Fermi LAT Observations of the Vela Pulsar
The Vela pulsar is the brightest persistent source in the GeV sky and thus is the traditional first target for new -ray observatories. We report here on initial Fermi Large Area Telescope observations during verification phase pointed exposure and early sky survey scanning. We have used the Vela signal to verify Fermi timing and angular resolution. The high quality pulse profile, with some 32,400 pulsed photons at GeV, shows new features, including pulse structure as fine as 0.3 ms and a distinct third peak, which shifts in phase with energy. We examine the high energy behavior of the pulsed emission; initial spectra suggest a phase-averaged power law index of with an exponential cut-off at GeV. Spectral fits with generalized cut-offs of the form require , which is inconsistent with magnetic pair attenuation, and thus favor outer magnetosphere emission models. Finally, we report on upper limits to any unpulsed component, as might be associated with a surrounding synchrotron wind nebula (PWN).
Subject headings:gamma rays: general; pulsars: individual: PSR B083345
Radio pulsations at =89 ms from PSR B083345 (=PSR J08354510) in the Vela supernova remnant were discovered by Large, Vaughan & Mills (1968). This pulsar is bright ( Jy), young (characteristic age kyr), and energetic (, for a neutron star moment of inertia ). It is embedded in a flat spectrum radio synchrotron nebula (see Frail et al. 1997) and is surrounded by a bright X-ray wind nebula displaying remarkable toroidal symmetry (Helfand et al. 2001; Pavlov et al. 2003). A VLBI parallax measurement provides a well-determined distance of pc (Dodson et al. 2003), improving on earlier optical HST measurements (Caraveo et al. 2001). This proximity means that the spindown energy flux at Earth, , is second only to that of the Crab pulsar and ensures that Vela is among the most intensely studied neutron stars, particularly at high energies.
Pulsed -ray emission from Vela was first detected during the SAS-2 mission (Thompson et al. 1975); it is, in fact, the brightest persistent source of celestial -rays and has been the proving ground of GeV observatories ever since. The basic source properties discovered by SAS-2 were elaborated using observations by COS-B (Kanbach et al. 1980), and CGRO/EGRET (Kanbach et al. 1994): the source is 100% pulsed with no convincing evidence for year-scale variations, the two -ray peaks are separated by 0.42 in pulsar phase and the first peak lags the radio peak by 0.12 in phase. The source spectrum is hard with an average photon index with evidence of a cut-off above 2–4 GeV. The most detailed study, to date, of the -ray pulsations is that of Fierro et al. (1998), which produced a high signal-to-noise pulse profile and evidence for phase-resolved spectral variations. As the present work was being prepared for submission, the AGILE team reported their initial results on -ray pulsars, including Vela (Pellizoni et al. 2008); the results, given the more limited count rates and energy resolution, are broadly consistent with our conclusions described below.
With the successful launch of the Fermi Gamma-ray Space Telescope, formerly GLAST, observatory on June 11, 2008, we have a new opportunity to examine the high energy behavior of the Vela pulsar and to study this archetype of the young, energetic pulsars in detail. During Launch and Early Operations phase (L&EO) the Fermi mission targeted the Vela pulsar for a number of pointed observations, in addition to coverage during initial tests of the sky survey mode. In the latter mode the instrument axis is offset North and South of the zenith during alternate orbits to provide near-uniform sky coverage every three hours. One main purpose of these early observations was to tune the Large Area Telescope (LAT) performance on celestial -ray sources. However, the initial results on Vela itself are of interest, including new high energy features in the pulse profile, an improved measurement of the high energy cut-off in the pulsar spectrum, and a search for associated pulsar wind nebula (PWN) emission at GeV energies.
