ABSTRACT
CTB109 is one of only three Galactic supernova remnants (SNRs) known to harbor an anomalous X-ray pulsar or magnetar. That makes this SNR an object of great importance and a prime target for high-energy astrophysics studies. Those studies rely heavily on the assumed distance to CTB109. There have been three major distance determinations over the last decade, all of which report completely different results. While chaotic distance determinations in the literature are not uncommon for SNRs as a class of object, the wild discrepancy in the distance to CTB109 makes it especially important to revisit and firmly resolve once and for all. In this Letter we bring to bear all available observational information and present a synthesis of evidence that consistently locates CTB109 within or close to the Perseus arm spiral shock, at a distance of 3.2 ± 0.2 kpc.
Export citation and abstract BibTeX RIS
1. INTRODUCTION
CTB109 is a radio and X-ray bright shell-type supernova remnant (SNR) discovered through its bright thermal X-ray signature by Gregory & Fahlman (1980) in data observed with the Einstein satellite. This SNR was classified by Downes (1983) as a shell-type SNR based on a comparison of its X-ray and radio properties. It is widely accepted that the peculiar semi-circular shape of this SNR is the result of its interaction with a neighboring giant molecular cloud (Tatematsu et al. 1987) which rapidly stalled the shockwave in its initially free expansion to the west. The supernova (SN) explosion left a compact object behind that was discovered as an X-ray pulsar (Fahlman & Gregory 1981), but there is no evidence for pulsed radio emission or a pulsar wind nebula (PWN). CTB109 is one of only three SNRs that hosts such an anomalous X-ray pulsar (AXP J2301+5852; also 1E2259+589) within its boundaries (Gaensler et al. 2001). AXP's are also known as "magnetars:" neutron stars with unusually high magnetic fields. This makes CTB109 a popular target of study. Its distance is thus a critical and highly relied-on measure, since the physical properties of the SNR and its AXP can be understood only if their distances are well determined.
Over the last few years there were quite a few attempts to determine the distance to CTB109, all of which led to different results. There was a thorough discussion about older distance estimates by Kothes et al. (2002) who placed the SNR in the Perseus spiral shock at a distance of 3.0 ± 0.5 kpc using data from the Canadian Galactic Plane Survey (CGPS; Taylor et al. 2003). Two more attempts to determine the distance to CTB109 followed (Durant & van Kerkwijk 2006; Tian et al. 2010). In our estimation these studies are not sufficiently conclusive, with the latter two neglecting or misinterpreting important information and results. Given the importance of the distance to future studies of this enigmatic SNR/AXP, we have revisited the distance determination of CTB109, and settle the question with a robust (and correct) distance estimate arrived at by a thorough investigation considering all relevant information.
Distance estimates to CTB109 rely heavily on the kinematic approach: SNRs as a class of objects have few alternative methods. Several authors have already shown that a substantial CO cloud complex that is spatially and dynamically coherent over a full degree of longitude and more than 10 km s−1 in velocity (near vLSR ∼ −50 km s−1) is physically associated with the SNR and the two nearby H ii regions (Tatematsu et al. 1987, 1990; Kothes et al. 2002). This is also the velocity of the first major H i concentration in this part of the sky, known to be the Perseus spiral arm. For Perseus arm objects, however, kinematically determined distances are usually a significant overestimate. A spiral shock in the Perseus arm decelerates the interstellar medium (ISM) gas (and objects like H ii regions, SNRs, and PWNe) as they approach the arm's potential minimum, giving them—from our perspective—a higher negative radial velocity, which makes them appear to be farther away than they actually are (Roberts 1972). Examples for this effect on distance estimates of SNRs and PWNe located in the Perseus arm can be found, e.g., in Kothes et al. (2003) and Foster et al. (2004). We will use the new method of Foster & MacWilliams (2006) to determine a new distance–velocity relation toward CTB109 and combine this with a thorough discussion of other available information to determine the most accurate distance possible.
