The Green Bank Northern Celestial Cap Pulsar Survey. VI. Timing and Discovery of PSR J1759+5036: A Double Neutron Star Binary Pulsar
Gabriella Agazie, Michael Mingyar, Maura McLaughlin, Joseph Swiggum, David Kaplan, Harsha Blumer, Pragya Chawla, Megan DeCesar, Paul Demorest, William Fiore, Emmanuel Fonseca, Joseph Gelfand, Victoria Kaspi, Vladislav Kondratiev, Malcolm LaRose, Joeri van Leeuwen, Lina Levin, Evan Lewis, Ryan Lynch, Alexander McEwen, Hind Al Noori, Emilie Parent, Scott Ransom, Mallory Roberts, Ann Schmiedekamp, Carl Schmiedekamp, Xavier Siemens, Renée Spiewak, Ingrid Stairs, Mayuresh Surnis
DDraft version February 23, 2021
Typeset using L A TEX twocolumn style in AASTeX63
The Green Bank Northern Celestial Cap Pulsar Survey. VI. Timing and Discovery of PSRJ1759+5036: A Double Neutron Star Binary Pulsar
G. Y. Agazie,
1, 2, 3
M. G. Mingyar,
1, 2, 4
M. A. McLaughlin,
1, 2
J. K. Swiggum, D. L. Kaplan, H. Blumer,
1, 2
P. Chawla, M. DeCesar, P. B. Demorest, W. Fiore,
1, 2
E. Fonseca, J. D. Gelfand,
8, 9
V. M. Kaspi, V. I. Kondratiev, M. LaRose,
1, 2
J. van Leeuwen, L. Levin, E. F. Lewis,
1, 2
R. S. Lynch, A. E. McEwen, H. Al Noori, E. Parent, S. M. Ransom, M. S. E. Roberts,
8, 15
A. Schmiedekamp, C. Schmiedekamp, X. Siemens, R. Spiewak,
11, 18
I. H. Stairs, and M. Surnis Dept. of Physics and Astronomy, West Virginia University, Morgantown, WV 26501 Center for Gravitational Waves and Cosmology, West Virginia University, Chestnut Ridge Research Building, Morgantown, WV 26505,USA Center for Gravitation, Cosmology, and Astrophysics, Dept. of Physics, University of Wisconsin-Milwaukee, P.O. Box 413, Milwaukee,WI 53201, USA Dept. of Physics, Montana State University, Bozeman, MT, 59717, USA Dept. of Physics, 730 High St., Lafayette College, Easton, PA 18042, USA Dept. of Physics and McGill Space Institute, McGill Univ., Montreal, QC H3A 2T8, Canada National Radio Astronomy Observatory, P.O. Box O, Socorro, NM 87801, USA Center for Astro, Particle, and Planetary Physics, New York University Abu Dhabi, PO Box 129188, Abu Dhabi, UAE Affiliated Member, Center for Cosmology and Particle Physics, New York University, New York, NY, 10276, USA ASTRON, the Netherlands Institute for Radio Astronomy, Oude Hoogeveensedijk 4, 7991 PD Dwingeloo, The Netherlands Jodrell Bank Centre for Astrophysics, School of Physics and Astronomy, The University of Manchester, Manchester, M13 9PL, UK Green Bank Observatory, P.O. Box 2, Green Bank, WV 24494, USA Dept. of Physics, University of California, Santa Barbara, CA 93106, USA National Radio Astronomy Observatory, 520 Edgemont Rd., Charlottesville, VA 22903, USA Eureka Scientific, Inc. 2452 Delmer Street Suite 100 Oakland, CA 94602, USA Dept. of Physics, The Pennsylvania State University, Ogontz Campus, Abington, Pennsylvania 19001, USA Dept. of Physics, Oregon State University, Corvallis, OR 97331, USA Centre for Astrophysics and Supercomputing, Swinburne University of Technology, PO Box 218, Hawthorn, VIC 3122, Australia Dept. of Physics and Astronomy, University of British Columbia, 6224 Agricultural Road, Vancouver, BC V6T 1Z1 Canada