|2008 Jun 30–Jul 4||54647.40–54651.36||Sky survey||1859||LAT First light.|
|2008 Jul 4–Jul 15||54651.38–54662.10||Survey+Pointings||5172||LAT Calibrations.|
|2008 Jul 15–Jul 19||54662.12–54666.08||Pointings+Survey+limb Following||1170||Pointed Obs. Tuning|
|2008 Jul 19–Jul 22*||54666.10–54669.14||Pointed Vela+ 2 Target||7751||Pointed Obs. Tuning|
|2008 Jul 22–Jul 24||54669.16–54671.41||Pointed Vela+ 2 Target||3212||Pointed Obs. Tuning|
|2008 Jul 24–Jul 30*||54671.36–54677.45||Pointed Vela+ 2 Target||7607||Pointed Obs. Tuning|
|2008 Jul 30–Aug 3||54677.45–54681.66||Polar study + Sky Survey Tuning||886||Nominal Ops.|
|2008 Aug 3–Sep 14*||54687.68–54723.91||Sky Survey||26680||Nominal Ops.|
|All photons (pulsed and background) within 5 of PSR B0833-45, zenith angle 105, event class Pass6-Diffuse.|
|Observations marked by asterisk are those used for spectral studies.|
|LEO ended on Aug.11.|
2. Radio Timing Observations
The Vela pulsar is young and exhibits substantial timing irregularities. This means that the optimal use of the Vela -ray photons, which arrive at an orbit-averaged rate of one every 4 minutes during LAT sky survey observations, requires a simultaneous radio ephemeris. The radio ephemeris is obtained using observations made with the 64-m Parkes radio telescope as part of the overall program for pulsar monitoring in support of the Fermi mission (Smith et al. 2008). A total of 27 times-of-arrival (TOAs) were obtained at a frequency of 1.4 GHz over the period of the LAT observations. The median error in each TOA is 1.7 s. Fits to the TOAs were made with the pulsar’s rotational frequency and frequency derivative as free parameters and the ephemeris created from these data maintains phase with an rms residual of 90 s, or of pulsar phase, throughout the LAT observations. No large timing irregularities or glitches were detected. An accurate determination of the pulsar’s dispersion measure of was obtained by measuring the delay in the TOAs of the pulse across the 256 MHz bandpass of the 1.4 GHz receiver. This allows extrapolation of the radio ephemeris to infinite frequency with an error of 60 s. Photon arrival times were referred to the Solar-System barycenter and pulse phases were assigned using the standard pulsar timing software TEMPO2 (Hobbs et al. 2006).
We thus can assign a pulsar phase to the -ray photons referenced to the radio with high confidence. Gamma-ray events recorded with the LAT have timestamps that derive from a GPS clock on the Fermi satellite. Pre-launch ground timing measurements using cosmic muons seen coincidentally in the LAT and in an independent counting system demonstrated that the LAT measures event times relative to the spacecraft clock with a resolution of better than 300 ns. On orbit, satellite telemetry indicates a comparable resolution. The contribution to the barycenter times from uncertainty in the LAT’s position is negligible. It is also useful to confirm the absolute timing using celestial signals; as of this writing the six classical pulsars detected by EGRET have provided good quality LAT pulse profiles. The phasing of the radio and -ray peaks is consistent with the best EGRET results, ensuring absolute timing relative to the radio time system better than ms.
3. Gamma-Ray Observations
The LAT instrument on Fermi is described by Atwood et al. (2008) and references therein. The LAT is an electron-positron pair production telescope featuring solid state silicon trackers sensitive to photons from MeV to GeV. It has a large 2.4sr field of view, and compared to earlier -ray missions, has a large effective area ( on axis), improved resolution (, 68% containment radius at 1 GeV for events collected in the ‘front’ section with thin radiator foils) and small dead time (s/event). We report here on the LAT’s initial observations of the Vela pulsar, using data collected during 35 days of on-orbit verification tests and the initial days of the on-going first year sky survey. These data already suffice to illustrate the power of the LAT for astronomical observations and, indeed, show several new features in the radiation of this well-known -ray pulsar.
Table 1 contains a journal of the Vela coverage, along with numbers of photons with measured energy above GeV recorded within a radius of 5 of the pulsar, including background photons. For both pointed and survey observations we exclude events within of the Earth’s limb to minimize contamination by ‘Earth albedo’ photons.
During the L&EO period, the instrument configuration was being tuned for optimum performance. Accordingly, the knowledge of the energy scale and effective area are more limited than for routine operations. We discuss how these data verify the LAT photon selection, effective area, timing, photon energy measurement and the variation of the point spread function (PSF) with energy. These results may be of use to researchers seeking to predict Fermi capabilities during longer exposures of fainter sources, in both sky survey and pointed mode.
