Six New Millisecond Pulsars from Arecibo Searches of Fermi Gamma-Ray Sources

0.85 MB·

by  H. T. Cromartie

DRAFTVERSIONJANUARY21,2016
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
SIXNEWMILLISECONDPULSARSFROMARECIBOSEARCHESOFFERMI GAMMA-RAYSOURCES
H.T.CROMARTIE1,2,F.CAMILO3,M.KERR4,J.S.DENEVA5,S.M.RANSOM6,P.S.RAY5,E.C.FERRARA7,P.F.MICHELSON8,AND
K.S.WOOD5
DraftversionJanuary21,2016
ABSTRACT
We have discovered six radio millisecond pulsars (MSPs) in a search with the Arecibo telescope of 34
unidentified gamma-ray sources from the Fermi Large Area Telescope (LAT) 4-year point source catalog.
Amongthe34sources,wealsodetectedtwoMSPspreviouslydiscoveredelsewhere.Eachsourcewasobserved
6
atacenterfrequencyof327MHz,typicallyatthreeepochswithindividualintegrationtimesof15minutes.The
1
newMSPspinperiodsrangefrom1.99to4.66ms. Fiveofthesixpulsarsareininteractingcompactbinaries
0
(period ≤ 8.1 hr), while the sixth is a more typical neutron star-white dwarf binary with an 83-day orbital
2
period. ThisisahigherproportionofinteractingbinariesthanforequivalentFermi-LATsearcheselsewhere.
n
The reason is that Arecibo’s large gain afforded us the opportunity to limit integration times to 15 minutes,
a
whichsignificantlyincreasedoursensitivitytothesehighlyacceleratedsystems. Seventeenoftheremaining
J
26gamma-raysourcesarestillcategorizedasstrongMSPcandidates,andwillbere-searched.
0
Subject headings: stars — pulsars: individual (PSR J0251+26, PSR J1048+2339, PSR J1805+06,
2
PSRJ1824+10,PSRJ1909+21,PSRJ2052+1218)

]
E
H 1. INTRODUCTION radiosearchtargets.
Before 2013, no Fermi MSPs had been discovered using
Ofthe230millisecondpulsars(MSPs)currentlyknownin
.
h the Galactic disk9, 30% have been discovered in previously the 305-m Arecibo radio telescope in Puerto Rico. In con-
p unidentified sources of gamma rays detected by the Fermi- trast, the Green Bank (GBT), Parkes, Nançay, Giant Metre-
- LAT instrument10 (Atwood et al. 2009). While only around wave (GMRT), and Effelsberg telescopes had been used to
o
discoverdozensofnewMSPsusingthe1FGLand2FGLcat-
r 10% of all known pulsars rotate at millisecond rates, MSPs
t makeuphalfofallpulsarsobservedtoemitgammarays(Car- alogs(Abdoetal.2010;Nolanetal.2012)asguides. Inthis
s work,wepresentthefirstsixMSPsdiscoveredinunidentified
a aveo2014). TheLATsourcecatalogshavebeeninstrumental
LATsourcesusingtheArecibotelescope,alongwithprelim-
[ inthesearchfornewMSPs,providingspectraldatatoaidin
inaryorbitalparametersgleanedfromradiotiming. Wethen
distinguishingpossibleMSPsfromothergamma-ray-emitting
1 quantitatively discuss the relative sensitivities — both in the
objects,suchasactivegalacticnuclei(AGNs). OnceanMSP
v fluxdensityandaccelerationregimes—betweentheFermi-
is discovered in a radio search and a phase-connected tim-
3 LAT MSP searches conducted at Arecibo and those done at
ing solution is available, the sparse gamma-ray photons are
4 theGBTandParkes.
foldedusingtheradioephemerisinordertoglimpsegamma-
3
ray pulsations(e.g., Cognardet al.2011). While itwas pos-
5
0 sible to search for radio pulsars in gamma-ray sources prior 2. CANDIDATESELECTION
1. to the Fermi era (e.g., Champion et al. 2005; Roberts et al. All but two of the 34 candidates observed came directly
2002), the small positional uncertainty of LAT gamma-ray
0 from early versions of the third Fermi-LAT catalog (3FGL,
sources has enabled single-pointing radio searches. Overall,
6 alsoknownasthe4-yearcatalog),whichwaslaterpublished
the search for MSPs in the Galactic disk has been made ex-
1 byAceroetal.(2015). Thetworemainingsourceswerede-
tremely efficient by employing Fermi-LAT data in selecting
: tectedwiththeLATbutwerebelowthesignificancethreshold
v
requiredforinclusioninthefinalcatalog. WhileLATsource
i 1DepartmentofAstronomy,UniversityofVirginia,Charlottesville,VA
X 22903,USA listscontain>1000unidentifiedgamma-rayemitters,several
2email:[email protected] constraintsdramaticallylimitthenumberofsourcesappropri-
r
a 3ColumbiaAstrophysicsLaboratory,ColumbiaUniversity,NewYork, ateforoursearches.
NY10027,USA Everysourcehadtobelocatedwithinthedeclinationrange
4CSIROAstronomyandSpaceScience,AustraliaTelescopeNational of Arecibo (−1◦ < δ < 39◦). Justification for picking the
Facility,Epping,NSW1710,Australia
5SpaceScienceDivision,NavalResearchLaboratory,Washington,DC 327-MHz receiver over L-band, for example, was two-fold.
20375-5352,USA First,thetargetsourceerrorcircleswererequiredtofitwithin
6NationalRadioAstronomyObservatory(NRAO),Charlottesville,VA Arecibo’sbeam,allowingforsingle,long-durationpointings.
