SNR G349.7+0.2: A γ-RAY SOURCE IN THE FAR 3 kpc ARM OF THE GALACTIC CENTER

Methods for extracting H i absorption spectra have been introduced in earlier papers (Tianet al. 2007; Leahy & Tian 2008; Leahy & Tian 2010). The H i emission spectrum of SNR G349.7+0.2 is shown in the upper panel of Figure 1. In this spectrum, there are several H i emission peaks from −113 km s⁻¹ to 35 km s⁻¹. The H i absorption and CO emission spectra are shown in the lower panel of Figure 1. SNR G349.7+0.2 is a strong radio source, which results in a strong H i absorption signal. An estimate of the error in exp(−τ(v)) of Figure 1 is the rms for velocity channels with no emission (>+50 km s⁻¹), which is 0.0063. Artifacts in the continuum image cause another type of error. We have analyzed absorption spectra both with the same source region and with different backgrounds. We find that the artifacts only affect the normalization and not the shape of the H i absorption spectra. We find that the error in normalization is about 20% in depth of the H i absorption. We find that each CO emission peak has an associated H i absorption feature except for the 16.5 km s⁻¹ CO peak. The highest positive velocity absorption feature is at 6 ± 7 km s⁻¹. The lowest negative velocity absorption feature is likely at −110 ± 10 km s⁻¹ (a weak absorption feature appears at −189 ± 2 km s⁻¹ which is real; see Section 3.1). We calculate the absorption column density from the H i absorption spectrum, N$_{{\rm H}\,{\scriptsize {I}}}$ ∼ 1.44 × 10²² cm⁻², taking a value of T_s = 100 K (using N$_{{\rm H}\,{\scriptsize {I}}}$ = 1.9 × 10¹⁸∫τdvT_s cm⁻²; Dickey & Lockman 1990).

Zoom In Zoom Out Reset image size

We use the H i channel maps to confirm the reality of all abovementioned absorption features, including the 189 km s⁻¹ absorption feature. Absorption continues to appear from the channel map at a velocity of ∼−123 km s⁻¹ s through the map at a velocity of ∼13 km s⁻¹. Figure 2 shows the H i channel maps at 8.3 km s⁻¹ and 16.5 km s⁻¹ toward the SNR. The left panel of Figure 2 shows clear absorption from against the SNR. The right panel reveals that the remnant sits on a bright H i patch which is associated with the 16.5 km s⁻¹ CO cloud. This is consistent with the above absorption spectrum analysis, i.e., no H i absorption associated with the H i and CO clouds at 16.5 km s⁻¹. This directly confirms that the 16.5 km s⁻¹ clouds are behind the remnant.

Zoom In Zoom Out Reset image size

The H i terminal velocity at l = 349° to 350° is about −190 km s⁻¹ (see Figure 2 of Weiner & Sellwood 1999 and of Burton & Liszt 1993). We detect a weak absorption feature at −189 km s⁻¹ in the spectrum of SNR G349.7+0.2, which is likely caused by clumps of H i possibly accelerated by an SNR or H ii region along this line of sight. The lowest-velocity major absorption feature, except for the −189 km s⁻¹ feature, is at ∼ −110 km s⁻¹ in the SNR spectrum (Figure 2). This velocity is consistent with the expected radial velocity (101.6 ± 8.7 km s⁻¹) of the near 3 kpc arm in the line of sight to l = 349° to 350° (Equation (3) of Jones et al. 2013 see also Figure 1 of Dame & Thaddeus 2008). This is convincing evidence that the remnant is beyond the near 3 kpc arm.

In addition, the highest velocity absorption in the spectrum of G349.7+0.2 is at 6 ± 7 km s⁻¹. This is consistent with the expected radial velocity (16.3 ± 15.6 km s⁻¹) of the far 3 kpc arm toward l ∼−10.3° (Figure 1 and Equation (4) of Jones et al. 2013). However, in the SNR spectrum we found no H i absorption associated with the H i and CO emission clouds at 16.5 km s⁻¹, which lie exactly in the far 3 kpc arm extending at least 20° in longitude, starting at l = −12° (Dame & Thaddeus 2008). These two pieces of evidence argue that the remnant is in the far 3 kpc arm; the cold H i gas at 6 ± 7 km s⁻¹ is in front of the remnant, and the H i and CO clouds at 16.5 km s⁻¹ are behind the remnant (see Figure 3).