4.1. Pulse Profiles
After calibration of the LAT pointing axis on a large set of identified high latitude -ray sources, we find that the best fit position of the Vela point source is within 0.5 of the radio pulsar position with a statistical error of 0.4(95%); the LAT has unprecedented accuracy for localization of bright hard -ray sources, although some systematic uncertainties remain.
The Vela pulsar is embedded in the bright -ray emission of the Galactic plane, which is particularly strong at low energies. Further, the LAT, like all pair production telescopes, has an angular resolution dominated by scattering at low photon energies (see below). Thus selection of photons from an energy-dependent region of interest (ROI) around Vela is important; the best selection depends on the desired product. Here, we seek pulse profiles with good signal to noise over a broad energy range, so we select photons within an angle deg of the pulsar position. This includes a larger fraction of the PSF at high energies, where the background is relatively faint. We use ‘Diffuse’ class events, those reconstructed events having the highest probability of being photons.
In this energy-dependent aperture we have 32,400242 pulsed photons and background photons with measured energy GeV. Figure 1 shows the 0.1-10 GeV pulse profile from this ROI with the peak of the radio pulsar signal at phase 0. To highlight the fine structure we plot in Figure 1 the pulse profile using variable-width bins, each containing 200 counts. These counts, divided by the bin width, give the photon flux in each phase interval; these phase bin fluxes thus have a Poisson statistical error of 7%. The pulse profile is normalized to 100 at the pulse peak. Three insets show structure near the first peak (P1), second peak (P2) and ‘off-pulse’ window at a finer scale. These peaks are at pulsar phases and , respectively.
The peaks are asymmetric. In particular P2 has a slow rise and a fast fall. P1 also has a steep outer edge with the fall somewhat slower. We fit each peak in Figure 1 with two half-Lorentzian functions (with different widths for the leading and trailing sides) over phase intervals which avoided the complex bridge flux ( for P1, for P2). The outer edges of the two peaks had consistent Lorentzian half-widths of . Extra structure at the pulse peaks made the half maximum widths somewhat smaller (; 0.8 ms). The falling edge of peak 1 has a Lorentzian half width of , while the rising edge of P2 has a width . Both peaks show apparently significant structure on scales as small as (s), but additional counts are required to fully probe the pulse profile at this scale. We also show an inset of the ‘off pulse region’ (), with 50 counts/bin to show fine structure. Here and in the main panel, the dashed line shows the estimated residual unpulsed background counts in our energy-dependent ROI, as measured from the pulse minimum. This is in good agreement with a background level estimated from around the pulsar. We see that, as for other non-radio Vela pulse profiles, there is a faint tail of pulsed emission in this region, reaching non-detectability only near . We estimate that this ‘off pulse’ window contains 23515 pulsed photons ( of the pulsed flux).
There is dramatic energy evolution in the -ray pulse profile. Figure 2 shows pulse profiles from six energy bins, covering three decades of the LAT data, drawn from our energy-dependent aperture. Note that the relatively large aperture at low energies leads to larger off-pulse flux from the surrounding background. We should also caution that, with the present limited accuracy in energy measurement below 0.1 GeV, some higher energy photons may leak into the lowest energy bin. The most prominent pulse feature is a decrease of P1 relative to P2 with increasing energy; P1 is not detectable above 10 GeV (Figure 2). This confirms a trend seen in the EGRET data for the Crab, Vela and Geminga pulsars – the second pulse dominates at the highest energies (Thompson 2001). Interestingly, at the lower energies, below MeV, the trend is reversed with P1 weakening again with respect to P2. We do not find any statistically significant evidence for shifts in the phases of the narrow P1 and P2 pulse components with energy. These structures, spanning together of the rotational phase, contain emission dominating over three decades of the pulsar photon output.
The ‘bridge’ region between the peaks also shows appreciable structure with a trailing shoulder of P1 shifting to later phase. At 1 GeV it is clear that this is a distinct pulse component (P3) with a peak at in the 3-10 GeV band (Figure 2). Note that this peak shifts in phase by between 0.2 and 15 GeV (Figure 3).