22903,USA The 327-MHz system, with a relatively large FWHM=15(cid:48),
7NASAGoddardSpaceFlightCenter,Greenbelt,MD20771,USA
8W. W. Hansen Experimental Physics Laboratory, Kavli Institute wasthebestchoice. Second, pulsarshavesteepspectra, and
for Particle Astrophysics and Cosmology, Department of Physics and thereforearebrighteratsucharelativelylowfrequency. Very
SLACNationalAcceleratorLaboratory,StanfordUniversity,Stanford,CA fewpulsarsareknowntobevariableingammarays(Rayetal.
94305,USA 2012); thus, only non-variable LAT sources were selected.
9 Public list of Galactic MSPs: http://astro.phys.wvu.edu/
Also, each of the selected sources had a spectral energy dis-
GalacticMSPs/
10 For a list of all LAT pulsars, see https://confluence. tributionconsistentwiththoseofknowngamma-raypulsars,
slac.stanford.edu/display/GLAMCOG/Public+List+of+ whichtypicallyhaveexponentiallycut-offpower-lawspectra
LAT-Detected+Gamma-Ray+Pulsars (Abdoetal.2013). BecauseitisdifficulttocharacterizeLAT
2 CROMARTIEETAL.
sources amid the Galaxy’s diffuse gamma-ray background
(see, e.g., Geringer-Sameth & Koushiappas 2012), and be-
cause the effects of dispersion, scattering, and synchrotron
emission inhibit radio pulsar observations at low frequency
along the Galactic plane, only sources with |b| > 5◦ were
considered. After whittling the list down, we obtained the
34sourcesinTable1.
3. OBSERVATIONSANDDATAANALYSIS
Observationsofthe34Fermi-LATsourceswereconducted
during 12 sessions between 2013 June and September using
theArecibotelescope. Inordertocombattheeffectsofscin-
tillation, orbital acceleration, and eclipses (discussed further
below), we aimed to observe each source three times for 15
minutes per pointing, though the exact number of pointings
persourcechangedasdatawereanalyzed.
Sessions in the months of June, July and early August
wereconductedin-personattheobservatory,whilelaterses-
sionsoccurredremotely. Ineithercase,thestandardCIMA11
telescope control software was used in conjunction with
command-linecontrolofthePuertoricanUltimatePulsarPro-
cessingInstrument(PUPPI).ThePUPPIbackend(areplicaof
GUPPI12 at the GBT) was configured for the settings shown
inTable2. Alldataweretakenintotal-intensity,summedpo- Figure1. SensitivityoffourradiopulsarsearchesofFermi-LATsources
larization mode. Once disks became full, they were shipped asafunctionofspinperiodforapulsarwith10%dutycycle. Allsurveys
from Arecibo to Columbia University for data reduction. A havebeenscaledto327MHzusingaspectralindexof−1.7. Thesurveys
presentedhere,inadditiontothiswork,are: GBT350MHz(Hesselsetal.
summaryofallobservationsisprovidedinTable1.
2011),GBT820MHz(Ransometal.2011),andParkes1390MHz(Camilo
Data were analyzed using the software package PRESTO etal.2015).SeeTable2fordetails.
(Ransom 2001). The data reduction process began with the
detectionandmaskingofsignificantradiofrequencyinterfer-
telescopegain,n isthenumberofpolarizationsrecorded(al-
enceinthedata. Dedispersionoccurreduptoaspecifieddis- p
ways2forthesearchesdiscussedhere),t istheintegration
persionmeasure(DM),whichwechosetobetwicethemaxi- int
time, and ∆f is the effective bandwidth. Relevant parame-
mumline-of-sightvaluegivenbytheNE2001model(Cordes
ters for the Arecibo survey are shown in Table 2 under “AO
&Lazio2002). PRESTOcanperformsearchesoverspinpe-
327”. Arecibo’s system equivalent flux density (SEFD) de-
riodvariationscausedbyorbitalmotion,searchingoverboth gradesforzenithanglesexceeding15degrees13,butthishad
period and period derivative. The extent of the acceleration
littleimpactformostofoursearches. Overalloursurveyhad
searchisspecifiedbythezmaxparameter,chosentobe200
anaverageSEFD=13Jy. ThesensitivityoftheArecibosur-
inourcase. Thismeansthatlinearpulsarspinfrequency(f )
0 vey in the context of other Fermi-LAT searches is discussed
driftsofupto200binsweresearchedinthehighestharmonic,
inSection5.
whichinouranalysiswastheeighth(Ransometal.2002). If
t is the total integration time and a is the maximum ac-
int max 4. RESULTS
celerationprobed,z =a t2 f /c(Ransometal.2000).
max max int 0 In the 34 sources searched, we discovered six new MSPs
(seeTable3). Pulseprofilesfromthediscoveryobservations
3.1. Sensitivity
are shown in Figure 2. The rotation periods range between
Figure1showstheminimumfluxdensitydetectablebyour 1.99 and 4.66 ms, and their DMs span 17–65pc cm−3. Or-
ArecibosearchesforarangeofMSPspinperiodsandDMs, bitalsolutionshavebeenobtainedforallnewdiscoveriesfrom
asdeterminedbytheradiometerequationforpulsars(Lorimer initial timing observations. However, phase-connected tim-
&Kramer2005,Appendix1.4): ingsolutions(includingprecisepositions,periodderivatives,
(cid:114) spin-downluminosities, andastudyofthegamma-rayprop-
S/N (T +T ) W
S =β min rec sky , (1) ertiesofthecoincidentgamma-raysources)arenotyetavail-
min G(cid:112)n t ∆f P−W ableformostoftheMSPs,andwillbepresentedelsewhere.A
p int
studyoftheredbackPSRJ1048+2339ispresentedinDeneva
where Smin is the minimum detectable flux density, β is a etal.(2015).
normalization factor including corrections for, e.g., system Five of the new MSPs are neutron stars with short orbital
digitization losses, S/Nmin is the threshold pulsar signal-to- periods. Three are “black widows”, in which much of the
noiseratio,Trec isthereceivertemperature(includingcontri- companion mass has been stripped away or accreted by the
butionsfromtheCMBandspillover),Tsky istheskytemper- pulsar,leavinga(partiallydegenerate)companionwithmass
ature, W is the effective pulse width (we assume the intrin- (cid:28)0.1M . Theremainingtwoshort-orbitsystemsare“red-
(cid:12)
sic pulse width to be P/10), P is the pulsar period, G is the backs”, where the pulsar is frequently eclipsed by outflows
11http://www.naic.edu/~cima 13 Detailed measurements for gain and system temperature of the 327-
12 https://safe.nrao.edu/wiki/bin/view/CICADA/ MHzGregorianreceiveratArecibocanbefoundathttp://www.naic.
