Discovery of a New Redback Millisecond Pulsar Candidate: 4FGL J0940.3-7610
Samuel J. Swihart, Jay Strader, Elias Aydi, Laura Chomiuk, Kristen C. Dage, Laura Shishkovsky
DDraft version March 2, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Discovery of a New Redback Millisecond Pulsar Candidate: 4FGL J0940.3–7610
Samuel J. Swihart,
1, 2
Jay Strader, Elias Aydi, Laura Chomiuk, Kristen C. Dage, and Laura Shishkovsky National Research Council Research Associate, National Academy of Sciences, Washington, DC 20001, USA,resident at Naval Research Laboratory, Washington, DC 20375, USA Center for Data Intensive and Time Domain Astronomy, Department of Physics and Astronomy,Michigan State University, East Lansing, MI 48824, USA
ABSTRACTWe have discovered a new candidate redback millisecond pulsar binary near the center of theerror ellipse of the bright unassociated
Fermi -LAT γ -ray source 4FGL J0940.3–7610. The candidatecounterpart is a variable optical source that also shows faint X-ray emission. Optical photometric andspectroscopic monitoring with the SOAR telescope indicates the companion is a low-mass star in a 6.5-hrorbit around an invisible primary, showing both ellipsoidal variations and irradiation and consistent withthe properties of known redback millisecond pulsar binaries. Given the orbital parameters, preliminarymodeling of the optical light curves suggests an edge-on inclination and a low-mass ( ∼ . . M (cid:12) )neutron star, along with a secondary mass somewhat more massive than typical (cid:38) . M (cid:12) . Thiscombination of inclination and secondary properties could make radio eclipses more likely for thissystem, explaining its previous non-discovery in radio pulsation searches. Hence 4FGL J0940.3–7610may be a strong candidate for a focused search for γ -ray pulsations to enable the future detection of amillisecond pulsar. INTRODUCTIONPulsars form the largest population of Galactic
Fermi -LAT γ -ray sources with clear associations (Abdollahiet al. 2020). Follow-up studies of as-yet unidentified γ -ray sources continue to reveal new compact binaries,especially those containing millisecond pulsars (MSPs)spun up to fast periods through accretion from a com-panion (e.g., Parent et al. 2019; Corongiu et al. 2021;Wang et al. 2020). γ -ray emission from these objectsmay be ubiquitous (Abdo et al. 2013) and can serve asa signpost for MSPs difficult to find or study at otherwavelengths, such as “spider” (black widow or redback)MSP binaries with non-degenerate companions that haveextensive eclipses in the radio (e.g., Polzin et al. 2019;Crowter et al. 2020; Kudale et al. 2020; Pan et al. 2020).Finding the multi-wavelength (optical, X-ray, or radio)counterpart of unidentified Fermi -LAT γ -ray sources canbe challenging. The γ -ray error ellipses are relativelylarge (often dozens of square arcminutes on the sky),potentially containing a substantial number of X-ray orradio sources and hundreds or more optical sources. Corresponding author: Samuel J. [email protected]
Characteristic multi-wavelength behavior can help nar-row down possible associations for spider MSPs. Mostof these show a characteristic hard X-ray spectrum withrapid stochastic and orbital variability, likely due to anintrabinary shock that occurs between the wind fromthe companion star and the pulsar wind (e.g., Gentileet al. 2014; Wadiasingh et al. 2017; Romani & Sanchez2016; Al Noori et al. 2018). In addition, γ -ray emittingcompact binaries with secondaries that substantially filltheir Roche lobes should have detectable periodic opticalvariability due to the tidal deformation (and sometimesirradiation) of the secondary star (e.g., Romani & Shaw2011; Strader et al. 2015; Halpern et al. 2017; Li et al.2018; Swihart et al. 2020).In this paper, we present the discovery of an X-rayand variable optical source that we argue is likely a newredback MSP binary associated with the unidentified Fermi source 4FGL J0940.3–7610. OBSERVATIONS2.1.
