The Interaction Between the Supernova Remnant W41 and the Filamentary Infrared Dark Cloud G23.33-0.30
Taylor G. Hogge, James M. Jackson, David Allingham, Andres E. Guzman, Nicholas Killerby-Smith, Kathleen E. Kraemer, Patricio Sanhueza, Ian W. Stephens, J. Scott Whitaker
DDraft version October 30, 2019
Preprint typeset using L A TEX style AASTeX6 v. 1.0
THE INTERACTION BETWEEN THE SUPERNOVA REMNANT W41 AND THEFILAMENTARY INFRARED DARK CLOUD G23.33-0.30
Taylor G. Hogge , James M. Jackson , David Allingham , Andres E. Guzman , NicholasKillerby-Smith , Kathleen E. Kraemer , Patricio Sanhueza , Ian W. Stephens , J. ScottWhitaker Institute for Astrophysical Research, 725 Commonwealth Ave., Boston University, Boston, MA 02215, USA;[email protected] SOFIA Science Center, USRA, NASA Ames Research Center, Moffett Field CA 94045, USA School of Mathematical and Physical Sciences, University of Newcastle, University Drive, Callaghan NSW 2308,Australia National Astronomical Observatory of Japan, National Institutes of Natural Sciences, 2-21-1 Osawa, Mitaka, Tokyo181-8588, Japan Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA Institute for Scientific Research, Boston College, 140 Commonwealth Avenue, Chestnut Hill, MA 02467, USA Physics Department, 590 Commonwealth Ave., Boston University, Boston, MA 02215, USA
ABSTRACTG23.33-0.30 is a 600 M (cid:12) infrared dark molecular filament that exhibits large NH velocity dispersions ( σ ∼ − ) and bright, narrow NH (3,3) line emission. Wehave probed G23.33-0.30 at the < . (3,3)line is emitted by four rare NH (3,3) masers, which are excited by a large-scale shock a r X i v : . [ a s t r o - ph . GA ] O c t Hogge et al. impacting the filament. G23.33-0.30 also displays a velocity gradient along its length,a velocity discontinuity across its width, shock-tracing SiO(5-4) emission extendedthroughout the filament, broad turbulent line widths in NH (1,1) through (6,6), CS(5-4), and SiO(5-4), as well as an increased NH rotational temperature ( T rot ) and velocitydispersion ( σ ) associated with the shocked, blueshifted component. The correlationsamong T rot , σ , and V LSR implies that the shock is accelerating, heating, and addingturbulent energy to the filament gas. Given G23.33-0.30’s location within the giantmolecular cloud G23.0-0.4, we speculate that the shock and NH (3,3) masers originatedfrom the supernova remnant W41, which exhibits additional evidence of an interactionwith G23.0-0.4. We have also detected the 1.3 mm dust continuum emission from atleast three embedded molecular cores associated with G23.33-0.30. Although the coreshave moderate gas masses ( M = 7 −
10 M (cid:12) ), their large virial parameters ( α = 4 − Keywords:
ISM: clouds − ISM: supernova remnants − stars: formation INTRODUCTIONHigh-mass stars (
M > (cid:12) ), though rare, have a profound impact on the evolution of the in-terstellar medium (ISM). Throughout their short lifetimes ( ∼ yr), high-mass stars will releasefast, radiation-driven stellar winds that carve out H ii regions into the surrounding molecular clouds(MCs). High-mass stars end their lives by releasing ∼ ergs of energy nearly instantaneously inthe form of supernovae (SNe). Shocks from expanding H ii regions and supernova remnants (SNRs)accelerate, heat, and add turbulence to their surrounding gas. While these feedback mechanisms arethought to play a significant role in regulating the star formation process, the exact role these shocks (3,3) inversion transition (McEwen et al. 2016). Dense molecular gas structures that exhibitevidence of broad molecular line widths, 1720 MHz OH maser emission, or NH3 (3,3) maser emission,but lack an obvious stellar source, are prime targets to investigate potential SNR-MC interactions.G23.33-0.30 is a dense molecular filament with extremely broad molecular line widths. Due tothe lack of any evidence at infrared wavelengths for embedded star formation, these broad linewidths are puzzling. The H O Southern Galactic Plane Survey (HOPS; Walsh et al. 2011) detectedextremely broad NH line widths and a bright, narrow feature in the NH (3,3) line profile towardG23.33-0.30. Figure 1 displays Radio Ammonia Mid-Plane Survey (RAMPS; Hogge et al. 2018)NH (1,1) through (4,4) spectra toward G23.33-0.30 at higher angular resolution ( θ ∼ (cid:48)(cid:48) ) thanthat of HOPS ( θ ∼ (cid:48)(cid:48) ). The spectra show the uncommonly broad line emission associated withG23.33-0.30, as well as emission from an unrelated background MC with a local standard of rest(LSR) velocity of V LSR = 103 km s − that exhibits more typical NH line profiles. G23.33-0.30’smolecular line emission has a large velocity dispersion ( σ ∼ − ), much larger than that of atypical dense MC ( σ ∼ . − − ; Sanhueza et al. 2012). Moreover, the NH (3,3) spectrumdisplays a bright, narrow line component superposed on a fainter, broad component. Unlike the Hogge et al.
50 75 100 125V
LSR [km s −1 ]0510 T m b [ K ] NH (1, 1)NH (2, 2)NH (3, 3)NH (4, 4) Figure 1 . RAMPS NH (1,1) through (4,4) spectra toward the peak of the narrow NH (3,3) emission. Thebroad emission from G23.33-0.30 is centered near V LSR = 67 km s − , while the more typical NH emissionfrom a presumably unrelated source peaks at V LSR = 103 km s − . The emission from G23.33-0.30 exhibitsextremely broad line widths, enhanced NH (3,3)/(1,1) brightness temperature ratio, and an NH (3,3) masercandidate. broad component, the bright, narrow line emission in the ortho ( K = 3 n ) NH (3,3) spectrum hasno corresponding component in the para ( K (cid:54) = 3 n ) transitions. Because masers emit at such smallspatial scales (e.g., Elitzur 1992), the single-dish observations lack the angular resolution necessaryto definitively determine whether the brightness temperature of this narrow feature is large enough toconfirm maser emission. Nevertheless, the narrow line width of this component and its appearancesolely in the NH (3,3) spectrum suggests that it is from an NH (3,3) maser. Compared to othermasing transitions, NH (3,3) masers are exceedingly rare. Indeed, to our knowledge only 15 sourceswith NH (3,3) masers have been discovered outside of the Galactic Center (Table 1). Althoughour understanding of NH (3,3) maser excitation is incomplete, studies have found that they can beexcited by shocks resulting from SNR-MC interactions (McEwen et al. 2016) or energetic outflows (3,3) masers have fluxes of (cid:46) Table 1 . Known NH (3,3) MasersSource Flux (Jy) ReferenceDR21(OH) 0.260 Mangum & Wootten (1994)W51 0.230 Zhang & Ho (1995)NGC 6334 V 0.114 Kraemer & Jackson (1995)NGC 6334 I 0.482 Kraemer & Jackson (1995)IRAS 20126+4104 0.079 Zhang et al. (1999)G5.89-0.39 0.031 Hunter et al. (2008)G20.08-0.14N 0.191 Galv´an-Madrid et al. (2009)G23.33-0.30 9.7 Walsh et al. (2011)G30.7206-00.0826 5 Urquhart et al. (2011)G35.03+0.35 0.065 Brogan et al. (2011)G28.34+0.06 0.03 Wang et al. (2012)W51C 1.4 McEwen et al. (2016) Table 1 continued on next page
Hogge et al.
Table 1 (continued)
Source Flux (Jy) ReferenceW44 0.07 McEwen et al. (2016)G5.7-0.0 0.35 McEwen et al. (2016)G1.4-0.1 0.58 McEwen et al. (2016)
G23.33-0.30’s broad NH line emission is associated with a filamentary infrared dark cloud (IRDC)that resides within the giant molecular cloud (GMC) G23.0-0.4, a large ( ∼ ×
15 pc), massive( ∼ × M (cid:12) ), and dense ( ∼ cm − ) filamentary GMC (Su et al. 2015) that hosts multiplegenerations of high-mass star formation (Messineo et al. 2014). In particular, there are severalnearby SNRs projected against G23.0-0.4, two of which, G22.7-0.2 (Su et al. 2014) and W41 (Frailet al. 2013; Su et al. 2015), may be interacting with the GMC. Furthermore, W41 exhibits 20 cmcontinuum emission, two 1720 MHz OH maser candidates, and extended TeV emission coincidentwith or adjacent to G23.33-0.30. G23.33-0.30’s large peak H column density of N H = 1 . × cm − (Peretto et al. 2016) is similar to that of other high-mass IRDCs, which are thought to be theformation sites of high-mass stars and stellar clusters (Rathborne et al. 2006). The Co-OrdinatedRadio ’N’ Infrared Survey for High-mass star formation’s (CORNISH; Hoare et al. 2012) non-detection of an H ii region likely indicates that any high-mass stars forming within G23.33-0.30 arein an embedded pre-stellar or protostellar phase. Considering GMC G23.0-0.4’s potential involvementin a SNR-MC interaction, as well as G23.33-0.30’s potential capacity for high-mass star formation,both stellar outflows or a SNR-MC interaction remain viable explanations for the excitation of the (3,3) maser emission.To confirm the suspected NH (3,3) maser emission, determine its excitation conditions, and investi-gate the nature of the broad NH line widths, we have performed followup observations of G23.33-0.30that probe the filament at the < . (1,1) through (6,6) inversion lines, Atacama Compact Array(ACA) observations of SiO(5-4), CS(5-4), and 1.3 mm continuum, and Submillimeter Array (SMA)observations of CO(2-1), C O(2-1), and 1.3 mm continuum. In Section 2 we describe these obser-vations and the reduction of the data, in Section 3 we present the results, in Section 4 we analyzethe data, in Section 5 we discuss the analysis, and in Section 6 we provide our conclusions. OBSERVATIONS AND DATA REDUCTIONWe have observed G23.33-0.30 using the VLA, operated by the National Radio Astronomy Obser-vatory , the ACA, and the SMA. Table 2 provides a summary of the continuum and spectral linedata analyzed in this work and Sections 2.1, 2.2, and 2.3 describe the calibration and reduction ofthese data. We also display archival data from several surveys, which are summarized in Table 3. Table 2 . New Observations
Telescope Date Transition ν θ P θ maj × θ min ∆ V BW ∆ V chan σ noise (GHz) (arcmin) (arcsec) (km s − ) (km s − ) (mJy beam − )SMA 2016 Jun 20 CO(2-1) 220.39868 0.8 4 . × . − − C O(2-1) 219.56035 0.8 4 . × . − − · · · . × . · · · Table 2 continued on next page The Submillimeter Array is a joint project between the Smithsonian Astrophysical Observatory and the AcademiaSinica Institute of Astronomy and Astrophysics, and is funded by the Smithsonian Institution and the AcademiaSinica. The National Radio Astronomy Observatory (NRAO) is a facility of the National Science Foundation operatedunder cooperative agreement by Associated Universities, Inc.
