CORRELATION OF SUPERNOVA REMNANT MASERS AND GAMMA-RAY SOURCES

It has long been speculated that the vast energy of supernova remnants (SNRs) is tapped to produce the bulk of Galactic cosmic rays. Those SNRs interacting with molecular clouds are excellent targets for detecting evidence of cosmic ray acceleration. The dense gas of the nearby molecular cloud acts as a target for hadronic cosmic rays, producing γ-rays via the decay of neutral pions (e.g., Drury et al. 1994).

Interest in cosmic rays from interacting SNRs has been reinvigorated by recent detections of γ-ray emission at GeV to multi-TeV energies from SNRs. A galactic plane survey detected 14 new TeV sources, of which 7 are coincident with SNRs (Aharonian et al. 2006). Targeted observations are now capable of reaching sufficient sensitivities to resolve the morphology of γ-ray emission associated with the remnant (Aharonian et al. 2008a, 2008b). In particular, a TeV-energy counterpart to IC 443 was observed to be marginally spatially offset from the GeV-energy γ-ray source (Albert et al. 2007; Acciari et al. 2009). If confirmed, spatial variations with energy may indicate the diffusion of cosmic rays accelerated by the SNR out into the adjacent molecular cloud (Torres et al. 2008).

Interacting SNRs represent a promising class of γ-ray sources which are likely to be uncovered in increasing numbers by the current generation of γ-ray observatories. However, to date, the identification of γ-ray counterparts to SNRs has been inhibited by the poor spatial resolution and sensitivity of γ-ray telescopes (Esposito et al. 1996; Torres et al. 2003). Often, the association of γ-ray sources with interacting SNRs rests on supplemental evidence from common tracers of interaction with dense gas: broad molecular lines, bright at infrared to millimeter wavelengths (e.g., Reach et al. 2005), and the hydroxyl (OH) maser at 1720 MHz (e.g., Frail et al. 1994).

Hydroxyl masers, when detected only at 1720 MHz, are a particularly powerful diagnostic associated only with SNR shocks. These "SNR masers" unambiguously trace cool (25–150 K), dense (∼10⁵ cm⁻³) gas in the wake of non-dissociative shocks and permit a direct measure of magnetic field strength (Lockett et al. 1999). However, SNR masers are somewhat rare, having been detected in only ∼10% of remnants. Prominent interacting SNRs W28, W44, and IC 443 have been noted for both bright masers and coincident EGRET sources, leading to the suggestion that masers may trace cosmic-ray acceleration sites (Claussen et al. 1997).

In this Letter we address the nature of interacting SNRs with γ-ray counterparts: first, we show that an excellent correlation exists between SNR masers and a subset of GeV- and TeV-energy sources toward SNRs. This strengthens the argument for a hadronic origin for the observed γ-rays. Second, if neutral pion decay is indeed the dominant emission mechanism, the implied cosmic ray enhancement is sufficient to explain the increased ionization needed to produce high columns of OH in the post-shock gas (Wardle 1999). This suggests there is a more intimate relationship between SNR masers and γ-ray sources than merely being complimentary tracers of dense gas interaction.

Here we assess potential γ-ray counterparts which may be directly associated with SNRs (hereafter, "γ-ray SNRs). It is not trivial to determine the origins of each γ-ray source. SNR shock waves are often coincident with pulsars or pulsar wind nebulae, which are also capable of producing high-energy γ-ray emission (Bartko & Bednarek 2008).

Potential γ-ray SNRs are drawn from three catalogs: EGRET sources (Torres et al. 2003), the Fermi Bright Source List (Abdo et al. 2009), and the TeVCat.⁴ Some sources reported by Torres et al. (2003) as potential EGRET counterparts to SNRs have since been associated with other astrophysical sources: 3EG J1410-6147, coincident with SNR G312.4-0.4, has recently been associated with the young pulsar J1410-6132 based on detected variability (O'Brien et al. 2008). Source 3EG J1824-1514 has been confirmed as the microquasar LS 5039 (Aharonian et al. 2005), ruling out an association with SNR G16.8-1. Sources 3EG J2016+3657 and 3EG J2020+4017, coincident with SNRs G74.9+1.2 and G78.2+2.1 respectively, are extragalactic blazars (Iyudin et al. 2007). Finally, the EGRET counterparts for SNRs G180.0-1.7, G355.6+0.0, G39.2-0.3, and G74.8+1.2 do not appear in the Fermi Bright Source List, and we therefore do not include them as detections in our analysis.

