Radio properties of the magnetar near Sagittarius A from observations with the Australia Telescope Compact Array.
We have carried out observations of the newly-discovered magnetar in the direction of Sagittarius A using the Australia Telescope Compact Array in four frequency bands from 4.5 to 20 GHz. Radio pulsations are clearly detected at all frequencies. We measure the pulsar’s dispersion measure to be cmpc, the highest of any of the known pulsars. Once Faraday rotation has been taken into account, the pulse profile is more than 80% linearly polarized at all frequencies and has a small degree (5%) of circular polarization. The rotation measure of rad m is the largest (in magnitude) ever measured for a pulsar but still a factor 8 smaller than Sgr A itself. The combination of the dispersion and rotation measures implies an integrated magnetic field strength of G along the line of sight. The flux density appears to have increased by about a factor of two between observations made 30 days apart. This object therefore joins the small class of radio emitting magnetars.
keywords:Pulsars: individual: PSR J1745–2900, ISM: Structure, Galaxy: centre
Magnetars are slowly-rotating neutron stars powered by the energy stored in their strong magnetic fields rather than their rotational energy. Typically magnetars have spin periods of 5 s, large spin down rates of order 10 and implied magnetic fields in excess of 10 G. They have been discovered via their high energy emission and their luminosities at X-ray energies and above exceed the spin-down energy by an order of magnitude (Duncan & Thompson 1992, Beloborodov 2009).
To date, of the more than 20 magnetars now known, only three have confirmed radio pulsations: XTE 1810–197 (Camilo et al. 2006), 1E1547–5408 (Camilo et al. 2008) and PSR J1622–4950 (Levin et al. 2010). The first two sources were detected at radio wavelengths following high energy outbursts; the latter source was first detected in the radio band. All three radio magnetars share similar characteristics in that they show complex pulse morphology, very high degree of polarization and a flat spectral index, with PSR J1622–4950 detected at frequencies up to 24 GHz (Keith et al. 2011) and XTE 1810–197 detected as high as 144 GHz (Camilo et al. 2007).
It has long been a goal of pulsar astronomers to find a pulsar in orbit about the black hole at the Galactic Centre (e.g., Kramer et al. 2004). In particular, even an isolated pulsar in a tight orbit around the black hole has the potential to provide exquisite tests of general relativity (Liu et al. 2012). Although it is expected that pulsars should exist at the Galactic Centre (Pfahl & Loeb 2004), the postulated high degree of scatter-broadening (Lazio & Cordes 1998) implies that searches need to be conducted at high radio frequencies. In spite of many attempts however (Johnston et al. 2006, Deneva et al. 2009, Macquart et al. 2010) no radio pulsars have yet been found within 10 minutes of arc of the Galactic Centre.
The recent discovery of a magnetar in the direction of Sagittarius A by the Swift satellite (Kennea et al. 2013) and the subsequent detection of pulsations by the NuStar satellite (Mori et al. 2013) was therefore of great interest to the radio pulsar community and has been the subject of follow-up observations in the radio band with single dishes (Burgay et al. 2013; Eatough et al. 2013a,b,c; Buttu et al. 2013). Here we report on interferometric observations made with the Australia Telescope Compact Array and discuss their implications.
Observations towards the Galactic Centre were made on two separate occasions with the Australia Telescope Compact Array (ATCA), an east-west synthesis telescope located near Narrabri, NSW, which consists of six 22-m antennas on a 6-km track. The configuration of the array was 6C (shortest baseline 153 m, longest baseline 6000 m), maximising the potential for point-source imaging. The observations took place on 2013 May 1 starting at UT 12:30 and 2013 May 31 at UT 09:30. The observing time was 10 hours and 4 hours respectively.
The ATCA has recently been equipped with broad-band feeds covering the spectral region between 4 and 11 GHz. The new broad-band correlator (CABB; described in Wilson et al. 2011) is capable of processing 4 frequency bands each with 2048 MHz bandwidth with 1 MHz frequency resolution and full polarization capability. In addition, it can perform ‘pulsar binning’, allowing high time resolution construction of pulse profiles (Wilson et al. 2011).
The observations switched between 2 frequencies with a cadence of 30 min. We observed from 4.5 to 8.5 GHz and from 16.0 to 20.0 GHz with 1024 spectral channels (i.e. 4 MHz spectral resolution) and 32 phase bins (i.e. 117 ms time resolution) folded at the spin period of 3.767 s given by the NuStar observations of Mori et al. (2013).
Data reduction and analysis were carried out with the miriad package111See http://www.atnf.csiro.au/computing/software/miriad using standard techniques. After flagging bad data, the primary calibrator (PKS 1934–638) was used for flux density and bandpass calibration and the secondary calibrators, PKS 1741–312 and PKS 1714–336, were used to solve for antenna gains, phases and polarization leakage terms in the low and high frequency observations respectively.