2. PREVIOUS DISTANCE ESTIMATES TO CTB109
Over the last decade, there have been three major distance determinations, all of which led to completely different and incompatible results. In particular, the studies by Kothes et al. (2002) and Tian et al. (2010) use in principle the same method, but interpret the available data differently. Here, we summarize each and highlight the strengths and weaknesses of those three determinations.
2.1. The Distance Estimate by Kothes et al. (2002)
The distance estimate by Kothes et al. (2002) was mainly based on data from the CGPS and also discussed the results of earlier studies. By comparing its radial velocity to those of nearby H ii regions (with known stellar distances) Kothes et al. (2002) found that CTB109 is co-located in the spiral shock of the Perseus spiral arm with these objects, whose ionizing stars average 3.0 ± 0.5 kpc in distance. It was argued that a spiral arm is a quite natural location for a source connected to a massive star. The high compression of the gas in the spiral shock creates the perfect environment for the formation of massive stars. Based on Perseus arm shock velocities in Roberts (1972) the systemic velocity of the molecular cloud complex interacting with CTB109 of about −50 km s−1 (Tatematsu et al. 1987) would support a location in the Perseus spiral arm shock at 3.0 kpc. The electron distribution models of our Galaxy by Taylor & Cordes (1993) and Cordes & Lazio (2003) also locate the Perseus spiral arm at this distance.
Kothes et al. (2002) also discussed the distance ambiguity caused by the spiral shock (Roberts 1972): the radial velocity of CTB109 is consistent with a location in the spiral shock and a more distant point far behind the Perseus spiral arm. The ambiguity was solved by an analysis of H i absorption profiles toward CTB109, the two nearby H ii regions Sh2-152 and Sh2-149, and an extragalactic source. While the SNR and these two regions display nearly identical absorption profiles, the profile of the extragalactic source contains an additional component at about −45 km s−1. According to the velocity–distance relations computed by Roberts (1972), the only possible location for a cloud with such a velocity is between the two possible distances for CTB109 where the rotation curve returns to its monotonic behavior. Since the radio continuum emission of the SNR and the H ii regions is not absorbed by that cloud at about −45 km s−1 they have to be in front of it.
Spectrophotometric distances to stars within H ii regions in the vicinity that have similar velocities then indicate a similar location of CTB109 in the Perseus spiral arm shock at 3.0 ± 0.5 kpc. This assumes that stellar distances are accurate, and that the objects are responding to the same dynamical forces (e.g., overall rotation, streaming from spiral arm potentials, cloud–cloud turbulence).
2.2. The Distance Estimate by Durant & van Kerkwijk (2006)
Durant & van Kerkwijk (2006) determine distance–extinction relationships based on the visual extinction Av extrapolated from Two Micron All Sky Survey data of core helium burning giants (K2III stars). A comparison of this relation to the visual extinction determined for AXPs based on measurements of their X-ray absorbing total neutral hydrogen foreground column density NH leads to a distance estimate for the pulsar. The correlation of Predehl & Schmitt (1995) is used to translate NH to Av for the AXPs. In the case of AXP 1E2259+589 and CTB109, they obtain a distance of 7.5 ± 1.0 kpc.
Vink (2008) already suggested using this new distance estimation method very cautiously. There is a large intrinsic spread in the NH–Av relation determined by Predehl & Schmitt (1995). A brief look at the determination of that translation function (Predehl & Schmitt 1995, their Figure 3) shows that fluctuations of about ±1 mag for Av are common for any given value of NH. This uncertainty was not included in the results of Durant & van Kerkwijk (2006).
Vink (2008) also pointed out that there is a large discrepancy between the absorbing H i column determined for CTB109 and its AXP. They then suggest comparing the X-ray absorption of the SNR (e.g., Sasaki et al. 2006) to the Av–distance relation instead of the AXP. The resulting extinction is consistent with the distance estimate of 3.0 kpc by Kothes et al. (2002).