Of the 26 remnants for which an association with the coincident γ-ray source has not been ruled out, 7 are young remnants (≲1 kyr), 12 are SNRs with evidence of interaction with dense clouds, and 7 are unclassified remnants. A few unclassified SNRs, such as G353.6-0.7, may be interacting with dense gas, but the evidence is not yet conclusive (Tian & Leahy 2008; Acero et al. 2009). Of the 12 remnants in our subset known to be interacting with dense gas, all but 2 (MSH 11-61A and W49B) have detected SNR masers. The presence of masers gives several advantages: (1) masers signpost interaction with dense (10⁵ cm⁻³) clouds, which will enhance the γ-ray luminosity from pion decay, (2) the velocity of the maser gives a kinematic distance, and (3) an established velocity allows the adjacent cloud to be isolated from confusing Galactic emission along the line of sight.

Table 1 lists all SNRs with masers and the properties of potentially associated γ-ray emission. In the first five columns the Galactic coordinates, name, diameter and kinematic distance are listed for each remnant. The γ-ray luminosity for all derived from the reported ∼100 MeV–10 GeV and ≳1 TeV fluxes in Columns 6 and 8. Spectral indices for are given in Columns 7 and 9. References for detections are given in the last column.

The 10 SNRs with coincident γ-ray sources are divided based on the certainty of their association: group A includes four SNRs for which γ-rays are established as related to the SNR; group B includes six SNRs with coincident γ-ray sources that have not been attributed to other astrophysical phenomenon (pulsars, blazars, etc.) but for which an association with the SNR is less than certain; group C lists the 16 SNRs with masers that do not yet have detected γ-ray sources. Given that both masers and γ-rays are detected for only 10% of SNRs, the large number of coincident detections makes an association between SNR masers and γ-ray emission as tracers of interaction quite plausible.

2.1. Contingency Table Analysis

2.2. Properties of γ-ray SNRs with Masers

For SNRs interacting with dense gas, γ-ray counterparts are likely to be dominated by emission from neutral pion decay originating from interactions between hadronic cosmic rays and gas nuclei. In contrast to young SNRs, older interacting SNRs have little, if any, detected X-ray synchrotron emission. An inverse-Compton scenario requires small magnetic field strengths <1 μG (Yamazaki et al. 2006) whereas Zeeman splitting of SNR masers measures field strengths of order ∼1 mG (Yusef-Zadeh et al. 1996; Claussen et al. 1997; Brogan et al. 2000). Electronic bremsstrahlung emission has also been proposed, but the expectation would be for γ-ray emission to trace the radio shell morphology (Bykov et al. 2000). TeV emission from W28 and IC 443 shows no correlation with the radio shell, and an excellent correlation with dense gas (Aharonian et al. 2008b; Acciari et al. 2009). Forthcoming Fermi observations will have sufficient resolution to resolve this question at GeV energies.

The presence of a large reservoir of dense gas around the SNR, the expectation that cosmic rays are accelerated by SNRs, and the relatively older ages of this subset of interacting SNRs support a hadronic origin for γ-ray counterparts. The morphology of γ-ray emission is seen to be well matched to that of the molecular cloud. In Figure 1, histograms of γ-ray luminosity show medians of 1.7 × 10³⁶ erg s⁻¹ and 1.5 × 10³³ erg s⁻¹ for GeV and TeV detections, respectively. This is consistent with luminosity estimates if a significant fraction of the SN energy (10%–20%) is diverted to cosmic-rays which are incident on the adjacent molecular cloud (Drury et al. 1994). The orders of magnitude differences in the luminosity reflect the expected energy spectrum of accelerated particles.

Zoom In Zoom Out Reset image size

On average, a hardening of the photon spectrum from GeV to TeV energies is also observed. The γ-ray spectral index, given in Columns 7 and 9 of Table 1, steepen at higher energies from a mean of α_GeV∼ 2.1 to α_TeV∼ 2.6. It has been suggested for nearby SNR IC 443 that this spectral steepening results either from the presence of both pion decay and a strong bremsstrahlung component (Bykov et al. 2000), or the diffusion of cosmic rays at different energies through the dense cloud (Torres et al. 2008). In addition to IC 443, the SNRs W28, Sgr A East, and G359.1-0.5 show this characteristic spectral hardening at higher energies, but CTB 37A does not. Future detailed spectral modeling of these sources can determine whether a spectral signature can be used to discriminate between different γ-ray emission mechanisms and accelerated cosmic ray populations.

The strong correlation observed between γ-ray-bright and maser SNRs may result from more than just both tracing interaction with high gas densities. The large post-shock OH abundances required for masing to occur are not produced in chemical models of slow, dense shocks; instead all gas-phase oxygen is rapidly converted to water before the gas cools to 50 K (Kaufman & Neufeld 1996).