Hogge et al.
Table 2 (continued)
Telescope Date Transition ν θ P θ maj × θ min ∆ V BW ∆ V chan σ noise (GHz) (arcmin) (arcsec) (km s − ) (km s − ) (mJy beam − )VLA-A 2016 Oct 7 NH (3,3) 23.87013 1.9 0 . × .
09 50 0.39 8.1VLA-D 2017 Apr 14 NH (1,1) 23.69450 1.9 3 . × . − − NH (2,2) 23.72263 1.9 3 . × . − − NH (3,3) 23.87013 1.9 3 . × . − − NH (4,4) 24.13942 1.9 3 . × . − − NH (5,5) 24.53299 1.8 3 . × . − − NH (6,6) 25.05603 1.8 3 . × . . × . − − · · · . × . · · · − . × . − − · · · . × . · · · Note — ν is the rest frequency of the spectral line, θ P is the full width at half maximum (FWHM) size of the primary beam, θ maj × θ min is the FWHM size of the synthesized beam, ∆ V BW is the spectral bandwidth, and ∆ V chan is the spectral resolution. Table 3 . Archival Data
Survey Telescope Wavelength/Energy Spatial Resolution ReferenceGLIMPSE
Spitzer µ m < (cid:48)(cid:48)
1, 2GLIMPSE
Spitzer µ m < (cid:48)(cid:48)
1, 2MIPSGAL
Spitzer µ m 6 (cid:48)(cid:48) Table 3 continued on next page Table 3 (continued)
Survey Telescope Wavelength/Energy Spatial Resolution ReferenceMAGPIS VLA 20 cm 5 . (cid:48)(cid:48) × . (cid:48)(cid:48) (cid:48) × (cid:48) CO(1-0) - 2.7 mm 46 (cid:48)(cid:48) . −
100 TeV ∼ . ◦ Note —References: (1) Benjamin et al. 2003; (2) Churchwell et al. 2009; (3) Carey et al. 2009; (4) Helfand et al.2006; (5) Stil et al. 2006; (6) Jackson et al. 2006; (7) Aharonian et al. 2006.
VLA Observations
We observed G23.33-0.30 using the VLA in the D array configuration for a seven hour track. Weperformed the bandpass and flux calibration of the data using observations of J1331+305 (3C286)and we performed the phase calibration using periodic observations of J1851+0035. The calibrationand imaging of the data were performed using CASA clean algorithm with Briggs weighting and the robustness parameter set to 0.5.G23.33-0.30 was also recently observed by Killerby-Smith (2018) using the VLA in the A arrayconfiguration, which only detected NH (3,3) emission. The A array observation used the same https://casa.nrao.edu/ Hogge et al. calibrators as the D array, but the bandpass/flux calibrator was partially resolved by the A array’slong baselines. Consequently, the fluxes measured from the A array data are lower limits. Thespectral band for the A array observations was shifted to lower V LSR compared to the D array, sothe D array observations were able to detect emission up to V LSR = 90 km s − , while the A arrayobservations could only detect emission up to V LSR = 75 km s − . The A array data were also reducedusing CASA 5.1.1-5 and were imaged using CASA’s clean algorithm with natural weighting.2.2. ACA Observations
We observed G23.33-0.30 with the ACA using three pointings and five 50 min execution blocksin two spectral setups. The first spectral setup was used to observe SiO(5-4) and executed fourtimes and the second spectral setup was used to observe CS(5-4) and was executed once. For bothspectral setups we performed the bandpass calibration using observations of J1924-2914 and the phasecalibration using periodic observations of J1743-0350. For the first spectral setup we performedthe flux calibration using observations of J1733-1304 and for the second spectral setup we usedobservations of J1751+0939. The ACA data were calibrated by the ALMA data reduction pipelineand imaged using CASA 4.7.2. We carried out imaging using CASA’s tclean algorithm with Briggsweighting and the robustness parameter set to 0.5.2.3.
SMA Observations
We observed G23.33-0.30 using the SMA for an eight hour track in the compact configuration.We performed the bandpass calibration using observations of 3C454.3, the flux calibration usingobservations of Neptune, and the phase calibration using periodic observations of 1743-038. Wecalibrated the data using MIR , an IDL-based data reduction software package, and converted the format for imaging using the mir2miriad procedure. We imaged thedata with MIRIAD 4.3.8 using MIRIAD’s clean algorithm with Briggs weighting and the robustnessparameter set to 1. RESULTSFigure 2 shows the 8 µ m Galactic Legacy Infrared Midplane Survey Extraordinaire (GLIMPSE;Benjamin et al. 2003; Churchwell et al. 2009) image of G23.33-0.30 with the VLA NH (2,2) integratedintensity overlaid as contours. Clearly, the thermal NH (2,2) emission traces the IRDC filament, whilea gap in the 8 µ m extinction ( l = 23 . ◦ ) corresponds to a gap in the NH (2,2) emission. The ∼ (cid:48)(cid:48) spatial resolution of the D array has resolved the bright, narrow NH (3,3) line at V LSR ∼
56 km s − detected by HOPS and RAMPS into two point-like sources, while also revealing another potentialmaser at V LSR ∼
76 km s − , < (cid:48)(cid:48) away from the other two sources. The VLA A array observationsof G23.33-0.30 (Killerby-Smith 2018) resolved one of the maser candidates detected by the D arrayinto two sources. The positions of the four maser candidates are shown with symbol markers inFigure 2, which also displays the NH (2,2) and NH (3,3) spectra toward each source. The spectrashow that the suspected NH (3,3) maser emission is much narrower and brighter than the thermalNH (2,2) and (3,3) emission. Three of the maser candidates have velocities near the peak of thenarrowest NH (2,2) component at 56 km s − , while the faintest is found near a component peakingat 77 km s − .Figure 3 illustrates the unusual kinematics in G23.33-0.30. The left panels show the NH (2,2),CS(5-4), and SiO(5-4) 1 st moment maps of the filament, with their respective integrated intensitycontours overlaid, while the right panels show spectra taken across the width of the filament. TheNH (2,2) and CS(5-4) data reveal a velocity discontinuity between a broad line component peaking Hogge et al. I n t e n s i t y [ J y b e a m − ] NH (2,2)NH (3,3)NH (3,3)) NH (2,2)NH (3,3)NH (3,3))
60 80 0.000.010.020.03 NH (2,2)NH (3,3)NH (3,3))
50 60 70 80V
LSR [km s −1 ] 0.000.010.020.03 NH (2,2)NH (3,3)NH (3,3)) Ga actic Longitude -00.31°-00.30°-00.29° G a l a c t i c L a t i t u d e N Figure 2 . Left: GLIMPSE 8 µ m map of G23.33-0.30 with VLA NH (2,2) integrated intensity contoursoverlaid at 10, 50, and 150 mJy beam − km s − . The VLA D Array beam is shown in the lower left cornerof the map and the arrow in the upper right corner points to the north celestial pole. The green rectangleshows the region used to make the position-velocity diagram in Figure 6. The symbol markers indicate thepositions of the NH (3,3) maser candidates. The thermal NH emission traces an IRDC filament. Right:VLA D array NH (2,2) and (3,3) spectra toward the four maser candidates, where the symbol markers inthe upper left of each plot correspond to those in the GLIMPSE 8 µ m map. The NH (3,3) spectra arepresented both at their true amplitude and scaled for comparison. at V LSR ∼
60 km s − that is associated with the left edge of the filament and a narrower componentat V LSR ∼
77 km s − that is associated with the right side. The NH (2,2) data exhibit an additionalnarrow velocity component at V LSR ∼
56 km s − , peaking near the velocities of three of the masercandidates. The SiO(5-4) emission is extremely broad and peaks primarily at V LSR <
76 km s − .With the ACA’s large spectral bandwidth, we are also able to detect emission near V LSR = 100 kms − . We do not show this emission in any figures, since it is associated with two background sources −0.305−0.300−0.295 G a l a c t i c L a t i t u e NH (2,2) −0.305−0.300−0.295 G a l a c t i c L a t i t u e CS(5,4) G a l a c t i c L a t i t u e SiO(5,4) V L A I n t e n ) i t , [ J , b m − ] NH (2,2)CS(5,4)SiO(5,4)0.000.010.020.03 V L A I n t e n ) i t , [ J , b m − ] V L A I n t e n ) i t , [ J , b m − ] V L A I n t e n ) i t , [ J , b m − ]
