Swift Discovery of a New Soft Gamma Repeater, SGR J1745-29, near Sagittarius A*

I. Introduction

Gillessen et al. 2012 recently reported that a gas cloud referred to as “G2” is expected to pass within 3100 $R_{G}$ of Sagittarius (Sgr) A* as early as mid-2013 (Gillessen et al. 2013). If G2 is indeed a gas cloud (Phifer et al. 2013, however, see), its tidal disruption may result in accretion onto Sgr A*, which in turn could lead to Sgr A* entering an X-ray active state. The anticipation of this event has led to monitoring programs of Sgr A* over a broad range of wavelengths starting early 2013.

NASA’s Swift satellite has a unique rapid slewing capability, which allows its moderate sensitivity X-Ray Telescope (XRT; Burrows et al. 2005a) to perform short (approximately 1 ks) daily monitoring of Sgr A*. Daily observations are being carried out between 2013 February 2 and 2013 November 2, except for a monthly 2–3 day drop out when Sgr A* is too close to the Moon. Using XRT data from an observation at 17:34 UT on 2013 April 24, Degenaar et al. 2013a reported an increase in the X-ray flux from the vicinity of Sgr A* by an order of magnitude above its quiescent level. An XRT observation on the previous day did not show any evidence of enhanced emission from this region. A follow-up observation on 2013 April 25 at 15:58 UT (Reynolds et al. 2013) showed that the enhanced emission persisted much longer than typical Sgr A* flare events, which only last tens of minutes to hours (e.g., Baganoff et al. 2001; Nowak et al. 2012), making this an unusual flaring episode.

At 19:15 UT on 2013 April 25, during a scheduled observation of Sgr A*, the Swift Burst Alert Telescope (BAT; Barthelmy et al. 2005) triggered on a short ($\sim 30$ ms), hard X-ray burst at a position consistent with Sgr A* (Barthelmy et al. 2013). Kennea et al. 2013 reported that the characteristics of this burst were consistent with Soft Gamma Repeater (SGR) bursts seen by BAT, and therefore suggested that both burst and enhanced emission were from a new SGR source too close to Sgr A* for the XRT (18″ HPD, 7″ FWHM) to resolve.

SGRs are members of a very small group of sources (26 known to date http://www.physics.mcgill.ca/$\sim$pulsar/magnetar/main.html), which are suggested to be magnetars (slowly rotating neutron stars with extreme surface dipole magnetic fields of $>10^{14}$ G); Duncan & Thompson 1992; Kouveliotou et al. 1998. Historically, SGRs have been discovered when they entered a burst active period emitting multiple hard X-ray/soft $\gamma-$ray bursts at irregular intervals; the first such source was discovered in 1987 (for reviews on magnetars see Woods & Thompson 2006; Mereghetti 2008 and references therein). All but two magnetars lie on the Galactic plane with approximately half of their population concentrated between $\sim 7^{\circ}$ and $\sim 30^{\circ}$ from the Galactic center.

A NuSTAR follow-up observation on 2013 April 26 found a $\sim 3.76$ s periodicity (Mori et al. 2013a), well within the range of magnetar periods (Woods & Thompson 2006; Mereghetti 2008, 2–12 s;), further confirming this source as a likely new SGR. A subsequent Chandra observation on 2013 April 29 found a new X-ray source $\sim 3^{\prime\prime}$ away from Sgr A* (Rea et al. 2013) and confirmed the presence of the 3.76 s period. Later Swift observations in Windowed Timing mode allowed a measurement of $\dot{P}=2.5\pm 1.1\times 10^{-11}$, implying a dipole magnetic field of $B=3\times 10^{14}$ G, consistent with this source being a magnetar (Gotthelf et al. 2013). The source was designated SGR J1745$-$29 (Gehrels et al. 2013).

SGR J1745$-$29 was observed with the Effelsberg, Green Bank, Parkes and Sardinia radio telescopes (Eatough et al. 2013; Burgay et al. 2013; Buttu et al. 2013), which also confirmed the 3.76s period, making SGR J1745$-$29 the fourth magnetar detected in radio wavelengths. Eatough et al. 2013 find a dispersion measure consistent with Sgr A*.

In this letter we discuss the discovery of SGR J1745$-$29 by Swift, reporting on the pre- and post-burst X-ray emission from the source, including spectral and temporal analyses, and a detailed report on the BAT detection of the SGR burst. Finally we discuss similarities between SGR J1745$-$29 and the overall magnetar population, focusing on the bursting behavior and flux evolution, and the implications of finding an SGR at the center of our galaxy. This letter is a companion to Mori et al. 2013b, which describes pulsar timing and broad band spectral analysis of SGR J1745$-$29 utilizing primarily NuSTAR data.

II. Observations

Swift observed the region around Sgr A* on an approximately daily basis beginning 2013 February 03. The detection of increased emission from the region initiated additional observations through the Swift Target of Opportunity program. As of 2013 May 5, a total of 70.6 ks of time has been devoted to observing the Sgr A* region with XRT. Further, the BAT trigger on the SGR J1745$-$29 burst, resulted in $\sim 20$ ks of automated follow-up observations. Note that observations were not performed on 2013 April 28 to 2013 April 30 due to the field being too close to the Moon for Swift to observe. A summary of the Swift observations used in this letter is given in Table 1.

