Discovery of PSR J0523-7125 as a Circularly Polarized Variable Radio Source in the Large Magellanic Cloud

We observed two 30 deg² fields (field names VAST 0530−68A and VAST 0127−73A) centered on the Magellanic Clouds, six times between 2019 August and 2020 January. Each observation had an integration time of 12 minutes, achieving a typical rms sensitivity of 0.25 mJy beam⁻¹ and angular resolution of 12'' at a central frequency of 888 MHz with a bandwidth of 288 MHz. VAST-P1 incorporated data from the Rapid ASKAP Continuum Survey (RACS; McConnell et al. 2020) as the first epoch, which has the same observing frequency, but a longer integration time of 15 minutes, resulting in an rms sensitivity of 0.20 mJy beam⁻¹ for regions near the Magellanic Clouds. Details of data reduction for these surveys are given in McConnell et al. (2020) and Murphy et al. (2021).

We conducted a search for highly variable sources in the two fields using the VAST transient detection pipeline (Pintaldi et al. 2021; Murphy et al. 2021). We selected candidates with a variability index ${ \mathcal V }\gt 1.0{\sigma }_{{ \mathcal V }}$ (equivalent to the fractional variability used by other surveys) and reduced chi-square η > 2.0σ_η (equating to values of ${ \mathcal V }\gt 0.295$ and η > 6.479), where σ is the standard deviation measured by fitting a Gaussian function to the distributions of both metrics in logarithmic space (following the calculations in Rowlinson et al. 2019). We identified ∼27 candidates as highly variable, compact sources. Two of them were detected with strong circular polarization, and VAST J052348.6−712552 was the only one lacking a clear multiwavelength association/identification (the other one is associated with a stellar object). The fractional circular polarization of VAST J052348.6−712552 (f_p = ∣S_V ∣/S_I ) is about 15%–30% (see the cutout in Figure 1). No linear polarization was detected for this source in VAST-P1, placing a 3σ upper limit of ∼0.9 mJy beam⁻¹ (≲10% fractional polarization) on the total linear polarization ($P=\sqrt{{Q}^{2}+{U}^{2}}$). The radio observations are summarized in Table 1 and shown in Figure 2. Other variable sources and the general survey results will be published in a subsequent paper.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We observed VAST J052348.6−712552 with the MeerKAT radio telescope (pulsar search mode and continuum imaging mode simultaneously) for 2.5 hr at a central frequency of 1284 MHz (bandwidth of 856 MHz) on 2021 August 25. We searched the MeerKAT (pulsar mode) data for pulsar candidates using the standard Fourier domain search procedure. This was done using pulsar_miner ²³ (see Ridolfi et al. 2021 for more details), an automated pipeline based on the presto ²⁴ pulsar searching package (Ransom 2011). After cleaning the observing band by removing the frequency channels affected by strong radio frequency interference (RFI), we generated de-dispersed time series, correcting for the interstellar dispersion with dispersion measure (DM) trial values in the range 2.0–300 pc cm⁻³, with steps of 0.05 pc cm⁻³. Each time series was Fourier transformed and the resulting power spectrum searched for prominent periodicities using presto's accelsearch routine. The latter is sensitive to both isolated and binary pulsars, by accounting for possible Doppler shifts of the pulsar spin frequency in the Fourier domain due to orbital motion (see Ransom et al. 2002 for a detailed discussion of the acceleration search technique). For our search, we allowed a maximum drift of up to 200 Fourier bins, using the -zmax 200 option of accelsearch. We identified a strong pulsar candidate, PSR J0523−7125, with a period of 322.5 ms at a DM of 157.5 pc cm⁻³. No significant acceleration was detected within the 2.5 hr of the observation, indicating that the candidate pulsar is likely isolated, although further timing is ongoing. Figure 3 shows the initial discovery plot of the pulsar. It has a very steep spectrum, consistent with the measurement in continuum imaging (see described below) and a wide pulse profile with duty cycle ∼35% (using the pulse width at 50% maximum, W₅₀). Note the duty cycle may be underestimated as it is difficult to identify the off-pulse baseline for a (wide-pulsed) pulsar like this. We also measured a RM = +456 ± 6 rad m⁻² using the rmfit tool in psrchive (Hotan et al. 2004) and found a significant linear polarization (≳50%) after RM correction (see pulse profiles in Figure 4). Further flux calibration is necessary to measure precise polarization properties. This pulsar has not been recorded in any known pulsar catalog (Manchester et al. 2005) or in the Pulsar Survey Scraper, ²⁵ which records newly discovered pulsars prior to publication.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also processed the MeerKAT (simultaneous) continuum data using oxkat (Heywood 2020), which uses: casa for basic flagging, cross calibration, and splitting out measurement sets; tricolour ²⁶ for further flagging; cubical (Kenyon et al. 2018) for self-calibration; and wsclean (Offringa et al. 2014) for continuum imaging. We used PMN J0408−6545 for bandpass and flux calibration, and PMN J0420−6223 for phase calibration. We detected VAST J052348.6−712552 with a total flux density (Stokes I) of 860 ± 7 μJy beam⁻¹ (at ≳100σ confidence level), and circularly polarized flux density (Stokes ∣V∣) of 137 ± 5 μJy beam⁻¹. The source is unsolved in the image, which has an angular resolution of 75. In contrast to the pulsar mode result, no linear polarization was detected above 3σ threshold, and a rotation measure (RM) synthesis found no significant polarized intensity in ∣RM∣ < 1200 rad m⁻². We will discuss this inconsistency in Section 3.1. We obtained the (Stokes I) spectrum from 8 sub-band images and noticed an unusual upturn at higher frequency (see Figure 5). We fit a smoothly broken power law ²⁷ to the spectrum, and found an upturn at 1.36 ± 0.02 GHz with a spectral index α = −4.40 ± 0.11 in the lower frequency range and α =+2.34 ± 0.51 in the upper frequency range. We also obtained lightcurves (with 10 minutes, 2 minutes, and 8 s resolution, respectively) within the 2.5 hr observation, and found no significant variability on those timescales.