GUPPiUsersGuide edu/~astro/RXstatus/327/327greg.shtml.
NEWARECIBOMILLISECONDPULSARSINFERMISOURCES 3
Table1
SummaryofAreciboSearchesofUnidentifiedFermi-LATSources
Name R.A.a Decl.a l b IntegrationTime DMmaxb
(J2000.0) (J2000.0) (deg) (deg) (minutes) (pccm−3)
3FGLJ0103.7+1323 01h03m46s 13◦23(cid:48)33(cid:48)(cid:48) 127.5 −49.4 15,15,15 72
3FGLJ0134.5+2638 01h34m31s 26◦38(cid:48)15(cid:48)(cid:48) 134.7 −35.2 15,15,15 92
3FGLJ0232.9+2606 02h32m56s 26◦06(cid:48)13(cid:48)(cid:48) 149.7 −31.4 15,15,15 130
3FGLJ0251.1+2603 02h51m08s 26◦04(cid:48)48(cid:48)(cid:48) 153.9 −29.5 15,15 124
3FGLJ0318.1+0252 03h18m09s 02◦52(cid:48)10(cid:48)(cid:48) 178.4 −43.6 15,15,15 900
3FGLJ0330.6+0437 03h30m40s 04◦37(cid:48)32(cid:48)(cid:48) 179.5 −40.1 15,15,15 680
3FGLJ0342.3+3148c 03h42m18s 31◦48(cid:48)33(cid:48)(cid:48) 160.3 −18.4 15,15,15 100
3FGLJ0421.6+1950 04h21m37s 19◦50(cid:48)49(cid:48)(cid:48) 175.9 −20.7 15,15,15 150
3FGLJ0517.1+2628c 05h17m10s 26◦28(cid:48)44(cid:48)(cid:48) 178.6 −6.6 10,15,15,15,15 120
3FGLJ0539.8+1434 05h39m48s 14◦33(cid:48)53(cid:48)(cid:48) 191.6 −8.6 5,5 200
3FGLJ1048.6+2338 10h48m41s 23◦38(cid:48)29(cid:48)(cid:48) 213.2 62.1 15,15 66
3FGLJ1049.7+1548 10h49m44s 15◦48(cid:48)25(cid:48)(cid:48) 228.5 59.6 10,15 68
3FGLJ1200.4+0202 12h00m27s 02◦02(cid:48)31(cid:48)(cid:48) 274.8 62.1 15,15,15 64
3FGLJ1225.9+2953 12h25m59s 29◦53(cid:48)25(cid:48)(cid:48) 185.2 83.8 15,15,15 40
P7R4J1250+3118e 12h50m52s 31◦18(cid:48)18(cid:48)(cid:48) 124.6 85.8 15,15,15 40
3FGLJ1309.0+0347 13h09m02s 03◦47(cid:48)27(cid:48)(cid:48) 313.9 66.3 15,15,10,15,10,15 60
3FGLJ1322.3+0839 13h22m20s 08◦39(cid:48)27(cid:48)(cid:48) 325.9 70.1 15,15,15 52
3FGLJ1601.9+2306 16h01m57s 23◦06(cid:48)39(cid:48)(cid:48) 38.5 46.9 15,15,15 60
3FGLJ1627.8+3217 16h27m52s 32◦17(cid:48)56(cid:48)(cid:48) 53.0 43.2 15,15 70
3FGLJ1704.1+1234 17h04m08s 12◦34(cid:48)25(cid:48)(cid:48) 32.5 29.4 15,15,15 116
3FGLJ1720.7+0711 17h20m46s 07◦11(cid:48)21(cid:48)(cid:48) 29.0 23.4 15,15,15 156
3FGLJ1805.9+0614 18h05m55s 06◦14(cid:48)15(cid:48)(cid:48) 33.4 13.0 15,15 304
3FGLJ1824.0+1017 18h24m05s 10◦17(cid:48)27(cid:48)(cid:48) 39.1 10.7 15,15 356
3FGLJ1827.7+1141 18h27m42s 11◦41(cid:48)50(cid:48)(cid:48) 40.8 10.5 15,15 356
3FGLJ1829.2+3229 18h29m08s 32◦30(cid:48)42(cid:48)(cid:48) 60.7 18.5 15,15,15 158
3FGLJ1842.2+2742 18h42m15s 27◦42(cid:48)09(cid:48)(cid:48) 57.1 14.1 15,15,15 216
P7R4J1909+2102e 19h09m32s 21◦02(cid:48)56(cid:48)(cid:48) 53.7 5.6 15,15 564
3FGLJ1921.2+0136d 19h21m14s 01◦36(cid:48)26(cid:48)(cid:48) 37.8 −5.9 5,10,15,15 670
3FGLJ2026.3+1430 20h26m21s 14◦30(cid:48)53(cid:48)(cid:48) 57.3 −13.4 15,15,15 226
3FGLJ2042.1+0247d 20h42m09s 02◦47(cid:48)35(cid:48)(cid:48) 49.0 −23.0 15,15 140
3FGLJ2052.7+1217 20h52m47s 12◦17(cid:48)51(cid:48)(cid:48) 59.1 −20.0 15,15,15 148
3FGLJ2108.0+3654 21h08m02s 36◦55(cid:48)19(cid:48)(cid:48) 81.1 −7.2 15,15,15 360
3FGLJ2212.5+0703 22h12m35s 07◦03(cid:48)35(cid:48)(cid:48) 68.7 −38.6 15,30,15,15,15,5 84
3FGLJ2352.0+1752 23h52m04s 17◦52(cid:48)50(cid:48)(cid:48) 103.5 −42.7 15,15,15,15 74
Note.—BoldfacedentriesdenoteobservationsyieldingMSPdetections.