The γ -ray Source The γ -ray source that is the subject of this work wasreported in the first full catalog of Fermi -LAT sources,based on the first eleven months of survey data, and waslisted as 1FGL J0940.2–7605 (Abdo et al. 2010). It hasappeared in each subsequent catalog, most recently as a r X i v : . [ a s t r o - ph . H E ] M a r Swihart et al. NE Optical: DSS10 arcmin
X-raypositionOptical Variable NE Optical: DSS 3 arcmin
10 arcsec
Figure 1.
Left: Optical Digitized Sky Survey image of the field showing the positions and overlapping 95% error ellipses fromthe 1FGL, 2FGL, and 3FGL catalogs corresponding to the γ -ray source 4FGL J0940.3–7610 (magenta), along with the positionof the Swift
X-ray source (blue circle, see § i (cid:48) images along with the Swift
X-ray position and 90% confidence region (blue).The optical variable that is the subject of this work is marked with a red circle, and is ∼ (cid:48)(cid:48) away from a nearby, brighter starthat is unrelated to the γ -ray source. The large streak in the inset is a diffraction spike from a nearby bright star. ∼ (cid:48) . The γ -raysource shows a spectrum with significant curvature (6 . σ for the LogParabola model; 7 . σ for a cut-off power law).The 0.1–100 GeV energy flux is (6 . ± . × − ergs − cm − , which is a strong detection as expected fora source present from the 1FGL catalog. There is nosignificant evidence for variability (Ballet et al. 2020).The spectral curvature and lack of evidence for vari-ability are typical of γ -ray observations of pulsars (Abdoet al. 2013). Indeed, this source is listed as a likely MSPin several papers that used machine learning to classifyunassociated LAT sources from 3FGL or 4FGL (e.g., SazParkinson et al. 2016; Luo et al. 2020).2.2. X-rays
The field of 4FGL J0940.3–7610 was observed with theSwift X-ray Telescope (XRT) nine times between 2011and 2015 as part of a program following-up unidentified
Fermi -LAT sources (Stroh & Falcone 2013). The totalgood exposure time on source is ∼ γ -ray error ellipse, only45 (cid:48)(cid:48) from its center, with an ICRS (R.A., Dec.) of (09:40:23.54, –76:10:01.0) and a 90% positional uncer-tainty of 4.6 (cid:48)(cid:48) (Figure 1). The more marginal X-raysource is near the edge of the 4FGL error ellipse and hasno apparent optical counterpart.2SXPS J094023.5–761001 has a 0.3–10 keV count rateof (6 . ± . × − ct s − . Assuming a power law spec-trum with standard redback photon index of Γ = 1 . N H = 9 . × cm − (HI4PI Collabora-tion et al. 2016) this corresponds to an unabsorbed fluxof (3 . ± . × − erg s − cm − . The photon indexinferred from the hardness ratios is Γ = 0 . +1 . − . , whichis nominally hard but with large enough uncertainties toadmit nearly any interpretation.2.3. Optical Spectroscopy
There are two optical sources in
Gaia
DR2 (Gaia Col-laboration et al. 2018) within the astrometric uncertaintyof 2SXPS J094023.5–761001. These optical sources areseparated by only 1.4 (cid:48)(cid:48) (Figure 1) and so in principleeither could be the counterpart to the X-ray source. Wenote that these two
Gaia sources have very differentproper motions and hence are not comoving.We obtained optical spectroscopy of both these targetsusing the Goodman spectrograph (Clemens et al. 2004)on a two-night run with the SOAR telescope on 2018Dec 10/11. The brighter of these two sources ( G = 18 . G = 19 . − ) shifts over just afew hr on 2018 Dec 11, proving that it is a close binary. new candidate redback MSP Gaia