Cas A is another SNR nearby (∼2 deg away) that displays similar properties for the absorbing H i column density, which Durant & van Kerkwijk (2006) use as an example of the validity of their method. They compared the visual extinction observed for this SNR (and not the neutron star) with the distance–Av relation determined in the direction of Cas A. The large variation of Av over the face of SNR Cas A, which is similar to that over CTB109 (2–6 mag; Sasaki et al. 2004, 2006), makes the assumption of Durant & van Kerkwijk that Av measured towards the neutron star is the same as the average over a large area very poor. When compared to the distance–Av relation in Durant & van Kerkwijk (2006), this variation predicts distances from about 3 kpc out to 8 kpc. For Cas A the spread in Av is 4.5–11 mag (e.g., Laming & Hwang 2003), which predicts distances from ∼3.5 kpc to the edge of the Galaxy. A better distance measure to Cas A by Reed et al. (1995) based on proper motions and age data is 3.4+0.3 − 0.1 kpc. Interestingly, the neutron star inside Cas A has a very high value for NH. According to Pavlov & Luna (2009) NH varies depending on the model that is fit to the X-ray spectrum of the neutron star, but is consistently found to be ⩾1.2 × 1022 cm−2, which translates to a lower limit of Av = 6.7 mag (using the dust to gas ratio of Predehl & Schmitt 1995). This is at the upper limit of the Av–distance relation in Durant & van Kerkwijk (2006) and is based on only one point in that relation. The indicated distance of ⩾7 kpc must be seriously questioned.
The major problem of Durant & van Kerkwijk (2006) is in comparing the foreground extinction Av determined over a very large angular extent with that of a pulsar, which is an unresolved source. This Av determination of the K2III stars is heavily biased toward (lower) average column densities of the diffuse foreground. However, extinction from unseen line-of-sight (LOS) molecular clouds toward the AXP would quickly exceed the diffuse foreground extinction, and it is very unclear whether the extinction to the AXP can be considered as representative of the diffuse extinction, or is inflated by a "cloudy" LOS. Durant & van Kerkwijk (2006) do note a significant jump in extinction at 3 kpc toward the AXP, consistent with an LOS passing through the myriad compressed dense clouds that are produced by a Galactic spiral shock. Consequently, SNR/AXP distances resulting from comparison of Av and an extinction–distance relation calculated from a different direction should be considered with reserve.
2.3. The Distance Estimate by Tian et al. (2010)
Tian et al. (2010) use the method of Foster & MacWilliams (2006) to kinematically determine whether the velocity of CTB109 places it either near the Perseus arm shock or beyond the arm. However, they conclude that CTB109 is at the far kinematic distance of 4 kpc, rather than the near one (at the Perseus arm shock) by solving the kinematic distance ambiguity (KDA) with the method of Roman-Duval et al. (2009), which is based on testing whether given molecular clouds are related to observed H i self-absorption (HISA) features or not. The method requires that cold H i and molecular hydrogen (traced by CO) are always present together, and the ISM is filled with warm H i. If a molecular cloud is not associated with HISA, no warm background H i can be present at the same radial velocity, indicating a far distance location. Conversely, if HISA is found, there must be a warm background, indicating a near location for the molecular cloud. The result of Tian, Leahy, & Li rests on their determination that no HISA features are seen relating to the large molecular cloud complex interacting with CTB109 (and star-forming H ii regions Sh2-152 and Sh2-149), and hence the associated CO is at the far distance of 4 kpc.