AbstractThe Green Bank North Celestial Cap (GBNCC) survey is a 350-MHz all-sky survey for pulsarsand fast radio transients using the Robert C. Byrd Green Bank Telescope. To date, the survey hasdiscovered over 190 pulsars, including 33 millisecond pulsars (MSPs) and 24 rotating radio transients(RRATs). Several exotic pulsars have been discovered in the survey, including PSR J1759+5036, abinary pulsar with a 176-ms spin period in an orbit with a period of 2.04 days, an eccentricity of 0.3,and a projected semi-major axis of 6.8 light seconds. Using seven years of timing data, we are ableto measure one post-Keplerian parameter, advance of periastron, which has allowed us to constrainthe total system mass to 2 . ± . M (cid:12) . This constraint, along with the spin period and orbitalparameters, suggests that this is a double neutron star system, although we cannot entirely rule out apulsar-white dwarf binary. This pulsar is only detectable in roughly 45% of observations, most likelydue to scintillation. However, additional observations are required to determine whether there may beother contributing effects. Keywords: pulsars, binary pulsar INTRODUCTION The Green Bank North Celestial Cap (GBNCC) pul-sar survey is a comprehensive survey of the northern http://astro.phys.wvu.edu/GBNCC/ a r X i v : . [ a s t r o - ph . H E ] F e b celestial sky ( δ > − ◦ ) at 350 MHz (Stovall et al. 2014)with the Robert C. Byrd Green Bank Telescope (GBT)in West Virginia. The GBNCC survey is projected tocover ∼
80% of the entire sky with sensitivity to mil-lisecond pulsars (MSPs), canonical pulsars, and sourcesof isolated dispersed pulses such as rotating radio tran-sients (RRATs) and fast radio bursts (FRBs). The nom-inal sensitivity to pulsars is 0.74 mJy with an 6% dutycycle (McEwen et al. 2020).Pulsars with a binary companion comprise about 6%of all known pulsars (ATNF Pulsar Catalogue v1.63;Manchester et al. 2005). Most MSPs are in binary sys-tems, as they are spun-up to millisecond periods throughangular momentum transfer from a companion throughRoche Lobe overflow (Alpar et al. 1982; Lorimer 2008).This accretion process typically results in an MSP-whitedwarf system in an extremely circular orbit, with eccen-tricity of 10 − (cid:47) e (cid:47) P ) ranging from16.9 ms to 185 ms and period derivative ( ˙ P ) range of2 . × − s s − to 1 . × − s s − (Tauris et al. 2017;Pol et al. 2019).Currently, 15 DNS systems are known; in 13 of them,we have detected the recycled pulsar, with P rangingfrom 16.9 ms to 185 ms (Pol et al. 2019). In two sys-tems, however, the younger canonical pulsar has beendetected, with P of 144 ms and 2.7 s, respectively. Thesesystems have eccentricities ranging from 0.08 to 0.83(Pol et al. 2019). Some DNS systems have measurablepost-Keplerian (PK) parameters, which can constrainthe pulsar and companion masses. Measurement of asingle PK parameter can constrain some combinationof the total system mass, pulsar mass, and companionmass. Two can fully constrain the pulsar and compan-ion masses and other system parameters such as orbitalinclination. A third makes the system over-determined,meaning it can be used to test theories of gravity by cal-culating the predicted value for the third PK parametergiven the first two and comparing it to the measuredvalue (Kramer et al. 2006).In this paper, we report the discovery and propertiesof PSR J1759+5036, a 176-ms binary pulsar in a 2-day,eccentric orbit. Its small ˙ P suggests partial recycling,which is consistent with the properties of DNS systems. Time frame Frequency (MHz) N obs
Detections2013–2014 820 24 102017–2019 820/350 54 2010/2019 350 7 512/2019 – 01/2020 820 24 15
Table 1.
Summary of observations for PSR J1759+5036with the number of observations (N obs ), observing frequen-cies, and detections per observation run indicated.
