KINEMATICS OF ULTRA-HIGH-VELOCITY GAS IN THE EXPANDING MOLECULAR SHELL ADJACENT TO THE W44 SUPERNOVA REMNANT

We mapped the Bullet in the CO J = 3–2 (345.796 GHz), HCO⁺ J = 4–3 (356.734 GHz), CO J = 4–3 (461.041 GHz), and C^{0 3}P₁–³P₀ (492.161 GHz) lines using the Atacama Submillimeter Telescope Experiment (ASTE) 10 m telescope. For comparison, we also mapped Wing 1 (Seta et al. 2004) as a representative of shocked molecular gas adjacent to the W44 SNR. The observations were conducted on July 30 and August 5–8 and 25–26 in 2014, as well as July 21–24 and 28 and September 12–13 and 21–24 in 2015. The main-beam efficiency (η_MB) and FWHM beam-size of the telescope are 0.6 and 22'' at 350 GHz, 0.45 and 17'' at 492 GHz, respectively. The pointing of the telescope was checked and corrected every 2 hr by observing CO emission from a late-type star R Aql, and its accuracy was maintained within 2'' (rms). The observations were made with a cartridge-type single-polarization side-band separating (2SB) mixer receiver CATS345 in 2014, and a 2SB mixer receiver DASH345 and dual-polarization 2SB mixer receiver ASTE BAND8 in 2015. All observations were made in OTF mapping mode. The reference position was taken at (l, b) = (+33750, −1509). During the observations, the system noise temperatures ranged between 200 and 600 K (SSB) with CATS345 and DASH345, 600 and 1500 K (SSB) with ASTE BAND8. The antenna temperature (${{T}_{{\rm{A}}}}^{* }$) was obtained by the standard chopper-wheel technique.

We used an XF-type digital spectro-correlator MAC in the 512 MHz bandwidth (0.5 MHz resolution) mode. This combination of the frequency bandwidth and resolution corresponds to 444 km s⁻¹ velocity coverage with a 0.43 km s⁻¹ velocity resolution at 346 GHz and 333 km s⁻¹ coverage with a 0.33 km s⁻¹ resolution at 461 GHz. The obtained data were reduced using the NOSTAR reduction package. We subtracted the baselines of all spectra by fitting first- or third-order polynomial lines. We scaled the antenna temperature by multiplying it by 1/η_MB to obtain the main-beam temperature, T_MB. The data were smoothed with a Bessel–Gaussian function and resampled on an 75 × 75 × 1.0 km s⁻¹ grid to obtain the final maps.

The Bullet and Wing 1 were also mapped in the CO J = 1–0 (115.271 GHz) and HCO⁺ J = 1–0 (89.1885 GHz) lines with the Nobeyama Radio Observatory (NRO) 45 m radio telescope. The NRO 45 m observations were made on 2015 January 8–15 and March 5–7. The main-beam efficiencies and FWHM beam sizes of the telescope are 0.5 and 20'' at 86 GHz, 0.4 and 15'' at 115 GHz, respectively. The telescope pointing was checked and corrected every 2 h by observing the SiO maser source V1418 Aql at 43 GHz, and its accuracy was maintained within 2'' (rms). All data were obtained with 2SB receivers TZ1V/H in OTF mapping mode. The reference position was taken at (l, b) = (+33750, −1509). During the observations, the system noise temperatures ranged between 100 and 400 K (SSB) with TZ1V/H. The antenna temperature was calibrated by the standard chopper-wheel technique.

We used the highly flexible FX-type correlator SAM45 in the 500 MHz bandwidth (122 kHz resolution) mode. This combination of the frequency bandwidth and resolution corresponds to 1300 km s⁻¹ velocity coverage with a 0.32 km s⁻¹ velocity resolution at 115 GHz, and 1700 km s⁻¹ coverage with a 0.41 km s⁻¹ resolution at 89 GHz, respectively. The NRO 45 m data were reduced using the NOSTAR reduction package, and the final maps were generated with the same parameters as those of the ASTE maps.