aArecibotelescopepointingposition.
bThemaximumdispersionmeasure(DM)uptowhichwesearchedcorrespondsapproximatelytotwicethemaximumvaluepredictedforeachlineofsightby
theNE2001electrondensitymodel(Cordes&Lazio2002),withtheexceptionof3FGLJ0318.1+0252andJ0330.6+0437,whichwereunintentionallysearched
tohigherDMs.
dDiscoveredattheGBT(S.Sanpa-Arsaetal.2016,inpreparation).
eSourcenotincludedin3FGLcatalog.
fromanon-degeneratecompanionwithmass(cid:38)0.1M . The vatecommunication);however,onlyaquicksearchofthefirst
(cid:12)
finalMSPisamoreclassicalneutronstar-whitedwarfbinary. fiveminutesofdatawasperformedandthepulsarwasnotde-
Foradiagramoforbitalperiodvs.companionmassforsuch tected. Searchingthefulldatasetfollowingourdiscoveryat
highlyacceleratedsystems,seeRoberts(2012). Arecibo,theMSPisclearlydetected. Ransometal.alsoob-
Figure3showsthedistributionoforbitalperiodsvs.mini- served this source twice with the GBT at 820 MHz, but the
mumcompanionmassesforfiveofthenewMSPspresented MSPdidnotshowupinapreliminaryanalysisofthefirstob-
inthiswork. Minimumcompanionmassesarecalculatedus- servationandtheseconddatasetwasnotsearched. Usingthe
ing Keplarian parameters derived from orbital timing solu- knownDMandapproximateperiodfromourArecibodetec-
tions. NotethatPSRJ1909+21isclassifiedasaredback,de- tionsrevealsthepulsarinbothGBTdatasets.
spite its minimum companion mass being less than 0.1 M . PSRJ2052+1218isanintriguingsystemduetothepulsar’s
(cid:12)
This is firstly because the 0.055 M value is the minimum veryfastrotation(1.99ms)anditsshortbinaryperiod(2.8hr).
(cid:12)
companionmass,anditislargerthananyknownblackwidow Evenaftersearchingoveracceleration,residualdriftsinphase
minimumcompanionmass;secondly,itseclipseslastforap- vs.timecanbeseeninthisandotherblackwidowandredback
proximatelyhalfoftheorbit,whichischaracteristicofared- systems(seeFigure2,especially(b)and(h)).
backsystemwithadensecircumstellarenvironment. We searched the sources containing PSRs J1921+01 and
PSR J1805+06 is in a black widow system with an or- J2042+02 and detected the MSPs, unaware that they had al-
bital period of 8.1 hr. The approximate position of 3FGL ready been discovered at the GBT. These will be published
J1805.9+0614 was observed in 2009 at the Robert C. Byrd in a forthcoming paper detailing Fermi-LAT searches at the
Green Bank Telescope (GBT) at 350 MHz (M. Roberts, pri- GBT(S.Sanpa-Arsaetal.2016,inpreparation).
4 CROMARTIEETAL.
Table2
ObservingParametersforFourRadioSurveysofFermi-LATSources
Parameter AO327-MHzSurvey GBT350-MHzSurveya GBT820-MHzSurveyb Parkes1390-MHzSurveyc
Detectionfractiond 8/34(24%);5/6 13/50(26%);3/13 3/25(12%);1/3 11/56(20%);2/11
Centerfrequency(MHz) 327 350 820 1390
Bandwidth,∆f (MHz) 68.75e 100 200 256
Numberofchannels 2816f 4096 2048 512
Sampletime(µs) 81.92 81.92 61.44 125
Receivertemperatureg,Trec(K) 62 20 18 25
Averageskytemperatureh,Tsky(K) 64 65 15 5
Telescopegain,G(K/Jy) 10 2 2 0.735
EffectivethresholdS/N(β·S/Nmin) 10 10 10 12
Integrationtime,tint(minutes) 15 32 45 60
aHesselsetal.(2011).
bRansometal.(2011).
cKerretal.(2012);Camiloetal.(2015).
dNumberofdetectedMSPsdividedbythetotalnumberofsourcesobserved;numberofblackwidowplusredbacksystemsdiscovereddividedbythetotalnumberofMSPsdiscovered.
eRecordedbandwidth:100MHzweresampledbyPUPPIbutweonlyrecordedthesectioncoveringthereceiverbandwidth.
fNumberofrecordedchannels;4096channelsweresampledacrosstheentire100-MHzbandwidth.
gReceivertemperatureincludingspilloverbutexcludingGalactic/CMBcontribution.ValuesforArecibo327MHzfromNAICa.GBTvaluesarefrompage11oftheproposer’sguideb.
SeealsoLynchetal.(2013)for350MHz.Parkesvaluesarebasedonthosefromtheusersguidec.
hWecalculatedskytemperaturesbyscalingtheHaslametal.(1981)408-MHzmaptoeachsurvey’sobservingfrequencyusingaspectralindexof−2.6(Lawsonetal.1987).ForAO
327,welisttheaverageTskyforeachofthe34targetlocations.ForGBT350andParkes,wecalculatedtheaveragetemperatureatanevenlyspacedgridofpointsencompassingthe
searchregions.ForGBT820weaveragedtheRansometal.(2011)valuesfortheirsearchesexcludingtheGalacticplane.