We present evidence, including a mass constraint, tosuggest this system is likely a DNS system. DISCOVERY AND TIMING OBSERVATIONSWe first detected PSR J1759+5036 in single-pulsesearch output from the discovery observation of PSRJ1800+5034, an unrelated 578-ms pulsar with a disper-sion measure (DM) of 22.7 pc cm − published in 2018(Lynch et al. 2018). The single pulses were detectedat a DM of 7.77 pc cm − two years after timing ob-servations ceased for the former source. The pulsarwas subsequently detectable in a periodicity search witha spin period of 176 ms, and timed as part of thePSR J1800+5034 observation campaign, consisting of24 observations taken on the GBT from February 2013to January 2014 at 820 MHz with time resolution of81.92 µ s and 2048 frequency channels. Dedispersing atthe DM of 7.77 pc cm − and refolding at the 176-ms spinperiod of PSR J1759+5036 produced 10 detections. AsPSR J1759+5036 was not the source of interest for theseobservations, it was not at the center of the beam whendata were being taken, possibly reducing the detectedflux.The entire timing dataset for PSR J1759+5036 spans2013–2020 with 109 total observations of which 50 weredetections. A summary of the timeline of observationruns can be found in Table 1. For all observations,the Green Bank Ultimate Pulsar Processing Instrument(GUPPI) backend was used, and radio frequency inter-ference (RFI) was excised using the pazi tool in the PSRCHIVE software package . We then summed each ob-servation in the time and frequency domain to generatebest profiles to compute times of arrival (TOAs) usingthe tool pat . For most observations, we calculated 4–7 TOAs, depending on detection signal-to-noise ratio(S/N), with a few marginal detections only producing1–3 useful TOAs. We made composite profiles for 350and 820-MHz data by adding together all the observa-tions of a particular frequency using the psradd tool. http://psrchive.sourceforge.net/ F l u x D e n s i t y ( m J y ) F l u x D e n s i t y ( m J y ) Figure 1.
Composite profiles of PSR J1759+5036 at 350(left) and 820 (right) MHz with integration times of 4.2 hrand 6.4 hr respectively. Both profiles have 256 phase binsand were used as standard profiles to recalculate TOAs inorder to refine the timing ephemeris.
We then used paas to generate a standard profile andremoved noise using psrsmooth (Figure 1).From 2017–2019 we obtained 54 observations ofPSR J1759+5036 on the GBT at 820 MHz with 40.96- µ s time resolution and 2048 frequency channels. Initialtiming efforts were complicated by the fact that the pul-sar is in a binary system, and was also undetectable at34 of the 54 epochs. We do not believe that extrin-sic factors, such as RFI, were a significant contributorto the lack of detections. Since timing required initialestimates of the Keplerian binary parameters, we mea-sured the barycentric P at each epoch, and used meth-ods from Bhattacharyya & Nityananda (2008) to cal-culate rough approximations. These were then refinedusing Tempo to build a phase-connected timing solutionand to study the intermittent nature of the detections.We show in Figure 2 that there is no apparent connec-tion between the binary orbital phase and the epochsof non-detections. This suggests that the lack of detec-tions is either intrinsic to the pulsar emission mechanismor due to external interstellar medium effects, such asscintillation, possibly in combination with a poorly con-strained position. A discussion of the scintillation willbe presented in Section 4.3.In May and June 2019 we conducted several observa-tions with the Karl G. Jansky Very Large Array (seeSection 2.1), through which we determined a more ac-curate position which was 7 . (cid:48) from our initial position(i.e. that of PSR J1800+5034). Given that the FWHMof the GBT 820-MHZ receiver is only 15 (cid:48) , our earlierobservations suffered significantly from degraded sensi-tivity due to a reduced gain. The offset caused a 10%and 44% reduction in sensitivity for 350-MHz and 820-MHz observations, respectively.We also carried out a short, high-cadence tim-ing campaign with the GBT at 350 MHz. At this https://sourceforge.net/projects/tempo/ M e a s u r e d S p i n P e r i o d . ( m s ) Figure 2.