Zoom In Zoom Out Reset image size

We observed VAST J052348.6−712552 with the 64 m Parkes telescope on 2020 April 13, 2020 July 1, and 2020 November 18 using the pulsar searching mode with the Ultra-Wideband Low (UWL) receiver (which provides simultaneous frequency coverage from 704 to 4032 MHz; Hobbs et al. 2020). We used presto (Ransom 2001) to perform a standard pulsar search of DMs spanning 0−300 pc cm⁻³, plus an acceleration search up to ∼200 m s⁻². Our initial search of the Parkes observations found some candidate pulsars, but none were convincing. After detecting the pulsar with MeerKAT we reinspected our data and found a pulsar candidate with a similar period and DM. It had not been identified as a convincing candidate in our initial search since the signal is very weak (near our searching threshold SNR of ∼8), possibly due to the wide pulse profile and the steep spectrum (so it is only bright at low frequencies). We then reanalyzed the Parkes observations using the low-frequency data only, and successfully detected this pulsar at frequencies ≲2 GHz. Our Parkes data also measured a similar RM = +457.6 ± 0.1 rad m⁻² using rmfit.

We note that this sky region was searched as part of the survey conducted by Ridley et al. (2013) using the Parkes telescope, who claimed a flux density limit of 0.05 mJy at 1400 MHz, well below the flux density of PSR J0523−7125. However, we were able to identify a weak candidate in their data as well, with parameters consistent with PSR J0523−7125. Note that the Ridley et al. (2013) limit assumed a 5% pulse width (compared to 35% for PSR J0523−7125) and assumed that sources were at the centers of the beams of the Parkes Multi-Beam Receiver. Accounting for our pulse width increases the limit by ∼3 while accounting for the position (as the pulsar is close to the half-power point of one of the beams) increases it by a further factor of ∼2. This gives a limiting flux density of about 0.3 mJy, still less than what we observe, but the remaining differences may be due to a combination of scintillation, lost bandwidth due to radio frequency interference, and similar effects.