III. Data Analysis

We analyzed the Swift data with the standard Swift analysis tools version 4.0, part of HEAsoft 6.13. XRT spectral fitting was performed in XSPEC (Arnaud 1996) with the v13 CALDB XRT Photon Counting (PC) mode RMFs and ARFs. The ARF files used time-dependent exposure maps to correct for the presence of hot columns and hot pixels on the total exposure. All errors are quoted at 90% confidence, and coordinates are given in the J2000 epoch.

We extracted a light-curve of the region that includes SGR J1745$-$29 and Sgr A*, using an extraction region of radius $10^{\prime\prime}$ centered on the position of Sgr A*. Compared to the previous quiescent count rates seen from this region http://www.swift-sgra.com, the data taken between 2013 February 2 and 2013 April 23 show no evidence of enhanced emission, with an XRT count rate consistent with a non-background subtracted mean of 0.011 s-1.

Starting with the observation taken on 2013 April 24 at 17:32 UT, approximately 1.1 days before the BAT-detected burst, the XRT count rate from this region had risen to $0.11\pm 0.1$ s-1. The previous observation ending 2013 April 16:28 UT showed no evidence of enhanced emission, and, therefore, we conclude that SGR J1745$-$29 became active within a period of $\sim 25$ hr.

III.2. Localization of SGR J1745−-29

III.3. Detection of the SGR Burst

The characteristics of this burst are very similar to those of other SGR bursts seen by BAT, e.g., those seen from SGR J1833$-$0832 (Göğüş et al. 2010) and Swift J1834.9$-$0846 (Kargaltsev et al. 2012). As there exists no known SGR within or near the BAT error circle, we conclude that this burst is from a previously undiscovered SGR in the Galactic Center.

At the time of writing only one burst from SGR J1745$-$29 has been seen by BAT. However, given that SGR J1745$-$29 turned on between 25 and 50 hr before this burst, it is possible that there were earlier bursts, not seen by BAT, which precipitated this turn-on. We examined the Swift observing plan to determine the BAT temporal coverage between the XRT observations on April 23 and 24. During this time SGR J1745$-$29 was only inside the BAT $>50\%$ coded field of view $\sim 4\%$ of the time, and therefore earlier bursts cannot be ruled out.

We have also searched for untriggered events from SGR J1745$-$29 in the Fermi/Gamma Ray Burst Monitor (GBM; Meegan et al. 2009) Time-Tagged Event data with 16 ms time resolution. We did not find any burst in the data taken pre-outburst or post-outburst from the SGR J1745$-$29 direction. However, the search also did not reveal any detection at the time of the BAT burst, suggesting that GBM may be insensitive to such weak bursts.

III.4. Swift/XRT Spectral Analysis

To characterize the XRT spectrum of SGR J1745$-$29 we extracted a region centered on the best fitted position in XRT detector coordinates of the transient, using a radius of $10^{\prime\prime}$. This follows the method of Degenaar et al. 2013b, to maximize the signal from SGR J1745$-$29 and minimize the effect of the bright complex diffuse emission near Sgr A*. The background region was taken from an annulus with inner and outer radii of $20^{\prime\prime}$ and $60^{\prime\prime}$, respectively. We fit absorption using the TBabs model with the abundances set to the those of Wilms et al. 2000 and the cross-sections set to the values of Verner et al. 1996.

We fit the time-averaged spectrum for the longest single exposure taken post-burst (ObsID 00554491001, with an exposure time of 19564 s). The XRT spectrum is dominated by the effects of high absorption, with negligible X-ray emission below $\sim 2$ keV. The spectrum is well fit with either an absorbed blackbody model or an absorbed power-law model. However, the best-fit photon index for the power-law model ($\Gamma=3.5\pm 0.3$) is very soft, suggesting that the spectrum is likely thermal in nature.

The average observed flux in the 0.3–10 keV band is $2.15^{+0.09}_{-0.08}\times 10^{-11}$ erg s-1 cm-1 ($4.77^{+0.40}_{-0.34}\times 10^{-11}$ erg s-1 cm-2, corrected for absorption). Assuming a distance of 8 kpc, this gives a luminosity of $3.6\pm 0.3\times 10^{35}$ erg s-1 (0.3–10 keV). The corresponding blackbody emission radius is equal to $1.44^{+0.20}_{-0.16}$ km, with the caveat that unfitted hard continuum components may be contributing to the XRT flux. We note, however, that this radius is consistent within errors to the value derived from the BAT burst spectral fit.

III.5. Investigation of Spectral and Flux Evolution

To determine if there is any spectral or flux evolution detectable in the Swift observations, we performed time resolved spectroscopy of the XRT data in Table 1. To maximize sensitivity to any changes in the blackbody temperature and emission radius, we fixed the absorption to the value reported in Section III.4 and utilized Cash statistics (Cash 1979), which generally provide more accurate fit parameters for low counts spectra.