Identifying HISA features in a spectrum requires careful comparison with neighboring spectra, and a sensitive high-resolution data set. The catalog of HISA in the CGPS (Gibson et al. 2005a) only lists the major large HISA features, none of which are on top of the molecular cloud associated with CTB109. However, the maps of the sky in Gibson et al. (2005a) show many isolated patches of HISA observed toward several directions atop this cloud. In Figure 1, we display four sample CO and H i profiles. Plotted are contours of isolated HISA patches found by Gibson et al. (2005a) clearly showing absorption in the CO cloud's velocity range in three separate directions. The direction not detected by Gibson et al. (2005a) shows a very small absorption amplitude that is below the cutoff used for the automatic detection program in Gibson et al. (2005a). The relative weakness of these features is likely due to the neighborhood's star-forming activity and massive stars. The UV environment slightly heats cold neutral hydrogen clouds, thus making their absorption signatures weak and creating the "patchy" appearance of bright HISA along only some directions, as most clouds would skirt the signal-to-noise limit for detection in the CGPS (|Ton − Toff| ≳ 10 K). The KDA resolution in Roman-Duval et al. (2009) does not change with either weak or strong HISA: what is important is the positive identification of HISA associated with the molecular cloud, which in this case implies a location near the Perseus spiral arm.
Figure 1. H i and CO spectra toward the massive molecular cloud interacting with CTB109. In the center a radio continuum image at 1420 MHz taken from the CGPS (Taylor et al. 2003) is displayed in gray scale. The relevant radio sources are marked. The white contours represent the 12CO(1−0) emission from the massive molecular cloud integrated between vLSR of −48 and −56 km s−1. The black contours represent all HISA features detected by Gibson et al. (2005a) with the automated HISA detection method described in Gibson et al. (2005b) over the same velocity range. The locations of the H i and CO spectra are indicated.
Download figure:
Standard image High-resolution image3. THE DISTANCE TO CTB109
The studies of Tatematsu et al. (1987), Kothes et al. (2002), and Sasaki et al. (2006) show molecular and atomic material interacting with the SNR in a wide velocity range of −48 to −56 km s−1. This is not surprising; this region has a very dynamic appearance, with many objects (including two very bright H ii regions) also interacting with the same cloud of gas. Such active star-forming regions input a substantial amount of energy into their environments and create a large cloud–cloud velocity dispersion. We conservatively use this entire velocity range, and compare it with a newly calculated distance–velocity relation to kinematically determine the distance to CTB109.
The standard approach to kinematic distances is to solve for the Galactocentric radius R (and hence the heliocentric distance d for a given longitude) where the observed LSR velocity is expected to occur due to a model of differential circular rotation and our viewing geometry. However, core-collapse produced SNRs are predominantly spiral arm objects, and this approach to their distances is often fraught with errors, as non-circular motions due to density wave streaming are significant, and the rotation model and the solar constants are themselves in error (though these have a secondary impact). To obtain a more accurate distance to CTB109, we use the distance modeling technique of Foster & MacWilliams (2006), which fits (via least squares) the shape of the 21 cm brightness temperature profile with a multi-parameter model of a large-scale Galactic structure and rotation, and includes density and velocity perturbations from a model spiral pattern. This technique is an extension of the kinematic approach, and has demonstrated tremendous success at reproducing distances to Galactic spiral arm objects with otherwise independently known positions.
For this study, H i profiles along eight separate LOSs (within 0
5 of the CTB109's center) are fitted independently with the model by Foster & MacWilliams (2006). H i profiles are obtained from a 10° × 10° CGPS mosaic that has first been processed by a "cloud-filter" technique (described in Foster & Cooper 2010), which produces a datacube of the smooth inter-cloud medium. Additionally, two of these LOSs are each independently modeled 10 times using random starting positions in the model's parameter space, which produces a family of solutions from which the rms variation in each model parameter (from solution to solution) is calculated. This is added in quadrature to the rms variation in each parameter from LOS-to-LOS, giving the final estimated uncertainty in the model and hence the uncertainty in the distance to CTB109 calculated from it. Solar constants (R0, θ0) = (8.4 kpc, 254 km s−1) from Reid et al. (2009) are used.