Prior to having identified the pulsations in PSR J0523−7125, we carried out follow-up observations of VAST J052348.6−712552 with the Australia Telescope Compact Array (ATCA) on 2020 August 12, ²⁸ 2021 April 25, 2021 June 4, and 2021 September 10 respectively, with a 2 GHz bandwidth centered at 2.1 GHz, 5.5 GHz, and 9.0 GHz (project code C3363, PI: Murphy; and project code C3431, PI: Pritchard). We reduced the data using miriad (Sault et al. 1995) with PKS B1934−638 and B0530−727 as the flux and phase calibrators, respectively, and imaged the data using casa (McMullin et al. 2007). We detected the source with a flux density of 0.3−0.9 mJy at 2.1 GHz, but did not detect it at 5.5 GHz or 9.0 GHz in any epoch. We measured a circular polarization flux density of 0.076 ± 0.013 mJy beam⁻¹ at 2.1 GHz for the observation on 2021 June 4, implying a polarization fraction of ∼20%. The source is unresolved in all detected ATCA images, with an angular resolution of 3–6''. We measured the spectral index by splitting the L-band (2.1 GHz) into four sub-bands, and found it to be very steep, varying between α ∼ −4 to ∼−2. The spectra can be fitted using a single power law, and their distributions are shown in Figure 5. The ATCA observations allowed us to measure a more accurate position compared to ASKAP (α = 05^h23^m4866, δ = −71°25'5258 in J2000 coordinates), with positional uncertainties of ∼0.15 arcsec in each coordinate.

We also detected VAST J052348.6−712552 in archival radio observations including the Sydney University Molonglo Sky Survey (SUMSS; Mauch et al. 2003) and an ASKAP commissioning observation (SB8532; Pennock et al. 2021). We obtained a higher-time resolution lightcurve from the latter (by imaging 9 × 7 minutes scans separated by 1.2 hr) and found no significant variability within the 12 hr 40 minutes observation. This area is also observed by the GaLactic and Extragalactic All-sky Murchison Widefield Array (GLEAM) survey (Hurley-Walker et al. 2017), and the nondetection yields a 3σ upper limit of ∼40 mJy at 72–231 MHz. All of these observations are listed in Table 1.

Prior to initiating our pulsar searches we searched archival multiwavelength data including Gaia (Gaia Collaboration et al. 2018), the Wide-field Infrared Survey Explorer (WISE; Wright et al. 2010), the ROSAT All-Sky Survey (Boller et al. 2016), and the Fermi Large Area Telescope source catalog (Acero et al. 2015). No multiwavelength counterparts were found within a 5'' radius in Gaia or WISE, or within a 2' radius in ROSAT or Fermi.

The most sensitive infrared data available for this region is the VISTA Magellanic survey catalog (VMC; Cioni et al. 2011). We found no sources within a 5σ position uncertainty, leading to 3σ limiting magnitudes Y > 21.1 mag, J > 20.9 mag, and K_s > 20.4 mag in this region. The position is also covered by the Magellanic Clouds Photometric Survey (MCPS; Zaritsky et al. 2004), and its nondetections of the source indicate U > 20.7 mag, B > 22.6 mag, V > 22.5 mag, and I > 21.2 mag.

We conducted a Neil Gehrels Swift Observatory Target of Opportunity (ToO) observation (target ID: 14338) on 2021 May 29 using the X-ray Telescope (XRT) in the photon counting mode (exposure time of 2250 s). There is one count within a 15 arcsec radius of VAST J052348.6−712552, implying a count rate upper limit of ∼0.0013 s⁻¹ (0.2–10 keV). We therefore estimated a tentative upper limit to the unabsorbed flux of ∼1.3 × 10⁻¹³ erg s⁻¹ cm⁻² (0.2–10 keV) based on an HI column density N_H = 0.6 × 10²² cm⁻² (similar to the LMC magnetar SGR 0526−66; Park et al. 2012) and a power-law photon index of Γ = 2 using the HEASARC web-based PIMMS. Details of these multiwavelength observations are summarized in Table 1.

As shown in Figure 1, we know the source is located in the direction of the LMC. The maximum DM contribution from the Milky Way along this line of sight is ∼50–60 pc cm⁻³ (NE2001 and YMW16; Cordes & Lazio 2002; Yao et al. 2017), and there is no known Galactic H ii region along this line of sight. With a DM of 157.5 pc cm⁻³, PSR J0523−7125 almost certainly resides in the LMC. We also note this DM is consistent with the DMs of other pulsars found in the LMC (∼50–270 pc cm⁻³; e.g., McConnell et al. 1991; Manchester et al. 2006; Ridley et al. 2013; Johnston et al. 2022).