Zoom In Zoom Out Reset image size

SNR G349.7+0.2 is known to be interacting with nearby molecular clouds because of clear evidence: there are five 1720 MHz OH masers along the radio emission ridge of the SNR (Frail et al. 1996), shock-excited near-infrared H₂ emission has been found toward the center of the remnant as well as OH absorption (1665 and 1667 MHz) in the remnant (Lazendic et al. 2010), and CO expansion caused by the SNR shock has been observed (Reynoso & Mangum 2001). Because most 1720 MHz OH masers are seen as signposts of SNR–molecular-cloud interaction (Wardle & Yusef-Zadeh 2002), their velocities have been used to determine kinematic distances to associated SNRs based on a rotation curve model. The velocities of the 1720 MHz OH masers (14.3–16.9 km s⁻¹) are nicely consistent with the velocity of shocked CO clouds (16.5 km s⁻¹) and with the far 3 kpc arm velocity of ∼16.3 km s⁻¹. Combined with the H i absorption analysis, we conclude that the 16.5 km s⁻¹ CO clouds are behind but shocked by SNR G349.7+0.2 and this SNR–cloud interaction excites the near-infrared H₂ emission and the 1720 MHz OH maser emission in the region.

An H i 21 cm absorption line study of G347.9+0.2 has previously been done using the Parkes hydrogen line interferometer (Caswell et al. 1975). Our absorption spectrum confirms the previous major absorption features near −110 km s⁻¹, −62.5 km s⁻¹, and 6 km s⁻¹, but we find more fine structure because of the better quality of the SGPS data (higher continuum sensitivity and H i spectral resolution) and the improved methods. We find new absorption features at −35 and −189 km s⁻¹, no absorption at 20 and 38 km s⁻¹, and much lower noise for baselines above 20 km s⁻¹ and below −125 km s⁻¹. Caswell et al. (1975) explained the −62.5 km s⁻¹ feature as from the far ring of the 3–4 kpc arm and the 6 km s⁻¹ feature as from local gas, based on 1970s knowledge of the GC, so they suggested a distance of 13.7 kpc ⩽ d ⩽ 23 kpc for SNR G349.7-0.2. However, our analysis utilizes new information that the near and the far 3 kpc arms have kinematic velocities of −110 and −16.5 km s⁻¹, respectively, toward G347.9+0.2, combined with our new result of reliable absorption at 6 ± 7 km s⁻¹. Thus, we find the SNR has no absorption associated with the 16.5 km s⁻¹ clouds in the far 3 kpc arm. In this case, the −62.5 km s⁻¹ feature is likely caused by much closer H i clouds whose motion follows the normal rotation curve model. Our conclusion is strongly supported by other evidence: the H i absorption at −62.5 km s⁻¹ is also clearly seen in absorption spectra of the nearby SNRs G348.5+0.1 and G348.5-0.0, both of which have distances of less than 9.5 kpc (Tian & Leahy 2012); and the remnant is interacting with 16.5 km s⁻¹ clouds in the far 3 kpc arm. Previously, the 1720 MHz OH masers at 14.3 to 16.9 km s⁻¹ were assumed to be outside of the solar circle based on the circular rotation curve model, so SNR G349.7+0.2 was suggested to have a kinematic distance of ∼22 kpc. Now we know that the circular rotation curve model is not valid near the GC, so we conclude that the SNR and its nearby clouds are located in the far 3 kpc arm, well within the solar circle.

SNR G349.7+0.2 is located at the near edge of the far 3 kpc arm because there is no absorption from the main part of the far 3 kpc arm (which has a center velocity of 16.5 km s⁻¹). The far 3 kpc arm is at a distance of ∼11.5 kpc (Dame & Thaddeus 2008; Jones et al. 2013). No precise thickness of the 3 kpc arm scale is available so far, so we here give an approximate distance of ∼11.5 kpc for SNR G349.7-0.2.

