HARD X-RAY FLUX UPPER LIMITS OF CENTRAL COMPACT OBJECTS IN SUPERNOVA REMNANTS

X-ray observations have revealed groups of radio-quiet young (isolated) neutron stars: anomalous X-ray pulsars (AXPs), soft gamma repeaters (SGRs), X-ray dim neutron stars (XDNSs), and central compact objects (CCOs). AXPs are known to emit pulsed X-rays below 10 keV with periods in the range ∼5–12 s (see Kaspi & Gavriil 2004; Mereghetti 1999, for a review). Their rate of rotational kinetic energy loss is well below the soft X-ray luminosity. They are believed to have exceptionally strong (≳10¹⁴ G) surface magnetic field strengths, and are powered by magnetic field decay, hence they are called magnetars (Thompson et al. 2002; Güver et al. 2008).

Another model attributes the soft X-ray emission and spin-down of persistent AXPs and SGRs and their newly discovered transient members to accretion torques from active fallback disks (Alpar 2001; Chatterjee et al. 2000; Ertan et al. 2007a; Ertan & Erkut 2008). During the supernova explosion, fallback of material from the progenitor may establish an accretion disk around the neutron star. Indeed a debris disk has been found around a young isolated neutron star, 4U 0142+61 (Wang et al. 2006, but also see Wang et al. 2007), whose properties explain the overall broadband data of the source (Ertan et al. 2007b).

Hard X-ray emission has been discovered from four AXPs between 20 keV and 300 keV (Molkov et al. 2004; Kuiper et al. 2004, 2006; den Hartog et al. 2004; Revnivtsev et al. 2004). The current models used to explain this hard X-ray emission often invoke magnetar fields (Heyl & Hernquist 2005; Thompson & Beloborodov 2005; Baring & Harding 2007). Discovery of high-energy emission from AXPs, which were formerly known to have only soft emission, prompted us to search for high-energy emission from similar isolated neutron stars, especially those that may also have high magnetic fields.

One such group of young isolated neutron stars that have circumstantial evidence for having high magnetic fields is the CCOs. These are X-ray point sources which are found near the center of supernova remnants (SNRs), show no radio counterpart, no pulsar wind nebula, and have thermal-like soft spectra. The soft X-ray spectra usually consist of thermal components with blackbody temperatures of 0.2–0.5 keV, X-ray luminosities L_x ∼ 10³³–10³⁴ erg s⁻¹, and characteristic sizes of 0.3–3 km (see Pavlov et al. 2004, for a review).

Some properties of these sources have been interpreted as a result of high magnetic fields in the past. For example, if the line features in 1E 1207.4−5209 are due to proton cyclotron absorption, the magnetic field of this source must be greater than 10¹⁴ G (Bignami et al. 2003). The same line features were also interpreted as fine structure splitting of He⁺ under magnetar fields (Pavlov & Bezchastnov 2005). Krause et al. (2005) observed the SNR Cassiopeia A (Cas A) with the Spitzer Space Telescope and detected infrared echoes. One possible explanation for those features is magnetar-type explosions and flares from the compact object J232327.9+584843 near the center of the SNR. Li (2007) studied 1E 161348−5055 in SNR RCW 103 and suggested that it is either an isolated magnetar or a low-mass X-ray binary (LMXB) that survived the supernova event with a fallback disk.

On the other hand, there is also evidence for CCOs having normal/weak magnetic fields. Gotthelf & Halpern (2008) reported that measured spin-down rate of 1E 1207.4−5209 implies a magnetic field strength of B_p < 3.3 ×10¹¹ G. The X-ray luminosity exceeds its rotational energy-loss rate. This indicates that luminosity arises from residual cooling and perhaps partly from accretion of supernova debris. Gotthelf & Halpern (2008) also suggested that the neutron star J232327.9+584843 may have undergone a one-time phase transition which powered the light echo.

Most of the properties of the CCOs can be understood if they are propellers (Alpar 2001). According to evolutionary scenarios with fallback disks, their periods are expected to be P < 1 s corresponding to high mass inflow rate. These periods that are smaller than those of AXPs suggest that they may be predecessors of AXPs (Alpar 2001). Although periods of most CCOs have not been observed yet, measured periods of two CCOs are consistent with this interpretation (period of the CCO 1E 1207.4−5209 is about 424 ms (Zavlin et al. 2000) and period of the CCO J185238.6+004020 is 105 ms, Gotthelf et al. 2005). The period measurement, P = 7.5 s for the source J160103.1−513353 (Park et al. 2006), has not been confirmed yet.