We can calculate the monochromatic radio luminosity density of PSR J0523−7125 at 400 MHz and 1400 MHz respectively (L_ν = S_ν d², where S_ν is the flux density and d is the pulsar distance). Our ATCA, MeerKAT, and ASKAP midband observations covered 1400 MHz, and the measured average flux density is S₁₄₀₀ ≈ 1 mJy. The flux density at 400 MHz is therefore S₄₀₀ ≈ 25 mJy assuming a median spectral index α ∼ −2.6 (Figure 5). The flux density at 400 MHz would be ten times higher for the steepest spectral index that we measure (α ∼ −4.4). The pulsar distance d is assumed to be the LMC distance of 50 kpc (Pietrzyński et al. 2013), which is reasonable since the uncertainty in luminosity should be dominated by the uncertainty in the flux density. We can then estimate L₄₀₀ ≈ 6.3 × 10⁴ mJy kpc² and L₁₄₀₀ ≈ 2.5 × 10³ mJy kpc², which is one of the most luminous pulsars to date (see Figure 6). Only one known pulsar with a well-constrained distance, B1641−45 (Komesaroff et al. 1973; Frail et al. 1991), has a comparable luminosity to PSR J0523−7125 (∼7.6 × 10³ mJy kpc² at 400 MHz and ∼6.1 × 10³ mJy kpc² at 1400 MHz). We exclude other high-luminosity pulsars, like PSR J0134−2937, from consideration due to their significant distance uncertainties.

Zoom In Zoom Out Reset image size

Ridley et al. (2013) calculated the luminosity function of the LMC pulsars and noticed a discrepancy between the Galactic and LMC luminosity distributions. They attributed the discrepancy to a bias due to the small sample size of the LMC pulsars and concluded that the luminosity function for the LMC pulsars is consistent with its counterpart in the Galactic disk. They also suggested that the maximum 1400 MHz radio luminosity for LMC pulsars is approximately 1000 mJy kpc² based on their sample. Our pulsar is about a factor of ∼2 more luminous than this limit. With the discovery of PSR J0523−7125, we can revisit the luminosity function of the LMC pulsars by fitting the high-luminosity tail of the distribution as a power law, i.e., $\mathrm{log}N=F\mathrm{log}L+G$, where F is the slope of the distribution (Lorimer et al. 2006). Based on the ATNF pulsar catalog (Manchester et al. 2005), we estimate F ≈ −1.30 ± 0.03 for Galactic pulsars with luminosities above 30 mJy kpc² (consistent with other work, e.g., Lorimer 2004; Ridley et al. 2013), and F ≈ −1.6 ± 0.2 for LMC pulsars with luminosities above 125 mJy kpc². Although the calculation is still limited by small number statistics for the LMC pulsars and the distance uncertainties for Galactic pulsars, our result shows that the LMC pulsars are consistent with the luminosity function of their Galactic counterparts (as suggested by Manchester et al. 2006; Ridley et al. 2013).

PSR J0523−7125 has a wide pulse profile with W₅₀ ≈ 100 ms at ∼1 GHz. Figure 6 shows the duty cycle distribution of known pulsars, and PSR J0523−7125 generally has a larger duty cycle than others (either Galactic or extragalactic pulsars). The observed pulse width W can be decomposed as $W=\sqrt{{\tau }_{\mathrm{int}}^{2}+{\tau }_{\mathrm{sc}}^{2}+{\tau }_{\mathrm{DM}}^{2}}$, where τ_int is the intrinsic pulse width, τ_sc is the pulse broadening due to multipath scattering, and τ_DM is the pulse broadening due to dispersion (Lorimer & Kramer 2012). Both τ_sc and τ_DM are related to free electrons in the interstellar medium (i.e., the DM; Bhat et al. 2004). Following the YMW16 model (which includes the distribution of free electrons in the Galaxy and the Magellanic Clouds), the predicted scattering delay for Magellanic Clouds pulsars along this line of sight is τ_sc ≪ 1 ms at 1100 MHz. This suggests that the wide pulse profile is largely intrinsic. As noted in Section 2.2, the pulse width may be further underestimated due to the difficulty of identifying the off-pulse baseline. If this is the case, PSR J0523−7125 could potentially have a nearly 100% duty cycle, i.e., a possible nearly aligned rotator (the magnetic and rotation axis of the star are aligned; e.g., Young et al. 2010). This assumption could also explain why the pulsar is so bright in continuum images, but hard to detect in a pulsar search. The pulsar calibration is ongoing and we will discuss it further in a later paper.