DR2 ICRS position of this source is (R.A., Dec.) =(09:40:23.787, –76:10:00.13), which we take as the bestposition available.We performed spectroscopic monitoring of this sourceover seven nights from 2018 Dec 11 to 2019 Mar 25. Foreach spectrum we used a 400 l mm − grating and a0.95 (cid:48)(cid:48) slit, yielding a resolution of about 5.6 ˚A (full widthat half-maximum). All spectra covered a wavelengthrange of ∼ ◦ to avoidcontamination from the nearby brighter source.Each spectrum was reduced and optimally extractedin the normal manner using standard packages in IRAF (Tody 1986). We measured barycentric radial velocitiesthrough cross-correlation with bright template stars ofsimilar spectral type, with the cross-correlations donesimultaneously in a bluer region around Mg b and a red-der region from ∼ Optical Photometry
To obtain light curves of the optical source, we alsoperformed imaging observations with SOAR/Goodmanon two nights in 2019, with the CCD binned 2x2 to apixel scale of 0.3 (cid:48)(cid:48) per pixel. On 2019 Apr 08 we useda series of alternating exposures with the g (cid:48) (exposuretime 300 s) and i (cid:48) (180 or 240 s) filters, with medianseeing of about 1.5 (cid:48)(cid:48) . On 2019 Apr 21 we observed onlyin i (cid:48) (300 s per exposure), and the median seeing wasabout 1.2 (cid:48)(cid:48) .The reduction was done as outlined in Swihart et al.(2020), including bias correction and flat fielding. Weperformed differential aperture photometry with respectto nearby, non-variable comparison stars, using 36 starsin g (cid:48) and 32 stars in i (cid:48) . We then calibrated these mag-nitudes to a standard system, using Gaia
DR2 G magfor g (cid:48) and Skymapper for i (cid:48) (Wolf et al. 2018). Finally,we made a frame-by-frame correction to account for the“excess” flux from the non-variable nearby (1.4 (cid:48)(cid:48) in projec-tion) source, by measuring its magnitude in both a smallaperture and the combined magnitude of both sourcesin a large (10 (cid:48)(cid:48) ) aperture. https://gea.esac.esa.int/archive/documentation/GDR2/Data processing/chap cu5pho/sec cu5pho calibr/ssec cu5pho PhotTransf.html Table 1.
Modified Barycentric RadialVelocities of 4FGL J0940.3–7610MBJD RV err.(d) (km s − ) (km s − )58463.2613142 322.2 22.958463.2787955 334.6 24.458463.2998051 184.1 23.158463.3174170 125.2 24.458484.1968846 –158.4 23.158484.2147938 –227.9 27.458484.2360225 –291.2 21.958484.2534997 –266.3 24.858484.2922107 –41.4 26.158491.2160776 –65.7 24.758491.2336100 –178.8 25.258491.2564964 –193.9 22.858491.2957546 –166.2 21.958491.3132683 –154.9 22.958525.0872665 –270.6 25.058525.1047505 –297.7 25.358525.1238465 –218.9 23.858525.1378938 –151.2 24.258525.1563986 –32.0 22.358525.1742462 39.1 20.758525.2992549 94.6 24.158525.3167359 –25.6 20.958537.2547892 –237.7 24.558537.2725452 –268.0 22.158547.1325566 254.7 24.558547.1506098 301.1 21.858547.1730925 302.8 27.658547.1905521 252.7 23.058547.2080537 129.4 24.958567.1326788 117.4 27.458567.1501386 216.6 26.858567.1896342 338.9 25.758567.2305353 175.5 29.358567.2527752 19.6 24.858567.2702688 –67.7 22.0 Our final sample consists of 36 and 102 photometricmeasurements in g (cid:48) and i (cid:48) , respectively. The mean ob-served magnitudes of our target are g (cid:48) = 19 .
754 mag and i (cid:48) = 18 .
832 mag, with median uncertainties (statistical+ systematic) of ∼ ∼ g (cid:48) and i (cid:48) ,respectively. Swihart et al. RESULTS AND ANALYSIS3.1.