Toward ℓ = 1091, b = −1, we obtain a distance to the Perseus arm shock of 3.0 ± 0.2 kpc (standard error). Among all fitted models the maximum and minimum distance to the shock is 3.24 kpc and 2.62 kpc, respectively. The distance to the density peak of H i associated with the arm is similar at 3.2 ± 0.2 kpc, varying between 3.6 kpc (max) and 2.9 kpc (min).
Figure 2 is the (fitted) position–velocity plot from the model, and shows three possible distances based on CTB109's velocity. The far one (d between 3.9 and 4.3 kpc) places CTB109 in the interarm region beyond the Perseus arm, but a simple calculation shows that this distance is very unlikely. Assuming the progenitor of CTB109 was a massive star that formed in the shock, the far distance suggests it migrated downstream and out of the arm before exploding. This would have taken some Δt = Δϕ/(Ω − Ωp) = 50–111 Myr, where Δϕ is the angular distance between the shock and CTB109 along a circle of radius R = 10.6 kpc (for d = 4.3 kpc), and (Ω − Ωp) = 3.3–7.3 km s−1 kpc−1 is the angular speed of Galactic rotation at the circle relative to the spiral shock/arm with pitch angle i = 12° and pattern speed range of Ωp = 16–20 km s−1 kpc−1 appropriate for the outer Galaxy (Foster & Cooper 2010). This is far too long for a typical massive star (earlier than about B2.5V) to exist before expiring in an SN. In addition, the presence of HISA related to the molecular cloud and the H i absorption study of the SNR, the two H ii regions, and the nearby extragalactic source negate a location at the far distance.
Figure 2. Distance–velocity relation toward CTB109. The diagram shows a flat rotation curve and the rotation curve by Foster & MacWilliams (2006) in relation to the systemic velocity interval determined for the SNR and its related H ii regions.
Download figure:
Standard image High-resolution image
The likely minimum distance to CTB109 is that of the shock (3.0 ± 0.2 kpc), with the more likely possibility of d = 3.2 ± 0.2 kpc slightly downstream. The ambiguity can be somewhat resolved by considering the observation that CTB109 resides in an H i depression, and not an H i shell or bubble, as would have been created by the stellar winds of a massive progenitor. This limits the progenitor to being closer to the low-mass, low-wind end of luminous stars that can produce SNe. If such a progenitor formed in the shock, its longevity compared to stars earlier than ∼B2 would allow it to migrate to the downstream distance of d = 3.2 ± 0.2 kpc, a journey that would take some 5–50 million years (Roberts 1972).
4. CONCLUSIONS
All available information that we found in the literature points to a location of the SNR CTB109 with its AXP in the center of the Perseus spiral arm density peak at a distance of 3.2 ± 0.2 kpc. The main arguments for this location are as follows.
- 1.A spiral arm location provides the natural environment for the formation of massive stars.
- 2.Nearby H ii regions and SNRs with similar radial velocities have average distances consistent with a location in Perseus arm.
- 3.The radial velocity of CTB109 is consistent with a location in Perseus arm and beyond it at about 4 kpc. The ambiguity was solved by
- 4.NH values determined for the SNR with X-ray studies in comparison with the Av–distance relation of Durant & van Kerkwijk (2006) are consistent with a Perseus arm location.
- 5.The absence of a stellar wind bubble in H i data, which indicates a less massive SN progenitor and longer lifetime, enabling the star to migrate beyond the shock by a few hundred parsecs before its core collapsed.
We conclude that the SNR CTB109 is located in the Perseus spiral arm at a distance of 3.2 ± 0.2 kpc.
We thank Steve Gibson for providing the data determined with his automated HISA detection algorithm. The Dominion Radio Astrophysical Observatory is a National Facility operated by the National Research Council. The Canadian Galactic Plane Survey is a Canadian project with international partners, and is supported by the Natural Sciences and Engineering Research Council (NSERC). T.F. thanks the Brandon University Research Committee (BURC) for their grant support which made this research possible.