The observed Doppler-modulated P of each epochvs orbital phase. Red lines signify epochs of non-detectionand do not appear to be associated with any particular phaseof orbit. Overlaid in black is a calculated model of the varia-tion of P across the orbital phase. Orbital phases have beencalculated in reference to the epoch of periastron. lower frequency, we had more success at detectingPSR J1759+5036 than at 820 MHz, with five of seven ob-servations resulting in a detection. This pulsar was alsoobserved during 350-MHz GBNCC survey observationsas a test source. We conducted another high-cadencecampaign at 820 MHz in December of 2019 that con-cluded in January of 2020. This gave us an additional15 detections out of 24 observations.2.1. VLA Observations
We observed PSR J1759+5036 using the VLA duringMay and June 2019. Since the source was previouslyobserved to be intermittent, we scheduled four sepa-rate observations to improve our chances of detectingit. These took place on MJDs 58614, 58615, 58652 and58655. Each session was 2.5 hr long, with about 2 totalon the pulsar. During this time the VLA was in the Bconfiguration, with maximum baseline length of 11 km.We acquired data using the VLA’s interferometric pul-sar binning mode, which averages visibility data into anumber of separate pulse phase bins following a timingephemeris. We used 32 pulse phase bins to record dataspanning 1–2 GHz, with 1-MHz frequency resolution and5-s dump time.The data were processed as follows: we used sdmpy to re-align the data in pulse phase using the best avail-able timing solution, dedisperse at the known DM ofthe pulsar, and split the original multi-bin data intoseparate data sets per bin as well as a bin-averaged ver- http://github.com/demorest/sdmpy sion. All further calibration and imaging was done usingthe Common Astronomy Software Applications package(CASA) (McMullin et al. 2007) version 5.6.2. The datawere calibrated using the standard VLA CASA calibra-tion pipeline. The flux scale was referenced to 3C286(Perley & Butler 2017), and interferometric phases werereferenced to J1740+5211. Calibration solutions weredetermined from the bin-averaged data set and appliedto each individual bin, which were then imaged sepa-rately. By subtracting the bin-averaged data from thesingle-bin data, all confusing background sources are re-moved, resulting in an unambiguous detection of thepulsar.The pulsar was detected in all four observations, withflux density ranging from 0.08 to 0.21 mJy on differentdays. From the brightest observation (MJD 58614) wedetermined a source position of RA = 17:59:45.66 ± ± (cid:48)(cid:48) . This observationhad a synthesized beamwidth of 4.9 (cid:48)(cid:48) by 4.1 (cid:48)(cid:48) at a posi-tion angle of 70 ◦ . The uncertainties quoted here reflectonly the statistical uncertainty due to noise in the data.It is possible that systematic effects may be present atthe sub-beam level, however this position is consistentwith values obtained from the other three observations,as well as with the later-determined timing position pre-sented in Table 2. TIMING ANALYSISOur total data set spanned roughly seven years with athree year gap between archived timing data taken dur-ing 2013–2014 for PSR J1800+5034 and the 2017–2020data for PSR J1759+5036. This can be seen in our tim-ing residuals, shown in Figure 3. Our ephemeris, shownin Table 2, uses the JPL DE436 solar system ephemerisand the DD binary model (Damour & Deruelle 1985,1986). We fit for DM by splitting five closely spaced in-dividual observations into four frequency subbands andcomputing a single TOA from each subband.We were able to measure the Post-Keplerian (PK) pa-rameter advance of periastron ( ˙ ω ) with high significance.Our reduced χ ( χ red ) was initially 1.5 with 175 degreesof freedom (DOF), so we used a multiplicative error fac-tor (EFAC) that accounts for random radiometer noiseby applying a constant multiplier to all TOA error bars.Our applied EFAC was 1.25, which gave a χ of 1.0.The root-mean-square (RMS) of our timing solution is246 µ s, or roughly 0.1% of the pulsar spin period.We calculated the average flux density at 350 ( S )and 820 ( S ) MHz, using the respective composite pro-files integrated across pulse phase, using the radiometerequation (Dewey et al. 1985). For S , we assumed abandwidth of 80 MHz, a system temperature ( T sys ) of Table 2.
