&White2001). That the torque and luminosity in1E1048.1?5937 do not vary simultaneously or in a clearly correlated way is thus not necessarily problematic if the magne-tospheric current con?guration is changing,although it does demonstrate that the magnetar model in this par-ticular regard is not strongly predictive.

9

Although the magnetar model is favored,it has also been suggested that accretion from a fossil debris disk could explain AXP spin characteristics(Chatterjee et al. 2000;Alpar2001)as well as all aspects of the broad band emission(Ertan&C?al??s kan2006).While it is now ev-ident that accretion alone cannot be responsible for all observed properties,most notably the energetic X-ray bursts,a“hybrid”model has been invoked that puts a thin debris disk around a highly magnetized pulsar (Ek?s?&Alpar2003).In this hybrid case,both the per-sistent luminosity and pulsar spin down are related to the mass transfer rate,˙M.According to GK04,who compare1E1048.1?5937’s torque changes with changes in RXTE pulsed?ux,the scale of their variability does not obey an expected relationship for an accreting pul-sar undergoing spin-down,thus presenting a challenge to models of fossil disk accretion.It could be argued, however,that since only the pulsed?ux P X was being monitored,the total X-ray luminosity L X is still an un-known quantity.We address this by using the quanti-tative correlation we establish between pulsed fraction P F and total unabsorbed?ux F X(discussed in§3.2.2 and shown in Fig.6)to simulate a well sampled set of phase-averaged?ux data as a function of RXTE pulsed ?ux P X,given that P F is simply P X/F X.The resulting “new”total?ux,F X,is shown in Figure1d,along with ˙νin Figure1b for comparison.At its most variable in 2002-2004,the absolute value of˙νchanged by a factor of>10in less than a year,while the maximal change in total unabsorbed?ux is a factor of～6from peak to qui-escence.Note that these variations are not simultaneous. For a pulsar experiencing a spin-down torque due to mass accretion while in a quasi-equilibrium“tracking”phase (i.e.the AXP phase),a strong correlation described by L X∝|˙ν|7/3can be derived from Chatterjee et al.(2000, equation3),where the magnitudes of torque and˙νare proportional,the radius of the magnetosphere is de?ned by the Alf`e n radius which is also a function of˙M,and L X∝˙M.Thus,a factor of>10change in˙νshould be re?ected by a factor of>200simultaneous change in L X,and thus F X,if this model is correct.Clearly this is not observed,rendering this particular accretion scenario unlikely.This point is further emphasized by the signi?-cant time o?set between the changes in torque and?ux. Overall,the fact that1E1048.1?5937’s long history of highly irregular spin-down is mirrored by trends in its X-ray emission only in a broad,rather than strict,sense strongly suggests that active accretion is not happening in this case.

3.1.3.Near-IR Quiescence

Our original purpose for making simultaneous CXO/HST observations was to compare low-level near-IR and X-ray?ux and spectral changes,with each observed with the same instrument,in order to look for correlations.Given that the near-IR source had faded considerably in the HST observations,render-ing it only marginally detected in2006,this was not e1ff7b4d767f5acfa1c7cde8pared to the handful of near-IR detec-tions made throughout2002-2003(see Fig.5)it is clear that2005-2006marked a period of near-IR quiescence in 1E1048.1?5937.At the time of the2005VLT observa-tions in which the AXP was faintly seen in K S-band,its ?ux was consistent with the last detection in mid-2003. With HST in2006we?nd that its H and J magnitudes have dropped lower than ever before observed.There is marginal evidence for near-IR?ux variability during qui-escence on comparably short timescales to the variabil-ity in X-ray spectral parameters,although closely spaced near-IR monitoring observations are required to con?rm this.

That the near-IR faded roughly in concert with the X-ray?ux is notable,suggesting a correlation similar to that seen in AXP1E2259+586(Tam et al.2004). This is discussed further below.However such a cor-relation can be argued to be expected in both the magnetar and disk models.In the magnetar model it would imply that the near-IR emission is magnetospheric and hence varies,as do the X-rays,when the magne-tospheric con?guration varies(Heyl&Hernquist2005; Beloborodov&Thompson2007).In the disk model,a correlation between near-IR emission and X-ray?ux is naturally expected since the putative disk is heated via X-ray illumination(Ertan et al.2006;Wang et al.2006), however why the torque should have varied strongly and non-simultaneously is a puzzle(see above).

