DISCOVERY OF PULSED γ-RAYS FROM THE YOUNG RADIO PULSAR PSR J1028−5819 WITH THE FERMI LARGE AREA TELESCOPE

One of the most intriguing legacies of the Energetic Gamma-Ray Experiment Telescope (EGRET) on the Compton Gamma-Ray Observatory (CGRO) was the group of 150 sources (Hartman et al. 1999) that could not be firmly identified with any known counterparts. The majority of these (about 100) are in a population clustered toward the Galactic plane, and many were presumed to be γ-ray pulsars. Indeed, a number of radio pulsars were discovered to lie in or near EGRET unidentified source error boxes after the end of CGRO observations in 2000, both in large surveys such as the Parkes Multibeam Survey (Kramer et al. 2003) or in deep observations in some of the EGRET error boxes (e.g., Roberts et al. 2002; Halpern et al. 2001). Unfortunately, searches for pulsations at the radio periods in the EGRET archival data were not feasible, given the small number of γ-ray photons in these sources and the difficulty of predicting pulsar phase in the presence of rotational instabilities (also known as timing noise).

The situation has improved dramatically with the recent launch of Fermi on 2008 June 11, which has been operating successfully through early calibration and now in the sky survey mode. The Fermi LAT has a sensitivity that is more than an order of magnitude that of EGRET, and it is already possible to better perform searches for γ-ray pulsations in many of the EGRET sources. This Letter reports results on one such search in the error circle of the EGRET source 3EG J1027−5817, at the location of the radio pulsar PSR J1028−5819, discovered (Keith et al. 2008) just a few months prior to the launch of Fermi as part of a search of three EGRET sources at high frequency. PSR J1028−5819 is a young pulsar, with period P = 91.4 ms, period derivative $\dot{P} = 1.61 \times 10^{-14}\,\rm s\,s^{-1}$ and characteristic age of 9.21 × 10⁴ yr. The derived spin-down power, $\dot{E}_{\rm sd} = 8.43 \times 10^{35} \,\rm erg\,s^{-1}$ combined with its dispersion measure-derived distance of 2.3 kpc makes it a plausible counterpart for the EGRET source with flux of 6.6 ± 0.7 × 10⁻⁷ ph cm^-2 s⁻¹ (E > 100 MeV). The radio pulse profile is extremely narrow, consisting of two highly linearly polarized components. The full width is only 560 μs, giving it the smallest duty cycle of any known pulsar and an order of magnitude smaller duty cycle than other young pulsars (e.g., Weltevrede & Johnston 2008). Fermi LAT has now discovered the γ-ray pulsations from PSR J1028−5819, and we can thus confirm that some of the photons attributed to 3EG J1027−5817 originate from PSR J1028−5819.

Fermi was launched into low-Earth orbit and, after a six-week commissioning phase, began nominal sky-survey observations on 2008 August 11. The LAT, the main instrument on Fermi, is a pair-production telescope (Atwood et al. 2009) sensitive to γ rays from 20 MeV to at least 300 GeV with on-axis effective area >1 GeV of ∼8000 cm², exceeding that of EGRET by a factor of about 5. It has a large field-of-view of 2.4 sr and in the survey mode observes the entire sky every 3 hr. We report here on observations using data collected during the initial 35 days in on-orbit verification that included sky-survey tuning and pointed-mode tuning on Vela, from June 30–2008 August 3, as well as the initial 15 weeks of sky survey, from August 3–2008 November 16. The Diffuse class events (LAT event class having the tightest background rejection) from these periods total 6014 photons with energy >100 MeV, within a radius of 15 surrounding PSR J1028−5819. We excluded periods when the pulsar was viewed at zenith angle >105° to the detector axis where the Earth's albedo photons gave excessive background contamination.

Radio observations of PSR J1028−5819 were carried out at the Parkes 64 m radio telescope at frequencies near 1.4 and 3.1 GHz with typical durations of ∼5 minutes. Timing observations commenced on 2008 April 7, shortly after the discovery of the radio pulsar. Since then 24 independent timing measurements have been made with typical uncertainties in the times-of-arrival of ∼30 μs or better. The fit to the timing points was carried out using TEMPO2 (Hobbs et al. 2006); the position of the pulsar was held fixed at the value given in Keith et al. (2008) and the rotation frequency and frequency derivative were fit. The dominant contribution to the resulting residual of 270 μs is timing noise intrinsic to the pulsar. Note that the timing solution used to derive the narrow radio profiles contains an extra fit parameter that is not included in the Fermi software tools used for the γ-ray profile. Including this additional parameter leads to at most a phase shift of 0.01 over the data set used for the γ-ray pulsation analysis.