Our MeerKAT pulsar data measured a RM = +456 rad m⁻², and the Parkes data confirmed this result. However, we cannot find any significant linear polarization in MeerKAT continuum data in an RM range from −1200 to 1200 rad m⁻². This discrepancy can be attributed to the potential swing of the polarization position angle in the pulse phase (e.g., following the characteristic S-shaped curve of the rotating vector model; Radhakrishnan & Cooke 1969), making the effective linear polarization fraction very diluted when integrated into the continuum image. Further precise polarimetry can confirm this. We noticed this ∣RM∣ is about twice the most extreme ∣RM∣ of known LMC pulsars, PSR J0540−6919 with RM = −246 rad m⁻² (Johnston et al. 2022), which is located in a supernova remnant. We checked the UM/CTIO Magellanic Cloud emission-line survey (MCELS; Smith & MCELS Team 1998) and found there is some faint, diffuse emission around our pulsar in the Hα map (with an angular scale ∼13 arcmin). We got a Hα intensity I_Hα ≈ 1.3 × 10⁻¹⁶ erg cm⁻² s⁻¹ arcsec⁻² from the MCELS map after extinction correction (following the method in Gaensler et al. 2005), and thus the emission measure (EM) is ∼70 pc cm⁻⁶. The foreground contribution to the DM in this line of sight is about 60 pc cm⁻³ (Yao et al. 2017), and to the RM is about +31 rad m⁻² (Mao et al. 2012). After subtracting the Galactic foreground contribution, the estimated parallel magnetic field B_∥ = 1.23 RM_LMC/DM_LMC ≈ 5 μG, and the estimated electron column density n_e = EM/DM_LMC ≈ 0.7 cm⁻³. The occupation length, fL, of ionized gas is ${\mathrm{DM}}_{\mathrm{LMC}}^{2}/\mathrm{EM}\approx 150$ pc, where L is the projected length and f is the filling factor (the fraction of the line of sight for free electron density). For reasonable values of f, L/d is of the order of the angular scale of the Hα emission (where d = 50 kpc for the LMC). If we assume the angular size is 13 arcmin (as we observed from the MCELS image), the inferred filling factor f ≈ 0.8 and the length L ≈ 180 pc. Those values are reasonable for an evolved HII region. If the pulsar turns out to be young in future timing analysis, it is likely embedded in this diffuse gas region.

In our continuum observations, VAST J052348.6−712552 shows strong variability with a modulation index ∼43% at 888 MHz over timescales of ∼days (see the lightcurve in Figure 2). There is no significant continuum variability within the 2.5 hr MeerKAT observation or the 12 hr ASKAP EMU observation. A structure–function analysis shows that the variability timescale is ∼17 d, though this is poorly constrained due to the absence of samples between 1 d and 14 d. This strong variability could be intrinsic (e.g., pulsar nulling; Backer 1970) or related to propagation effects (e.g., diffractive scintillation; Rickett 1990). The diffractive scintillation bandwidth follows Δf_DISS ∼ C₁/2π τ_sc MHz (Cordes & Lazio 2002), where C₁ = 1.16 assuming a Kolmogorov spectrum and τ_sc is the scattering time. The predicted scattering delay for the LMC pulsars in this line of sight is τ_sc ∼ 10⁻² ms based on the YMW16 model, and thus Δf_DISS ∼ 10⁻² MHz. The scintillation strength, $u=\sqrt{f/{\rm{\Delta }}{f}_{\mathrm{DISS}}}$ where f is the observing frequency, is therefore u ≫ 1, implying a strong scattering regime (i.e., diffractive scintillation and/or refractive scintillation). Diffractive scintillation in this case is an unlikely explanation, as the calculated Δf_DISS is about 10⁻² MHz, far lower than the observing frequency channel width ∼1 MHz. For refractive scintillation, the calculated modulation index $m=\sqrt{{u}^{5/3}}\sim 16 \%$, and the timescale ∼10² days. Walker (1998) estimated the effects of interstellar scintillation (arising from the Milky Way) on extragalactic sources using the TC93 model (Taylor & Cordes 1993), and we find a different modulation index of m ∼ 42% and timescale of ∼2 day based on this model. These estimations are not too far from our observed results, and we therefore think the variability can be explained by propagation effects. However, we cannot rule out the possibility of contribution from intrinsic variation. As we mentioned before, the large duty cycle of the pulsar suggests that the pulsar could be an aligned rotator, which is known to show larger levels of nulling (Cordes & Shannon 2008). Aligned rotating pulsars also show mode-changing behavior (e.g., PSR J1107-5907, Hobbs et al. 2016).