To compare with the lower energy structure, we show at bottom left of Figure 2 the 8-16 keV non-thermal X-ray pulse measured by RXTE (Harding et al. 2002). On the right we similarly show the 4.1-6.5 eV NUV HST STIS/MAMA pulse profile of Romani et al. (2005). The 1.4 GHz radio pulse profile, whose peak defines phase , runs across the bottom two panels. Both the optical/UV and the hard X-rays are dominated by non-thermal magnetospheric emission. In contrast, the keV soft X-ray flux is dominated by thermal emission from the neutron star surface (e.g. Manzali, de Luca & Caraveo 2008), making it much more difficult to trace the non-thermal peaks through this intermediate energy band.
We may compare components in the UV and hard X-ray pulse profiles shown with those of the non-thermal -ray pulsations. In particular, we should note that the -ray P1 component is dominant in the non-thermal X-rays. This component is absent in the optical/UV, but pulse profiles at these energies have a strong peak in the bridge region at , well matched in phase to the P3 structure in the GeV pulse profiles. This possible connection between optical pulses and the -ray bridge emission was already noted by Kanbach et al. (1980). Note that the profile for UV energies has a distinct sharp pulse coincident with the radio at . While the general slope of the faint -ray emission matches that of the UV, no sharp -ray pulse components are yet visible in this phase interval.
4.2. The PSF and a Search for Unpulsed Emission
Figure 1 makes it clear that the Vela extraction region is strongly dominated by the pulsed emission. However there are substantial phase windows where the pulsar is very faint. Figure 4 presents Fermi -ray images of the Vela pulsar in three energy bands in the LAT range separated into the on- () and off- () pulse phase regions. The images (except for the low count high energy off-pulse frame) are smoothed by Gaussians (with , and for the low to high energy images, respectively). The off-pulse images show the bright diagonal band of the Galactic diffuse emission, with appreciable structure. Comparison with the PSF contours (red) show that this is resolved in the vicinity of Vela at the higher energies. While a clump of extended emission surrounds the pulsar position on the few degree scale, it is at present not possible to associate this flux with known Vela PWN structures (e.g. the Vela-X TeV PWN, contours in the middle energy band).
We have subtracted these off- pulse images (scaled by to normalize exposure) from the on- pulse images and plotted (error bars) the profile of the resulting point source (bottom panels). The lines show simulations of the expected PSF from pre-launch calibrations (‘Pass6’), computed for a source with a Vela-like spectrum and off-axis angles similar to the actual observations. The agreement is quite good.
As a first attempt at constraining unpulsed (e.g. PWN) emission from Vela we have attempted to fit for a point source in the off-pulse phase window, fixed at the position of the pulsar. Using 0.1–10 GeV photons in an ROI in the phase interval, we derive a 95% CL upper limit on the flux of . After subtracting the estimated remnant pulsed flux in this window (0.73% of the phase-averaged flux) and scaling to the full pulse phase, this provides a limit on an unpulsed point source at the position of Vela of 2.8% of the MeV pulsed emission count-rate. More photons, especially at higher energy, will be required to search for a resolved PWN correlated with the TeV or radio structures.
4.3. Energy Spectrum
To study the phase-averaged spectrum of Vela, we used the standard maximum-likelihood spectral estimator ‘gtlike’ to be provided with the Fermi SSC science tools. This fits a source model to the data, along with models for the isotropic (instrumental and extragalactic) and structured Galactic backgrounds. The instrumental background was comparable to estimates from pre-launch simulation (Atwood et al. 2008). We used data from the observation spans indicated in Table 1, selecting photons with GeV within 15 of Vela. Our basic model for the spectrum of Vela is a simple power-law with an exponential cut-off.
With the large number of events collected for Vela, the statistical errors are very small. Systematic errors for the LAT are still under investigation (Abdo et al. 2009). Here we adopt conservative estimates of the systematic uncertainty in the LAT effective area, derived from the on-orbit estimation of the photon selection efficiency as function of energy and off-axis angle. This estimation has been made by comparing Vela on-pulse event selection to event selection in the off-phase bins. This allows an estimate of the efficiency for a pure photon sample, which is compared with our current simulation of the LAT response. We conducted this exercise for a range of energy and off-axis angle bins, and use the precision of the resulting effective area estimates and the difference from the simulated effective area as an estimate for the uncertainty in the LAT effective area as a function of incident photon energy. This varies from near 1 GeV to as much as 20% for energies below 0.1 GeV and 30% for energies greater than 10 GeV.