For the detection of pulsations, Diffuse class events with energy >100 MeV and within a radius of 15 of the radio position were corrected to the solar system barycenter using the JPL DE405 solar system ephemeris and folded with the radio period using the Parkes ephemeris. The Fermi LAT timing is derived from a GPS clock on the satellite and photons are timestamped to an accuracy better than 300 ns. The LAT software tools for pulsars have been shown to be accurate to a few μs for isolated pulsars (Smith et al. 2008). We detect γ-ray pulsations at the radio period with chance probability 2 × 10⁻²⁷ using a Z_n-test with two harmonics (De Jager et al. 1989). Within two weeks from the onset of data collection, a Z_n significance of 3.5σ was found for pulsations, and this was improved to a better than 10σ pulsed signal with less than four months of LAT data. The γ-ray pulsations at the same period were independently found by a blind search in P and $\dot{P}$ using a time-differencing technique (Atwood et al. 2006; Ziegler et al. 2008).

The LAT has an angular resolution with a dependence on reconstructed event energy E of E^−0.75, with a 68% containment radius of ∼05 at 1 GeV for near on-axis events which convert in one of the first 12 layers of the tracker (the thin section) and that increases with incidence angle as detailed in Atwood et al. (2009). Events converting in one of the last four layers of the tracker (the thick section) have a 68% containment radius which is ∼2 times that of the thin section. In order to explore the energy dependence of the light curve, the event selection was refined to be θ_c(E/100 MeV)^−0.75, where θ_c = 3° for thin events and θ_c = 41 for thick events are the containment radii at 100 MeV chosen to maximize the detection significance with this energy-dependent cut. The histogram of folded counts at energies 0.1–13 GeV is shown in Figure 1. The γ-ray light curve shows two strong peaks, P1 at phase 0.200 ± 0.003 and P2 at phase 0.661 ± 0.002, where phase 0 is defined by the dedispersed radio pulse. The phase separation of P1 and P2 is 0.460 ± 0.004. The peaks are fairly narrow, with Lorentzian FWHM for P1 of 0.040 ± 0.011 and for P2 of 0.035 ± 0.007. This light curve is very similar to that of Vela, as seen with EGRET (Kanbach et al. 1994) and now with Fermi (Abdo et al. 2009a). Figure 2 shows light curves in four different energy ranges, 100–300 MeV, 300 MeV to 1 GeV, 1–3 GeV, and greater than 3 GeV that do not exhibit significant evolution in shape. In particular, we have measured the widths of the peaks as a function of energy and for P1 there is a deviation at only the 2.19σ level and for P2 at the 1.63σ level. There is also no evidence for significant evolution of the P1/P2 ratio. A fit of the P1/P2 values to a constant function of energy gives a χ² per degree of freedom of 0.54, which is consistent with no variation. This contrasts to the decrease in the P1/P2 ratio with energy seen in the Crab, Vela, Geminga, and B1951+32 pulsars by EGRET (Thompson 2004), and is now confirmed in the Vela pulsar by Fermi (Abdo et al. 2009a). We have measured the pulsed significance of PSR J1028+5819 as a function of increasing energy, successively raising the low energy threshold. For all events above 4 GeV (with the energy-dependent cut) the pulsed detection is 3.5σ, but above 5 GeV it is only 1.8σ, indicating that the maximum energy of pulsations is around 4 GeV.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The LAT point source 0FGL J1028.6−5817 from the Fermi LAT bright source list (Abdo et al. 2009b) corresponding to PSR J1028−5819 is located at (R.A., decl.) = (157.166, −58.292) with a 95% confidence level radius of 0.079 deg. There are two other LAT point sources nearby, 0FGL J1024.0−5754 and 0FGL J1018.2−5858, 073 and 152 away respectively. The COS-B source 2CG 284-00 (Swanenburg et al. 1981) was apparently made up of contributions from all three LAT sources, while the EGRET source 3EG J1027−5817 has now been resolved by the Fermi LAT into contributions from the two sources, 0FGL J1028.6−5817 and 0FGL J1024.0−5754. The LAT source associated with PSR J1028−5819 is therefore somewhat confused with the nearby point sources and contains significant photon flux from 0FGL J1024.0−5754. This adds to the unpulsed background and must be taken into account in computing the phase-averaged flux from the pulsar. A Swift XRT (Burrows et al. 2005) observation on 2008 November 23–24, (9.6 ks cleaned exposure) yielded a 4.1σ detection of an X-ray source that we tentatively associate with the pulsar. Although the X-ray source is positionally coincident with the radio pulsar, it is poorly localized (∼20'' total extent) and is faint, with a net count rate (0.3–10 keV) of 2.3 × 10⁻³ counts s⁻¹ which corresponds to an absorbed (unabsorbed) flux of 1.5 (3.4) × 10⁻¹³ erg cm⁻² s⁻¹, assuming a power law with Γ = 2 and N_H = 1.59 × 10²² cm⁻².