Previous Chandra X-ray observations of SNR G349.7+0.3 gave a column density of 7 × 10²² cm⁻². This is consistent with the estimate from adding H i (2.8 × 10²² cm⁻², assuming T_s = 140 K) and H₂ (4.6 × 10²² cm⁻², assuming CO-to-H₂ conversion factor X = 1.8 × 10²⁰ cm⁻² K⁻¹ km s⁻¹) emission (Lazendic et al. 2005). Our calculation of the column density against G349.7+0.2 by H i absorption (using T_s = 100 K) is ∼1.4 × 10²² cm⁻²), similar to Lazendic et al. (2005). The H i column density measures neutral hydrogen, whereas the X-ray column density is sensitive to the heavier element component in the total gas (H i, H₂, and H ii) plus dust. So the X-ray column density could easily be three or more times higher than the H i column density. In addition, it is generally known that the column density value is strongly related to the direction of a source because Galactic H i has a different distribution along different lines of sight. For example, SNR G18.8+0.3 has a distance of 12 kpc and column densities of 2 × 10²² cm⁻² from H i and 3.2 10²² cm⁻² from H₂ (Tian et al. 2007). Generally, there are more clouds toward the GC than in other directions, so there is no inconsistency in column densities with G349.7+0.2 being at 11.5 kpc.

It should be noted that kinematic distances to Galactic objects always include uncertainty due to uncertain parameters of different rotation curve models and widespread non-circular streaming motions. This uncertainty may reach up to 20% at the particular longitude of some objects (Jones et al. 2013). SNR G349.7+0.2 is within the far 3 kpc arm: its distance's uncertainty is from the R_o value and the width of the arm. The IAU standard is R_o = 8.5 kpc, but recent measurements show a smaller value, e.g., 8.4 ± 0.6 kpc from very long baseline interferometry (VLBI) trigonometric parallax measurements (Reid et al. 2009), 8.05 ± 0.45 kpc from VLBI astrometry (Honma et al. 2013), 8.33 ± 0.19 kpc from optical and near-infrared observation of known RR Lyrae stars in the bulge (Dekany et al. 2013), etc. The uncertainty of R_o is about 0.5 kpc. The width of the far 3 kpc arm is not yet known but should be smaller than the error of R_o. We therefore estimate a distance of 11.5 ± 0.7 kpc for the far 3 kpc arm.

In summary, we significantly revise the distance to G349.7+0.2 from a previous 22 kpc to 11.5 kpc, based on better understanding of the GC structure. We previously revised distances to three other SNRs near G349.7+0.2 (Tian & Leahy 2012). One should use caution when considering old kinematic distances of Galactic objects (e.g., SNRs, pulsars, and H ii regions) in the range of −12° ⩽ l ⩽ 12° and having distance estimates of ⩾5.5 kpc.

G349.7+0.2 has been suggested to be one of the brightest Galactic sources in the radio (Shaver et al. 1985), X-ray (Slane et al. 2002), and GeV γ-ray wavebands (Castro & Slane 2010) because of its large distance. We give its luminosity based on the new distance measurement of 11.5 kpc obtained here: L_1 GHz ∼ 3.2 × 10¹⁷ Watt Hz⁻¹, L_{X(0.5 − 10.0 keV)} ∼ 1.0 × 10³⁷ erg s⁻¹, L_{γ(0.1 − 100 GeV)} ∼3.6 × 10³⁴ erg s⁻¹.

The angular diameter of G349.7+0.2 (∼2') from Chandra (Lazendic et al. 2005) yields a radius of R = 3.3 pc (d = 11.5 kpc) for the SNR. Lazendic et al. (2005) find a shock velocity of ∼710 km s⁻¹ based on the X-ray measured plasma temperature. The small remnant size and fast shock velocity indicate that the remnant is still evolving in the Sedov phase. So we estimate its age of ∼1800 yr by applying the Sedov model (Cox 1972; also see Equations (4) and (5) of Reynoso & Mangum 2001) using the known remnant radius and shock velocity. The supernova explosion energy depends on the environment density n₀. Using the X-ray emission measure from the Chandra spectrum of 9.9 × 10⁵⁹ cm⁻³ (Lazendic et al. 2005) and a Sedov interior density profile (see Leahy et al. 2013), the interstellar medium density is n₀ ∼ 10 cm⁻³ (d = 11.5 kpc). G349.7+0.2 is then found to have a low explosion energy of 2.5 × 10⁵⁰ erg.