The gtlike fit for GeV is unbinned, and results in a spectrum of the form
with in GeV, and GeV. The first errors are the statistical values for the fit parameters, while the second errors are our propagated systematic uncertainties. The latter are particularly large for our spectral index parameter because of the large uncertainty at present in the effective area at high and low energies. To explore the stability of this result, we also fit using three other spectral estimators. We used a standard XSPEC analysis with our best model response matrices, a binned maximum likelihood estimator which computes the on-pulse photon counts in a point source weighted aperture in excess of off-pulse background counts (‘ptlike’) and a method which propagates the model spectrum through simulated instrument response to compare with observed pulsed source counts. For additional tests of the stability of the spectral fits, the data were also analyzed separately for ‘pointed’ and ‘survey’ observations, as well as for events in the front (thin radiator foil) and back (thick radiator foil) sections of the LAT detector. The fit parameters for the various data sets and analysis techniques, do have some statistically significant variation, but all were well within our presently-estimated systematic errors, as listed in Equation (1). We believe that as our understanding of the LAT performance improves we should be able to substantially decrease these systematic uncertainties.
Figure 5 presents the spectral power along with this best fit model. The binned spectral points are drawn from the ptlike analysis and show both the statistical error flags (bars with caps) and the inferred systematic error flags (bars without caps); the systematic errors dominate for all energies below 7 GeV. We also plot the EGRET data points of Kanbach et al. (1994). These are discrepant with the LAT points, particularly above a few GeV. Some studies have indicated that the EGRET response was incorrectly estimated (Baughman 2007, Stecker et al 2008).
We have also attempted to fit the data with a generalized cut-off model of the form Exp. We find so that models with a hyper-exponential behavior are well excluded. Taking into account the systematic errors the spectrum is fully consistent with the simple exponential cut-off. To check the exclusion of a given hyper-exponential model we compute the probability of incorrect rejection of a given value using the likelihood ratio test. For example, if only statistical errors are included is rejected at 16.5; inflating the errors to the level of our estimated systematic uncertainty leads to a small, but not negligible, 0.29% chance of incorrectly rejecting the model.
The analyses of Kanbach et al. (1980), Grenier et al. (1988) and Fierro et al. (1995) already showed that there is significant spectral variation through the pulse, with the bridge region harder than the peaks. Clearly, in Figure 2 we see that the situation is even more complicated, with three or more overlapping physical components contributing to the energy-dependent pulse profiles. Also, in Figure 5, one sees that there may be emission in the highest energy bins in excess of a simple exponentially cut-off model. This likely arises from the extended hard spectrum of the P2 component, and drives the best-fit value for below 1. Clearly the phase-averaged spectrum is a superposition of emission from a variety of distinct physical regions, and we cannot expect a simple cut-off power-law spectral model to be an exact representation of the data. Reduced systematic errors and the photons of a year or more of LAT data will be required to explore the phase-resolved spectral structure.
5.1. Light Curve
The early LAT data have already greatly clarified the -ray emission of the Vela pulsar and, even before detailed phase-resolved spectral measurements and model comparisons, significantly constrain the origin of high energy pulsar emission. Pulsar particle acceleration is believed to occur in the open zone, on the field lines above each magnetic pole which extend through the light cylinder. These subtend of pulsar phase at high altitude, so with a total phase in the narrow pulses of one would expect a comparable fraction of the open zone, and hence fraction of the spin-down power, to contribute to the -ray emission. The narrow peaks are often interpreted as caustics (Morini 1983; Romani & Yadigaroglu 1995, Dyks et al. 2004), where the combination of field retardation, aberration and time delay causes emission from the boundary of an acceleration zone (gap) to pile up in pulsar phase. A simple pulse from a 1-D fold caustic would increase to a maximum as as one approaches the destruction of an image pair at phase or fall off in mirror form after image pair creation. Vela’s P2 structure with a slow rise and abrupt fall looks like a trailing (image pair destruction) caustic, albeit with a steeper increase to the maximum, suggesting intensity variation along the emission surface or a higher-order catastrophe. The shape of P1 is less clear, but the apparent fast rise and slow fall would indicate a leading (image pair creation) caustic (Figure 1).