To obtain the phase-averaged flux of the pulsar, we have performed a maximum likelihood spectral analysis using the LAT tool gtlike⁵⁰ on counts within a radius of 15° from the LAT source position. The pulsar spectrum was fit with a power law with an exponential cutoff, giving an index 1.22 ±  0.2 ±  0.12 and cutoff energy of 2.5 ±  0.6 ±  0.5 GeV (the first errors are statistical and the second are systematic). With the current statistics in this crowded region, a differentiation between a simple and a hyper-exponential cutoff is not yet possible and, therefore, was not attempted. From this fit, we obtain an integral photon flux at 0.1–30 GeV of 1.62 ± 0.27 ± 0.32 × 10⁻⁷ ph cm^-2 s⁻¹ and integral energy flux of 1.78 ±  0.15 ±  0.35 × 10⁻¹⁰ erg cm^-2 s⁻¹. This flux is a quarter of the EGRET source flux, which apparently included contributions from both PSR J1028−5819 and its neighboring source 0FGL J1024.0−5754. In addition to the pulsar, the diffuse γ-ray emission from the Milky Way was modeled, while the extragalactic diffuse emission plus instrumental residual background and other LAT point sources within the region of interest were fit with power law spectra. The LAT source 0FGL J1024.0−5754 which likely contributed to the EGRET source was fit with an integrated flux from 0.1–30 GeV of 3.0 × 10⁻⁷ ph cm^-2 s⁻¹. The flux from this source plus that from PSR J1028−5819 add up to ∼4.6 × 10⁻⁷ ph cm^-2 s⁻¹ which is consistent, within statistical and systematic uncertainties, with the flux 6.6 ± 0.7 × 10⁻⁷ ph cm^-2 s⁻¹ of 3EG J1027−5817. The 3EG source flux was derived by assuming a power law with index 2 (Hartman et al. 1999) which would have preferentially weighted the spectrum with a larger PSF, adding more counts from the background at low energy to produce a higher flux. The LAT source 0FGL J1018.2−5858, which likely contributed to the flux of the 2CG source, yielded an integrated flux at 0.1–30 GeV of 5.41 × 10⁻⁷ ph cm^-2 s⁻¹ from a power-law fit. Adding the flux from PSR J1028−5819, 0FGL J1024.0−5754, and 0FGL J1018.2−5858 then could account for the 2CG flux of ∼2.7 × 10⁻⁶ ph cm^-2 s⁻¹.

The detection of pulsed γ rays from the recently discovered pulsar PSR J1028−5819 within the first few weeks of the Fermi mission confirms the promise that the Fermi LAT will be an important instrument for pulsar studies. The γ-ray pulses of PSR J1028−5819 cover a wide phase range and neither of the peaks is aligned with the very narrow radio pulse. This strongly suggests that the γ-ray beam covers a large fraction of solid angle of the sky and favors its interpretation in outer magnetosphere models such as the outer gap (OG; Cheng et al. 1986; Romani & Yadigaroglu 1995) or slot gap (SG; Muslimov & Harding 2004). Furthermore, the maximum observed energy of pulsations, _max ≃ 4 GeV must lie below any γ–B pair production turnover threshold, thereby providing a lower bound to the altitude of emission, even though we have not been able to rule out the hyperexponential spectral cutoff expected for pair production attenuation. Using a standard polar cap model estimate for the minimum emission height of r ≳ (_maxB₁₂/1.76 GeV)^2/7 P^−1/7 R_* (inverting Equation (1) of Baring 2004), for a surface polar field strength of B₁₂ = B/10¹² G and neutron star radius R_*, the PSR J1028−5819 spin-down parameters (P = 0.0914s, B₁₂ = 1.015) yield r ≳ 1.8R_*. This bound precludes emission very near the stellar surface.

In order to estimate the γ-ray efficiency of a pulsar, one needs to know the total luminosity radiated, L^tot_γ = 4πf(α, ζ_E)Φ_obsd², where f is a correction factor that contains information about the beaming geometry, Φ_obs is the observed phase-averaged energy flux, and d is the distance to the source. The factor f(α, ζ_E) is a function of the pulsar magnetic inclination angle α and the viewing angle to the rotation axis, ζ_E, is very model sensitive and can be computed for any particular model geometry by,

$$\begin{equation} f(\alpha, \zeta _E) = {\int F(\alpha, \zeta, \phi) d\Omega \over 4\pi \Phi _{\rm obs}} \end{equation} \tag{ 1 }$$