We present the hard X-ray flux measurements of all CCOs and CCO candidates that have distances less than 10 kpc for an unbiased analysis (1E 1207.4−5209, 1WGA J1713−3949, J0002−6246, J082157.5−430017, J085201.4− 461753, J160103.1−513353, J161348−5055, J181852.0− 150213, J185238.6+004020, J232327.9+584843) on the International Gamma-Ray Astrophysics Laboratory (INTEGRAL) with the INTEGRAL Soft Gamma-Ray Imager (ISGRI) detector. This leaves out only J171801.0−372617 which has a distance of 22 kpc. Except Cas A, the contribution of the SNR to hard X-rays is negligible for all sources. SNRs G296.5+10.0, G347.3−0.5, Puppis A (G260.4−3.4) and Vela Junior (G266.1−1.2) are not only large compared to the pixel size of ISGRI, but they are also soft. Similarly, X-ray spectra of Kes 79 (G33.6+0.1) (Sun et al. 2002) and RCW 103 (G332.4−0.4) (Gotthelf et al. 1999) are so soft that their contribution to hard X-rays would be negligible compared to our upper limits. As an example, G330.2+1.0, according to the srcut model parameters reported in Torii et al. (2006), can at most contribute 1% of the upper-limit flux we derived for the source. In this work, we aim to understand the nature of the emission for CCOs, and by comparing the hard X-ray properties to that of AXPs we want to investigate the connections between different types of isolated neutron stars.

One of the best investigated CCOs is 1E 1207.4−5209 in the SNR G296.5+10.0. It was discovered by Helfand & Becker (1984). It is located about 6' off the center of G296.5+10, at a distance of about 2 kpc. The soft X-ray energy spectrum is best modeled with a continuum blackbody component with temperature kT∼ 0.14 keV, and at least two broad absorption lines centered at 0.7 keV and 1.4 keV. Zavlin et al. (2000) observed 1E 1207.4−5209 with the Chandra X-ray Observatory and discovered a period of about 424 ms. The second Chandra observation provided an estimate of the period derivative, $\dot{P}$ ∼ (0.7–3) × 10⁻¹⁴ s s⁻¹ (Pavlov et al. 2002). The corresponding characteristic age of the pulsar, P/2$\dot{P}$ ∼ 200–900 kyr, is much larger than the estimated age of the SNR, ∼7 kyr. It has been proposed that this discrepancy is due to glitches. Alternatively, P/2$\dot{P}$ may be irrelevant as an age estimate because the $\dot{P}$ is due to a fallback disk or a low-mass binary companion. Using the XMM-Newton data, Woods et al. (2006) claimed that the observed spin irregularities originated from the presence of a binary companion to 1E 1207.4−5209.

RX J1713−3946 is a shell-type SNR, embedded in the galactic plane. The remnant was discovered in X-rays with ROSAT, which also identified two point sources within the boundaries of the SNR shell (Pfeffermann & Aschenbach 1996). 1WGA J1713.4−3949 is located at the geometrical center of the SNR and no optical counterpart has been found within 10'' of the ROSAT position (Slane et al. 1999). Chandra, XMM-Newton, and RXTE observations showed that the X-ray spectrum of the source is well fitted by the sum of a blackbody component with a temperature of ∼0.4 keV plus a power-law component with a photon index of ∼4 (Lazendic et al. 2003).

The source J082157.5−430017, located about 6' off the center of SNR the Puppis A, was discovered with Einstein by Petre et al. (1982). Hui & Becker (2006) presented spectral analysis results using XMM-Newton and Chandra data. The observed point source is best described with a double blackbody model with temperatures T₁ = (2.35–2.91) × 10⁶ K, T₂ = (4.84–5.3) × 10⁶ K.

Pavlov et al. (2001a) found a relatively bright source (J085201.4−461753) in the SNR Vela Junior (G266.1−1.2) with a flux of ∼2 ×10⁻¹² erg s⁻¹ cm⁻² in the 0.5–10 keV energy band using Chandra data. The spectrum of the source is best fitted with a blackbody model with a temperature of kT = 404 ± 5 eV and radius of the emitting region R = 0.28 ± 0.01 km at a distance of 1 kpc. Becker et al. (2006) did not find any pulsations to a 3σ upper limit for any pulsed fraction with XMM-Newton.