Optical Spectroscopy and Orbital Parameters
Overall, the spectra of the optical source are consistentwith that of a moderately cool, late-G to early-K typedwarf star (we reach similar conclusions from our lightcurve fitting in § TheJoker (Price-Whelan et al. 2017). We initially fit for the binary period P , time of ascending node T , orbital semi-amplitude K , systemic velocity γ , and the eccentricity e . However,we found no significant evidence for non-zero eccentricity,so for the remainder of our analysis we assume a circularorbit. This is expected, since in the absence of an ad-ditional perturbing body, the binary is expected to becircularized in a very short timescale ( (cid:46) yr) at theobserved orbital period for a typical redback mass ratio(Zahn 1977).For our circular model, we find P = 0 . T = 58525 . K = 293 . ± . − , and γ = 28 . ± . − . Overall the fit isvery good statistically, with a rms of 22.6 km s − and a χ /d.o.f. = 33.2/31. We plot this model along with theradial velocity data in Figure 2.We derive the binary mass function f ( M ) using poste-rior samples from our radial velocity modeling: f ( M ) = P K (1 − e ) / πG = M sin i (1 + q ) , (1)where M is the mass of the primary, q = M /M isthe mass ratio, and i is the system inclination. Usingour spectroscopic results, f ( M ) = 0 . ± . M (cid:12) , whichrepresents a lower limit on the primary mass. Any normalstar of this minimum mass would be apparent in theoptical spectra; since it is not seen, the clear implicationis that the primary is a compact object. We note that thisvalue of the mass function is fairly typical for redbacks(Strader et al. 2019). Figure 2.
Circular Keplerian fit to the SOAR/Goodmanbarycentric radial velocities of the optical counterpart to4FGL J0940.3–7610.
Neither q nor i can be well-constrained from our low-resolution optical spectroscopy alone, but some infor-mation can be obtained from modeling the light curves,which we do in the next subsection.3.2. Light Curve Modeling
Folding the g (cid:48) and i (cid:48) photometry on the spectroscopicorbital period shows the characteristic double-peakedmorphology expected for a tidally-distorted secondary,with two equally bright maxima when the system isviewed at quadrature (Figure 3). For simple ellipsoidalvariations the expectation is that, due to gravity dark-ening, the secondary should be fainter at φ = 0 .
75 (atits superior conjunction) than at φ = 0 .
25. For 4FGLJ0940.3–7610 we see the opposite, implying that the sec-ondary is heated on its tidally-locked “dayside”. Forredbacks, this heating is typically inferred to be eitherdirectly from the pulsar wind or more indirectly by re-processed emission from an intrabinary shock (Romani& Sanchez 2016; Sanchez & Romani 2017; Wadiasinghet al. 2018).We modeled our photometry using the Eclipsing LightCurve (ELC, Orosz & Hauschildt 2000) code, assuminga compact invisible primary with no accretion disk, anda tidally distorted secondary with the orbital propertiesderived from our spectroscopic results.Since the mass ratio is unknown and this parameteris very challenging to tightly constrain with a modestamount of optical photometry, we instead chose to do a new candidate redback MSP phase (P=0.270639 d) m ag M M M Figure 3.
SOAR g (cid:48) (blue) and i (cid:48) (red) photometry of the op-tical counterpart to 4FGL J0940.3–7610 folded on the best-fitperiod and ephemeris from our radial velocity measurements.The best-fit ELC light curve models with varying secondarymasses are shown with dashed lines. The differences betweenthe models are small, but only models with a secondary mass (cid:38) . M (cid:12) can accommodate a nuetron star-mass primary. series of fits with a broad range of typical redback com-panion masses ( ∼ . − . M (cid:12) ) and report the results fordifferent assumed companion masses, requiring that theobserved K be matched. For each assumed companionmass we ran models fitting for the system inclination i ,the Roche lobe filling factor f , the effective temperature T of the companion, and the irradiating luminosity.We note that ELC models the irradiation in a rela-tively simple manner, and for some other redbacks morecomplex heating models have been explored. These takeinto account reprocessed or magnetically-ducted emissionfrom an extended intrabinary shock between the pulsarand companion, or redistibution of heat on the stellarsurface due to diffusion and convection within the photo-sphere (e.g., Romani & Sanchez 2016; Sanchez & Romani2017; Voisin et al. 2020). Given the poorly constrainedmass ratio of the binary and the modest amount of ourphotometry, for the purpose of this discovery paper westick to these simpler models, while emphasizing theneed for more sophisticated modeling once better dataare available.Now to the light curve modeling results. First, in allcases, we find the best fits come from models that are rela-tively edge-on, with inclinations (cid:38) ◦ . In all the modelsthat were good statistical fits to the data, the effectivetemperature of the companion was ∼ ±
200 K, consis-tent with the spectral types inferred from the optical spec-tra. The Roche lobe filling factors were relatively well-constrained, with values ranging from f ∼ . − .