In any case,the concurrence of the near-IR and X-ray quiescence and the timing stability is unlikely to be an accident.Similarly,the presence of the two?ares and the subsequent strong torque changes are also not likely to be by chance.

3.2.The2007March Event and its Aftermath

1E1048.1?5937reactivated in2007March,when a sudden pulsed?ux increase was observed(Fig.1c),ac-companied by a large spin-up glitch,the details of which will appear elsewhere(Dib et al.,in preparation).Our follow-up X-ray observations with CXO and Swift show that the spectral parameters,pulse shape,total?ux,and pulsed fraction are drastically changed since this?are oc-curred(Figs.2and4).In particular,the total X-ray?ux was>7times greater in the2?10keV band compared with quiescence,while the pulsed fraction decreased from ～75%to～20%.Compared to the previous long-term ?ares from1E1048.1?5937,the onset of this event took place more quickly by a factor of>4,the peak pulsed?ux is～10%greater,and the“high”state appears to be last-ing longer.Optical and near-IR follow-up with the Mag-ellan Telescope and VLT(Wang et al.2007;Israel et al. 2007)reveal an increase in both I and K S-band of～1.3 mag.Very recently,Rea et al.(2007)reported on XMM ToO observations,and revealed that nearly three months after the?are onset,the total X-ray?ux has decreased slightly but is still～5times brighter than in quiescence.

3.2.1.Near-IR Enhancement Previously,an anti-correlation between near-IR and X-ray?ux in this AXP was suggested,based on the correspondence of the highest near-IR detection with a low point in X-ray pulsed?ux17(Durant&van Kerkwijk 2005).Given the new measurements from both before and after the recent event,such an anti-correlation is highly questionable.The signi?cant rise in near-IR?ux 17Note that an error in Figure4of Durant&van Kerkwijk (2005)places the brightest K S?ux on the wrong date;we have corrected this in our Figure5c.

10

now appears correlated with the most recent X-ray?are. Interestingly,in2002,a high near-IR detection was be-tween the two X-ray?ares;this suggests that the near-IR emission is not exactly correlated with the X-rays,but rather an enhancement can precede,follow,or last longer than an X-ray enhancement.A near-IR brightening pre-ceding an X-ray enhancement would rule out illuminated disk models for this source;frequent mid-or near-IR ob-servations are required to check this.As discussed above, AXP1E2259+586also demonstrated strongly correlated near-IR and X-ray?ux decay following its large2002 outburst(Tam et al.2004),but in contrast,4U0142+61 was highly variable in the near-IR despite X-ray stability (Durant&van Kerkwijk2006c),and has shown no ev-idence of near-IR changes coinciding with X-ray bursts (Gonzalez et al.,in preparation).Note that the claim of correlated?ux decay in the case of XTE J1810?197 (Rea et al.2004)is under dispute(Camilo et al.2007). Such inconsistent behaviour is puzzling for both the mag-netar and disk models.

3.2.2.Correlation between Pulsed Fraction and Total Flux One conclusion to be drawn from Figures1d and1e is that1E1048.1?5937’s pulsed fraction P F is strongly anti-correlated with the total X-ray?ux F X,a trend al-ready noticed by Tiengo et al.(2005)and Gavriil et al. (2006).Now,with our larger sample of data covering a wider dynamic range,we can quantify this correla-tion:we?nd a power-law dependence of P F∝F n X where n=?0.46±0.02(see Fig.6),with F X in the 2?10keV range.In?tting,we set the index as a free parameter,ie.not tied to any model;therefore,some scatter is expected,as evident in the largeχ2ν=9.36 produced by the?t(ν=14dof).However,in?nd-ing the uncertainty on n,we have scaled the P F data uncertainties to forceχ2ν=1,e?ectively scaling upσn as well.A similar correlation has been proposed in an-other AXP,1RXS J170849.0?400910(Dib et al.2007a), although in that case it is such that the pulsed?ux re-mains nearly constant in the presence of>50%total?ux changes(Campana et al.2007).