ahttp://www.naic.edu/~phil/cal327/327Calib.html
bhttps://science.nrao.edu/facilities/gbt/proposing/GBTpg.pdf
chttp://www.parkes.atnf.csiro.au/observing/documentation/user_guide/pks_ug_3.html#Receivers-and-Correlators
Table3
PulsarsdiscoveredinAreciboSearchesofFermi-LATSources
Namea P DM Distanceb Porbit MinimumCompanion Typed Eclipses? DiscoveryFluxDensities
(ms) (pccm−3) (kpc) (hr) Massc(M(cid:12)) (mJy)
J0251+26 2.54 20 0.8 4.9 0.024 BW Yes 0.3,0.3
J1048+2339 4.66 17 0.7 6.0 0.30 RB Yes 2.4
J1805+06 2.13 65 2.5 8.1 0.023 BW Noe 1.1,1.5
J1824+10 4.07 60 2.5 1980.0 0.26 NSWD No 0.09,0.15
J1909+21 2.56 62 3.2 3.5 0.055 RB Yes 0.6
J2052+1218 1.99 42 2.4 2.6 0.033 BW Yes 1.3,1.0,1.2
aNameswithfourdigitsofdeclinationhavebeengivenonlytoMSPswithphase-connectedtimingsolutions.
bFromtheNE2001model(Cordes&Lazio2002).
cAssumingapulsarmassof1.35M(cid:12)(Özeletal.2012).
dBW=blackwidow,RB=redback,NSWD=neutronstar-whitedwarf.
ePSRJ1805+06hasnotyetshownanyeclipsingbehavior;however,thereisagapinorbitalcoverageatphases0.2–0.27,soeclipsescannotberuledout.
ThesixnewfindingsmarkthefirstFermiMSPsdiscovered gesttheyarenotpulsars.Aninabilitytomakeadetectiondoes
usingtheArecibotelescopeandbrokethe50-pulsarthreshold notprecludethepresenceofanMSP;rather,itmaybeduetoa
fortotalLAT-guidedradioMSPdiscoveries(whichasof2015 pulsar’sfaintness,eclipses,scintillation,orextremeorbitalor
Decemberstandsat69). spin parameters. The large number of black widow and red-
backsystemsthathavebeendiscoveredinFermi-LATsources
5. DISCUSSION
makeeclipsesadistinctpossibilityforthiscollectionofcan-
5.1. PossibleCandidatesforRe-Observation didates. Forexample,bothPSRsJ1048+2339andJ1909+21
In searching 34 unidentified Fermi-LAT gamma-ray wereonlydetectedinthesecondoftwosearchobservations,
sources at Arecibo, we detected 8 MSPs, for a 24% success owingtoeclipses(Table1). Additionalobservationsofthe17
rate.ThisisinlinewiththesuccessrateforLAT-guidedradio remaining“good”sourcesinTable4mayresultinthedetec-
surveysattheGBTandParkes(Hesselsetal.2011;Ransom tionofnewMSPs.
et al. 2011; Camilo et al. 2015, see Table 2). While we find
5.2. UncertainGamma-RayAssociations
this to bea satisfying result, it ispossible that some remain-
ingsourcesinTable4couldstillbepulsars. Seventeenofthe Two new MSPs, PSRs J1048+2339 and J1909+21, may
26 sources currently without a known pulsar counterpart are have been “lucky” discoveries within the error circle of a
spectrallyconsistentwithpulsars(denotedbyarankingof1, gamma-ray source, but not necessarily associated with that
2,or3inthe“SpectrumNotes”column)andhavenoknown source. In3FGL,J1048.6+2338islistedasbeingpossiblyas-
AGNassociation. Sourcesranked1or2areverylikelytobe sociatedwithaBLLacertae-typeblazar. Blazarassociations
pulsars,whilerank-3sourceslackdefinitiveevidencetosug- aregenerallyspatial,andaccidentalcoincidenceisacommon
NEWARECIBOMILLISECONDPULSARSINFERMISOURCES 5
(a)J0251+26 (b)J1048+2339 (c)J1805+06 (d)J1824+10
(e)J1909+21 (f)J2052+1218 (g)J2052+1218 (h)J2052+1218
(2013Jun25) (2013Jul04) (2013Sep12)
Figure2. ThebestdetectionsfromthesearchobservationsofsixnewMSPs,foldedmodulotheperiodandperiodderivativereturnedbythesoftware(two
rotationsareshown).ThethreesearchobservationsoftheeclipsingPSRJ2052+1218showsomeeffectsrelatedtolikelyeclipseegress(panelf)andingress(g).
cause for reclassification of non-variable sources. Until it is culttoclassifyspectrally. Inaddition, somerelativelybright
possibletofoldthegamma-rayphotonsmodulotheparame- MSPs, particularly those not subject to large accelerations,
ters obtained with a radio timing solution, it will remain un- wouldhavealreadybeendiscoveredinpreviousArecibo“all-
clearwhethertheFermi-LATsourceisanMSPorpossiblya sky”surveys.
blazar. PSRJ1909+21isnotassociatedwithanearby3FGL Figure 1 presents minimum detectable flux densities for
source. We selected it for observation because in a prelim- four radio searches of Fermi-LAT sources, including our
inary source list internal to the LAT collaboration there ap- Arecibo work. Parameters for each of the surveys are pro-
pearedtobeapromisingsource. AsforalltheMSPswehave vided in Table 2. Each of the sensitivity curves has been
discovered,wewillknowwhetherthisoneisassociatedwith scaledto327MHzusinganassumedMSPspectralindexof
a LAT source once we have rotational ephemerides and can α=−1.7 (Stovall et al. 2014). As an example of their rela-
foldthegamma-rayphotons. tive power, for spin period P=1.8ms the Arecibo searches
areassensitiveatDM=100pccm−3 astheGBTsurveysare
5.3. SensitivityintheContextofOtherLATRadioSurveys forDM=10pccm−3.ForidenticallowDMs,theArecibosur-
veysareabouttwiceassensitiveastheGBTsearches.Inother
IfnobreakexistsinthelogN−logSdistributionofMSPs,
words,integrationtimeattheGBTwouldhavetobequadru-
one might expect that an increase in sensitivity would yield
pled to reach comparable raw sensitivities to Arecibo — but
highersurveysuccessrates. Instead,ourdiscoveryrateswere
such an increase in integration time would have deleterious
comparable to those of other lower-sensitivity LAT-guided
consequencesforthedetectabilityofcompactbinaries.