95, corresponding to stellar radii of R ∼ . − . R (cid:12) , de-pending on the model (the more massive companions arealso larger). As revealed by the light curves, ellipsoidalvariations from the partially Roche lobe-filling secondaryare apparent, suggesting the level of heating is not asextreme as observed in some redbacks (e.g., Schroeder& Halpern 2014; Cho et al. 2018; Linares 2018; Swihartet al. 2020). Our modeling supports this: we find themaximum “dayside” temperature of the secondary to beonly ∼ . M (cid:12) , the lowest known for aredback (PSR J1622–0315; Strader et al. 2019) yields abest-fit model with i ∼ ◦ and a resulting primary massof ∼ . M (cid:12) , too low for a neutron star. Such a systemcould in principle have a white dwarf primary, althoughX-ray emission would be difficult to explain since thereis no evidence of accretion (see § . M (cid:12) ; Strader et al. 2019), the bestfit inclination is 81 . +7 . − . , corresponding to a primarymass of 1 . +0 . − . M (cid:12) .If the companion is on the more massive end of theredback distribution, approaching 0 . M (cid:12) , such as theconfirmed redback PSR J1306–40 (Keane et al. 2018;Linares 2018; Swihart et al. 2019) or the candidate red-back 3FGL J0212.1+5320 (Li et al. 2016; Shahbaz et al.2017), then the best fit inclination is i = 87 . +2 . − . corre-sponding to M = 1 . +0 . − . M (cid:12) .These models are very good fits to the data and arenearly identical statistically, with χ /d.o.f. = 79.1/133and 81.2/133 for the models with 0 . M (cid:12) and 0 . M (cid:12) secondaries, respectively. We show these models withthe data, along with a model for a 0 . M (cid:12) secondary inFigure 3, where it is clear that variations in the secondarymass results in only small differences between the models.The models perform most poorly in g (cid:48) near compan-ion superior conjunction ( φ = 0 . Swihart et al.
The next largest excursions from the model appear inthe g (cid:48) data just after quadrature, near φ = 1 .
0, wherethe data appear brighter than expected from these mod-els. Given the evidence for irradiative heating and thelikely strong magnetic fields induced on the tidally-locked(and thereby rapidly-rotating) companion’s surface, itis possible that accelerated intrabinary shock particlesare magnetically-ducted to the surface, mimicking oneor more hot-spots (e.g., Sanchez & Romani 2017). Thesespots are one way to cause irregular heating patternslike these in the light curves of redback secondaries if thespots are shifted in azimuth to be off the line connectingthe primary and secondary (e.g., Strader et al. 2019;Swihart et al. 2019). We do not attempt to model thesespots in this discovery paper, but we suggest these morecomplex heating models be explored when future databecome available.Overall we conclude that, given the spectroscopic or-bital parameters, the light curves suggest an edge-onorientation with a relatively low neutron star mass ( ∼ . . M (cid:12) ). 3.3. Distance
The optical binary has a marginally significant paral-lax listed in
Gaia
DR2: (cid:36) = 0 . ± .