How could an increase in phase-averaged?ux be met with a simultaneous decrease in pulsed fraction?A grow-ing hot spot on the neutron-star surface,a result of ei-ther a changing magnitude or con?guration of returning magnetospheric currents or internal processes,could at least in principle produce such an e?ect,if the initial hot spot size were large enough that its size as viewed from a distant observer were a signi?cant fraction of the stellar surface.We note that¨Ozel&Guver(2007)suggest that magnetar afterglows may be dominated by surface ther-mal,rather than magnetospheric,changes;detailed simu-lations of AXP pulsations using gravitational light bend-ing and appropriate radiation beaming functions(e.g. Dedeo et al.2001)in addition to magnetospheric scat-tering(Thompson et al.2002;Fern′a ndez&Thompson 2007)are needed to see if the observed correlation can be reproduced.

3.2.3.Correlation between Hardness and Total Flux

A prediction made by the twisted magnetosphere model of Thompson et al.(2002)is that enhanced emis-sion should also be spectrally harder,since both are asso-ciated with larger magnetospheric twist angles.Indeed,we observed a correlation between the total X-ray?ux and spectral hardness:see Figure6.Similar behaviour has also been reported in1RXS J170849.0?400910 (Rea et al.2005a;Campana et al.2007).This can be interpreted as a con?rmation of an important magnetar model prediction,namely that both the spectral hard-ness and the total?ux should increase for an increas-ing magnetospheric twist angle(Thompson et al.2002). Hardening of the spectrum might also be an expected e?ect of increased luminosity if a large injection of ther-mal seed photons is repeatedly up-scattered due to reso-nant cyclotron scattering feedback processes in the mag-netosphere,thereby shifting the photon energies higher. Recently¨Ozel&Guver(2007)have suggested that the hardness-intensity correlation in magnetars is a result of purely surface thermal changes,with no change in the magnetospheric con?guration.It would be interesting to apply their model to the1E1048.1?5937data as they span a much greater dynamic range in?ux than did the sources studied by¨Ozel&Guver(2007).This is beyond the scope of our paper,however.

3.2.

4.2007Pulse Pro?le Changes

As seen in the CXO data obtained after2007March, the pulse pro?le of1E1048.1?5937changed abruptly after the X-ray?ux enhancement and glitch.Several new harmonics are clearly visible,and are present in a largely energy-independent way(Fig.4).This extra power in the higher harmonics clearly signals a com-plication of the surface and/or magnetospheric con?g-uration.This change contrasts with that seen after the 2002outburst of1E2259+586,which involved mainly an exchange of powers between the fundamental and?rst harmonic,with higher harmonics remaining relatively unchanged(Woods et al.2004).Also,the change con-trasts with that seen during the1998SGR1900+14giant ?are,in which the SGR pulse pro?le simpli?ed greatly immediately post-?are(G¨o?g¨u?s et al.2002).However, the opposite occured followed the2004giant?are from SGR1806?20(Palmer et al.2005),when a sinusoidal-to-complex evolution in pulse morphology took place, similar to that of1E1048.1?5937.This persity of behaviors in magnetars post-outburst is problematic for understanding the underlying physics;indeed it indicates a wide variety of phenomenon phase space,which itself must be explained by models.The peculiar suppression of the4th harmonic in the post-glitch1E1048.1?5937 pro?le is particularly puzzling to us;perhaps it indicates an important and unchangeable symmetry in the com-bined emission and viewing geometries,or perhaps it is a chance occurence and will not occur again in future events.

3.2.5.2007April Spectral Feature?