Timing solution for PSR J1759+5036.Measured Parameter ValueRight Ascension (J2000) 17:59:45.672(3)Declination (J2000) +50:36:56.96(2) P (s) 0.17601634721733(8)˙ P (s s − ) 2 . × − Dispersion Measure (pc cm − ) 7.775(3)Statistic and Model ParametersTiming Data Span (MJD) 56406–58859RMS Residual ( µ s) 246.54EFAC 1.251Number of TOAs 187Binary Model DDSolar System Ephemeris DE436Reference Epoch (MJD) 57633Binary ParametersOrbital Period, P b (days) 2.04298385(3)Orbital Eccentricity, e ω (deg) 92.142(2)Epoch of Periastron (MJD) 57633.09399(14)Advance of Periastron, ˙ ω (deg yr − ) 0.127(10)Derived ParametersSurface Magnetic Field (10 Gauss) 9.5Spin-down Luminosity (10 erg s − ) 9.0Characteristic Age (Gyr) 50Dist DM NE2001 (pc) 711Dist DM YMW16 (pc) 542Mass Function 0.081768(1)Minimum Companion Mass ( M (cid:12) ) 0.7006Total Mass ( M (cid:12) ) 2.62(3) S (mJy) 0.38(8) S (mJy) 0.112(3) S (mJy) 0.12(2)In parentheses are the TEMPO-reported uncertainties inthe last significant digit. The EFAC reported was used toachieve a reduced χ of 1.0. Position uncertainty is deter-mined from the timing fit to position. The VLA observa-tion position is consistent with timing position.
46 K, and a sky temperature ( T sky ) of 38 K (Haslamet al. 1982a). For S we assumed a bandwidth of200 MHz, a T sys of 29 K, and a T sky of 4.2 K. T sky valueswere calculated from the Haslam et al. (1982b) sky-mapsand scaled to the appropriate frequency with a spectralindex of –2.6. We calculated the off-pulse noise of eachprofile and then used the expected radiometer noise to R e s i d u a l s ( P h a s e )
350 MHz820 MHz
Figure 3.
MJD vs post-fit residuals for 187 TOAs over 48epochs. The TOAs from before MJD 58101 are from archiveddata taken from the PSR J1800+5034 timing campaign. RedTOAs are calculated from 350-MHz data and blue TOAs arefrom 820-MHz data. scale the entire profile accordingly to determine the fluxdensity. We estimate an approximate error of 20% foreach flux measurement made using the radiometer equa-tion. From fitting a power law to the three points weget a spectral index of − . ± . Single-Pulse Analysis
Using the
PRESTO package single_pulse_search.py (Ransom 2001), we searched for single pulses with S/Ngreater than 7 at every epoch. We determined whetherpulses were indeed from PSR J1759+5036 and not dueto RFI by first requiring that the S/N be modeled asa Gaussian distribution with peak S/N at a DM be-tween 7 and 8.25 pc cm − , a range determined by thedistribution of measured DMs on the detectable epochs.Through this method we found that 20% of observationsshowed detectable single pulses.Since this pulsar was originally detected only throughits single pulses, we explored whether it is typicallybetter detected in this way. In Figure 4 we show theratio of the S/N of the composite profile along withthe S/N of the brightest single pulse. This showsthat PSR J1759+5036 is almost always detected athigher S/N values through periodicity searches thanthrough single pulses, indicating that it is not an RRAT(Shapiro-Albert et al. 2018). Only in the first searchobservation was the pulsar detected only through singlepulses.We also wanted to see if the single pulse S/N distri-bution had a power-law tail, like those seen in RRATsand systems that emit giant pulses (Mickaliger et al.2018). In Figure 5, we show a histogram of the detected Figure 4.
Periodicity search S/N vs the S/N of the bright-est detectable single pulse at each epoch of the timing ob-servations, with the line representing a slope of 1. On alldays, the pulsar is equally or better detected in a period-icity search. Here, a value of zero indicates the pulse wasnot detectable using that method. The original detection isomitted in this plot. C o un t s log-normalnormalpower-lawlog-normal+power-law Figure 5.