Because Swift is in a low Earth orbit, observations longer than $\sim 1.8$ ks are broken into multiple “snapshots”, with start times separated roughly by the Swift orbital period (96 minutes). We extracted XRT spectra for all snapshots longer than 100s. To maximize the quality of the data, we grouped adjacent snapshots within a single observation to achieve a minimum exposure time of 2 ks whenever possible.

IV. Discussion

Swift has observed the sudden turn-on of a new transient source near Sgr A*. This, combined with the BAT detection of a short hard X-ray burst from a position consistent with the new transient, suggests this transient is a new SGR in the Galactic Center, SGR J1745$-$29.

The $3^{\prime\prime}$ separation puts SGR J1745$-$29 at a projected distance of $\sim 0.1$ pc from Sgr A*. Given the apparently similar absorption column, we argue that SGR J1745$-$29 is likely located close to Sgr A*. At the projected distance, the effects of Sgr A* on SGR J1745$-$29 will be small: we estimate that the gravitational acceleration due to Sgr A* will contribute no more than $\pm 2\times 10^{-13}$ to the apparent $\dot{P}$, assuming nominal values of $4\times 10^{6}\,M_{\odot}$ and 8 kpc distance for the central black hole. We note that even with the constraints from the absorption column, the true distance of SGR J1745$-$29 from Sgr A* remains highly uncertain.

As $\sim 50\%$ of known magnetars lie within 30 deg of the Galactic Center, discovering a new SGR in this region area was not unexpected, but the close proximity of this source to Sgr A*, combined with the temporal coincidence with the anticipated encounter of G2 with Sgr A*, made this event intriguing. However, it seems unlikely that the turn-on of SGR J1745$-$29 is related to any interaction with G2, as G2 is currently within $1^{\prime\prime}$ of Sgr A* (Gillessen et al. 2013), whereas SGR J1745$-$29 is $3^{\prime\prime}$ away. We conclude that the onset of emission from SGR J1745$-$29 at this time is coincidental.

The Galactic Center is very well studied in X-rays, allowing us to place limits on burst and soft X-ray outburst emission from SGR J1745$-$29 in the recent past. We estimate that Swift/BAT spends approximately 3.5 Ms year-1 covering the Sgr A* region, meaning that any repeated flaring activity in the past eight years would likely have been seen. Swift monitoring observations of the Galactic Center region with XRT have been on-going since 2006 February 24, so we can rule out any similar outburst with high confidence for the past $\sim 7$ yr.

Chandra has performed regular observations of this region starting 1999 September 21, and did not detect SGR J1745$-$29 previously (Muno et al. 2009).

The excess diffuse emission, that is likely produced by colliding winds of IRS 16SW and other nearby windy stars, makes it hard to estimate an upper limit for the quiescent source state. However, Mori et al. 2013b conservatively estimate $\sim 10^{32}$ erg s-1 (2-10 keV), based on the quiescent limit on CXOGC J174540.0$-$290031 from Muno et al. 2005. We note that SGR J1833$-$0832 was also observed pre-outburst by Chandra and was not detected, with an upper limit of $3.4\times 10^{-13}$ erg cm-2 s-1 (Göğüş et al. 2010), which is equivalent to a luminosity of $4\times 10^{32}$ erg s-1 for an assumed distance of 5.7 kpc, close to the Chandra limit on SGR J1745$-$29.

Since 2008 August, five new magnetar candidates have been discovered by Swift and Fermi/GBM. Four of these have intriguing differences from the previous members of the magnetar family: they are all transient sources discovered by emitting typical magnetar short bursts, which became burst inactive after exhibiting one or two relatively dim events, and their persistent X-ray spectra are different than the rest of the magnetar sources; they are typically well described by a single blackbody function with a temperature around 1 keV ($0.3-10$ keV). SGR J1745$-$29 shares these common properties. In Figure 4 we present the unabsorbed flux trend of the persistent X-ray emission from SGR J1745$-$29 following the outburst onset, along with that of a set of transient and persistent magnetars. It is striking to note that the X-ray flux of both SGR J1745$-$29 and SGR J1833$-$0832 remained fairly constant in the first 10–20 days into the outburst, while that of other transient magnetars (such as, SGR J1627$-$41 or SGR J1550$-$5418) declined steadily following the outburst onset. We, therefore, suggest that SGRs with low bursting rates possess slightly different characteristics than the bulk of the population. We know from the spin and spin-down rates of these sources that their dipole (or more local multi-pole magnetic field) is in the magnetar regime. It is, however, possible that these sources cannot efficiently radiate away the energy released by events leading to bursts, therefore, cannot appear as prolific bursters. Instead, the energy released in a burst event could be trapped within the system, which could then result in crustal heating near the poles. It is, then plausible that further energy release from the neutron star, possibly as bursts, is continuously trapped, resulting in the constant persistent X-ray flux seen in SGR J1745$-$29 and other SGRs with apparent low bursting rates. In this scenario, we would expect the SGR J1745$-$29 flux to decline when the active episode ends, typically after 1-2 weeks.

References