The broad structure of the P3 interpulse region suggests a beam of emission from a bundle of field lines bracketed by the two caustic-forming zones. The similarity between the -ray P3 phase structure and that of the lower frequency optical/UV emission at suggests a common origin – they both cover similar phase intervals and the peak shifts to later phase at higher energy. One appealing possibility is to associate the optical/UV P3 with synchrotron emission from relatively low altitude pair cascades produced by the few-GeV curvature radiation dominating the spectral energy distribution (SED). These e pairs, with electron energy , will produce synchrotron emission with characteristic peak energy keV as they radiate away their pitch angle radians in a magnetic field of G. For Vela, we have , so to produce eV photons the pairs would reside at , where is the neutron star radius. If the synchrotron emission is outward beamed, along with the pairs, then the Compton scattered flux would appear at GeV, matching the LAT P3 component. This suggests that the pulse of seed photons at IR to UV energies should have energy-dependent phase shifts tracking their Compton-up-scattered P3 progeny.
5.2. Luminosity and Beaming
While the VLBI parallax provides a well-determined distance of pc, the pattern of the -ray beam and hence the true -ray luminosity are uncertain. To compute the true luminosity
we need a correction factor from the observed phase-averaged energy flux F at the Earth line of sight (at angle to the rotation axis) to an average over the sky. For a given model, the correction depends on the geometry, so we have
where the model for a given magnetic inclination from the rotation axis gives , the pulse profile as a function of phase seen at viewing angle . is the expected phase-averaged flux for the actual Earth line-of-sight . In general highly-aligned models, such as the polar cap scenario, have small sky coverage and , while outer-magnetosphere models have much larger (Romani & Yadigaroglu 1995). For Vela, the viewing angle is estimated to be from the geometry of the Chandra X-ray Observatory-measured X-ray torus (Ng & Romani 2008). The magnetic impact angle is constrained by the radio polarization data. Together, these allow us to compute the high energy beam shape (assuming uniform emissivity along gap surfaces). For the outer gap (OG) picture we compute for . For the two-pole caustic (TPC) model appropriate to the slot gap picture, we find (Watters et al. 2008).
The observed energy flux obtained by integrating times the photon number flux in Equation (1) over the range 0.110 GeV is (7.87)10 ergs/cm/s. For a distance of 0.287 kpc, the -ray luminosity of Vela is then 7.8 10 ergs/s. Thus the observed -ray flux implies an efficiency of with for both the OG and TPC models. Note that this is a substantial fraction of the geometrical bound on the efficiency from the peak widths noted above.
5.3. Gamma-Ray Pulsar Models
Three general classes of models have been discussed for -ray pulsars. In polar cap models (Daugherty & Harding 1994, 1996; Sturner & Dermer 1995; Harding & Muslimov 2004), the particle acceleration and -ray production takes place in the open field line region within one stellar radius of the magnetic pole of the neutron star. In outer gap models (Cheng, Ho & Ruderman 1986; Romani 1996; Hirotani 2005), the interaction region lies in the outer magnetosphere in vacuum gaps associated with the last open field line. Other recent models hark back to the ‘slot-gap’ picture of Arons (1983), with the polar cap rim acceleration extending to many stellar radii (Harding & Muslimov 2004).
The appreciable offset from the radio peak, coupled with the presence of only one radio pulse, suggests that the -rays arise at high altitude. This is also supported by the lack of hyper-exponential absorption in the spectrum, which is a signature of -B pair attenuation of low altitude emission. The remaining two geometrical models can produce the general double peak profile of Vela (and many other -ray pulsars). In ‘Outer gap’ models emission starts near the ‘null charge’ surface at radius and extends toward the light cylinder . One magnetic pole dominates the emission in each hemisphere and the two peaks represent leading and trailing edges of the hollow cone of emission from this pole. If on the other hand emission extends well below the null charge surface toward the neutron star, both magnetic poles can contribute toward emission in a given hemisphere. This is the two-pole caustic model of Dyks & Rudak (2003) which might be realized in ‘slot gap’ acceleration models. In this case the leading ‘image pair creation’ caustic from high altitudes should not be visible, and the first -ray pulse represents the trailing caustic from emission at below the null-charge surface (the radio magnetic pole) while the second pulse represents a trailing caustic at higher altitudes from the opposite pole; both pulses should generally show a slow rise and steeper fall. Detailed -ray pulse profiles and, especially, phase resolved spectra can help distinguish these scenarios.