In the CXO observation made on2007April6,the one closest to and just after the2007March event, the X-ray spectrum of1E1048.1?5937could not be well?t by a simple continuum model,in contrast with the other datasets.We identi?ed a possible absorp-tion line near2.7keV,though we note that the spec-tral model including this line is still not a good?t to the data.This suggests that other low-level features may be present as well,including a possible feature at

11

Fig.6.—X-ray ?ux dependent properties of 1E 1048.1?5937,from 6years of observations with CXO (squares ),XMM (triangles ),and Swift (circles ).Left:Pulsed fraction vs.total unabsorbed ?ux,both in 2?10keV.The solid black line shows the best-?t power law that describes the correlation between the two quantities.Right:Hardness ratio vs.total unabsorbed 2?10keV ?ux.The hardness ratio is de?ned as (S ?H )/(S +H )where S and H are the phase-averaged count rates in the 1?3and 3?10keV bands,respectively.roughly half the above line’s energy,～1.3keV.The pu-tative 2.7keV line joins the growing list of puzzling and sometimes statistically marginal AXP/SGR spec-tral features (Ibrahim et al.2002,2003;Rea et al.2003,2005a).Examples of some feature detections that are clearly statistically signi?cant are,interestingly,repeat-edly near ～13keV,e.g.in 1E 1048.1?5937(Gavriil et al.2002,2006),in XTE J1810?197(Woods et al.2005),and very recently AXP 4U 0142+61(Gavriil et al.2007,and Gavriil et al.,in preparation).The feature we see in the 2007April data for 1E 1048.1?5937are unlike any of these.It is perhaps more reminiscent of the absorp-tion features seen in some “isolated neutron stars”(e.g.Bignami et al.2003;van Kerkwijk et al.2004),and,if similarly interpreted as an electron cyclotron line,im-plies a magnetic ?eld of ～3×1011G,or ～6×1014G for a proton cyclotron line.The latter value is of course more in line with the ?eld expected near the surface of a magnetar.However,the feature could also be an atomic transition (e.g.Mori &Ho 2007);it is di?cult to know given the present data.Regardless,the spectrum 10days later,on 2007April 16,showed no such features,indicat-ing that whatever their origin,it had subsided,and on a shorter timescale than that of any pro?le or ?ux relax-ation.

e1ff7b4d767f5acfa1c7cde8parison of the 20071E 1048.1?5937Event with

Other Similar Events

The 2001-2002?ares observed in 1E 1048.1?5937are unique in that they had very well resolved rise times of many weeks (GK04).The 2007April event,by contrast,involved an X-ray pulsed ?ux enhancement that rose at least 4times faster.In this sense it is perhaps more sim-ilar to the 2002outburst of 1E 2259+586(Kaspi et al.2003),although in that case a brief rise could have been missed by the sparse monitoring.During that outburst,1E 2259+586exhibited an order-of-magnitude increase in pulsed and phase-averaged ?ux,as well as numerous other changes including (but not limited to)～80SGR-like bursts.Similarly to 1E 1048.1?5937,both the to-tal and pulsed ?ux increased during the 1E 2259+586outburst,and the pulsed fraction decreased,going from 23%in quiescence down to 15%(Woods et al.2004).However,the pulsed fraction was not as tightly cor-related to the pulsed ?ux:the pulsed fraction recov-ered within ～3days (Zhu et al.2007),whereas the ?ux took several months to return to its pre-outburst value (Woods et al.2004).In terms of energetics,the total energy in 2?10keV released in the 1E 2259+586out-burst was 3×1039erg and 2×1040erg during the rapid and gradual decay components,respectively (see Table 5of Woods et al.2004).For 1E 1048.1?5937,we estimate that up to 2007April 28,the date of the last CXO obser-vation,the total energy emitted is comparable,roughly 7×1039erg (2?10keV),based on the pulsed fraction-total ?ux correlation and assuming a distance of 2.7kpc (Gaensler et al.2005),or 8×1040erg assuming 9.0kpc (Durant &van Kerkwijk 2006a).Near-IR ?ux variabil-ity,possibly (in the case of 1E 1048.1?5937)and likely (in 1E 2259+586)correlated to X-ray ?ux,was observed following both events.That the 20071E 1048.1?5937event clearly involved a large rotational glitch,as did the 20021E 2259+586event,is also a commonality,al-though glitches at the time of the 2001-2002?ares could easily have been missed because of the large amount of timing noise and our inability to phase connect the data at that time (Dib et al.,in preparation).SGR-like bursts from 1E 1048.1?5937have not been observed thus far in 2007,but that does not preclude their exis-tence given that our observing duty cycle is so low.Sim-ilar to 1E 1048.1?5937,1E 2259+586exhibited a pulse pro?le change,albeit one involving only the fundamen-tal and ?rst harmonic,in constrast to the appearance of higher harmonic structure in 1E 1048.1?5937.Finally,as in 1E 1048.1?5937,spectral hardening was observed during the 1E 2259+586?are.Thus overall,the 20071E 1048.1?5937event has practically all properties con-sistent with the 20021E 2259+586event,but di?ers from

12

its earlier?ares primarily by rise time.