Park et al. (2006) discovered a pointlike source J160103.1− 513353 at the center of Galactic SNR G330.2+1.0 with Chandra X-ray Observatory. The X-ray spectrum is fitted with a blackbody model with kT ∼ 0.49 keV, implying a small emission region R ∼ 0.4 km at distance of 5 kpc and estimated X-ray luminosity is L_x ∼ 1 × 10³³ erg s⁻¹ in the 1–10 keV energy band. They found a possible X-ray period of the source P ≈ 7.5 s.

The first discovered radio-quiet X-ray point source, J161348−5055, was found in the young SNR RCW 103 by Tuohy & Garmire (1980). A period of 6 hr was found with both Chandra and ASCA (Garmire et al. 2000) and the X-ray observation has shown that its flux varies through a wide range. Sanwal et al. (2002) have found a 6.4 hr period and multiple dips in the X-ray light curve using Chandra ACIS data. Becker & Aschenbach (2002) reported the evidence of an eclipse from a light curve observed by XMM-Newton. All of these results suggest that J161348−5055 is an accreting object in a binary system. On the other hand, De Luca et al. (2006) suggest that J161348−5055 may be a magnetar rotating at 6.67 hr with B > 10¹⁵ G. Even with such magnetic field, a neutron star cannot spin down to 6.7 hr via magnetic dipole radiation during the lifetime of the SNR. A supernova fallback disk is proposed to interact with the magnetar, providing additional propeller spin-down torques (Li 2007).

Lazendic et al. (2005) found an X-ray point source J171801.0−372617 inside SNR G349.7+0.2 with Chandra. The SNR distance is 22 kpc and estimated to be 4000 years old. The source's flux is F_X(0.5–10 keV) ∼ (0.4–4.1) × 10⁻¹³ erg cm⁻² s⁻¹ and its luminosity L_X ∼ (0.2–2.3) × 10³⁴ erg s⁻¹. Although its luminosity is close to the CCO luminosity we did not add this source to our analysis because of its distance.

Reynolds et al. (2006) discovered an X-ray point source J181852.0−150213 within SNR G15.9+0.2 with Chandra. The X-ray spectrum of the source is consistent with either a steep Γ∼ 4 power law or a kT ∼ 0.4 keV blackbody and the neutral hydrogen column density is N_H∼ 4 × 10²² cm⁻².

A pointlike source J185238.6+004020 was found at the center of SNR Kes 79 (G33.6+0.1) by Seward et al. (2003) with Chandra ACIS. Gotthelf et al. (2005) discovered 105 ms X-ray pulsations from the source using XMM-Newton data, and using two observations of the pulsar, 6 days apart, they calculated $\dot{P}$ < 7 × 10⁻¹⁴ s s⁻¹. The X-ray spectrum of the source is well fitted with a blackbody model with temperature kT = 0.44 ± 0.03 keV, radius R ≈ 0.9 km, and L_bol = 3.7 ×10³³ erg s⁻¹ assuming d = 7.1 kpc (Gotthelf et al. 2005).

The prototype of CCO, J232327.9+584843, was discovered near the center of SNR Cass A in the first-light Chandra observation by Tananbaum (1999), and found later in the archival ROSAT and Einstein images. A blackbody fit to the observed Chandra ACIS spectrum shows a high temperature kT ≈ 0.5 keV and a small effective radius R≈ 0.4 km, at the remnant's estimated distance of 3.4 kpc (Reed et al. 1995), leading to suggestions of a neutron star with hot spots (Pavlov et al. 2001b). The properties of these CCOs are shown in Table 1.

The ISGRI is the low-energy camera of the IBIS telescope (Lebrun 2003) on INTEGRAL and is sensitive between 20 keV and 1 MeV. Its detector area is 2621 cm² made up of 16,384 CdTe pixels. The fully coded field of view is 9°, and the angular resolution is ∼13' (Gros et al. 2003).

The INTEGRAL IBIS ISGRI imaging analysis (OSA 5.1 and OSA 7.0) was applied in our study. Table 2 shows the observations that are used in this study for imaging and spectral analysis. "Begin Date" and "End Date" show the observation date of the first and last revolutions, respectively. We produced images in two energy bands (20–75 keV and 75–300 keV) for each source while restricting our science windows within 10° radius from the source. For eliminating images with poor quality, we automatically searched for science windows with large variations in the variance maps. Then we manually investigated flagged science windows and eliminated poor images. After that we combined all clean science windows producing single sky mosaics for each source in these two energy bands. Comparing significance maps with flux maps in these sky mosaics, we obtained 3σ flux upper limits for each source. For calculating upper-limit fluxes, we assumed a photon index of 1, based on the ISGRI hard X-ray spectra of AXPs.