Figures 1(c) to 1(h) show the velocity-integrated maps of the 3' × 3' (2.6 × 2.6 pc²) area around the Bullet in the six observed lines. The Bullet appears as a compact clump with a size of 0.5 × 0.8 pc² in the CO J = 1–0, CO J = 3–2, CO J = 4–3, and HCO⁺ J = 1–0 maps. The detected line emissions peak at (l, b) ≃ (3473, −047). The entity of the Bullet is well defined by CO J = 3–2 and CO J = 4–3 emissions, while C^{0 3}P₁–³P₀ emission is absent. A blob of 20 cm continuum emission traces the Galactic eastern rim of the Bullet (Figure 1(a)). A nebulosity of H₂ ro-vibrational transition line emission overlaps the Galactic eastern half of the Bullet (Reach et al. 2005). An inspection of the infrared (Yamauchi et al. 2011) and X-ray (Muno et al. 2006) point-source catalogs reveals that no counterpart of the Bullet is detected in those wavelengths.

Longitude–velocity maps illustrate the broad-velocity nature of the Bullet (Figure 2). Its velocity width in FWZI well exceeds 100 km s⁻¹. The narrow velocity width features at V_LSR ≃ 12 km s⁻¹ and 30 km s⁻¹ are Galactic spiral arms irrelevant to the W44 molecular cloud. The Bullet is very compact (∼0.8 pc) in its negative high-velocity half (V_LSR ≤ −10 km s⁻¹). It bifurcates in the lower velocity range (V_LSR ≥ −10 km s⁻¹), exhibiting a "Y" shape, and merges into the W44 molecular cloud at V_LSR ≃ 40 km s⁻¹. CO J = 3–2, CO J = 4–3, and HCO⁺ J = 1–0 emissions decline at the footpoint of the Bullet (V_LSR ∼ +30 km s⁻¹).

Zoom In Zoom Out Reset image size

We estimated the physical conditions in the Bullet and Wing 1 by employing large velocity gradient (LVG) model calculations (e.g., Goldreich & Kwan 1974; Scoville & Solomon 1974). Table 1 summarizes the line intensities and the best-fit value of physical conditions. Beam-filling factors were assumed to be unity. We used the main-beam temperatures at (l, b, V_LSR) = (34725, −0472, −57.5 km s⁻¹) and (l, b, V_LSR) = (34754, −0406, +36.8 km s⁻¹) as representative positions of the Bullet and Wing 1, respectively. We assumed the fractional abundance [CO]/[H₂] = 10^−4.1 (Frerking et al. 1982). The 90% confidence intervals are also shown in Table 1. We see that the Bullet clearly has a higher temperature and lower density than Wing 1, which represent shocked gas in the W44 molecular cloud.

Table 1.  LVG Model Calculations of Wing 1 and Bullet

Source		TMB [K]		Tk	log(n(H2))	log(N(H2)/ΔV)
	CO J = 1–0	CO J = 3–2	CO J = 4–3	[K]	[cm−3]	[cm−2 (km s−1)−1]
Bullet	0.40 ± 0.07	2.13 ± 0.03	1.82 ± 0.08	77−43a	${4.3}_{-0.2}^{+1.1}$	${19.00}_{-0.02}^{+0.06}$
Wing 1	4.29 ± 0.23	12.55 ± 0.63	12.00 ± 0.60	${33}_{-3}^{+11}$	9.0−4.4a	${20.00}_{-0.04}^{+0.02}$

Note.

^aThe upper bounds could not be found within the parameter ranges of our calculations: 10 [K] ≤ T_k ≤ 100 [K] and 10¹ [cm⁻³] ≤ n(H₂) ≤ 10⁹ [cm⁻³].

Download table as:  ASCIITypeset image

Using the physical conditions derived for the Bullet and assuming the optically thin limit, we estimated the total mass of the Bullet from CO J = 3–2 intensity to be 7.5 M_⊙. If the Bullet arises from the W44 molecular cloud at V_LSR ≃ 40 km s⁻¹, its kinetic energy amounts to 10^48.0 erg. This is approximately 1.5 orders of magnitude greater than the kinetic energy shared to the small solid angle of the Bullet, (1–3) × 10⁵⁰ erg × (ΔΩ/4π) = 10^46.3–46.8 erg. Obviously, the energy injection by an SN explosion alone cannot account for the kinetic energy of the Bullet.