(Watters et al. 2009). Here, F(α, ζ, ϕ) is the radiated flux from a pulsar as a function of inclination angle, viewing angle, and rotation phase, ϕ. The factor f(α, ζ_E) is very important because it quantifies the amount of the emission we may be missing with our limited sweep over the pulsar beam. For polar cap models where the emission originates within several stellar radii of the neutron star surface, the effective emission solid angle is small and thus the factor f ≪ 1. For outer magnetosphere models, the traditional solid angle measure is not appropriate since the emission is radiated over a large fraction of 4π. The beaming correction factor f must therefore be computed numerically using Equation (1) and it is found that f ≳ 1 for both OG and SG models (Watters et al. 2009). For PSR J1028−5819, we obtain a total luminosity of L^tot_γ = 1.1 × 10³⁵ f erg s^-1 from the observed energy flux and source distance, d = 2.3 kpc. The γ-ray efficiency is thus $\eta _{\gamma } = L_{\gamma }^{\rm tot}/\dot{E}_{sd} = 0.13\,f/I_{45}$, where I₄₅ = I/10⁴⁵ g cm² is the neutron star moment of inertia. Since the distance could have as much as a 40% error from fluctuation in the free electron density (Brisken et al. 2002), there is an uncertainty of about a factor of 2 in the derived luminosity and the efficiency. If the outer magnetosphere interpretation of the pulsed γ rays from PSR J1028−5819 and other young pulsars is correct, then these pulsars have efficiencies that are larger by about an order of magnitude than would be deduced using the previously standard 1 sr solid angle.

The full geometry of PSR J1028−5819 is not easily determined from the radio data alone because the narrowness of the pulse makes it difficult to derive good solutions from polarization position angle variation using the rotating vector model. But the very narrow pulse does argue for a large α (in agreement with the constraints below). In the outer-magnetosphere geometry now favored by the γ-ray emission, one can derive constraints on α and ζ from the measured separation of the γ-ray peaks and γ-ray efficiency alone. Using the γ-ray light curve "Atlas" of Watters et al. (2009), we estimate an allowed range of α ∼ 70°–90°, ζ ∼ 75°–80°, and f ∼ 1.1 for the OG model and α ∼ 65°–80°, ζ ∼ 60°–80°, and f ∼ 0.9–1.0 for the two-pole caustic (TPC) or SG models. Both of these models therefore have a good range for viable solutions. However, the above estimates assume the γ-ray efficiency relation $\eta _{\gamma } \simeq (10^{33}\,{\rm erg\,s^{-1}}/\dot{E}_{sd})^{1/2}$ which gives η_γ = 0.03, a factor of 4 smaller than that derived for this pulsar from its measured luminosity. Using our derived η_γ = 0.13, the range allowed for α shrinks to ∼80°–90° for the OG model, but remains about the same for the TPC model. Detection of a pulsar wind nebula (which can give an estimate of ζ_E) or radio polarization measurement (which might give an α estimate) would help further constrain the model. The phase lag of the first γ-ray peak relative to the radio pulse can be explained in either OG or SG (or generally TPC; Dyks & Rudak 2003) models, where the radio phase crossing associated with a magnetic pole occurs before the first high-energy emission caustic formed at high altitude.

If many of the other young radio pulsars discovered in EGRET error boxes are similar to PSR J1028−5819 and if the γ-ray beams of young pulsars can be seen from a wide range of viewing angles, then Fermi searches for their pulsed γ rays promise to be very fruitful. The idea of wide γ-ray beams recently gained strong support from the Fermi LAT discovery of a radio-quiet pulsar in the young supernova remnant CTA 1 (Abdo et al. 2008) whose pulsations have been detected only in γ rays, a blind search result secured promptly during the Fermi commissioning phase. The discovery of γ-ray pulsations from a young radio pulsar coincident with an EGRET source, together with the discovery of γ-ray pulsations from CTA 1, confirms expectations from before the Fermi launch that many γ-ray pulsars remain to be discovered. Even more exciting are the prospects that the all-sky survey of the Fermi LAT will identify many new sources, and that searches in both the γ-ray and radio bands will uncover an entirely new population of pulsars.

The Fermi LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat à l'Energie Atomique and the Centre National de la Recherche Scientifique/Institut National de Physique Nucléaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), High Energy Accelerator Research Organization (KEK) and Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden.

Additional support for science analysis during the operations phase from the following agencies is also gratefully acknowledged: the Istituto Nazionale di Astrofisica in Italy and the K. A. Wallenberg Foundation in Sweden for providing a grant in support of a Royal Swedish Academy of Sciences Research fellowship for J.C.

The Parkes radio telescope is part of the Australia Telescope which is funded by the Commonwealth of Australia for operation as a National Facility managed by the CSIRO.