MSPsurveys. However, wehavebasedourtargetlistonthe
We list the radio flux densities for all discovery obser-
3FGLcatalog,whileprevioussurveyshavebeenbasedlargely
vations in Table 3 (these were obtained from an applica-
on earlier catalogs, and newer Fermi-LAT catalogs include
tion of the radiometer equation and we estimate they have
weaker, less well characterized sources that are more diffi-
6 CROMARTIEETAL.
Table4
AreciboSearches:UnidentifiedGamma-RaySourceInformationfrom3FGLCatalog
3FGLnamea r95b Classc Sigd Curvee Varf Spectrumg Nobsh
(deg) (σ) (σ) Notes
J0103.7+1323 0.08 bcu 7.1 2.4 53 3lh 3
J0134.5+2638 0.06 bcu 12.0 3.0 58 4lh 3
J0232.9+2606 0.09 bcu 4.3 1.6 34 3h 3
J0251.1+2603 0.11 psr 7.9 3.4 36 2cp 2
J0318.1+0252 0.09 ··· 12.8 5.7 50 1Pc 3
J0330.6+0437 0.11 ··· 7.4 2.4 62 2cd 3
J0342.3+3148c 0.10 ··· 6.8 2.5 48 3ld 3
J0421.6+1950 0.12 ··· 6.1 1.9 47 2ld 3
J0517.1+2628c 0.12 ··· 7.0 1.6 50 3ld 5
J0539.8+1434 0.08 fsrq 7.6 3.1 299 5lVd 2
J1048.6+2338 0.12 bll 9.0 2.5 50 3LD 2
J1049.7+1548 0.09 ··· 7.6 1.1 58 3lh 2
J1200.4+0202 0.07 ··· 8.8 1.6 56 4Lh 3
J1225.9+2953 0.05 ··· 17.4 5.2 58 1Cp 3
P7R4J1250+3118i ··· ··· ··· ··· ··· ··· 3
J1309.0+0347 0.15 ··· 4.1 2.7 43 2?l 6
J1322.3+0839 0.12 bcu 7.8 0.4 74 3ld 3
J1601.9+2306 0.11 ··· 7.8 4.3 47 2P 3
J1627.8+3217 0.07 ··· 10.2 3.8 33 2C 2
J1704.1+1234 0.07 ··· 9.4 0.5 47 4LD 3
J1720.7+0711 0.09 ··· 9.6 1.6 44 3cD 3
J1805.9+0614 0.09 psr 9.3 4.3 40 1CP 2
J1824.0+1017 0.09 psr 6.7 3.5 48 2lc 2
J1827.7+1141 0.10 ··· 6.5 3.8 39 2lc 2
J1829.2+3229 0.15 ··· 5.7 3.5 49 2ld 3
J1842.2+2742 0.08 ··· 8.3 2.4 39 2c 3
P7R4J1909+2102i ··· ··· ··· ··· ··· ··· 2
J1921.2+0136 0.10 psr 9.3 1.7 35 2cD 4
J2026.3+1430 0.09 ··· 7.1 2.1 49 2c 3
J2042.1+0247 0.14 PSR 7.5 4.8 28 1CP 2
J2052.7+1217 0.11 psr 7.2 1.6 39 3lD 3
J2108.0+3654 0.06 bcu 6.4 1.5 40 4Hl 3
J2212.5+0703 0.10 ··· 14.3 4.2 57 1Cp 6
J2352.0+1752 0.07 bll 8.9 1.4 56 4Hl 4
Note.—3FGLsourcepropertiesarefromtheAceroetal.(2015)catalog.
aThesixnewAreciboMSPsandtwoMSPsindependentlydiscoveredattheGBT(S.Sanpa-Arsaetal.2016,inpreparation)areshowninbold.
b3FGLsourceerrorcircleradiusatthe95%confidencelevel.
c3FGLpipelineclassificationscheme. PSRandpsrarepulsarswithandwithoutLATpulsations,respectively. TheblldesignationsignifiesaBLLacobject,bcuisanunclassified
blazar,andfsrqisaflat-spectrumradioquasar.
d3FGLsourcesignificance.
eCurvaturesignificancefor3FGLsourcespectrumwhenfittoalog-parabolicmodel.
fVariabilityindexforsource,whereanindex>73denotesvariabilityat>99%confidencelevel.
gForafulldescriptionofSpectrumNotes,seeCamiloetal.(2015).Thefirstnumberinthisschemeisaratingofhowlikelythesourceistobeapulsar.A“1”meansitisverylikely,
whilesourceswitha“4”or“5”rating(or“3”withapossibleAGNassociation)havebeencrossedoffthelistandwillnotbereobservedbecausetheyareunlikelytobepulsars.The
sourcecharacteristics,onwhichtheratingisbased,areobtainedfrominspectionofasource’sspectralenergydistribution.
hNumberoftimeseach3FGLsourcewasobserved(fromTable1).
iNotincludedinthe3FGLcatalog.P7R4designatorsrefertounpublishedsourcelists.