The large observed OH column densities are thought to be produced by dissociation of the abundant post-shock water (Wardle 1999). In this model energetic electrons produced by ionizations in the molecular gas will excite H₂, which will subsequently generate a weak flux of far-ultraviolet photons which photo-dissociate water molecules producing OH (Prasad & Tarafdar 1983). The high flux of X-rays from the hot interior characteristic of mixed morphology SNRs has been shown to be a viable ionization mechanism (Yusef-Zadeh et al. 2003). Cosmic rays and X-rays are notionally lumped together in this model.

The γ-ray luminosity can be used to discriminate the relative importance of ionizations by cosmic rays. The γ-ray spectrum resulting from pion decay of hadronic cosmic ray interactions in SNRs has been described by many authors (Drury et al. 1994; Baring et al. 1999; Bykov et al. 2000). Here we follow Aharonian (1991), formulating the γ-ray flux as a function of the cosmic ray enhancement in the cloud adjacent to SNR, neglecting any possible gradients through the cloud as a simplification. The γ-ray flux above energy E_γ is given by the equation

$$\begin{eqnarray} &&{\rm F(\ge E_\gamma) \simeq 3\times 10^{-13} \left(\frac{E_\gamma }{\rm TeV}\right)^{-1.6} \omega _{CR} \left(\frac{M_{5}}{d_{\rm kpc}^2} \right) ph\ cm^{-2}\ s^{-1} },\nonumber\\ \end{eqnarray} \tag{ 1 }$$

where d_kpc is the distance to the SNR in kpc, E_γ is the threshold above which the γ-ray flux is totaled, ω_CR is the ratio of cosmic ray density at the SNR to the local solar value assuming the same γ-ray emissivity, and M₅ is the total mass of the SNR shell and adjacent molecular cloud in units of 10⁵ M_☉. We note that Drury et al. (1994) established that spectral variations between 2.0 and 2.7 changes this estimate by less than 20%.

Assuming the observed γ-ray flux from the SNR is due to hadronic interactions with the surrounding dense gas, it is straightforward to roughly estimate the local energy density of cosmic rays, provided the distance and molecular environment are known. For EGRET and Fermi detections, E_γ > 100 MeV, Equation (1) can be rewritten to express the local cosmic ray enhancement ω_CR:

$$\begin{equation} \omega _{\rm{CR}} \simeq \ \frac{F_{\gamma }(> 100 \ {\rm MeV})}{7 \times \ 10^{-7} {\rm \ ph \ cm^{-2}\ s^{-1}}} \ \left(\frac{d_{\rm kpc}^2}{M_5}\right). \end{equation} \tag{ 2 }$$

The cosmic ray ionization scales as the cosmic ray density, so we obtain the ionization rate due to cosmic rays ζ_CR ≃  ω_CR ζ_☉, where ζ_☉ is the local cosmic ray ionization rate of ∼4×10⁻¹⁷ s⁻¹ (Webber 1998).

In Table 3 we list the observed properties of SNRs from which the local enhancement of cosmic rays is estimated. We find cosmic ray densities are typically increased by 30–150 times that observed for quiescent molecular clouds. This provides an ionization rate sufficient to produce OH abundances of ∼10⁻⁷–10⁻⁶ in the post-shock gas. For comparison we also list the ionization rate from interior X-ray emission derived by Yusef-Zadeh et al. (2003). The two sources of ionization are generally found to be comparable. Either could be the dominant ionization mechanism depending on the location of the gas with respect to the interior X-ray emitting plasma and cosmic ray acceleration sites, and may vary on a source-by-source basis.

We caution that our estimates of cosmic ray ionization rate are somewhat crude. Even if the dominant emission mechanism is hadronic, there may still be a significant bremsstrahlung component. Detailed spectral modeling with data from the Fermi γ-ray Space Telescope will permit a much better estimate of the local cosmic ray density. Observations of direct chemical tracers of cosmic ray ionization rates, such as H⁺₃ and He⁺ (Oka 2006; Dalgarno 2006; Indriolo et al. 2009), could be used to confirm the cosmic ray enhancements in the immediate environment of these interacting SNRs.

We have explored the class of γ-ray sources that can be explained by enhanced cosmic ray densities at the interaction sites between SNRs and molecular clouds. By correlating SNR masers with known GeV and TeV sources, we have identified an emerging class of γ-ray-bright interacting SNRs. Of the 24 known maser SNRs currently identified there are 10 with GeV- to TeV-energy γ-ray associations, and 6 with both. If γ-ray emission from these sources is largely due to hadronic cosmic rays, the enhanced local cosmic ray ionization rates in these clouds can explain the production of OH molecules behind a C-type shock, suggested by Wardle (1999). Furthermore, cosmic ray ionization is typically comparable to or dominant over ionizations from interior thermal X-rays, though without detailed knowledge of the cosmic ray spectrum these results have large uncertainties. Interacting SNRs represent a growing class of γ-ray sources likely to be uncovered by future γ-ray observations.