The transient AXP XTE J1810?197presents an-other interesting example for comparison.This AXP,which“turned-on”in late2002or early2003 (Ibrahim et al.2004),has been steadily declining in X-ray?ux since,reaching apparent quiescence in2005-2006(Gotthelf&Halpern2007).Unfortunately,very little is known about the onset of enhanced emission, ie.,whether it was accompanied by a glitch/burst,or what the timescale of the rise was.While the ex-ponential decay timescale of several years does resem-ble that of1E1048.1?5937’s2001-2002post-?are be-havior,how the most recent?are will fade is as yet undetermined.Furthermore,the peak and quiescent ?uxes of XTE J1810?197di?ered by a much greater amount,nearly2orders of magnitude(Gotthelf et al. 2004),while the recent total2?10keV?ux increase in1E1048.1?5937was by a factor of～7.Spectrally, XTE J1810?197was harder in outburst and softer in qui-escence,similar to1E1048.1?5937,when modelled as2-temperature blackbody(Gotthelf&Halpern2007)or as a blackbody plus power law(W.Zhu,private communi-cation).One major distinction,however,is the clear ev-idence for a positive correlation between pulsed fraction and total?ux in XTE J1810?197(Gotthelf&Halpern 2007).While it may be tempting to link the transience of XTE J1810?197to1E1048.1?5937-like post-?aring behaviour,the dissimilarities are important and remind us that we may be comparing mere coincidences. Compared to the SGR giant?ares,rare events in which an enormous amount of broad band energy (1044?1046erg;Hurley et al.2005)is output,the 1E1048.1?5937event is orders of magnitude less en-ergetic.Giant?ares release most of their energy in a short(<1s),initialγ-ray spike,a property not observed in this event,although a small initial burst could have been missed.Furthermore,the2007event is prolonged, and while giant?ares do gradually decay,the amount of energy in the tail is usually small compared to that of the spike(Woods et al.2004;Hurley et al.2005).

On the other hand,the gradual evolution of the persis-tent properties seen before and after the2004December giant?are of SGR1806?20do show some resemblance to1E1048.1?5937’s behaviour.During this pre-?are period of enhanced burst activity in SGR1806?20that began in mid-2003,the pulsar torque,pulsed?ux,to-tal?ux,and hardness increased on a～year timescale, peaking several months before the giant?are occurred in 2004December,and continuing to decline well after the ?are epoch(Woods et al.2007).The gradual changes, particularly the well-resolved factor of>2increase in total unabsorbed?ux,perhaps resemble the2001-2002 events of1E1048.1?5937more than the2007event. As with1E1048.1?5937,a correlation was found be-tween spectral hardness and intensity in SGR1806?20, and there appeared to be some sort of physical con-nection between frequency derivative(or torque)evo-lution and variability in the spectrum and pulsed?ux (Mereghetti et al.2005;Woods et al.2007).However,in SGR1806?20,there was no evidence for a correlation between the phase-averaged?ux,which likely peaked around2004October,and pulsed fraction,which was sta-ble in the?ux-enhanced pre-?are period but was lower than average in early2005when the persistent?ux had approached quiescent levels(Rea et al.2005b).Further-more,SGR1806?20’s pulse pro?le was noticeably more sinusoidal while brightest(Woods et al.2007),although the pulse pro?le likely changed as a direct results of the giant?are,rather than the slow evolution of persistent ?ux.