50 60 70 80V
LSR [km ) −1 ]0.000.010.02 V L A I n t e n ) i t , [ J , b m − ] ) t M ( m e n t [ k m ) − ] ) t M ( m e n t [ k m ) − ] ) t M ( m e n t [ k m ) − ] A C A I n t e n ) i t , [ J , b m − ] A C A I n t e n ) i t , [ J , b m − ] A C A I n t e n ) i t , [ J , b m − ] A C A I n t e n ) i t , [ J , b m − ] A C A I n t e n ) i t , [ J , b m − ] Figure 3 . Left: NH (2,2) (top), CS(5-4) (middle), and SiO(5-4) (bottom) 1 st moment maps overlaid withintegrated intensity (moment 0) contours at 0.05 and 0.15, 2 and 8, and 1 and 4 Jy beam − km s − ,respectively. The symbols overlaid correspond to the locations of the spectra to the right. Right: NH (2,2)(black), CS(5-4) (magenta), and SiO(5-4) (orange) spectra from the positions indicated by the symbolmarkers in the left panels. The left axis corresponds to the NH (2,2) spectra, while the right axis correspondsto the CS(5-4) and SiO(5-4) spectra. These data reveal a velocity discontinuity across the width of thefilament. Hogge et al. unrelated to G23.33-0.30. G a l a c t i c L a t i t u d e NH (1,1) NH (2,2) NH (3,3) G a l a c t i c L a t i t u d e NH (4,4) NH (5,5) NH (6,6) s t M m e n t [ k m s − ] s t M m e n t [ k m s − ] Figure 4 . NH (1,1) (top left), (2,2) (top middle), (3,3) (top right), (4,4) (bottom left), (5,5) (bottommiddle), and (6,6) (bottom right) 1 st moment maps overlaid with integrated intensity contours at 25, 150,and 250 mJy beam − km s − . The silver symbol markers indicate the positions of the NH (3,3) masers andthe synthesized beam for each transition is shown in the lower left of the map. G23.33-0.30 exhibits bright emission from all of the observed NH inversion transitions. Figure 4displays the NH (1,1) through (6,6) 1 st moment maps with integrated intensity contours overlaid.The linear features in the NH (3,3) map are a result of cleaning artifacts, not real emission. Theemission from the higher energy transitions (NH (4,4) through (6,6)) is strongest at lower V LSR anddisplays similarly broad line widths as the 60 km s − NH (2,2) velocity component. The highestenergy transition, NH (6,6), features particularly bright emission compared to NH (5,5), and evendisplays amplitude ratios of NH (6,6)/(4,4) > (6,6) emission. In contrastto the NH (1,1) and (2,2) data, the NH (4,4) through (6,6) emission peaks only at V LSR ≤
76 km − , but exhibits a velocity discontinuity between components peaking at V LSR ∼
59 and 75 km s − . −0.305−0.300−0.295 G a l a c t i c L a t i t u d e G a l a c t i c L a t i t u d e I n t e n s i t y [ J y b e a − ] MM1 CO(2-1)C O(2-1)012 I n t e n s i t y [ J y b e a m − ] MM250 60 70 80V
LSR [k s −1 ]01 I n t e n s i t y [ J y b e a − ] MM301020 I n t e g r a t e d I n t e n s i t y [ J y b e a − k s − ] I n t e g r a t e d I n t e n s i t y [ J y b e a − k s − ] Figure 5 . Left: Color shows the SMA CO (top) and C O (bottom) integrated intensity maps. Overlaidare 1.3 mm continuum contours from the ACA (cyan) at 4 and 30 mJy beam − and the SMA (green) at 3.5,7, and 10.5 mJy beam − . The lowest contours correspond to a 5 σ detection in both cases. The silver symbolmarkers indicate the positions of the masers. The continuum data reveal three compact cores associatedwith the filament: MM1, MM2, and MM3. Right: SMA CO (black) and C O (orange) spectra towardMM1, MM2, and MM3. The brightest line in each spectrum peaks at V LSR ∼
77 km s − . Figure 5 shows the SMA CO(2-1), C O(2-1), 1.3 mm continuum, and ACA 1.3 mm continuumdata. Due to the ACA’s good uv coverage at shorter baselines, the continuum emission detectedby the ACA traces the larger scale filament, while the SMA continuum observations are primarily6 Hogge et al. sensitive to the compact continuum cores. Three of the continuum cores, MM1, MM2, and MM3,lie along the filament. Because the bright mm core west of the filament is coincident with emissionnear V LSR = 100 km s − , it is not associated with G23.33-0.30. The SMA CO(2-1) and C O(2-1)emission peaks at V LSR ∼
77 km s − between MM1 and MM2. The CO(2-1) and C O(2-1) spectrapeak at V LSR ∼
77 km s − toward each of the cores associated with G23.33-0.30. There is also CO(2-1) emission at lower V LSR appearing mainly on the eastern edge of the filament in the range V LSR = 56 −
70 km s − .
50 60 70 80V
LSR [km s −1 ]20"-8 ∘ D e c ∘ ( J ) I n t .∘ I n t e n s i t ∘ [ J ∘ b m − a r c s e c ] Figure 6 . Plot of Declination vs. V LSR made from the NH (2,2) intensity integrated over the Right As-cension axis. The region over which we performed the integration is shown with a green box in Figure 2.Because G23.33-0.30 is oriented roughly north to south, this displays the gas kinematics along the filament’slength. The cyan symbol markers indicate the positions and velocities of the NH (3,3) masers and the whitehorizontal lines mark the positions of the molecular cores MM1, MM2, and MM3. Figure 6 displays the NH (2,2) intensity integrated along lines of constant Right Ascension. Because (3,3) masercandidates and the positions of the continuum sources MM1, MM2, and MM3. While the vastmajority of the emission is at V LSR <
78 km s − and displays broad line widths, a small portion ofthe filament between MM2 and MM3 exhibits narrow line emission peaking at V LSR ∼
78 km s − . Thisnarrow line emission corresponds to the southernmost section of the filament detected in NH (2,2),but not SiO(5-4). The gas north of this narrow line emission is at lower V LSR and composed of twocomponents: a narrow component at V LSR ∼
56 km s − that is spatially and spectrally coincident withthree of the NH (3,3) maser candidates (Fig. 2), and a turbulent component peaking at V LSR = 60 − − . The section of the filament associated with the NH (3,3) maser candidate at V LSR = 76 . − is less turbulent and less blueshifted than the northern part of the filament. Figure 6 also revealsslight deficits in emission in the range 67 < V LSR <
75 km s − at the positions of MM1 and MM2,while the NH emission toward MM3 reveals a velocity discontinuity between components peakingat V LSR = 74 km s − and V LSR = 78 km s − . ANALYSIS4.1. NH (3,3) Maser Emission We determined the position and flux of each NH (3,3) maser candidate by first fitting spectrato determine line amplitudes and then fitting these amplitude maps to estimate the positions. Weused the Python Markov Chain Monte Carlo (MCMC) fitting package emcee (Foreman-Mackey etal. 2013) to fit both the narrow line emission in the data cube and the resulting maps of the lineamplitudes. We fit the narrow line emission with a Gaussian model and fit the amplitude maps witha model of the synthesized beam, since we expect maser emission to be unresolved. Table 4 presentsthe A array fit results for the sources peaking at 56 −
57 km s − and the D array fit results for the8 Hogge et al. source peaking at 76 km s − , as well as the symbol markers corresponding to each source shown inFigures 2, 4, 5, and 6. We display the best-fit values for each source’s Galactic coordinate position( l, b ), flux ( I ν ), LSR velocity ( V LSR ), velocity dispersion ( σ ), and line brightness temperature (∆ T B ). Table 4 . VLA NH (3,3) Maser Properties Maser 1 Maser 2 Maser 3 Maser 4Symbol (cid:35) (cid:52) (cid:50) (cid:51)
Array A A A D l (deg) 23.325713 ± ± ± ± b (deg) -0.303063 ± . ± ± ± I ν (Jy) 1 . ± .
086 0 . ± .
049 0 . ± .
033 0 . ± . V LSR (km s − ) 56 . ± .
002 57 . ± .
007 55 . ± .
008 76 . ± . σ (km s − ) 0 . ± .
003 0 . ± .
007 0 . ± .