Except for a point source in the Cas A region in the 20–75 keV range, none of the CCOs have been significantly detected with ISGRI despite large exposure times. For Cas A, we probably detected the young SNR. This SNR has already been detected between 40 keV and 120 keV energy range with Compton Gamma Ray Observatory (CGRO)/OSSE (The et al. 1996), with HEXTE on RXTE (Rothschild & Lingenfelter 2003), and with ISGRI (Renaud et al. 2006). With ISGRI's imaging capabilities, we cannot resolve the compact object from the SNR and therefore cannot provide reliable flux limits for this source in the 25–75 keV band. Hard X-ray flux upper limits of all these CCOs are shown in Table 3.

To compare the soft X-ray to hard X-ray emission, we took model parameters of the eight sources from the literature (Table 4) and simulated the soft spectra using XSPEC11. Then, we combine these with flux upper limits. Two samples of the overall spectra are shown in Figures 1 and 2. We chose the nearest and the farthest CCOs as a sample for comparing.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also compared CCO upper-limit fluxes with the measured AXP high-energy fluxes. To achieve a valid comparison, we normalized all AXP distances to those of CCOs and recalculated all flux values in the 20–75 keV and 75–300 keV energy band (we assumed a Crab spectrum of power-law photon index Γ = 2.15 and normalization N= 10.4). We also calculated the hard X-ray luminosities of AXPs to compare with the CCOs in Table 5.

There are a few models to explain hard X-ray emission from AXPs and SGRs, all invoking magnetar fields. One such mechanism is the fast-mode breakdown under very strong magnetic fields (Heyl & Hernquist 2005). An alternative model is proposed by Thompson & Beloborodov (2005), again invoking an energy release by ultra strong magnetic fields. In this case, there may be a heating of the surface layer through Langmuir turbulence producing high-energy bremsstrahlung radiation. Another possibility for the source of high-energy emission is the acceleration of positrons under strong electric fields ∼100 km from the star, and subsequent pair production and synchrotron emission. These models, while invoking high magnetic fields, also assume gaps that would create electric fields accelerating charged particles, and producing gamma-ray emission. Finally, according to Baring & Harding (2007), for a range of magnetic colatitudes proximate to the neutron star surface, resonant Compton scattering in strong magnetic fields can be extremely efficient for powering the AXP in hard X-rays.

In this work, we have shown that the nearby (distance < 4 kpc) CCOs have hard X-ray luminosities which are at least an order of magnitude lower than the hard X-ray luminosities of AXPs. If hard X-ray emissions from the isolated neutron stars occur only under strong magnetic fields as the current models claim, our work may indicate low magnetic fields for CCOs. Note that AXPs have comparable luminosities in soft and hard X-rays. If the hard X-ray emission is powered by a mechanism that also powers soft X-ray emission, we should have detected hard X-rays from nearby CCOs.

According to Alpar (2001), the similarity of rotation periods of the AXPs, SGRs, and the XDNSs suggests a common mechanism with an asymptotic spin-down phase. This model proposes that AXPs, SGRs, XDNSs, and CCOs constitute alternative subclasses of young neutron stars. The differences between these are due to mass inflow rates and possibly also due to differences in the dipole component of the magnetic field, independently of whether the total field at the surface is at magnetar values. The pulse periods may not be observable if the optical thickness to electron scattering of the material surrounding the star destroys the X-ray beaming (but also see Göğüş et al. 2007 in context of LMXBs), or if in the strong mass inflow regime destroys the modulation of the accretion and the accretion luminosity over the star's surface. This scenario may also explain the lack of hard X-ray emission from the CCOs, as strong mass accretion rate would short out any gaps that could provide acceleration of particles in the magnetosphere.

This work is a part of the PhD thesis titled "High-Energy Analysis of CCOs in SNRs", prosecuted by Istanbul University, Institute of Science, and was supported by the research funds of the Istanbul University, project number T-84/15122006. This work was also supported by the European Commission 6th Framework Projects INDAM (International Reintegration Grant, MIRG-CT-2005-017203) and ASTRONS (Transfer of Knowledge Project, MTKD-2006-42722). E.K. is partially supported by TÜBİTAK. E.K. is also supported by the Turkish Academy of Sciences (TÜBA) with a Young and Successful Scientist Award. Authors thank Steve Reynolds and Hakan Erkut for valuable discussions.