Histogram of the S/N of single pulses with 1- σ error bars, using a minimum S/N cutoff of 7. Overlaidare the log-normal, normal, power-law, and log-normal pluspower-law distributions that were fit to the data. single-pulse S/N values recorded at a central frequencyof 820 MHz. We show 1- σ error bars as calculated inGehrels (1986). To study the behavior of the S/N am-plitudes we fit log-normal, normal, power-law, and com-bined log-normal/power-law distributions to the data.Given the large error bars on each bin, all four fits hadlow chi-squared values, but due to the low number ofcounts per bin we can not make any firm conclusions.With future data-sets we may be able to distinguishbetween these models and better understand how thesingle-pulse properties of this pulsar compare to thoseof other pulsars and RRATs. DISCUSSION P e r i o d D e r i v a t i v e ( i . e S p i n - D o w n R a t e ) G G G G e r g / s e r g / s e r g / s e r g / s K y r K y r M y r G y r G y r Radio PSRsBinariesDNS SystemsJ1759+5036
Figure 6. P vs ˙ P of all known pulsars, with binary pulsarsin blue, DNS systems in green, and PSR J1759+5036 in red.The parallel lines correspond to constant values of the char-acteristic age (dashed dotted), spin-down energy loss rate(dashed), and surface dipole inferred magnetic field (solid)(see Lorimer & Kramer 2012 for definitions). The small ˙ P and large P place PSR J1759+5036 ina sparsely populated region of P and ˙ P space (see Fig-ure 6). The relatively low magnetic field of about 10 G suggests that this pulsar is partially recycled, butsome process, such as a companion supernova explosion,halted accretion from the companion before spin-up tovery short periods and circularizing of the orbit (Tauriset al. 2017; Pol et al. 2019). The high eccentricity is alsoconsistent with those seen in DNS systems.The time to merger due to gravitational wave radia-tion is 182 Gyr, much longer than a Hubble time, mean-ing that PSR J1759+5036’s discovery will not impactDNS merger rate estimates (Lorimer 2008; Pol et al.2019). 4.1.
Nature of the Binary System
The mass function implies a minimum companionmass of 0.7 M (cid:12) , assuming a system inclination of 90 ◦ (Lorimer 2008). Our measurement of ˙ ω corresponds toa total system mass of 2.62 ± M (cid:12) . In Figure 7 wehave overlaid the possible pulsar and companion massesallowed by ˙ ω with those forbidden by the mass func-tion. This allowed us to determine the maximum pulsarmass to be 1.8 M (cid:12) . The probability distributions forpulsar and companion mass in the figure have been cal-culated using the measured ˙ ω assuming random orbitalinclinations. The 2 σ mass ranges are 1.26–1.79 M (cid:12) forthe pulsar and 0.84–1.37 M (cid:12) for the companion, with P r o b a b ili t y D e n s i t y M )0.60.81.01.21.41.6 C o m p a n i o n s m a ss ( M ) Mass FunctionAdvance of Periastron
Figure 7.
Allowed values of pulsar mass vs companionmass given the rate of advance of periastron (between thered dashed lines) and the mass function (above the greendotted line). The red shaded region indicates possible valuesof pulsar mass and companion mass. Above and to the rightare the probability distributions for the pulsar and compan-ion mass respectively. a 0.16 probability that the companion mass is greaterthan 1.2 M (cid:12) . If the companion is indeed a NS, as seemslikely given its spindown and orbital properties, the in-clination angle must be less than 44 ◦ (assuming the NSmass must be greater than 1.2 M (cid:12) ).In Figure 8 we show projected values for the inclina-tion angle of the binary system using the range of pos-sible pulsar and companion masses determined by thetotal system mass and the mass function. We can inferfrom this that the inclination angle must be between 35 ◦ and 75 ◦ .4.2. Optical Constraints on a Companion
We conducted optical observations with the Las Cum-bres Observatory 2-meter telescope on Haleakala on 30Apr 2020. Three consecutive 500 s exposures were takenwith the SDSS r (cid:48) filter (median MJD 58969.5637447)and then another three 500s exposures with the SDSS g (cid:48) filter (median MJD 58969.5457716). These were thenmedian added for each filter and aperture photometrywas attempted on them. We used images whose darksubtraction, bias correction, flat-fielding and plate solu-tions were generated by the LCO Banzai pipeline. Inneither filter was a counterpart detected at the posi-tion of the pulsar (Figure 9). By measuring the faintestnearby stars using two SDSS catalog (release 12) stars M )304050607080 I n c li n a t i o n A n g l e ( d e g ) M )304050607080 I n c li n a t i o n A n g l e ( d e g ) Figure 8.