005 0 . ± . T B (K) 171300 ± ± ± . ± . Maser emission occurs in gas with a population inversion, with more molecules in the upper state ofa transition than expected in local thermodynamic equilibrium (LTE). Population inversions in theNH (3,3) transition are a result of collisions with H molecules (Walmsley & Ungerechts 1983). Thebrightness temperature of a spectral line is given by ∆ T B = ( T ex − T bkg )(1 − e − τ ν ), where T ex is thetransition’s excitation temperature, T bkg is the background temperature, and τ ν is the optical depth ν . If the molecular transition is in LTE with gas at temperature T gas and the backgroundradiation is dominated by the Cosmic Microwave Background (CMB), then ∆ T B ≤ ( T gas − T CMB ).On the other hand, a nonthermal population inversion produces negative values of τ ν and T ex , result-ing in ∆ T B (cid:29) T gas . Given the low temperatures expected in G23.33-0.30 ( T gas (cid:28) K), the threemasers detected by the A array all exhibit ∆ T B (cid:29) T gas , confirming their nonthermal nature. Becausethe source at V LSR = 76 km s − was outside of the velocity range of the A array data, the D arraydata provide our only measurement of its brightness temperature. This source is much fainter thanthe NH (3,3) masers at V LSR ∼ −
57 km s − , exhibiting a D array brightness temperature of only17 K, which is comparable to the temperatures typically measured in molecular clouds. Althoughthe D array brightness temperature cannot prove this source’s nonthermal origin, it is approximatelytwice as bright as the peak thermal emission and exhibits a narrow line width, similar to those of theconfirmed masers. Moreover, the NH (3,3) maser candidate has no corresponding velocity featurein any of the other NH spectra. Consequently, we assume that the emission is nonthermal and willrefer to this source as an NH (3,3) maser. Thus, G23.33-0.30 hosts four NH (3,3) masers: threeassociated with gas corresponding to the narrow NH (2,2) velocity component near 57 km s − , anda fourth associated with the asymmetric line emission peaking at 77 km s − .4.2. Thermal NH Emission
In addition to accelerating gas, shocks can heat and add turbulence to the entrained gas component.Using the NH modeling methods described by Hogge et al. (2018), we employed a PySpecKit(Ginsburg & Mirocha 2011) LTE NH model to investigate whether the gas properties in G23.33-0.30 indicate a shock. We first derived NH rotational temperatures ( T rot ), velocity dispersions ( σ ),and LSR velocities ( V LSR ) from the NH (1,1) and (2,2) data cubes. In order to exclude emissionthat would be too faint to provide accurate derived quantities, we only fit pixels that had NH (2,2)0 Hogge et al. integrated intensities greater than 10 mJy beam − km s − . Figure 7 displays maps of the best-fit parameter values for the velocity component with the larger NH (1,1) through (2,2) integratedintensity. The maps are overlaid with the positions of the NH (3,3) masers, which we expect to resideat the locations of shock fronts. Although the presence of broad, overlapping, and asymmetric lineshapes did not allow for accurate fit results over the full map, it is clear that the gas at higher V LSR is generally colder and has a lower velocity dispersion than that at lower V LSR . The southernmostsection of the filament, which has the largest measured V LSR , was detected in NH (2,2) but notSiO(5-4). It also corresponds to the region that is coldest and has the lowest velocity dispersion,likely indicating that the gas in this portion of the filament is unshocked and that V LSR ∼
78 km s − isthe filament’s pre-shock LSR velocity. While our analysis of the NH (1,1) and (2,2) data implied onlymoderate heating of the turbulent component (∆ T ∼ −
40 K), the relatively bright emission fromthe higher energy transitions NH (4-4) through (6,6) indicates the presence of a hotter component towhich the NH (1,1) and (2,2) amplitudes are insensitive. Consequently, we also performed fits usingall of the observed para-NH lines: NH (1,1), (2,2), (4,4), and (5,5). However, the best-fit modelsof the para-NH lines often featured NH (1,1) amplitudes that were smaller than those observedin the data and NH (2,2) amplitudes larger than observed in the data. The fact that a single-temperature NH model could not reproduce the para-NH amplitudes suggests that there exists atleast two temperature components. Thus, we performed fits using only NH (4,4) and (5,5) to betterdetermine the temperature of the hotter component. The rotational temperatures based only on theNH (4,4) and (5,5) emission are T rot (4 ,
4; 5 , ∼ −
200 K, much higher than those derived fromNH (1,1) and (2,2). Thus, the shock has deposited significant thermal ( T pre − shock ∼ −
20 K vs. T post − shock ∼ −
200 K) as well as turbulent ( σ pre − shock < − vs. σ post − shock ∼ − − )energy into the filament. G a l a c t i c L a t i t u d e T r o t [ K ] σ [ k m s − ] V L S R [ k m s − ] Figure 7 . Maps of the NH model best-fit values of T rot , σ , and V LSR with the positions of the NH (3,3)masers overlaid. The gas at the pre-shock velocity is cold and has a low velocity dispersion, while theturbulent component is hot and has σ > − . Virial Analysis
We detected the 1.3 mm dust continuum emission from three compact sources associated with thefilament, MM1, MM2, and MM3 (Fig. 5), which represent molecular cores embedded within G23.33-0.30. We used an MCMC routine to fit the positions and sizes of the cores. The compact continuumemission from these cores is superposed on more extended emission. Consequently, we modeled theSMA continuum emission as the superposition of two elliptical Gaussians convolved with the SMAsynthesized beam and estimated their positions ( l, b ) and sizes ( R × R ). This analysis implies thatMM1 and MM2 are at most barely resolved by the SMA synthesized beam, so their best-fit sizes areupper limits. Table 5 displays the fit results.The collapse of a molecular core depends on its turbulent energy content. Due to the added supportagainst gravity, gas that is highly turbulent will have more difficulty collapsing to form stars thanwill gas with lower levels of turbulence. Thus, the line of sight velocity dispersion, along with thecore masses, can inform whether the detected cores can collapse to form stars. In order to evaluatethe fate of G23.33-0.30 and its associated molecule cores, we performed a virial analysis using theSMA continuum data and CO(2-1) spectra. Neglecting magnetic fields and external pressure, the2
Hogge et al. gravitational stability of a molecular core is dictated by the core mass ( M core ) and the virial mass( M vir ) in the form of the virial parameter α = M vir M core , where α < M core using the following equation from Hildebrand (1983) M core = R F ν D κ ν B ν ( T ) , (1)where R is the gas-to-dust mass ratio, F ν is the source flux integrated within the FWHM boundary ofthe best-fit elliptical Gaussian, D is the distance to the source, κ ν is the dust opacity, and B ν ( T dust ) isthe Planck function at the dust temperature T dust . We assumed R = 100 and κ . = 0 . g − ,which is the opacity expected for dust with thin ice mantles at a number density of 10 cm − (Ossenkopf & Henning 1994). Because the values of R and κ . are uncertain, we assume a 30%uncertainty on our assumed values. We adopted D = 4 . +0 . − . kpc, the parallax distance to thehigh-mass star-forming region G23.01-0.41 that also resides within GMC 23.0-0.4 (Brunthaler et al.2009). To estimate the dust temperature, we analyzed Herschel sub-mm data (Molinari et al. 2010)using a single temperature graybody model and the methods and assumptions described by Guzm´anet al. (2015). We calculated dust temperatures in the range of T dust ∼ −
20 K and adopted T dust = 18 ± ∼ (cid:48)(cid:48) ). We measured F ν within the FWHM boundary of the best-fitmodels. Next we determined M vir using an equation given by MacLaren et al. (1988), M vir = 3 (cid:18) − n − n (cid:19) Rσ G , (2)where n specifies the density distribution ( ρ ( r ) ∝ r − n ), R is the radius of the core, σ is the line-of-sight velocity dispersion, and G is the gravitational constant. We assumed that n = 1 .
8, theaverage value for the radial density profile index found by Zhang et al. (2009) in IRDC cores. Weestimated R from the geometric mean of the best-fit core radii R = √ R R and measured σ fromthe SMA CO(2-1) spectra toward the best-fit locations of the cores. We also detected C O(2-1) σ (C O) ≈ σ ( CO). Because the continuum data provide nokinematic information, the association between the continuum emission from a molecular core and aparticular velocity component is sometimes ambiguous. Although we detected no obvious compactmolecular line emission associated solely with the continuum cores, the SMA CO emission peakingat V LSR ∼
77 km s − suggests that the cores are associated with the pre-shock velocity component.Consequently, we have measured σ from the bright velocity component at V LSR ∼
77 km s − . Wedisplay the results of our virial analysis in Table 5. Although we have only determined upper limitson the virial parameter for MM1 and MM2 due to the unresolved core radii, the upper limit radiiare similar to the expected size of pre-stellar molecular cores (Ward-Thompson et al. 1999), so thetrue virial parameters are unlikely to be much smaller than our upper limit estimates. Our analysisindicates that all of the cores embedded within G23.33-0.30 have α >
1, implying that they are notcurrently unstable to collapse.
Table 5 . Molecular Core PropertiesMM1 MM2 MM3 l (deg) 23.32384 ± ± ± b (deg) -0.30314 ± ± ± R (arcsec) 2 . ± . . ± . . ± . R (arcsec) < . < . . ± . R (pc) < . < .
035 0 . ± . Table 5 continued on next page Hogge et al.
Table 5 (continued)
MM1 MM2 MM3 F ν (mJy) 19.1 ± ± ± M (M (cid:12) ) 9 . ± . . ± . . ± . σ ( CO) (km s − ) 0 . ± .
07 0 . ± .
02 0 . ± . σ (C O) (km s − ) · · · . ± .