4.CONCLUSIONS

We have considered～10yrs of multiwavelength obser-vations of1E1048.1?5937.The source,in2004-2006, gradually relaxed apparently back to quiescence in most respects,only to“awaken”suddenly in the glitch/?are event of2007March.By observing1E1048.1?5937 while quiescent,it has become apparent that previous instabilities in timing,X-ray and near-IR?ux were likely all linked to the major long-term?aring events of2001-2002,whose nature is yet unknown but likely includes changes in the stellar magnetosphere’s current con?gu-ration.The asynchronous spin and X-ray?ux variability we have observed is incompatible with expectations of the fossil disk accretion model.Following the most re-cent2007event,we observed total X-ray?ux variability that is strongly correlated with X-ray pulsed fraction,X-ray spectral hardness,and changes in near-IR?ux.These observations largely agree with expectations of the mag-netar model.To date,the1E1048.1?59372007event is ongoing,as is continued multiwavelength monitoring.

We thank Joe Hill and Lorella Angelini for their help with the Swift XRT data analysis,and Zhongxiang Wang for his help with the near-IR analysis.We thank the anonymous referee for many judicious comments that have improved the quality of the paper.FPG is sup-ported by the NASA Postdoctoral Program administered by Oak Ridge Associated Universities at NASA God-dard Space Flight Center.PMW gratefully acknowl-edges support for this work from NASA/SAO through grant GO7-8077A.This research has made use of data obtained through the High Energy Astrophysics Science Archive Research Center Online Service,provided by the NASA/Goddard Space Flight Center,and is based on observations made with the NASA/ESA Hubble Space Telescope,obtained at the Space Telescope Science In-stitute,which is operated by the Association of Univer-sities for Research in Astronomy,Inc.,under NASA con-tract NAS5-26555.These observations are associated with program#10761.This work has been supported by SAO grant GO7-8077Z,an NSERC Discovery Grant, the Canadian Institute for Advanced Research,and Le Fonds Qu′e b′e cois de la Recherche sur la Nature et les Technologies.

REFERENCES

Alpar,M.A.2001,ApJ,554,1245

Anders,E.&Grevesse,N.1989,Geochim.Cosmochim.Acta,53, 197

Balucinska-Church,M.&McCammon,D.1992,ApJ,400,699Balucinska-Church,M.&McCammon,D.1998,ApJ,496,1044 Beloborodov,A.M.&Thompson,C.2007,ApJ,657,967 Bignami,G.F.,Caraveo,P.A.,De Luca,A.,&Mereghetti,S. 2003,Nature,423,725

13

Camilo,F.,Ransom,S.M.,Halpern,J.P.,Reynolds,J.,Helfand, D.J.,Zimmerman,N.,&Sarkissian,J.2006,Nature,442,892 Camilo,F.,Ransom,S.M.,Pe?n alver,J.,Karastergiou,A.,van Kerkwijk,M.H.,Durant,M.,Halpern,J.P.,Reynolds,J.,Thum, C.,Helfand,D.J.,Zimmerman,N.,&Cognard,I.2007,ApJ, 669,561

Campana,S.&Israel,G.L.2007,ATel,1043

Campana,S.,Rea,N.,Israel,G.L.,Turolla,R.,&Zane,S.2007, A&A,463,1047

Chatterjee,P.,Hernquist,L.,&Narayan,R.2000,ApJ,534,373 Cox,A.N.,ed.2000,Allen’s astrophysical quantities,4th ed.(New York:AIP Press;Springer)

Dedeo,S.,Psaltis,D.,&Narayan,R.2001,ApJ,559,346

Dib,R.,Kaspi,V.M.,&Gavriil,F.P.2007a,ApJ,submitted (arXiv:0706.4156)

Dib,R.,Kaspi,V.M.,Gavriil,F.P.,&Woods,P.M.2007b,ATel, 1041

Duncan,R.C.,&Thompson,C.1992,ApJ,392,L9

Durant,M.&van Kerkwijk,M.H.2005,ApJ,627,376

—.2006a,ApJ,650,1070

—.2006b,ApJ,650,1082

—.2006c,ApJ,652,576

Eichler,D.,Gedalin,M.,&Lyubarsky,Y.2002,ApJ,578,L121 Ek?s?,K.Y.&Alpar,M.A.2003,ApJ,599,450