The source shows very steep spectral indices (varying from α ∼ −4 to α ∼ −2) even compared to the pulsar population (average α of −1.4; Bates et al. 2013, and α < −2.5 for the fastest rotating millisecond pulsar; e.g., Frail et al. 2016a). The high-quality MeerKAT continuum data show an unusual radio spectral shape, with an upturn at 1.36 GHz from α = −4.4 to α = 2.34. The upturn at ∼1 GHz is hard to explain as most pulsars can be described using a simple power-law spectrum. Some pulsar spectra are known to show a turnover at low frequencies due to synchrotron self-absorption or thermal free–free absorption; however, this is the opposite of what we see (i.e., transitioning from a flat to a steeper spectrum; see examples in Jankowski et al. 2018). We also note that some pulsars can have an upturn spectrum at a higher frequency, but this usually occurs at millimeter wavelengths (≳10 GHz; Kramer et al. 1996). Therefore, we instead consider external effects. Tuntsov et al. (2017) found that scintillation can cause kinks, bumps, and wiggles in the broadband radio spectrum of a quasar. If PSR J0523−7125 is scintillating (as we suggested above), we might be able to see unusual spectra structures like these.

Further observations should be able to conclusively determine a timing solution, providing an age and spin-down luminosity of the source. We will discuss these properties in a later paper.

The Magellanic Clouds have been targeted several times with pulsar surveys using the Parkes telescope (e.g., McCulloch et al. 1983; McConnell et al. 1991; Crawford et al. 2001; Manchester et al. 2006; Ridley et al. 2013), identifying 31 extragalactic pulsars. PSR J0523−7125 is brighter than all of these but was not identified in these surveys. Its wide pulse profile and steep spectrum could be responsible for its nondetection since they would reduce the signal-to-noise ratio. This is especially true in surveys with the Parkes Multibeam receiver at 1400 MHz, where many recent searches have been conducted. As described in Section 2.2, we cannot easily detect the pulsar using traditional methods from targeted Parkes data. However, we can clearly identify it in Parkes using only low-frequency data from the UWL receiver, consistent with the spectral properties determined from continuum images.

There are many continuum-based pulsar searches using a steep spectral index as the selection metric (e.g., Damico et al. 1985; Camilo et al. 2000; Maan et al. 2018; Hyman et al. 2019), but only a limited number of new pulsars have been found (e.g., Bhakta et al. 2017; Frail et al. 2018). The steep-spectrum selection normally requires two observations at different frequencies to measure the spectral index. Moreover, there are also other types of sources that can have very steep spectra, such as high-redshift radio galaxies (O'Dea 1998). Circularly polarized sources without deep infrared/optical counterparts are highly likely to be pulsars. We also note that such sources may belong to an as yet unknown class or classes. For instance, the Galactic Center Radio Transient (GCRT; Hyman et al. 2002, 2005), two steep-spectrum, polarized sources near the Galactic bulge (C1748−2827 and C1709−3918; Hyman et al. 2021), and a polarized transient recently discovered in the VAST survey near the Galactic Centre (ASKAP J173608.2−321635; Wang et al. 2021) are circularly polarized sources without deep infrared/optical counterparts, whose natures are still unknown.