Another geometric difference in the two scenarios is the origin of the emission outside the main pulse. Low altitude field lines in the two-pole model tend to produce emission along all lines of sight. In the outer gap picture high altitude field lines can contribute faint emission extending through the off pulse region.
Further, as noted above, the energy spectrum also presents challenges for near-surface emission. In the polar cap models, a sharp turnover is expected in the few to 10 GeV energy range due to attenuation of the -ray flux in the magnetic field (Daugherty & Harding 1996). The spectral change at 2.9 GeV does not appear to fit this model. We conclude that low altitude radiation subject to -B pair production cannot account for the bulk of the Vela -ray emission. Indeed we can use the observed cut-off to estimate a minimum emission height as , where is the unabsorbed photon energy, is the spin period and the surface field is G (Baring 2004). Using Vela parameters ( s, ) we see that our cut-off implies that the bulk of the emission arises from . Since we see pulsed photons up to GeV, this emission must arise at . A similar conclusion has been recently made for the Crab pulsar by the MAGIC team, who observe pulsed GeV emission, using the imaging air Cerenkov technique (Aliu et al. 2008). Finally, it should be noted that while typical radio pulsar emission is inferred to arise at relatively low altitudes, for Vela-type pulsars recent radio models infer emission heights of as much as (Karastergiou & Johnston 2007). All of these factors point to a high-altitude origin of the Vela -ray emission.
The early LAT observations of Vela serve to show that the instrument is performing very well, with effective area comparable to expectations, good PSFs and excellent source localization, especially at high energies. The time-tagging of events is excellent and our pulse profile reconstruction is limited primarily by the accuracy of the radio-derived pulsar ephemeris. This presently has an accuracy of 100 s over our analyzed data span. The energy response and effective area calibration are currently being validated, but are at present known to % near 1 GeV, with uncertainties increasing to % at GeV and % at GeV.
These data also substantially improve our knowledge of the pulse properties of the Vela pulsar (PSR B083345) and are now placing important constraints on theoretical models. Already in these data we see:
1. The pulse profile is complex, with P1 and P2 dominated by very narrow components and substantial structure in the ‘bridge’ region.
2. Although the P1 and P2 phases are very stable across the -ray band, the the P1/P2 ratio decreases with energy. There is a distinct third peak in the ‘bridge’ component which sharpens and moves to later pulsar phase with increasing energy.
3. While faint emission appears within of the pulsar in the off-pulse phase, association if any with Vela is still unclear, due to the bright Galactic background in the vicinity. We thus quote an upper limit on the unpulsed point source flux, at 2.8% of the phase-averaged pulsed flux. A true PWN association will require additional data, allowing resolution at higher energies and a match to the radio and TeV images.
4. The phase-averaged -ray energy spectrum can be represented by a power law, with a exponential cut-off at GeV. The hyper-exponential cut-off index is not significantly different from the simple exponential value . Large values of , as expected for models radiating from the near-surface polar cap zone, are excluded.
During the first-year sky survey, we expect to collect approximately additional pulsed Vela photons. This will allow a substantial increase in sensitivity to narrow pulse components. We also expect improvements in our understanding of the instrument that will allow us to significantly reduce systematic errors. These two improvements will allow derivation of high accuracy spectral properties in more than 100 bins of pulsar phase. These are the observational results that should let us pin down the location and energy distribution of the radiating particles. In turn this should help us ‘reverse-engineer’ this -ray machine and should provide substantial insight into the physics of pulsar magnetospheres.
- Abdo et al. (2009) Abdo, A., et al. 2009, in prep.
- Aharonina et al. (2006) Aharonian, F., et al. 2006, A&A, 448, L43
- Aliu et al. (2008) Aliu, E., et al. 2008, Science Express, 10.1126
- Arons (1983) Arons, J. 1983, ApJ, 266, 215
- Atwood et al. (2008) Atwood, W. B., et al. 2008, ApJ, in prep.