09 0 . ± . M vir (M (cid:12) ) < <
66 26 ± α < . < . . ± . We also investigated the stability of the larger-scale filament by comparing G23.33-0.30’s mass tothe mass expected for a collapsing filament. We estimated G23.33-0.30’s mass using
Herschel sub-mm data and the methods and assumptions described in Guzm´an et al. (2015). We calculated agas mass of M ∼
600 M (cid:12) for the ∼ × . M/l ∼ M (cid:12) pc − . For a typical dense molecular clump(pc size scale), the minimum mass required to form an 8 M (cid:12) star is ∼
260 M (cid:12) (Sanhueza et al. 2017),assuming a star formation efficiency of 30% and a Kroupa (2001) initial mass function. Althoughthis places G23.33-0.30 in the category of potentially high-mass star-forming filaments, the filament’shighly blueshifted and turbulent gas make this less certain. The critical linear mass density abovewhich a molecular filament is unstable to collapse is given by (
M/l ) crit = 84(∆ V ) M (cid:12) pc − (Jacksonet al. 2010), where ∆ V is the FWHM line width. Although the velocity dispersion varies throughoutG23.33-0.30, a typical value is σ ∼ − , which corresponds to ∆ V ∼ − . This value M/l ) crit ∼ M (cid:12) pc − , much larger than our estimateof G23.33-0.30’s linear mass density. Thus, like G23.33-0.30’s embedded cores, the filament is notmassive enough to collapse given the turbulent gas. DISCUSSION5.1.
Evidence of a Large-Scale Shock
Although NH (3,3) maser emission typically indicates shocked gas, the positions and velocitiesof the masers alone cannot distinguish between the protostellar outflow and SNR-MC interactionscenarios. On the other hand, the sharp velocity discontinuity across the width of the filament,which corresponds to an increase in temperature and velocity dispersion, implies a large-scale shockand greatly favors a SNR-MC interaction scenario. Moreover, the SiO(5-4) emission, which tracesshocked gas (Caselli et al. 1997), is extended throughout the filament (Fig. 3). While shocks fromprotostellar outflow could be consistent with the NH (3,3) maser emission, the resulting SiO emissionwould be confined to narrow outflow jets emanating from the continuum cores. Since protostellaroutflows cannot account for such extended SiO emission, it is more likely that a large-scale shockfrom a SNR is responsible.Figure 8 shows plots of T rot (1,1; 2,2) and σ (1,1; 2,2) vs. V LSR and T rot (4,4; 5,5) and σ (4,4; 5,5)vs. V LSR . The NH (4,4) and (5,5) emission is clearly more sensitive to the hot gas component thanthe NH (1,1) and (2,2) emission. The plots of T rot (1,1; 2,2) and σ (1,1; 2,2) vs. V LSR show that thegas at the pre-shock velocity ( V LSR = 77 −
78 km s − ) is generally colder and has a lower velocitydispersion than the turbulent component at lower V LSR , but the correlations among the parametersis not particularly tight. On the other hand, T rot (4,4; 5,5) and σ (4,4; 5,5), which are sensitive to theshocked component, exhibit a more coherent relationship with V LSR . The emission at V LSR <
60 kms − displays positive correlations among T rot (4,4; 5,5), σ (4,4; 5,5), and V LSR , while the emission at6
Hogge et al.
60 70 80V
LSR [km s −1 ]1020304050 T r o t ( , ; , ) [ K ]
60 70 80V
LSR [km s −1 ]024 σ ( , ; , ) [ k m s − ]
60 70V
LSR [km s −1 ]50100150 T r o t ( , ; , ) [ K ]
60 70V
LSR [km s −1 ]024 σ ( , ; , ) [ k m s − ] Figure 8 . Plots of the NH model best-fit values of T rot and σ vs. V LSR , where the symbol color correspondsto the density of points. The upper plots display the parameter values derived from the NH (1,1) and (2,2)spectra, while the lower plots show the values derived from the NH (4,4) and (5,5) spectra. We show onlythe fit results that have parameter errors below the 75 th percentile and parameter values that are neitherpegged to their maximum nor minimum values. V LSR >
60 km s − displays negative correlations among these parameters. We speculate that thesetrends are a signature of the impact that triggered the NH (3,3) maser emission.In G23.33-0.30’s reference frame ( V LSR ∼
77 km s − ), the gas component associated with theNH (3,3) masers at V LSR ∼
56 km s − approaches the filament with a relative velocity of at least 20km s − . The impacting gas component forms a shock and excites maser emission as it interacts withthe filament. The impulse of the shock accelerates the filament gas, blueshifting it to lower V LSR , aswell as increasing its temperature and velocity dispersion. This shock acceleration of the filament T rot (4,4; 5,5), σ (4,4; 5,5), and V LSR for V LSR > − . Simultaneously, the impacting gas component is decelerated by the interaction with thefilament due to G23.33-0.30’s inertia. In the reference frame of the component at V LSR ∼
56 km s − ,it encounters the dense filament approaching quickly and is impacted, redshifting it to higher V LSR ,as well as increasing its temperature and velocity dispersion. This deceleration of the impactingcomponent due to G23.33-0.30’s inertia accounts for the positive correlations among T rot (4,4; 5,5), σ (4,4; 5,5), and V LSR for V LSR <
60 km s − . Consequently, the hot, highly turbulent component at V LSR = 60 −
68 km s − likely represents the turbulent wake of the shock (Pittard & Parkin 2016).Assuming the maser at V LSR = 76 . − represents genuine nonthermal emission, it likelysignals another shock front within the filament. This source is associated with gas at higher V LSR and is much fainter than the other NH (3,3) masers. Figure 6 suggests that the maser, MM1, andMM2 are all associated with a section of the filament that exhibits values of T rot , σ , and V LSR thatare intermediate between the turbulent component and the pre-shock component. Given that thisportion of the filament is situated between a highly shocked region and an unshocked region ofthe filament, we speculate that it is at an earlier stage of shock interaction than the hot, turbulentcomponent. The moderate temperature of this intermediately shocked component may in part explainthe faintness of the maser at V LSR = 76 . − . Figure 9 shows a non-LTE RADEX (van derTak et al. 2007) plot of the expected NH (3,3) optical depth ( τ (3 , ) and brightness temperature(∆ T B (3,3)) as a function of the gas kinetic temperature ( T k ) and number density ( n ) for the derivedbeam-averaged column density ( N NH = 10 . cm − ) and velocity dispersion ( σ = 1 . − ) ofthe intermediately shocked component. Large brightness temperatures and negative optical depths,which indicate strong masing, are only achieved for larger temperatures. At low temperatures, theRADEX model predicts that the gas should be either weakly masing or non-masing. Given the lowrotational temperature of the intermediately shocked component ( T rot (1,1; 2,2) ∼ −
25 K), the8
Hogge et al. lower ∆ T B (3,3) of the maser is expected. If the shock continues to heat this section of the filament,it is possible that the maser’s brightness temperature will increase.
100 200T K [K]10 n [ c m − ]
100 200T K [K]10 n [ c m − ] −3−2−10123 τ ( , ) Δ T B ( , ) Figure 9 . Plot of the RADEX NH (3,3) optical depth ( τ (3 , ) (top) and brightness temperature (∆ T B (3,3))(bottom) as a function of the kinetic temperature ( T k ) and number density ( n ) for a column density of N = 10 . cm − and a velocity dispersion of σ = 1 . − . The large rotational temperatures of the hot, turbulent component demonstrate that the shock hasheated the gas in the filament, but the NH (4,4) and (6,6) data also provide evidence of dust heating.NH modeling by Faure et al. (2013) suggests that ortho-NH ( K = 3 n ) forms preferentially overpara-NH ( K (cid:54) = 3 n ) on the surfaces of cold ( T <
30 K) dust grains. In addition, the ortho-NH ground state is at a lower energy than the para-NH ground state, resulting in a larger amount ofenergy needed to desorb para-NH than ortho-NH (Umemoto et al. 1999). Consequently, shocksthat heat the icy mantles of cold dust grains release ortho-enhanced NH into the gas phase and resultin an ortho-to-para abundance ratio (OPR) larger than the statistical equilibrium value of OPR = 1. (4,4) through (6,6) spectra toward the peak of theNH (6,6) emission. We calculated brightness temperature ratios of ∆ T B (5 , T B (4 , = 0 . ± . ∆ T B (6 , T B (5 , =2 . ± .
13, and ∆ T B (6 , T B (4 , = 1 . ± .
06. Figure 10 shows the NH (4,4) through (6,6) brightnesstemperature ratios predicted by RADEX for a range of temperatures and densities, assuming thecolumn density ( N = 10 . cm − ) and velocity dispersion ( σ = 1 .
75 km s − ) measured from theNH (4,4) and (5,5) spectra and OPR= 1. The temperatures and densities corresponding to ourmeasured value of ∆ T B (5 , T B (4 , are indicated by the white lines overlaid on each plot. If the turbulentgas component had OPR=1, the lines would intersect our measured values of ∆ T B (6 , T B (5 , and ∆ T B (6 , T B (4 , .However, our measured values for ∆ T B (6 , T B (5 , and ∆ T B (6 , T B (4 , are larger than any of the values on eitherplot, suggesting that OPR > (6,6)transition offers another possible explanation for the large ortho-to-para brightness ratios, the largevelocity dispersion of the NH (6,6) line ( σ ∼ . − ) make this unlikely. Thus, G23.33-0.30 isassociated with ortho-enhanced gas, implying that the shock has heated the filament’s cold dust andsublimated NH from their icy mantles.5.2. W41’s SNR-MC Interaction with GMC G23.0-0.4
Although molecular clouds can experience shocks due to protostellar jets or H ii regions, supernovashocks deliver a stronger impulse over a much shorter time span. Supernovae release roughly 10 ergs of energy nearly instantaneously, sending powerful shock waves over tens of parsecs. Despite thefact that only ∼ −
10% of the total energy is converted into kinetic energy in the shock (Walch& Naab 2015), the energy in a supernova shock can be sufficient to disrupt and disperse molecularclouds and cores. Using a simple E shock = M shocked ∆ V energy transfer analysis, the mass displaced0 Hogge et al.