≈ 25% uncertainty). We see by comparison to Figure 1 yieldsaminimumdetectablefluxdensitythatissubstantially
that PSRs J0251+26 and J1824+10 could only have been lowerthanthelongerintegrationselsewhere(seeFigure1).
discovered with Arecibo. Parkes could only have detected Thepopulation ofFermi-LATMSPs containsadispropor-
PSR J1048+2339. This only considers raw telescope sensi- tionately large number of interacting binary systems for rea-
tivity; it does not take into account sensitivity to high accel- sonsthatarecurrentlypoorlyunderstood. Foratime, itwas
eration(discussedlater),whichfurtheremphasizestheutility thought that a tendency for intrabinary shocks to produce
of large telescopes. For a discussion of selection effects re- high-energyradiationcouldbiasFermi-LATsearchestowards
latedtointerstellarscintillationandeclipses,seeCamiloetal. discoveringthesesystems(e.g.,Rayetal.2012). Morerecent
(2015). analyses, however, have found little evidence to support this
WhyhastheArecibosurveyturnedupsuchalargepropor- claim (Johnson 2015). The bias is likely due in part to pre-
tion(5/6)ofhighlyacceleratedinteractingbinaries,compared vioussurveys’biasesagainstfindingbinariesduetoeclipses
to the fractions found in other Fermi-LAT surveys? Though andacceleration.
small-number statistics is a possible explanation, the result The use of modern acceleration search techniques (as im-
can likely be attributed to the Arecibo telescope’s very large plemented within PRESTO in our case; Ransom 2001) was
gain,coupledwiththerelativelyshortintegrationtimesused, essentialforthedetectionofthefivecompactMSPsystems.
and the multiple-observation strategy used to search each Both Johnston & Kulkarni (1991) and Bagchi et al. (2013)
goodtarget. Anintegrationtimeofjust15minutesatArecibo have explored the detectability of binary pulsar systems, the
NEWARECIBOMILLISECONDPULSARSINFERMISOURCES 7
Table5
Valuesofγ2 forPSRJ2052+1218asaFunctionofIntegrationTime
2m
m(harmonic#) 15min 32min 45min 60min
1 0.745 0.315 0.188 0.120
4 0.358 0.138 0.099 0.052
8 0.253 0.087 0.051 0.032
Note.—SeeSection5.3foradiscussionofthiscomparisonofrelativesensitivityto
ahighlyacceleratedfast-spinningbinarypulsar.
withalargeandrapidlychangingacceleration,15-minuteob-
servations are significantly better at recovering power from
a range of harmonics than longer integrations. Comparing
γ2 inthefirstharmonicbetweenthe15-minuteAreciboob-
2m
servation and the next-longest (32-minute GBT) observation
shows that Arecibo yields a signal that is more than twice
the strength of the GBT’s (not considering differences in
Figure3. Thenewshort-orbitMSPsfromthisworkarepresentedinanor- telescope gain and system temperature). The difference be-
bitalperiodvs.minimumcompanionmassplot.MSPsinthelightgreyarea
comes even more dramatic for successively longer observa-
(leftmostblock)areblackwidows,theoneinthepinkarea(rightmostblock)
isaredback,andPSRJ1909+21isintermediatebetweenthetwo,butclassi- tions. Whilelongerintegrationtimesimprovesensitivity,the
fiedasaredback(seeSection4). effectisonlyproportionaltothesquarerootoftheobservation
length,whiletelescopegainisadirectlyproportionalparame-
latterhavingexpandedtheformer’sworktoincludeeccentric ter. Onestrategytocombattheeffectsofacceleration(useful
binaries. Johnston & Kulkarni (1991) provide a quantitative for relatively bright MSPs) is to take a long observation and
measure of the loss of power due to acceleration by way of apply acceleration searches to small subsections of the data,
an “efficiency factor”, γm. Squaring this value gives a ratio aswellassearchingtheentireobservation.
of the power in the mth harmonic, which includes degrada- The characteristics of the Arecibo telescope give it a two-
tionduetoaccelerationaandjerka˙,tothepowerthatwould foldadvantageoversimilarinstruments. First,ithasasignifi-
be present were the acceleration zero. Three such γ terms cantlybetterrawsensitivitythanboththeGBTandParkesfor
m
werereformulatedinBagchietal.(2013). Thefirst,γ2 ,de- similarFermi-LATsourcesearches. Wearethereforeableto
1m
scribes the ratio that would be found in a “standard” pulsar detectfainter systems, evenintheabsenceofconsiderations
search in which acceleration is not searched over. The term relating to binary systems. Second, its large gain allows for
γ2 describesthepowerthatwouldberecoveredinaconstant short observations, which in turn increases its sensitivity to
2m
accelerationsearch(liketheonesweperformed),andwillbe highly accelerated binaries, of which there are many among
employedhere. Itisformulatedasfollows: the Fermi source population. Short observations also allow
us to split observing time over multiple epochs, rather than
γ2m= t1 (cid:12)(cid:12)(cid:12)(cid:12)(cid:90) tintexp(cid:20)imcωp(cid:18)(cid:18)(cid:90) tvldt(cid:19)−αat2−αvt(cid:19)(cid:21)dt(cid:12)(cid:12)(cid:12)(cid:12), iinntgegoruartianbgilfiotyratolocnogmtbimateeactliapsseisngalnedepscoicnhti,llfautritohne.riTnhcoreuagsh-
int 0 0
(2) its declination range is limited, the Arecibo telescope’s raw
wheret is the integration time of the observation, v is the sensitivity firmly establishes its indispensability as an MSP-
int l
pulsar’s line-of-sight velocity, and ω is its angular spin fre- findingresource.
p
quency. Amodernsearchalgorithmyieldsvaluesofacceler-
ation α and velocity α that maximize γ . Here, γ2 =1
a v 2m 2m H. Thankful Cromartie would like to thank the NSF, Fer-
forasystemwithconstantacceleration. Thefinalterm, γ2 ,
3m nando Camilo, the staff of the Arecibo Observatory, and its
describesthepowerratiorecoveredinasearchovervelocity,
residentscientistsfortheopportunitytopursuethisresearch,
acceleration, and jerk. Such search algorithms are currently
andforanexperiencethatcompelledhertocontinueinastro-
beingdeveloped,buthavenotyetbeenimplemented.
physics.