- Baring (2004) Baring, M.G. 2004, Adv. Sp. Res., 33, 552
- Baughman (2007) Baughman, B. 2007, Ph.D. Thesis, USCS
- Caraveo et al. (2001) Caraveo, P. et al. 2001, ApJ, 561, 487
- Cheng, et al. (1986) Cheng, K. S., Ho, C. & Ruderman, M. 1986, ApJ, 300, 522
- Daugherty & Harding (1996) Daugherty, J. K., & Harding, A. K. 1996, A&AS, 120, 107
- Dodson et al. (2003) Dodson, R., et al. 2003, ApJ, 596, 1137
- Dyks & Rudak (2003) Dyks, J., & Rudak, B. 2003, ApJ, 598, 1201
- Dyks et al. (2004) Dyks, J., et al. 2004, ApJ, 607, 939
- Edwards et al. (2006) Edwards, R.T., et al. 2006, MNRAS, 372, 1549
- Fierro (1995) Fierro, J. M. 1995, Ph. D. Thesis, Stanford U.
- Fierro et al. (1998) Fierro, J. M., et al. 1998, ApJ, 494, 734
- Frail et al. (1997) Frail, D.A., et al. 1997, ApJ, 475, 224
- (Grenier et al. 1988) Grenier, I. A., Hermsen, W., & Clear, J. 1988, A&A, 204, 117
- Harding et al. (2002) Harding, A. K., et al. 2002, ApJ, 576, 376
- Helfand et al. (2001) Helfand, D.J., Gotthelf, E.V. & Halpern, J.P. 2001, ApJ, 556, 380
- Hirotani (2005) Hirotani, K. 2005, Ap&SS, 297, 81
- Hobbs et al. (2006) Hobbs, G. B., Edwards, R. T., & Manchester, R. N. 2006, MNRAS, 369, 655
- Johnston et al. (2008) Johnston, S., et al. 2008, ApJ, in prep.
- Kanbach et al. (1980) Kanbach, G. et al., 1980, A&A, 90, 163
- Kanbach et al. (1994) Kanbach, G. et al., 1994, A&A, 289, 855
- Karastergiou & Johnston (2007) Karastergiou, A. & Johnston, S. 2007, MNRaS, 380, 1678
- Large, Vaughan & Mills (1968) Large, M.I., Vaughan, A.E. & Mills, B.Y. 1968, Nature, 220 340.
- Manzali et al. (2008) Manzali, De Luca, A., Caraveo, P. A. 2008, ApJ, in prep.
- Morini (1983) Morini, M. 1983, MNRAS, 202, 495
- Muslimov & Harding (2004) Muslimov, A. G., & Harding, A. K. 2004, ApJ, 606, 1143
- Ng & Romani (2008) Ng, C.-Y., & Romani, R. W. 2008, A. K. 2004, ApJ, 606, 1143
- Pavlov et al. (2003) Pavlov, G.G., et al. 2003, ApJ, 591, 1157
- Pellizoni et al. (2008) Pellizoni, A., et al. 2008, ApJ, in press
- Romani & Yadigaroglu (1995) Romani, R. W. & Yadigaroglu, I.-A. 1995, ApJ, 438, 314
- Romani (1996) Romani, R. W. 1996, ApJ, 470, 469
- Romani et al. (2005) Romani, R. W., Kargaltsev, O., & Pavlov, G. G. 2005, ApJ, 627, 383
- Smith et al. (2008) Smith, D.A. et al. 2008. A&A, submitted; arXiv:0810.1637
- Stecker et al. (2008) Stecker, F.W., Hunter, S.D. & Kniffen, D.A. 2008, Astropart. Phys. 29, 25
- Sturner & Dermer (1995) Sturner, S. J., & Dermer, C. D. 1995, Adv. Sp. Res., 15, 77
- Thompson et al. (1975) Thompson, D. J. et al. 1975, ApJ, 200, 79L
- Thompson (2001) Thompson, D. J. 2001, in ”High Energy Gamma Ray Astronomy” ed. F. A. Aharonian, H. J. Volk, AIP Conf. Proc. 558 (Melville, NY), p. 103
- Watters et al. (2008) Watters, K. et al. 2008, ApJ, submitted
Data table for Electronic version.
|Energy (GeV)||Differential flux F (erg cms||F||F||F111F is the systematic error on the flux and F are the statistical errors on the flux, while F is the total error on the flux|