Left: Potential pulsar masses vs projected incli-nation angle implied by each mass as shown by the shadedregion. Right: Potential masses of the companion mass vsinclination angle, with the shaded region indicating possiblevalues. as references, we estimate a limiting magnitude of ∼ , withthe limits for each distance indicated (Bergeron et al.1995; Holberg & Bergeron 2006; Kowalski & Saumon2006; Bergeron et al. 2011; Tremblay et al. 2011; Blouinet al. 2018; B´edard et al. 2020). For this analysis weassumed a hydrogen WD atmosphere, used dustmapsto take extinction into account, and used cooling curvesfor masses of 0.8 M (cid:12) and 1.2 M (cid:12) , which are the lowerand upper mass limits for a WD companion for this sys-tem (Tremblay et al. 2011). For a 0.8 M (cid:12) companionat a distance of 700 pc the effective temperature is < > g (cid:48) limit. At the same distance, for a1.2 M (cid:12) companion, the effective temperature < > . ∼ bergeron/CoolingModels/ Figure 9.
Image from the LCO observation to look foran optical counterpart for PSR J1759+5036. The black cir-cle represents a 5 (cid:48)(cid:48) diameter region around the pulsar, muchlarger than the < (cid:48)(cid:48) error in position. ure 9). Our DM measurements indicate a distance ofapproximately 700 pc using the NE2001 electron densitymodel (Cordes & Lazio 2003), or 550 pc using YMW16model of electron density (Yao et al. 2017); any mainsequence star of comparable luminosity and size to theSun at these distances would have a magnitude of 13–15and thus would be easily visible. Optical observationsat a magnitude of 26–27 would be needed to confidentlyrule out or confirm a WD companion.While most pulsar-WD binaries are circular, thereare several systems with eccentric companions, suchas PSR J2305+4707 and PSR J1755 − Intermittency
The low DM of PSR J1759+5036 suggests that diffrac-tive interstellar scintillation may play a role in the inter-mittency we observed, shown in Figure 2. To study this,we created dynamic spectra of each detected epoch byfirst folding the data in time sub-integrations of 60 s us-ing fold_psrfits . Then, we used pam to fold the datainto frequency sub-integrations of 3 MHz and plotted Age (yr)05000100001500020000 E ff e c t i v e T e m p e r a t u r e ( K ) Figure 10.
Cooling curves for 0.8 M (cid:12) (blue) and1.2 M (cid:12) (orange) WD stars with effective temperature upperlimits of and age lower limits indicated for distances of 500 pc(red), 700 pc (green), and 1000 pc (blue). For an 0.8 M (cid:12) WDwe have limits of < > . < > . < > . M (cid:12) WD we have limits of < > . < > . < > . F r e q u e n c y ( M H z ) F r e q u e n c y ( M H z ) Figure 11.
Left: A dynamic spectrum on epoch 58131at 820 MHz. Intensity is plotted against frequency bins ofroughly 6 MHz, and time bins of 60 s. Right: Dynamic spec-trum on epoch 58466 at 350 MHz. On-pulse intensity is plot-ted vs frequency and time, with time bins of 60 s and fre-quency bins of roughly 3 MHz. The purple strip is a regionof bright RFI that was removed from the data. the intensity of each frequency/time bin and correctedthe bandpass for instrument variation using the pythonpackage
PyPulse (Lam 2017). In Figure 11 we haveshown dynamic spectra from two epochs where scintil-lation was observed.The NE2001 electron density model predicts a scintil-lation bandwidth of ∼
26 MHz at 350 MHz and ∼
40 MHz at 820 MHz at the position and DM of J1759+5036(Cordes & Lazio 2003).Given the measured timing residual and the 6-yearbaseline of our timing data, we can place an upper limiton the transverse velocity of 16 km s − . This is ratherlow, but consistent with other DNS systems such asPSR J0737-3039A/B ( ∼ − ) and PSR B1913+16( ∼
22 km s − ) (Deller et al. 2009, 2018). For this ve-locity we estimate a scintillation timescale of ∼ ∼ N u m b e r o f D e t e c t i o n s N u m b e r o f D e t e c t i o n s Figure 12.