UsingsoftwareprovidedbyBagchietal.(2013)14,wecal-
The Arecibo Observatory is operated by SRI Interna-
culatedvaluesofγ22mforPSRJ2052+1218duringa15-minute tional under a cooperative agreement with the National Sci-
integration, such as at Arecibo. We then recalculated these ence Foundation (AST-1100968), and in alliance with Ana
values using the integration times for surveys at the GBT G. Méndez-Universidad Metropolitana, and the Universities
andParkestocomparethedetectabilityofthisfast-spinning, SpaceResearchAssociation.
highlyacceleratedbinarypulsarbythefourdifferentsurveys. TheNationalRadioAstronomyObservatoryisafacilityof
ResultsaregiveninTable5. the National Science Foundation operated under cooperative
As expected, the power recovered in successively higher agreementbyAssociatedUniversities,Inc.
harmonicsdecreasesforeachofthefoursurveys. Thevalue TheFermiLATCollaborationacknowledgesgenerouson-
of γ2 in the first harmonic is a reasonable proxy for binary going support from a number of agencies and institutes that
2m
detectability; that is, the higher the fraction of power that is have supported both the development and the operation of
recoveredinanaccelerationsearch,themorelikelyoneisto theLATaswellasscientificdataanalysis. Theseincludethe
detecttheMSPinagivenobservation. ForPSRJ2052+1218, National Aeronautics and Space Administration and the De-
partmentofEnergyintheUnitedStates, theCommissariatà
14http://psrpop.phys.wvu.edu/binary l’EnergieAtomiqueandtheCentreNationaldelaRecherche
8 CROMARTIEETAL.
Scientifique / Institut National de Physique Nucléaire et de Cognard,I.,Guillemot,L.,Johnson,T.J.,etal.2011,ApJ,732,47
PhysiquedesParticulesinFrance,theAgenziaSpazialeItal- Cordes,J.M.,&Lazio,T.J.W.2002,preprint(arXiv:astro-ph/0207156)
Deneva,J.S.,Ray,P.S.,Camilo,F.,etal.2015,ApJ,submitted
ianaandtheIstitutoNazionalediFisicaNucleareinItaly,the
(arXiv:1601.03681)
MinistryofEducation,Culture,Sports,ScienceandTechnol- Geringer-Sameth,A.&Koushiappas,S.M.2012,MNRAS,421,1813
ogy (MEXT), High Energy Accelerator Research Organiza- Haslam,C.G.T.,Klein,U.,Salter,C.J.,etal.1981,A&A,100,209
Hessels,J.W.T.,Roberts,M.S.E.,McLaughlin,M.A.,etal.2011,inAIP
tion(KEK)andJapanAerospaceExplorationAgency(JAXA) Conf.Series,Vol.1357,RadioPulsars,ed.M.Burgay,N.D’Amico,
inJapan, andtheK.A.WallenbergFoundation,theSwedish P.Esposito,A.Pellizzoni,&A.Possenti,40–43
Johnson,T.2015,inEWASSSymposium11,Neutronstarsatthe
Research Council and the Swedish National Space Board in
crossroads:X-raybinariesandtransitionalmillisecondpulsars,
Sweden. Additional support for science analysis during the http://www.iac.es/congreso/ns-ewass-2015/
operations phase is gratefully acknowledged from the Isti- Johnston,H.M.,&Kulkarni,S.R.1991,ApJ,368,504
Kerr,M.,Camilo,F.,Johnson,T.J.,etal.2012,ApJ,748,L2
tutoNazionalediAstrofisicainItalyandtheCentreNational
Lawson,K.D.,Mayer,C.J.,Osborne,J.L.,ParkinsonM.L.1987,
d’ÉtudesSpatialesinFrance. MNRAS,225,307
Lorimer,D.R.,&Kramer,M.2005,HandbookofPulsarAstronomy
Facilities: Fermi (LAT), Arecibo Observatory (PUPPI),
(CambridgeUniversityPress)
GBT(GUPPI) Lynch,R.S.,Boyles,J.,Ransom,S.M.,etal.2013,ApJ,763,81
Nolan,P.L.,Abdo,A.A.,Ackermann,M.,etal.2012,ApJS,199,31
Özel,F.,Psaltis,D.,Narayan,R.,&Villarreal,A.S.2012,ApJ,757,55
REFERENCES Ransom,S.M.,Greenhill,L.J.,Herrnstein,J.R.,etal.2000,ApJ,546,L25
Ransom,S.M.2001,PhDthesis,HarvardUniversity
Ransom,S.M.,Eikenberry,S.S.,&Middleditch,J.2002,AJ,124,1788
Ransom,S.M.,Ray,P.S.,Camilo,F.,etal.2011,ApJ,727,L16
Abdo,A.A.,Ackermann,M.,Ajello,M.,etal.2010,ApJS,188,405
Ray,P.S.,Abdo,A.A.,Parent,D.,etal.2012,eConfC110509,
Abdo,A.A.,Ajello,M.,Allafort,A.,etal.2013,ApJS,208,17
arXiv:1205.3089
Acero,F.,Ackermann,M.,Ajello,M.,etal.2015,ApJS,218,23
Roberts,M.S.E.,Hessels,J.W.T.,Ransom,S.M.,&Kaspi,V.M.2002,
Atwood,W.B.,Abdo,A.A.,Ackermann,M.,etal.2009,ApJ,697,1071
ApJ,577,L19
Bagchi,M.,Lorimer,D.R.,&Wolfe,S.2013,MNRAS,432,1303
Roberts,M.S.E.2012,inProceedingsIAUSymposiumNo.291,Neutron
Camilo,F.,Kerr,M.,Ray,P.,etal.2015,ApJ,810,85
StarsandPulsars:ChallengesandOpportunitiesafter80years,ed.van
Caraveo,P.A.,2014,AnnualReviewofAstronomyandAstrophysicsvol.
Leeuwen,J.,arXiv:1210.6903
52,211
Stovall,K.,Lynch,R.S.,Ransom,S.M.,etal.2014,ApJ,791,67
Champion,D.J.,McLaughlin,M.A.,&Lorimer,D.R.2005,MNRAS,
364,1011