50 100 150 200 250T K [K]10 n [ c m − ]
50 100 150 200 250T K [K]10 n [ c m − ]
50 100 150 200 250T K [K]10 n [ c m − ] Δ T B Δ , ) Δ T B Δ , ) Δ T B Δ , ) Δ T B Δ , ) Δ T B Δ , ) Δ T B Δ , ) Figure 10 . Plot of RADEX brightness temperature ratios as a function of the kinetic temperature ( T k ) andnumber density ( n ): ∆ T B (5 , T B (4 , (top), ∆ T B (6 , T B (5 , (middle), and ∆ T B (6 , T B (4 , (bottom). We assume a column densityof N = 10 . cm − and a velocity dispersion of σ = 1 .
75 km s − . The solid white line in each plot marksthe temperatures and densities corresponding to the ∆ T B (5 , T B (4 , = 0 . ± .
026 measured towards the peak ofthe NH (6,6) emission. The dashed lines mark the 1- σ uncertainties on the measured amplitude ratio. M shocked ∼ f kin E SN π ∆ V , where Ω is the molecular cloud’ssolid angle at its distance from the supernova, f kin is the kinetic efficiency of the SNR shock, E SN isthe total energy of the supernova, and ∆ V is the change in velocity of the shocked gas. Assumingthat E SN = 10 ergs, f kin = 5%, ∆ V = 10 km s − , and Ω is the solid angle of a spherical cloud witha radius of 0.5 pc radius at 5, 10, or 20 pc away from the supernova, the gas mass displaced by theshock is 130, 35, or 9 M (cid:12) , respectively.Given the uncertainty in the shock properties, the mass in the shocked portion of G23.33-0.30 iscomparable to the mass able to be displaced by a SNR shock, implying that a SNR is a plausiblesource for the shock that is accelerating the molecular filament gas. Considering the suggestionsof a SNR-MC interaction between SNR W41 and GMC G23.0-0.4 (Frail et al. 2013), W41 is anattractive progenitor for the large-scale shock impacting G23.33-0.30. A SNR shock would likelysupply enough energy to explain G23.33-0.30’s highly blueshifted emission, broad turbulent linewidths, and increased temperature and velocity dispersion. Thus, we speculate that the turbulent,blueshifted gas observed in G23.33-0.30 is the result of a large scale shock originating from the nearbySNR W41.Figure 11 shows the Multi-Array Galactic Plane Imaging Survey (MAGPIS; Helfand et al. 2006)20 cm continuum emission from W41 in color and the VLA Galactic Plane Survey (VGPS; Stil etal. 2006) 21 cm data as contours. W41 is an asymmetric shell-type supernova (Green 1991; Kassim1992) suspected of interacting with the nearby GMC G23.0-0.4 at V LSR = 77 km s − (Su et al.2015). Frail et al. (2013) detected two 1720 MHz OH maser candidates, known to trace SNR-MCinteractions (Wardle & Yusef-Zadeh 2002), coincident with W41’s central continuum peak. W41 isalso coincident with HESS J1834-087 (Aharonian et al. 2006; Albert et al. 2006), a source of TeVemission thought to be triggered by W41’s interaction with a GMC (Tian et al. 2007). Deep follow-up observations with H.E.S.S. revealed that the TeV emission is composed of a point-like source2 Hogge et al. and an extended component (H. E. S. S. Collaboration et al. 2015). Although the pulsar candidateCXOU J183434.9-08444 (Misanovic et al. 2011) may account for the point-like component of theTeV emission, H. E. S. S. Collaboration et al. (2015) argued that the extended TeV emission is bestexplained by the SNR-MC interaction.The MIR emission toward the W41 region from GLIMPSE and the MIPS Galactic Plane Survey(MIPSGAL; Carey et al. 2009) is shown in Figure 12, displaying a rich and complicated star-formingcomplex. These data reveal filamentary IRDCs and several sources of bright MIR emission, but pro-vide no kinematic information about the molecular gas near W41. In order to better understand thekinematics of the molecular gas in the W41 region, we analyzed the Galactic Ring Survey (GRS;Jackson et al. 2006) CO(1-0) data. The VLA NH data revealed three prominent velocity compo-nents: a narrow component at V LSR ∼
56 km s − that appears to be associated with three of theNH (3,3) masers, a broad, turbulent component at V LSR ∼
60 km s − , and a relatively asymmetricline peaked at V LSR ∼
77 km s − , which is associated with the NH (3,3) maser at V LSR ∼
76 kms − . We inspected the CO data for emission associated with these velocity components and createdmaps of the integrated intensity in 2 km s − windows centered on these velocities. We also foundanother distinct CO(1-0) velocity component peaking at 81 −
82 km s − , which is spatially andspectrally adjacent to the other emission. Figure 13 shows the GRS CO(1-0) data integrated inthe velocity ranges indicated in each panel, with the MAGPIS continuum overlaid for comparison.The lower velocity emission centered at 56 and 60 km s − displays emission near the positions ofW41’s central 20 cm continuum peak and G23.33-0.30. On the other hand, the emission at 77 kms − clearly traces GMC G23.0-0.4, but exhibits a conspicuous deficit in emission where W41’s centralcontinuum emission peaks. Finally, the 82 km s − component traces a MC, seen as a collection ofIRDC filaments in Figure 12, that could represent a nearby background cloud or some other compo-nent of the GMC. Tian et al. (2007) also noted the association between the lower velocity emission, G a l a c t i c L a t i t u d e I n t e n s i t y [ m J y b e a m − ] Figure 11 . Color shows the MAGPIS 20 cm continuum data. The white contours show the VGPS 21 cmcontinuum data at 30 K and the cyan contours show the VLA NH (2,2) integrated intensity data fromG23.33-0.30 at 50 mJy beam − km s − . The cyan circles coincident with the central 20 cm continuum peakindicate the positions of two 1720 MHz OH maser candidates, which have velocities near 75 km s − , andthe cyan triangle shows the position of the pulsar candidate CXOU J183434.9-08444. The large yellow circledisplays the fitted position and FWHM size of the extended component of HESS J1834-087’s TeV γ -rayemission. W41’s continuum, and HESS J1834-087, but they assumed that the emission represented a separateGMC with which W41 was interacting. If this were true, the emission from GMC G23.0-0.4 wouldlikely be uncorrelated with the central 20 cm continuum peak, rather than anti-correlated. Moreover,4
Hogge et al.