Ertan,¨U.&C?al??s kan,S?.2006,ApJ,649,L87

Ertan,¨U.,G¨o?g¨u?s,E.,&Alpar,M.A.2006,ApJ,640,435

Fern′a ndez,R.&Thompson,C.2007,ApJ,660,615

G¨o?g¨u?s,E.,Kouveliotou,C.,Woods,P.M.,Finger,M.H.,&van der Klis,M.2002,ApJ,577,929

Gaensler,B.M.,McClure-Gri?ths,N.M.,Oey,M.S.,Haverkorn, M.,Dickey,J.M.,&Green,A.J.2005,ApJ,620,L95 Gavriil,F.P.,Dib,R.,Kaspi,V.M.,&Woods,P.M.2007,ATel, 993

Gavriil,F.P.&Kaspi,V.M.2004,ApJ,609,L67,;GK04 Gavriil,F.P.,Kaspi,V.M.,&Woods,P.M.2002,Nature,419, 142

—.2004,ApJ,607,959

—.2006,ApJ,641,418

Gotthelf, E.V.&Halpern,J.P.2007,Ap&SS,308,79, (astro-ph/0608473)

Gotthelf,E.V.,Halpern,J.P.,Buxton,M.,&Bailyn,C.2004, ApJ,605,368

Heyl,J.S.&Hernquist,L.2005,MNRAS,362,777

Hulleman,F.,van Kerkwijk,M.H.,Verbunt,F.W.M.,&Kulkarni, S.R.2000,A&A,358,605

Hurley,K.,Boggs,S.E.,Smith,D.M.,Duncan,R.C.,Lin,R., Zoglauer,A.,Krucker,S.,Hurford,G.,Hudson,H.,Wigger,C., Hajdas,W.,Thompson,C.,Mitrofanov,I.,Sanin,A.,Boynton, W.,Fellows,C.,von Kienlin,A.,Lichti,G.,Rau,A.,&Cline, T.2005,Nature,434,1098

Ibrahim, A.I.,Markwardt, C. B.,Swank,J.H.,Ransom,S., Roberts,M.,Kaspi,V.,Woods,P.M.,Sa?-Harb,S.,Balman, S.,Parke,W.C.,Kouveliotou,C.,Hurley,K.,&Cline,T.2004, ApJ,609,L21

Ibrahim,A.I.,Sa?-Harb,S.,Swank,J.H.,Parke,W.,Zane,S.,& Turolla,R.2002,ApJ,574,L51

Ibrahim,A.I.,Swank,J.H.,&Parke,W.2003,ApJ,584,L17 Israel,G.L.,Campana,S.,Testa,V.,Mereghetti,S.,Zane,S.,Rea, N.,Lo Curto,G.,&Stella,L.2007,ATel,1077

Israel,G.L.,Covino,S.,Stella,L.,Campana,S.,Marconi,G., Mereghetti,S.,Mignani,R.,Negueruela,I.,Oosterbroek,T., Parmar,A.N.,Burderi,L.,&Angelini,L.2002,ApJ,580,L143 Kaspi,V.M.2007,Ap&SS,in press(astro-ph/0610304)

Kaspi,V.M.,Chakrabarty,D.,&Steinberger,J.1999,ApJ,525, L33Kaspi,V.M.,Gavriil,F.P.,Chakrabarty,D.,Lackey,J.R.,& Muno,M.P.2001,ApJ,558,253

Kaspi,V.M.,Gavriil,F.P.,Woods,P.M.,Jensen,J.B.,Roberts, M.S.E.,&Chakrabarty,D.2003,ApJ,588,L93

Kuiper,L.,Hermsen,W.,den Hartog,P.,&Collmar,W.2006, ApJ,645,556

Marsden,D.&White,N.E.2001,ApJ,551,L155

Mereghetti,S.,Tiengo,A.,Esposito,P.,G¨o tz,D.,Stella,L.,Israel, G.L.,Rea,N.,Feroci,M.,Turolla,R.,&Zane,S.2005,ApJ, 628,938

Mereghetti,S.,Tiengo,A.,Stella,L.,Israel,G.L.,Rea,N.,Zane, S.,&Oosterbroek,T.2004,ApJ,608,427

Mori,K.&Ho,W.C.G.2007,MNRAS,377,905 Oosterbroek,T.,Parmar,A.N.,Mereghetti,S.,&Israel,G.L. 1998,A&A,334,925