The characteristics of the Bullet are summarized as follows:

• 1.  
  It has a compact appearance (0.5 × 0.8 pc²) and an extraordinarily broad-velocity width (100 km s⁻¹), giving a dynamical time of 5000–8000 years.
• 2.  
  Unlike previously detected broad emissions, it appears in the negative velocity side only.
• 3.  
  It bifurcates at V_LSR ≃ −10 km s⁻¹, showing a "Y" shape in the l–V plane.
• 4.  
  It has a mass of 7.5 M_⊙ and a kinetic energy of 10^48.0 erg. The kinetic energy cannot be supplied by the W44 SN shell.
• 5.  
  A radio continuum blob and an H₂ ro-vibrational line nebulosity are associated with it.

These characteristics indicate that the Bullet has been accelerated by a certain local event, possibly driven by a single source. If so, the kinetic energy supplied by the driving source must be greater than 10^48.0 erg. In addition, the short dynamical time and large kinetic energy indicate that the kinetic power of the driving source must be greater than 10^36.6 erg s⁻¹ (≃1000 L_⊙). Here, we present two possible formation scenarios of the Bullet.

3.3.1. Explosion Model

An additional explosive event near the expanding SN shell could account for the kinematics of the Bullet (Figure 3(a)). In this case, the driving source must be located behind the expanding shell since the Bullet is single-sided. For the same reason, the kinetic energy and kinetic power necessary for the Bullet formation are doubled, 10^48.3 erg and 10^36.9 erg s⁻¹, respectively. The upper part of the "Y" shape can be interpreted as the expanding shell due to the local explosion, while the lower part might be the broken tip of the shell in the low-density layer.

Zoom In Zoom Out Reset image size

A rotation-powered pulsar may be a candidate for the driving source. Some energetic pulsars have spin-down power higher than 10³⁷ erg s⁻¹ (e.g., Arumugasamy et al. 2014). However, no pulsar has been detected toward this position. The W44 pulsar is located at least ∼8 pc away from the Bullet. A bipolar outflow from a massive protostar is not a candidate for the driving source, because of the absence of a luminous stellar object, although some of them provide kinetic energies greater than 10^48.0 erg (Bally & Zinnecker 2005).

A completely separate supernova explosion can account for the kinetic energy of the Bullet. If we assume an isotropic disk (radius = 15 kpc, thickness = 200 pc) distribution of population I objects and that supernovae occur in spiral arms (volume filling factor ∼0.3), the supernova rate of the entire Galaxy (∼0.02 yr⁻¹; Tammann et al. 1994) provides a chance probability of another supernova in the W44 shell (radius = 15 pc) within 20,000 years as ∼10^−3.8. Considering that massive stars are formed in clusters, this chance probability may be further enhanced.

Despite the separate supernova scenario that cannot be ruled out, here we consider another possibility in detail. That is an isolated black hole (BH) activated by the passage of a high-density molecular gas layer. Such an object has already been suggested as a driving source of the Tornado nebula (Sakai et al. 2014). This process is formalized as the Bondi-Hoyle-Littleton (BHL) accretion (e.g., Edgar 2004). Assuming the standard accretion disk model onto a BH and that all the gravitational energy is converted to the kinetic energy, the BHL accretion with a duration time Δt provides

$$\begin{eqnarray}&&{E}_{\mathrm{HL}}=\displaystyle \frac{1}{12}{\dot{M}}_{\mathrm{HL}}{c}^{2}{\rm{\Delta }}t\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{\dot{M}}_{\mathrm{HL}}=\displaystyle \frac{4\pi {G}^{2}{M}^{2}\rho }{{({v}^{2}+{\sigma }^{2})}^{3/2}},\end{eqnarray} \tag{ 2 }$$

where M is the BH mass, ρ is the mass density of the ambient interstellar medium, v is the relative velocity, and σ is the velocity dispersion. The duration time of the accretion is calculated by Δt = l/v, where l is the thickness of the high-density layer. If we take l = 0.1 pc, which is about the thickness of thin filaments detected in W44 at 20 cm continuum emission (Jones et al. 1993), we have Δt = 7500 years. Employing n(H₂) = 10^4.5 cm⁻³, v = 13.2 km s⁻¹, σ = 6.8 km s⁻¹ (Sashida et al. 2013), and the mean molecular weight of 1.36, we have

$$\begin{eqnarray}&&{E}_{\mathrm{HL}}\,=\,{10}^{48.3}{\left(\displaystyle \frac{M}{3.4{M}_{\odot }}\right)}^{2}\ {\rm{erg}}.\end{eqnarray} \tag{ 3 }$$