Figure 14 shows that the lower velocity emission at 56 and 60 km s − and the GMC emission at 77km s − also appear to be anti-correlated. In light of the agreement between the velocities of the1720 MHz OH maser candidates and the GMC, as well as the apparent anti-correlation between thelower velocity emission and the GMC emission, we argue that these lower velocity components areassociated with GMC G23.0-0.4 and the SNR-MC interaction.Frail et al. (2013) also detected an OH absorption feature at 76 km s − , which places W41 within orbehind G23.0-0.4. In addition, Leahy & Tian (2008) measured a maximum H i absorption velocity of78 ± − toward W41, indicating that W41’s progenitor may have formed within G23.0-0.4. TheReid et al. (2014) near kinematic distance for V LSR = 77 km s − toward G23.33-0.30 is 4 . +0 . − . kpcand the maser parallax distance to the nearby HMSFR G23.01-0.41 is 4 . +0 . − . kpc (Brunthaler et al.2009). Considering that the GRS CO(1-0) data strongly imply that G23.01-0.41 and G23.33-0.30both reside within GMC G23.0-0.4, we adopt the maser parallax distance for W41. W41’s angularsize of ∼ . ◦ ∼
40 pc. This size is in agreement withestimates from the Sedov relation (Sedov 1959), assuming a reasonable average density ( ∼ − )and SNR age ( ∼ × yr) (Tian et al. 2007).Given that these data are consistent with an interaction between SNR W41 and GMC G23.0-0.4,we further speculate that the CO(1-0) emission with V LSR = 60 −
75 km s − (not shown) in thevicinity of the W41’s central 20 cm continuum peak represents gas from G23.0-0.4 that has beenshock-accelerated to its current velocity. The V LSR = 56 km − component seems to correspond tothe component associated with the NH (3,3) masers in G23.33-0.30. This velocity component, whichpresumably represents gas moving with the largest velocity relative to G23.33-0.30, appears to bestreaming past and interacting with the filament. The existence of unshocked gas within G23.33-0.30, as well as CO(1-0) emission at the pre-shock velocity adjacent to the filament, implies thatthe interaction between SNR W41 and G23.33-0.30 is ongoing. This agrees with our interpretation G a l a c t i c L a t i t u d e Figure 12 . Color shows the MIPSGAL 24 µ m (red), GLIMPSE 8 µ m (green) and 3.6 µ m (blue) MIR data.For reference we show the MAGPIS 20 cm continuum data with contours at 3 mJy beam − , as well as thedata overlaid in Figure 11. of the interferometric data, which exhibits emission at V LSR = 77 −
78 km s − (pre-shock gas), V LSR = 60 −
75 km s − (turbulent shock wake), and V LSR = 56 −
57 km s − (gas entrained in shockfront). A caveat to this interpretation is that W41 lies near the plane of the Galaxy, so the line ofsight towards the SNR is crowded with molecular clouds at various velocities, which could confuse6 Hogge et al. -00.5°-00.4°-00.3°-00.2°-00.1° G a l a c t i c L a t i t u d e −1 −1 Galactic Longitude -00.5°-00.4°-00.3°-00.2°-00.1° G a l a c t i c L a t i t u d e −1 Galactic Longitude −1 Figure 13 . Color shows the GRS CO(1-0) integrated intensity over the ranges specified in each panel. Weshow the MAGPIS 20 cm continuum data with contours at 3 mJy beam − for reference. The cyan contoursshow the VLA NH (2,2) integrated intensity data from G23.33-0.30 at 50 mJy beam − km s − . our interpretation. In addition, feedback from previous generations of high-mass stars can accelerategas and create molecular gas structures that are physically close, but have different V LSR . Thus, itis possible that the molecular gas with V LSR = 60 −
75 km s − is not currently associated with theGMC, but is in fact a remnant of the larger GMC structure that is being impacted by the SNR shockfrom W41. On the other hand, it is difficult to imagine such a scenario producing the striking anti-correlation between the GMC velocity component and the V LSR = 60 km s − velocity component.Although more observations are needed to determine the true relationship between these velocity G a l a c t i c L a t i t u d e Figure 14 . Color shows the GRS CO integrated intensity over the range 55 −
57 km s − (blue), 59 − − (green), and 76 −
78 km s − (red). Symbols and contours as in Figure 12. components, a SNR-MC interaction seems to best explain the large turbulent line widths, heating,NH (3,3) masers, and blueshifted emission observed in G23.33-0.30.An alternative explanation for G23.33-0.30’s large turbulent line widths is energy added by nearbyH ii regions. Figure 15 shows a zoomed view of the MIR emission toward W41’s central continuum8 Hogge et al. peak. The overlaid circles show the positions and sizes of several nearby H ii regions and candidateH ii regions from the WISE
Catalog of Galactic H ii regions (Anderson et al. 2014). While it isclear that these H ii regions cannot account for all of the 20 cm continuum emission toward W41’scenter, it is likely that they contribute a portion of the emission. The red circles indicate H ii regionsthat have known velocities from radio recombination lines (RRLs): G023.250-00.240 ( V LSR = 76 . − ), G023.265-00.301a ( V LSR = 73 . − ), and G023.295-00.280 ( V LSR = 61 . − ).Although these velocities may indicate an association with the GMC or the 60 km s − CO(1-0)component, it is unlikely that these sources could account for the significant energy input impliedby G23.33-0.30’s large turbulent line widths given their angular separation from G23.33-0.30. Thesource with the smallest angular separation from G23.33-0.30, the candidate H ii region G023.317-0.300, lacks a reliable velocity from a RRL detection, but its position behind G23.33-0.30 makes itan attractive alternative source for G23.33-0.30’s turbulence. On the other hand, our interferometricdata exhibit broad SiO(5-4) and CS(5-4) line emission peaking at V LSR = 97 km s − near the center ofthe candidate H ii region, potentially signifying its association with the background source HMSFRG23.44-0.18, which has a maser parallax distance of 5 . +1 . − . kpc. In addition, the velocity dispersionof G23.33-0.30’s turbulent velocity component is σ shocked = 2 − − , while the velocity dispersionsmeasured towards H ii regions in the RAMPS dataset (Hogge et al. 2018) are at most σ (cid:46) − . Thus, an H ii region would need to input an unusually large amount of energy into G23.33-0.30to reproduce the measured velocity dispersion. Furthermore, NH (3,3) masers have previously onlybeen associated with SNR-MC interactions (McEwen et al. 2016) or energetic outflows from high-massprotostars (Mangum & Wootten 1994; Kraemer & Jackson 1995; Zhang & Ho 1995), so NH (3,3)maser emission resulting from an interaction with an H ii region would also be unusual. Consequently,G23.33-0.30’s high levels of turbulence are most likely the result of a SNR-MC interaction. G a l a c t i c L a t i t u d e Figure 15 . Color shows the MIPSGAL 24 µ m (red), GLIMPSE 8 µ m (green) and 3.6 µ m (blue) MIR data.The MAGPIS 20 cm continuum data are shown with white contours at 3 mJy beam − and the magentacontours show the VLA NH (2,2) integrated intensity data of G23.33-0.30 at 50 mJy beam − km s − . Thecircles show the nearby sources from the WISE
HII Region Catalog. The red circles indicate known HIIregions with RRL detections, the cyan circles indicate candidate HII regions that exhibit continue emissionbut lack RRL detections, and the yellow circles indicate radio quiet candidate HII regions, which exhibitthe MIR characteristics of an HII region but lack radio continue emission and RRL emission. Hogge et al.
W41’s Potential Impact Geometry
Figure 13 shows that the diameter of W41’s shell is much larger than the radial extent of theGMC. If W41’s interaction is ongoing and it exploded within G23.0-0.4, then W41’s plane of skydiameter must be larger than its size along the line of sight. Dense molecular gas can slow or evenstall the expansion of a SNR shock into a MC (Tatematsu et al. 1987). Given that MCs can exhibitasymmetric density profiles, core-collapse supernovae shell structures evolving in these environmentscan also display asymmetries (Lopez et al. 2009). While this is plausible, we might expect W41’sexpansion out of a dense, massive GMC to sweep up more molecular gas in its shell, whereas thissignature is absent in the GRS data. An alternative explanation is that W41’s progenitor formedwithin a MC ≤
20 pc away from G23.0-0.4. The 82 km s − MC shown in Figure 13 could potentiallybe this cloud. The Reid et al. (2014) near kinematic distance for V LSR = 82 km s − is 4 . +0 . − . kpc,consistent with our adopted distance to G23.33-0.30. Although Leahy & Tian (2008) measured amaximum H i absorption velocity of 78 ± − toward W41, this may not preclude W41 from alsobeing associated with a background MC. If W41 expanded out of this background MC, the shockmay have blueshifted much of the foreground molecular gas that was previously associated with thebackground MC, confusing the interpretation. Hence, the uncertainty in the origin point of W41’sexpansion leaves the impact geometry ambiguous.We have searched for redshifted gas corresponding to the back side of the expanding supernova shellwithin G23.0-0.4 to help differentiate between these two scenarios. We detected CS emission peakingat V LSR = 98 and 103 km s − and SiO emission peaking at V LSR = 96 km s − near G23.33-0.30, but itis unclear whether this emission is associated with G23.0-0.4 or the interaction. Thus, we examinedthe GRS CO data for signatures of a redshifted component potentially associated with the higher V LSR gas. Although we found emission spanning V LSR = 93 −
106 km s − toward the left half of W41, . +1 . − . kpc and V LSR = 97 ± − , placing it in the Norma arm nearthe end of the long bar (Brunthaler et al. 2009; Reid et al. 2014). The second background source isthe H ii region G23.42-0.21, which seems to contribute much of the continuum emission near W41’sleft edge (Fig. 12). G23.42-0.21 has a recombination line velocity of 103 km s − and a maximum H i absorption velocity of 106 km s − ; hence, Leahy & Tian (2008) argued that the absorption velocitywas significantly higher than the recombination velocity and that G23.42-0.21 must be assigned tothe far distance of ∼ . Hogge et al. of the GMC. Another alternative is that projection affects are confusing our interpretation of thefilament’s velocity structure. Regardless of the uncertain impact geometry, W41’s association withG23.33-0.30’s blueshifted gas component and large velocity dispersions is well founded.
Expanding into dense gas, blueshifted Expanding into diffuse gas Expanding into dense gas, redshifted
W41 inside the filament lb Figure 16 . Illustration of top-down view of the scenario in which W41 formed near the center of GMCG23.0-0.4. The figure shows the GMC represented as a blue cylinder, W41’s shock bubble as a peach ellipse,G23.33-0.30 as red line segments, and the position of W41’s progenitor as a black circle. The filament isrotated in the plane of the Galaxy, such that it roughly matches the orientation of the Reid et al. (2014)Scutum arm, in which the GMC may reside (Brunthaler et al. 2009). Although this model does not dependon the exact orientation of the GMC, some degree of rotation would be required in order to detect blueshiftedemission at the location of G23.33-0.30.