263 mas (GaiaCollaboration et al. 2018). Taking into account a globalparallax offset of +0.029 mas (Lindegren et al. 2018), andusing a simple exponential length prior of 1.35 kpc (As-traatmadja & Bailer-Jones 2016), this parallax implies adistance of 1 . +1 . − . kpc.We independently estimate the distance using our lightcurve models following the procedures described in Swi-hart et al. (2017). For the model with a 0 . M (cid:12) sec-ondary (typical for a redback), the inferred distance is2 . +0 . − . kpc, where the uncertainties represent the rangeof distances found for different assumptions for the dustreddening and metallicity (see e.g., Swihart et al. 2019).Models with less (more) massive secondaries would givesmaller (larger) distances; for example, a 0 . M (cid:12) sec-ondary would give a distance in the range 1.4–1.7 kpc.At this point, both the light curve modeling and Gaia distances have substantial random and systemtic uncer-tainties, but also concur on a distance around ∼ Gaia parallax distanceshould improve in future data releases.3.4.
The X-ray and γ -ray emission The inferred 0.5–10 keV X-ray luminosity of 4FGLJ0940.3–7610 is (1 . ± . × ( d/ erg s − , avalue squarely in the range for that observed for knownredbacks in the pulsar state (e.g., Linares 2014; Hui & Li 2019; Strader et al. 2019). This is likewise the casefor the 0.1–100 GeV γ -ray luminosity of (3 . ± . × ( d/ erg s − , and as noted in § γ -rayspectrum and lack of variability is consistent with theproperties of MSPs.Miller et al. (2020) discuss the use of the ratio of X-ray to γ -ray flux ( F X /F γ ) to classify Galactic compactbinaries, focusing on the distinction between spider MSPsand transitional MSPs in the sub-luminous disk state ina distance-independent manner. The median value of F X /F γ for redbacks is 0.012, with a range from 0.003to 0.12. For 4FGL J0940.3–7610, F X /F γ = 0 . ± . F X /F γ in the range 0.28–0.43). Hence F X /F γ isconsistent with the classification of 4FGL J0940.3–7610as a normal redback in the pulsar state.The properties of the binary are not well-explainedas a chance alignment with a γ -ray source: while theorbital properties and light curve modeling could admita solution with a primary of white dwarf mass, there isno evidence in the optical spectra for accretion, withoutwhich it would be hard to reach the observed L X ∼ erg s − , or to produce the observed irradiation in thelight curves. DISCUSSION AND CONCLUSIONSWe have discovered a short-period (6.5 hr) compactbinary with X-ray emission near the center of the errorellipse of the unassociated
Fermi γ -ray source 4FGLJ0940.3–7610. The optical and X-ray properties of thesource are well-explained as a redback MSP but are notconsistent in a straigtforward manner with any otherclass of source. Hence we think the binary is likely tobe a redback millisecond pulsar and the counterpart to4FGL J0940.3–7610.Compared to known (or strong candidate) redbacks,the orbital period, high-energy properties, and distanceare typical. The inclination appears relatively edge-on.The mass of the neutron star appears to be low, closerto ∼ . M (cid:12) as for the redbacks PSR J1723–2837 (vanStaden & Antoniadis 2016) and PSR J2039–5617 (Clarket al. 2021) rather than ∼ . M (cid:12) as found for a typical(median) redback (Strader et al. 2019).There is also a hint that the secondary might be onthe more massive side for redback companions (perhaps (cid:38) . M (cid:12) ). This has potential relevance for the factthat a radio pulsar has not yet been detected towardthis region despite extensive searches (e.g., Camilo et al.2016). There are