This means a BH mass of greater than 3.4 M_⊙ could account for the kinetic energy of the Bullet. The putative BH is currently inactive since the accretion has already ceased.

3.3.2. Shooting Model

Conversely, a plunge of a high-velocity massive object into the high-density layer toward us could also explain the kinematics of the Bullet (Figure 3(b)). This process can be described by the BHL accretion formula as well. The upper part of the "Y" shape can be interpreted as the shock wave propagating in the high-density layer, while the lower part might correspond to the "wake" accreting to the plunging object (e.g., Edgar 2004). Again, the absence of a luminous stellar object favors a drifting, isolated BH as a candidate for the plunging object.

The plunging velocity must be higher than 13.2 km s⁻¹, but it is not necessarily as high as 100 km s⁻¹ since gas in the "wake" is accelerating toward the object. Of course, the initial kinetic energy of the plunging object, E_init = (1/2)Mv², must be greater than E_Bullet = 10^48.0 erg. Setting v to the plunging velocity, we can calculate the total energy available from the BHL accretion by using Equation (1). This BHL energy E_HL must also be greater than E_Bullet. In addition, the BHL luminosity, ${L}_{\mathrm{HL}}=(1/12){\dot{M}}_{\mathrm{HL}}{c}^{2}$, probably does not exceed the Eddington limit (L_Edd). Figure 4 shows the range of (M, v) permitted by the above conditions, indicating that the plunging object must have a mass greater than 30 M_⊙.

Zoom In Zoom Out Reset image size

This shooting model, as well as the explosion model, can reproduce the broad-velocity width and the enormous kinetic energy of the Bullet. A naive shooting model predicts that the size of the accreting zone should be of the same order as the Hoyle–Litteleton radius, r_HL = 2GM/(v² + σ²) ∼ 10⁻⁵ pc. Since the observed size of the negative high-velocity tail of the Bullet (∼0.3 pc) is comparable to the telescope beam-size (22''), the real spatial size might be much smaller. Moreover, the magneto-hydrodynamical effect can enlarge the accreting zone, which is roughly estimated by (v_A/v)l, where v_A is the Alfvén velocity and l is the thickness of the high-density layer (M. Nomura et al. 2016, in preparation). The magnetic field strength should be enhanced in the shocked gas layer, and thereby increases v_A and the size of the accreting zone accordingly.

Either scenario includes an isolated BH, which is currently inactive. But is it reasonable? Actually, there must be a number of invisible BHs in the Galaxy. For example, Agol & Kamionkowski (2002) estimated the total number of stellar mass BHs in the Galaxy to be ∼10⁹, which is far greater than the latest total number of the Galactic BH candidates (∼60). Assuming a spherically isotropic distribution with a radius of 30 kpc, this total BH number corresponds to a number density of 10^−5.1 pc⁻³. Thus, we obtain the expected value of the BH number within the W44 shell (15 pc radius) as 0.13, which is not significantly low. Note that this value may be a lower limit since an overdensity of BHs due to the disk population and a OB association (Ögelman & Maran 1976) is expected.

Considering the permitted mass ranges of an isolated BH, the explosion model seems to be more plausible. Such an isolated, currently inactive BH with a mass of >10 M_⊙ was suggested as the driving source of the Tornado nebula (Sakai et al. 2014). We also detected another extremely broad-velocity-width feature, CO–0.40–0.22, in the central molecular zone of our Galaxy (Oka et al. 2016). This was explained by a gravitational kick to the molecular cloud caused by an intermediate-mass BH. Therefore, an elaborate search for compact high-velocity features using molecular lines might be an effective method of seeking inactive BHs in the Galaxy and nearby galaxies.