The distribution of flux densities measured ateach epoch at 350 MHz (left) and 820 MHz (right). Fluxdensities that were calculated from observations impactedby the position offset were corrected to reflect the degradedsensitivity. more data are needed to determine whether a true peri-odicity might exist. CONCLUSIONS AND FUTURE WORKIn this paper we report on the discovery and timingcampaign for PSR J1759+5036, a 176-ms binary pul-sar with a 2.04-day eccentric orbit about a likely neu-tron star companion. Over our seven-year dataset, thispulsar has been highly intermittent with an approxi-mately 45% detection rate which we believe to be largelyattributed to scintillation, possibly combined with apoorly constrained position that we then determinedthrough observations taken with the VLA.We were able to measure all five Keplerian parame-ters as well as a single PK parameter, ˙ ω . This allowedus to determine the minimum companion mass and totalsystem mass, which we used to place limits on possiblepulsar and companion masses as well as system inclina-tion angle. Due to the eccentricity (0.308) of the systemand lack of optical counterpart, we believe the candi-date is most likely a neutron star. We have ruled out amain sequence companion and most white dwarf stars(see Section 4.2).Continued follow-up observations and timing ofPSR J1759+5036 are expected to eventually produce ameasurement of a second PK parameter, which wouldallow us to fully constrain the pulsar and companionmasses. Using simulated TOAs, we estimate that ap-proximately 12 years of data would be needed to mea-sure γ , the time dilation and gravitational redshift pa- rameter, which would also determine the inclination an-gle of the system. We also hope to be able to measure theproper motion, which would allow us to calculate moreaccurate estimates of distance and pulsar velocity. Cur-rently, follow-up observations are being conducted onthe Canadian Hydrogen Intensity Mapping Experimenttelescope with weekly cadence, the results from whichare expected to be included in future publications.ACKNOWLEDGMENTSMAM, JKS, HB, MD, PBD, WF, EF, DLK, RSL,AEM, SMR, AS, CS, XS, IHS, and MS are membersof the NANOGrav Physics Frontiers Center, supportedby NSF award number 1430284. VMK holds the LorneTrottier Chair in Astrophysics & Cosmology, a Dis-tinguished James McGill Chair, and receives supportfrom an NSERC Discovery Grant, the Herzberg Award,CIFAR, and from the FRQ-QNT Centre de Rechercheen Astrophysique du Quebec. MAM, GYA, and MGMare also supported by NSF award number 1458952. Wethank the WVU Research Corporation for their pur-chase of observing time on the GBT which has supportedsome of the observations for this project. Pulsar work atUBC is supported by an NSERC Discovery Grant and bythe Canadian Institute for Advanced Research. The Na-tional Radio Astronomy Observatory and Green BankObservatory are facilities of the National Science Foun-dation operated under cooperative agreement by Associ-ated Universities, Inc. Computations were made on thesupercomputer Guillimin at McGill University , man-aged by Calcul Quebec and Compute Canada. The op-eration of this supercomputer is funded by the CanadaFoundation for Innovation (CFI), NanoQuebec, RMGAand the Fonds de recherche du Quebec - Nature et tech-nologies (FRQ-NT). VLA observations were taken asproject 19A-387. GBT observations were taken underprojects AGBT13A-458, AGBT17B-285, AGBT17B-423, AGBT17B-292, AGBT18A-482, AGBT17B-325, AGBT18B-335, AGBT18B-360, AGBT19A-180,AGBT19A-486, AGBT19B-306, AGBT19B-327, andAGBT19B-320. Software : Astropy (Price-Whelan et al. 2018),
PRESTO (Ransom 2001),
PSRCHIVE (van Straten et al. 2012),
PINT (Luo et al. 2019),
NumPy (Harris et al. 2020),
PyPulse (Lam 2017),
TEMPO , SciPy , CASA (McMullin et al.2007),
DS9 (Joye & Mandel 2003), sdmpy (http://github.com/demorest/sdmpy)
Facilities : Robert C. Bryd Green Bank Telescope(GBT), Karl G. Jansky Very Large Array (VLA), LasCumbres Observatory (LCO)0
Figure 13.
Epochs of observation on which PSR J1759+5036 was detected (blue sold lines) and not detected (red dotted lines).Left: 2013–2014 dataset. Right: 2017–2020 dataset.
REFERENCES1