several other systems that have com-pelling optical and X-ray evidence for being redbacks new candidate redback MSP γ -ray pulsations.Although all the available evidence points towards aredback classification, this needs to be confirmed withadditional data. Ultimately this requires a detection ofa pulsar in either radio or γ -ray observations. However,much deeper X-ray data than the shallow Swift/XRTobservations presented here could allow the detectionof orbital variability or a hard X-ray spectrum, whichwould provide compelling supporting evidence for ourclassification, and we were recently approved for XMM -Newton observations for AO-20 (2021 May–2022 Apr).The discovery of yet another redback candidate as-sociated with a persistent γ -ray source that has beenknown since the first year after Fermi ’s launch suggests a substantial population of compact binaries still awaitsdetection, and that multi-wavelength follow-up of unas-sociated γ -ray sources remains a fruitful route to findnew candidate MSPs.ACKNOWLEDGEMENTSThis research was performed while SJS held a NRCResearch Associateship award at the Naval ResearchLaboratory. Work at the Naval Research Laboratory issupported by NASA DPR S-15633-Y.We also acknowledge support from NSF grant AST-1714825 and the Packard Foundation.Based on observations obtained at the Southern As-trophysical Research (SOAR) telescope, which is a jointproject of the Minist´erio da Ciˆencia, Tecnologia, In-ova¸c˜oes e Comunica¸c˜oes (MCTIC) do Brasil, the U.S.National Optical Astronomy Observatory (NOAO), theUniversity of North Carolina at Chapel Hill (UNC), andMichigan State University (MSU).We acknowledge the use of public data from the Swiftdata archive.REFERENCES Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010, ApJS,188, 405Abdo, A. A., Ajello, M., Allafort, A., et al. 2013, ApJS, 208,17Abdollahi, S., Acero, F., Ackermann, M., et al. 2020, ApJS,247, 33Al Noori, H., Roberts, M. S. E., Torres, R. A., et al. 2018,ApJ, 861, 89Astraatmadja, T. L., & Bailer-Jones, C. A. L. 2016, ApJ,832, 137Ballet, J., Burnett, T. H., Digel, S. W., & Lott, B. 2020,arXiv e-prints, arXiv:2005.11208Bellm, E. C., Kaplan, D. L., Breton, R. P., et al. 2016, ApJ,816, 74Camilo, F., Reynolds, J. E., Ransom, S. M., et al. 2016, ApJ,820, 6Cho, P. B., Halpern, J. P., & Bogdanov, S. 2018, ApJ, 866,71Clark, C. J., Nieder, L., Voisin, G., et al. 2021, MNRAS, 502,915Clemens, J. C., Crain, J. A., & Anderson, R. 2004, inProc. SPIE, Vol. 5492, Ground-based Instrumentation forAstronomy, ed. A. F. M. Moorwood & M. Iye, 331–340Corongiu, A., Mignani, R. P., Seyffert, A. S., et al. 2021,MNRAS, 502, 935 Crowter, K., Stairs, I. H., McPhee, C. A., et al. 2020,MNRAS, 495, 3052Eastman, J., Siverd, R., & Gaudi, B. S. 2010, PASP, 122,935Evans, P. A., Page, K. L., Osborne, J. P., et al. 2020, ApJS,247, 54Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al.2018, A&A, 616, A1Gentile, P. A., Roberts, M. S. E., McLaughlin, M. A., et al.2014, ApJ, 783, 69Halpern, J. P., Strader, J., & Li, M. 2017, ApJ, 844, 150HI4PI Collaboration, Ben Bekhti, N., Fl¨oer, L., et al. 2016,A&A, 594, A116Hui, C. Y., & Li, K. L. 2019, Galaxies, 7, 93Keane, E. F., Barr, E. D., Jameson, A., et al. 2018, MNRAS,473, 116Kudale, S., Roy, J., Bhattacharyya, B., Stappers, B., &Chengalur, J. 2020, ApJ, 900, 194Li, K.-L., Kong, A. K. H., Hou, X., et al. 2016, ApJ, 833, 143Li, K.-L., Hou, X., Strader, J., et al. 2018, ApJ, 863, 194Linares, M. 2014, ApJ, 795, 72—. 2018, MNRAS, 473, L50Linares, M., Miles-P´aez, P., Rodr´ıguez-Gil, P., et al. 2017,MNRAS, 465, 4602Linares, M., Shahbaz, T., & Casares, J. 2018, ApJ, 859, 54