There are an increasing number of large-scale radio continuum surveys, ranging from low-frequency surveys such as the Low-Frequency Array (LOFAR) Two-meter Sky Survey (LoTSS; Shimwell et al. 2017), the Giant Metrewave Radio Telescope (GMRT) 150 MHz All-Sky Survey (TGSS; Intema et al. 2017), and the GaLactic and Extragalactic All-sky Murchison Widefield Array survey (GLEAM; Hurley-Walker et al. 2017), to gigahertz-frequency surveys such as the ASKAP Rapid Commissioning Survey (RACS; McConnell et al. 2020), the Evolutionary Map of the Universe survey with ASKAP (EMU; Norris et al. 2021), the Polarization Sky Survey of the Universe's Magnetism (POSSUM; using the same sky coverage as EMU with full polarization; Gaensler et al. 2010), and the Karl G. Jansky Very Large Array (VLA) Sky Survey (VLASS; Lacy et al. 2020). With improved instruments in the Square Kilometre Array era, instantaneous large fields of view and great sensitivity will be even more common, leading to the detection of large numbers of radio sources across the sky. Some or most of the surveys include circular polarization measurements. Apart from radio wavelengths, there are significant improvements in multiwavelengths surveys (with better sensitivity and large sky coverage), e.g., the VISTA Variables in the Via Lactea (VVV; Minniti et al. 2010), the VISTA Hemisphere Survey (VHS; McMahon et al. 2013), the Dark Energy Survey (DES; Abbott et al. 2018), and the Gaia mission (Gaia Collaboration et al. 2016). These deep multiwavelength surveys can greatly help with the classification of a circularly polarized object (e.g., to check if there is any stellar counterpart). As shown in Figure 7, which listed known circular polarized objects including various stellar objects, the radio to near-infrared flux ratio (plus the fractional circular polarization) can be used as a good diagnostic tool to select strong pulsar candidates.

Zoom In Zoom Out Reset image size

One limitation of circular polarization selection is that not all pulsars have a high level of circular polarization (the median fraction ∼10%; Han et al. 1998; Johnston & Kerr 2018). As estimated by Kaplan et al. (2019), even deep ASKAP continuum searches (e.g., EMU, with sensitivity ∼50 μJy) for circularly polarized sources are unlikely to find large numbers of new pulsars. However, it may be critical to identify extreme pulsars that are missed in traditional pulsar surveys. For example, the high electron density in the Galactic Centre direction (which has much higher stellar densities) causes strong scattering and makes pulsars relatively difficult to discover through traditional periodicity searches.

Extragalactic pulsar searches in continuum images are also possible. To date, the Magellanic Clouds are still the only place where extragalactic pulsars have been detected, even after several attempts at searching for pulsars in other galaxies including M31, M33, and nearby dwarf galaxies (e.g., McLaughlin & Cordes 2003; Bhat et al. 2011; Rubio-Herrera et al. 2013; Mikhailov & van Leeuwen 2016; van Leeuwen et al. 2020). We can estimate the number of detectable pulsars in M31 (our neighbor galaxy) through continuum surveys, by putting all known Galactic pulsars at the distance of M31 (785 ± 25 kpc; McConnachie et al. 2005). We assume that the number of pulsars in M31 and our Galaxy is similar, though some work suggests there may be a smaller radio pulsar population in M31 (van Leeuwen et al. 2020). We consider using the Next Generation Very Large Array (ngVLA; Murphy et al. 2018), which is about 10 times more sensitive than the VLA and can achieve a resolution at ∼mas level. The ngVLA is located in the Northern hemisphere and therefore also ideal for observing M31 (and M33). The rms sensitivity of the ngVLA is expected to be ∼0.1 μJy beam⁻¹ for a 10 hr observation at ∼1–3 GHz. Assuming this sensitivity, more than ∼50 simulated M31 pulsars could be detected at the >5σ level in a survey of M31, and ∼5 could be detected in circular polarization. Although the estimated number is still very low, it is a promising possibility to detect the most luminous pulsars in M31 (e.g., those similar to PSR J0523−7125). With such an observation, we would also be able to measure their flux densities and spectral properties, providing valuable information for follow-up targeted periodicity searches.