¨Ozel,F.&Guver,T.2007,ApJ,submitted(astro-ph/0612561) Palmer,D.M.,Barthelmy,S.,Gehrels,N.,Kippen,R.M.,Cayton, T.,Kouveliotou,C.,Eichler,D.,Wijers,R.A.M.J.,Woods, P.M.,Granot,J.,Lyubarsky,Y.E.,Ramirez-Ruiz,E.,Barbier, L.,Chester,M.,Cummings,J.,Fenimore,E.E.,Finger,M.H., Gaensler,B.M.,Hullinger,D.,Krimm,H.,Markwardt,C.B., Nousek,J.A.,Parsons,A.,Patel,S.,Sakamoto,T.,Sato,G., Suzuki,M.,&Tueller,J.2005,Nature,434,1107

Rea,N.,Israel,G.L.,Stella,L.,Oosterbroek,T.,Mereghetti,S., Angelini,L.,Campana,S.,&Covino,S.2003,ApJ,586,L65 Rea,N.,Oosterbroek,T.,Zane,S.,Turolla,R.,M′e ndez,M.,Israel, G.L.,Stella,L.,&Haberl,F.2005a,MNRAS,361,710

Rea,N.,Testa,V.,Israel,G.L.,Mereghetti,S.,Perna,R.,Stella, L.,Tiengo, A.,Mangano,V.,Oosterbroek,T.,Mignani,R., Curto,G.L.,Campana,S.,&Covino,S.2004,A&A,425,L5 Rea,N.,Tiengo,A.,Israel,G.L.,&Campana,C.2007,ATel,1121 Rea,N.,Tiengo,A.,Mereghetti,S.,Israel,G.L.,Zane,S.,Turolla, R.,&Stella,L.2005b,ApJ,627,L133

Stetson,P.B.1987,PASP,99,191

Tam,C.R.,Kaspi,V.M.,van Kerkwijk,M.H.,&Durant,M. 2004,ApJ,617,L53

Thompson,C.&Duncan,R.C.1995,MNRAS,275,255 Thompson,C.&Duncan,R.C.1996,ApJ,473,322 Thompson,C.,Lyutikov,M.,&Kulkarni,S.R.2002,ApJ,574, 332

Tiengo,A.,Mereghetti,S.,Turolla,R.,Zane,S.,Rea,N.,Stella, L.,&Israel,G.L.2005,A&A,437,997

van Kerkwijk,M.H.,Kaplan,D.L.,Durant,M.,Kulkarni,S.R., &Paerels,F.2004,ApJ,608,432,in press(astro-ph/0402418) Wang,Z.,Bassa,C.,Kaspi,V.M.,Bryant,J.J.,&Morrell,N. 2007,ApJ,submitted(arXiv:0711.2290)

Wang,Z.&Chakrabarty,D.2002,ApJ,579,L33

Wang,Z.,Chakrabarty,D.,&Kaplan,D.L.2006,Nature,440, 772

Woods,P.M.,Kouveliotou, C.,Gavriil, F.P.,Kaspi,V.M., Roberts,M.S.E.,Ibrahim,A.,Markwardt,C.B.,Swank,J.H., &Finger,M.H.2005,ApJ,629,985

Woods,P.M.,Kaspi,V.M.,Thompson, C.,Gavriil, F.P., Marshall,H.L.,Chakrabarty, D.,Flanagan,K.,Heyl,J.,& Hernquist,L.2004,ApJ,605,378

Woods,P.M.,Kouveliotou,C.,Finger,M.H.,G¨o?g¨u?s,E.,Wilson, C.A.,Patel,S.K.,Hurley,K.,&Swank,J.H.2007,ApJ,654, 470

Woods,P.M.&Thompson,C.2006,in Compact Stellar X-ray Sources,ed.W.H.G.Lewin&M.van der Klis(UK:Cambridge University Press)

Zhu,W.,Kaspi,V.M.,Woods,P.M.,Gavriil,F.P.,&Dib,R. 2007,ApJ,submitted(arXiv:0710.1896)

本文来源：https://www.bwwdw.com/article/kkol.html