The Galactic Centre contains highly complex regions of bright radio emission, making it difficult to image given the sparse uv coverage of the ATCA observations. However, the advantage of pulsar binning is that we can create visibilities formed from the difference between the on-pulse and off-pulse bins (using the miriad task psrbl). This therefore removes all non time-varying emission from the visibilities, resulting in an image which contains only the point source pulsar.
We created such an image using multi-frequency synthesis of the data using the May 1 observations between 16.0 and 18.0 GHz, the band and epoch in which the most precise measurement of the position of the pulsar could be made. The resultant beam size is arcsec. The location of the pulsar, as determined with the miriad routine imfit is (J2000) 17:45:40.164, –29:00:29.818 with measurement uncertainty of 0.022 and 0.090 arcsec, respectively. This position is consistent with, but has smaller error bars than the Chandra position reported by Rea et al. (2013).
3.2 Pulse profile
At all frequencies, the pulsar is highly linearly polarized (see below) and possesses some degree of circular polarization. In the 4.5 to 8.5 GHz band, the pulse has a shallow rising edge and fast falling edge, reminiscent of the trailing partial cones identified in Lyne & Manchester (1998). The linear polarized fraction as measured is only 55% and the circular polarization is 22%. However, even across the 4 MHz channels there is a significant fraction of depolarization due to the high rotation measure. At the higher frequency there is evidence also for a weak trailing component to the pulse. The fractional circular polarization is lower at 5% but the linear polarized remains above 80%. It is difficult to draw any further conclusions given the low time resolution of our data. Pulse profiles are shown in Fig 1.
3.3 Dispersion Measure (DM)
Although our temporal resolution is rather crude with 117 ms per phase bin it is still possible to estimate the DM based on the comparison between the observations made over the frequency range between 4.5 and 8.5 GHz. Between these two frequencies the delay difference amounts to 15 ms per 100 DM units. We cross correlated pulse profiles made in 16 frequency bands between 4.5 and 8.5 GHz and determined the delay to be close to 2.0 phase bins. This implies a DM of 165050 cmpc, consistent with but more precise than the value reported by Eatough et al. (2013b).
3.4 Polarization and Rotation Measure (RM)
If the pulsar is in the Galactic centre region the electron density and magnetic field are expected to be high. We therefore searched for rotation measure using the 17 GHz data because at this frequency the pulsar is bright and the effects of Faraday rotation and interstellar scattering are smaller. We divided the 2 GHz bandwidth between 16 and 18 GHz into 16 MHz sub-bands.
Visual inspection of the pulse profiles showed a high degree of linear polarization in the individual frequency channels and clearly showed rotation of Stokes Q and U between channels. Figure 2 shows the Stokes parameters as a function of observing frequency measured at the profile peak using the miriad function psrplt. Two things can be noted from the figure. First, the fractional polarization is very high, approaching 100%. Secondly, there is a large rotation of Stokes Q and U, implying a high RM. We therefore computed the position angle () of the linear polarization as a function of observing frequency via
Figure 3 shows as a function of the square of the observing wavelength. The RM is determined from a linear fit to the data indicating an exceptionally high value of rad m.
The extremely high rotation measure explains the non-detection of polarization reported by Eatough et al. (2013b) from Effelsberg observations, because it causes depolarization of signals if averaged over wide bandwidths.
3.5 Flux density and spectral index
We measured the flux density of the pulsar across four frequency bands from 4.5 to 8.5 GHz and 16 to 20 GHz frequencies using the miriad task imfit at each of the two observing epochs. Figure 4 shows the results.
For the first set of observations, the flux density at 6.5 GHz is 0.8 mJy and at 18 GHz is 0.3 mJy. Between 4.5 and 8.5 GHz the flux densities are constant within the error bars, yielding a spectral index of zero. The drop off in flux density between the 6.5 and 18 GHz would seem to imply a spectral index of . However, we see an apparent steep flux density decrease through the 16 to 20 GHz band. Also, our measured values not easily reconciled with those reported in Eatough et al. (2013c) which show a virtually flat spectral index between 2.4 and 19 GHz. Our values are significantly higher at low frequencies and broadly consistent at the high frequencies. During the second set of observations the magnetar was considerably brighter by about a factor two between 16 and 20 GHz. Futhermore, unlike the earlier observations, the data show a steep decline in flux density between 4.5 and 8.5 GHz with a spectral index close to .
Magnetars are known to have variable flux densities as a function of time (e.g., Levin et al. 2011), which may explain the differences in the two sets measurements and the differences between our values and those in Eatough et al. (2013c). We can rule out scintillation as a cause. The scintillation bandwidth is extremely small, of order a few kHz at the low band and less than 100 kHz even in the higher band, well below our frequency resolution. We also see no sign of variability within each observing epoch.
The pulsar’s angular proximity to the Galactic Centre coupled with its high dispersion measure and large absorption in the X-ray band make it likely that it is indeed very close to the centre of our Galaxy. It is offset by 3 arcsec from Sgr A, implying a lower limit to the physical separation of only 0.1 pc. The pulsar shares characteristics with the other known radio magnetars. It has a high degree of linear polarization and remains bright at high observing frequencies. It may also have variable flux density.