Possible Negative Feedback from SNR W41
Although shocks from supernovae may drive much of the turbulence in the ISM (Padoan et al.2016), the influence of these shocks on the star formation process is an open question. Simulations ofshock-cloud interactions that include dense substructures demonstrate that the more diffuse moleculargas is efficiently stripped from MCs, while such an impact forms a bow shock around a sufficiently Expanding into dense gasExpanding into diffuse gas lb W41 outside the filament
Figure 17 . Illustration of top-down view of the scenario in which W41 formed behind GMC G23.0-0.4. Themeaning of the shapes is given in the caption of Figure 16. dense molecular core (Patnaude & Fesen 2005). It is possible that the emission deficits in Figure 6at lower V LSR represent a wake formed behind the cores, which would suggest that some dense gasassociated with the cores will remain relatively unperturbed by the passage of the shock. On the otherhand, the virial parameters derived from the CO emission suggest that the cores are too turbulentto collapse. Furthermore, given that the molecular cores are associated with the intermediatelyshocked component, they may not yet have experienced as significant an impulse from the shock ascompared to the northern section of G23.33-0.30. Indeed, MM3, the core furthest from the turbulentcomponent, has the largest C O(2 − CO(2 − brightness ratio. This indicates that MM3 has the highest CO(2-1) optical depth, implying that it is also the densest of the cores. If W41’s shock passes through thefilament and continues to interact with the molecular cores, their velocity dispersions may increase,potentially resulting in the dispersal of the cores. Although we cannot determine whether W41’sshock is responsible for the turbulent velocity dispersions of the cores, our data are consistent withnegative feedback from the SNR.4 Hogge et al. CONCLUSIONG23.33-0.30 is a massive IRDC filament that exhibits broad molecular line widths and narrowNH (3,3) line emission. We have imaged the filament using the VLA (NH (1,1) through (6,6)), theSMA ( CO, C O, 1.3 mm continuum), and the ACA (SiO(5-4), CS(5-4), 1.3 mm continuum) andwe have drawn the following conclusions from our data:1. We have confirmed the nonthermal nature of three NH (3,3) masers that peak near V LSR = 56km s − and discovered a fourth NH (3,3) maser at V LSR = 76 km s − .2. The ACA observations revealed broad SiO(5-4) emission throughout the filament, indicatingthe presence of a highly turbulent and extended shocked gas component. Because protostellaroutflows cannot reproduce the observed SiO emission, high-mass protostellar outflows do notproduce G23.33-0.30’s NH (3,3) maser emission. The widespread nature of the SiO emissionindicates a shock acting on larger scales.3. The NH emission displays a velocity gradient along the length of the filament, with a significantportion of the filament apparently blueshifted by ∼ −
17 km s − with respect to the rest ofthe filament. G23.33-0.30 also exhibits a velocity discontinuity across the width of the filament,which separates the shocked, turbulent component from the pre-shock component.4. Our LTE NH model fitting to the NH (1,1), (2,2), (4,4), and (5,5) data has provided mapsof the NH rotational temperature ( T rot ), velocity dispersion ( σ ), and LSR velocity ( V LSR ),which show that the shocked component is hotter, more turbulent, and blueshifted comparedto the pre-shock component. The correlation among T rot , σ , and V LSR implies that the shock issimultaneously accelerating, heating, and injecting turbulent energy into the shocked filamentgas. In addition, the increased ortho-NH abundance inferred from the large ∆ T B (6 , T B (5 , and ∆ T B (6 , T B (4 , brightness temperature ratios imply dust heating. CO dataprovide additional evidence of this interaction, which suggests that W41’s shock is the commoncause for the observed gas kinematics on large and small scales.6. Although W41’s impact geometry remains ambiguous, its interaction with G23.33-0.30 and itsplane-of-sky diameter imply that it is at a distance of d ≤
20 pc from G23.33-0.30.7. The SMA 1.3 mm continuum data revealed dust cores embedded within G23.33-0.30. AlthoughG23.33-0.30 appears to have sufficient mass ( M ∼
600 M (cid:12) ) to form a high-mass star, theobserved gas kinematics suggest that the filament is presently being displaced, and potentiallydispersed, by the SNR shock. Likewise, our virial analysis ( α = 4 −
9) suggests that the coresare also unlikely to collapse. Thus, our data are consistent with negative feedback from theSNR. ACKNOWLEDGMENTSWe thank the anonymous referee for useful comments which improved the paper. This researchwas supported by the National Science Foundation Grant AST-1616635. The National Radio As-tronomy Observatory is a facility of the National Science Foundation operated under cooperativeagreement by Associated Universities, Inc. This paper makes use of the following ALMA data:ADS/JAO.ALMA
Hogge et al. has made use of NASA’s Astrophysics Data System.
Software:
PySpecKit (Ginsburg et al. http://doi.org/10.5281/zenodo.12490), APLpy (Robitailleand Bressert, 2012) REFERENCES
Aharonian, F., Akhperjanian, A. G., Bazer-Bachi,A. R., et al. 2006, ApJ, 636, 777Albert, J., Aliu, E., Anderhub, H., et al. 2006, ApJL,638, L101Anderson, L. D., Bania, T. M., Balser, D. S., et al.2014, ApJS, 212, 1Benjamin, R. A.,Churchwell, E., Babler, B. L., et al.2003, PASP, 115, 953Brogan, C. L., Hunter, T. R., Cyganowski, C. J., et al.2011, ApJL, 739, L16Brunthaler, A., Reid, M. J., Menten, K. M., et al.2009, ApJ, 693, 424Carey, S. J., Noriega-Crespo, A., Mizuno, D. R., et al.2009, PASP, 121, 76Caselli, P., Hartquist, T. W., & Havnes, O. 1997,A&A, 322, 296Churchwell, E.,Babler, B. L., Meade, M. R., et al.2009, PASP, 121, 213de Wilt, P., Rowell, G., Walsh, A. J., et al. 2017,MNRAS, 468, 2093Elitzur, M. 1992, ARA&A, 30, 75Faure, A., Hily-Blant, P., Le Gal, R., Rist, C., &Pineau des Forˆets, G. 2013, ApJL, 770, L2Foreman-Mackey, D., Hogg, D. W., Lang, D., &Goodman, J. 2013, PASP, 125, 306Frail, D. A., Claussen, M. J., & M´ehault, J. 2013,ApJL, 773, L19Galv´an-Madrid, R., Keto, E., Zhang, Q., et al. 2009,ApJ, 706, 1036Ginsburg, A., & Mirocha, J. 2011, Astrophysics SourceCode Library, ascl:1109.001Green, D. A. 1991, Publications of the AstronomicalSociety of the Pacific, 103, 209Guzm´an, A. E., Sanhueza, P., Contreras, Y., et al.2015, ApJ, 815, 130Helfand, D. J., Becker, R. H., White, R. L., Fallon, A.,& Tuttle, S. 2006, AJ, 131, 2525H. E. S. S. Collaboration, Abramowski, A., Aharonian,F., et al. 2015, A&A, 574, A27Hildebrand, R. H. 1983, QJRAS, 24, 267Hoare, M. G., Purcell, C. R., Churchwell, E. B., et al.2012, PASP, 124, 939Hogge, T., Jackson, J., Stephens, I., et al. 2018, ApJS,237, 27Hunter, T. R., Brogan, C. L., Indebetouw, R., &Cyganowski, C. J. 2008, ApJ, 680, 1271Jackson, J. M., et al. 2006, ApJS, 163, 145Jackson, J. M., Finn, S. C., Chambers, E. T., et al.2010, ApJL, 719, L185Kassim, N. E. 1992, AJ, 103, 943 Killerby-Smith, N. 2018, Review of the Rare NH3(3,3)Astronomical Masers (Callaghan, NSW, AUS:University of Newcastle)Kraemer, K. E., & Jackson, J. M. 1995, ApJL, 439, L9Kroupa, P. 2001, MNRAS, 322, 231Leahy, D. A., & Tian, W. W. 2008, AJ, 135, 167Lopez, L. A., Ramirez-Ruiz, E., Badenes, C., et al.2009, ApJL, 706, L106MacLaren, I., Richardson, K. M., & Wolfendale, A. W.1988, ApJ, 333, 821Mangum, J. G., & Wootten, A. 1994, ApJL, 428, L33McEwen, B. C., Pihlstr¨om, Y. M., & Sjouwerman,L. O. 2016, ApJ, 826, 189Messineo, M., Menten, K. M., Figer, D. F., et al. 2014,A&A, 569, A20Misanovic, Z., Kargaltsev, O., & Pavlov, G. G. 2011,ApJ, 735, 33Molinari, S., Swinyard, B., Bally, J., et al. 2010, PASP,122, 314Orlando, S., Peres, G., Reale, F., et al. 2005, A&A,444, 505Ossenkopf, V., & Henning, T. 1994, A&A, 291, 943Padoan, P., Pan, L., Haugbølle, T., & Nordlund, ˚A.2016, ApJ, 822, 11Patnaude & Fesen 2005, ApJ, 633, 240Peretto, N., Lenfestey, C., Fuller, G. A., et al. 2016,A&A, 590, A72Pittard, J. M., & Parkin, E. R. 2016, MNRAS, 457,4470Rathborne, J. M., Jackson, J. M., & Simon, R. 2006,ApJ, 641, 389Reach, W. T., Rho, J., & Jarrett, T. H. 2005, ApJ,618, 297Reid, M. J., Menten, K. M., Brunthaler, A., et al.2014, ApJ, 783, 130Sanhueza, P., Jackson, J. M., Foster, J. B., et al. 2012,ApJ, 756, 60Sanhueza, P., Jackson, J. M., Zhang, Q., et al. 2017,ApJ, 841, 97Sedov, L. 1959, Journal of Applied Mathematics andMechanics, 23, 568Stil, J. M., Taylor, A. R., Dickey, J. M., et al. 2006,AJ, 132, 1158Su, Y., Yang, J., Zhou, X., Zhou, P., & Chen, Y. 2014,ApJ, 796, 122Su, Y., Zhang, S., Shao, X., & Yang, J. 2015, ApJ, 811,134Tatematsu, K., Fukui, Y., Nakano, M., et al. 1987,A&A, 184, 279Tian, W. W., Li, Z., Leahy, D. A., & Wang, Q. D.2007, ApJL, 657, L2547