The measurement of both dispersion and rotation measures allows us measure the average parallel magnetic field strength in along the line of sight to the pulsar via =1.2 RM/DM with RM and DM in conventional units, and measured in G. This value is 50 G, about an order of magnitude larger than typical line of sight measurements in the Galactic plane. We note that the RM of the magnetar has the same sign and is about a factor 8 smaller than the RM of Sgr A itself (Bower et al. 2003) but significantly larger than values of the non-thermal filaments. Bower et al. (2003) argue that the RM does not originate from the scattering screen but rather from material closer to Sgr A. This would then imply that the magnetar is within this same material but in front of Sgr A as seen from Earth.
The nature of the magnetic field in the Galactic Centre is unclear. Synchrotron filaments are present in the inner region with mG magnetic fields, but it remains the topic of debate as to whether fields this strong are typical of the region as a whole. Indeed current ideas indicate weaker fields of order 100 G might be more typical (Boldyrev & Yusef-Zadeh 2006). Indeed, Crocker et al. (2010) claim that the inner few hundred pc of the Galaxy has a field strength of order 50 - 100 G, in line with our measurement. Continued monitoring of the source will will yield rotation measure and dispersion measure variations, enabling improved characterization of the interstellar medium and magnetic field structure in the Galactic Centre region.
If the physical separation from the Galactic Centre is indeed 0.1 pc, then the orbital period is greater than 1500 yr assuming a circular orbit about Sgr A. This orbital period is too long to provide interesting constraints on gravitational theories (Liu et al. 2012). We note that the pulsar would easily have been detectable (if on at the time) in all the targetted surveys of the Galactic Centre carried out to date (Johnston et al. 2006, Deneva et al. 2009, Macquart et al. 2010). Although we cannot measure the scattering from our data, it appears to be low - only a few ms at 8 GHz (Eatough et al. 2013c). This gives optimism that observing frequencies around 10 GHz might well be optimal for future surveys. Continuing searches of the Galactic Centre for pulsars are therefore warranted.
The Australia Telescope Compact Array is part of the Australia Telescope, which is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO.
- [Belo] Beloborodov, A., 2009, ApJ, 703, 1044
- [Bold] Boldyrev, S., Yusef-Zadeh, F. 2006, ApJ, 637, L101
- [Bower] Bower, G., Wright, M., Falcke, F., Backer, D. 2003, ApJ, 588, 331
- [Burgay] Burgay, M. et al. 2013, ATel, 5035
- [Buttu] Buttu, M. et al. 2013, ATel, 5053
- [Camilo et al.(2006)] Camilo, F. et al. 2006, Nature, 442, 892
- [Camilo et al.(2007a)] Camilo, F., Ransom, S. M., Halpern, J. P., Reynolds, J. 2007, ApJ, 666, L93
- [Camilo et al.(2007b)] Camilo, F. et al. 2007, ApJ, 659, L37
- [Camilo et al.(2007c)] Camilo, F. et al. 2007, ApJ, 669, L561
- [Camilo et al.(2008)] Camilo, F. et al. 2008, ApJ, 679, L681
- [Crocker] Crocker, R., Jones, D., Melia, F., Ott, J., Protheroe, R. 2010, Nature, 463, 65
- [Deneva09] Deneva, J., Cordes, J., Lazio, T. 2009, ApJ, 702, L177.
- [Duncan & Thompson(1992)] Duncan, R. C., Thompson, C. 1992, ApJ, 392, L9
- [Eato1] Eatough, R. et al., 2013a, ATel, 5040
- [Eato2] Eatough, R. et al., 2013b, ATel, 5043
- [Eato3] Eatough, R. et al., 2013c, ATel, 5058
- [Johnston06] Johnston, S., et al. 2006, MNRAS, 373, L6.
- [Keith] Keith, M., Johnston, S., Levin, L., Bailes, M. 2011, MNRAS, 416, 346.
- [Kramer04] Kramer, M., et al. 2004, NewAR, 48, 993.
- [Kennea] Kennea, J. et al. 2013, ApJ, Submitted
- [lazio] Lazio, T., Cordes, J. 1998, ApJ 505, 715
- [levin1] Levin, L. et al. 2010, ApJ, 721, L33
- [levin2] Levin, L. et al. 2011, MNRAS, 422, 2489.
- [liu1] Liu, K., Wex, N., Kramer, M., Cordes, J., Lazio, T. 2012, ApJ, 747, 1
- [lm88] Lyne, A., Manchester, R. 1988, MNRAS, 234, 477.
- [macq] Macquart, J.-P., Kanekar, N., Frail, D., Ransom, S. 2010, ApJ, 715, 939
- [Mori] Mori, K. et al. 2013, ApJ, Submitted
- [Pfahl] Pfahl, E. & Loeb A. 2004, ApJ, 615, 253.
- [Rea] Rea, N. et al. 2013, ATel, 5032
- [Wilson] Wilson, W. et al. 2011, MNRAS, 416, 832.