Ammonia observations of bright-rimmed clouds: establishing a sample of triggered protostars
aa r X i v : . [ a s t r o - ph . GA ] J un Mon. Not. R. Astron. Soc. , 1–10 (2010) Printed 2 November 2018 (MN L A TEX style file v2.2)
Ammonia observations of bright-rimmed clouds: establishing asample of triggered protostars
L. K. Morgan ⋆ , C. C .Figura , J. S. Urquhart and M. A. Thompson Astrophysics Research Institute, Liverpool John Moores University, Twelve Quays House, Egerton Wharf, Birkenhead CH41 1LD Wartburg College, 100 Wartburg Blvd. Waverley, IA 50677, USA Australia Telescope National Facility, CSIRO, Sydney, NSW 2052, Australia Centre for Astrophysics Research, Science and Technology Research Institute,University of Hertfordshire, College Lane, Hatfield AL10 9AB
Accepted ??. Received ??; in original form ??
ABSTRACT
We observed 42 molecular condensations within previously identified bright-rimmed cloudsin the ammonia rotational inversion lines NH (1,1), (2,2), (3,3) and (4,4) using the Green BankTelescope in Green Bank, West Virginia. Using the relative peaks of the ammonia lines andtheir hyperfine satellites we have determined important parameters of these clouds, includingrotational temperatures and column densities.These observations confirm the presence of dense gas towards IRAS point sources detectedat submillimetre wavelengths. Derived physical properties allow us to refine the sample ofbright-rimmed clouds into those likely to be sites of star formation, triggered via the processof radiatively-driven implosion. An investigation of the physical properties of our sourcesshow that triggered sources are host to greater turbulent velocity dispersions, likely indicativeof shock motions within the cloud material. These may be attributed to the passage of triggeredshocks or simply the association of outflow activity with the sources.In all, we have refined the Sugitani et al. (1991) catalogue to 15 clouds which are clearlystar-forming and influenced by external photoionisation-induced shocks. These sources maybe said, with high confidence, to represent the best examples of triggering within bright-rimmed clouds. Key words:
Stars: formation – Stars: early-type – Stars: pre-main sequence.
Bright-rimmed clouds (BRCs) are small molecular clouds asso-ciated with old ( > ff ⋆ E-mail: [email protected]
These shocks are driven into the clouds; compressing and heatingthe molecular gas, leading to the formation of dense cores and po-tentially causing their collapse (Bertoldi 1989; Lefloch & Lazare ff c (cid:13) L. K. Morgan et al.
Figure 1.
Multi-wavelength composite image of SFO 13, Spitzer IRACbands at 4.5 and 8.0 µ m are coloured green and blue respectively, a SpitzerMIPS 24 µ m image is incorporated in red. 21 cm NVSS contours are over-laid in white, beginning at a level of two times the r.m.s. noise ( σ ) of theradio image and increasing in units of 1 σ . SCUBA 850 µ m contours areoverlaid in yellow, beginning at a level of 6 σ and increasing in units of 3 σ . We have therefore designed a multi-wavelength programme ofobservations to investigate the current level of star formation takingplace within the whole sample of clouds, and to ascertain whetherthe interaction with the HII region has been a contributing factorin that formation. The programme includes the use of radio con-tinuum observations (Thompson et al. 2004) and NVSS archivaldata (Morgan et al. 2004) to trace the ionised gas associated withthe IBL, CO molecular line observations to probe the kinematicsof the protostellar cores (Urquhart et al. 2009; Morgan et al. 2009),and submillimetre observations to trace the dust associated withthe protostellar envelopes and obtain core temperatures and masses(Morgan et al. 2008). In addition to our programme of observationswe have obtained archival near- to far-infrared imaging and pho-tometry (2MASS, MSX, IRAS, Spitzer IRAC and MIPS) in an ef-fort to build up a detailed picture of the structure within BRCs anda census of the current state of star formation within them.A key element of our campaign is to answer the question ofwhether the observed star formation within BRCs has been trig-gered by the action of the OB star or is pre-existing and is simplybeing unveiled by the dispersal of the cloud. In order to addressthis question, we must first confirm the existence of protostellarcores within BRCs and characterise their properties. The sampleof 44 BRCs catalogued by Sugitani et al. (1991) forms the basisfor our campaign in the northern hemisphere. A high proportion ofthese clouds have been found to be undergoing a strong interac-tion with their surrounding HII region (Morgan et al. 2004) and areassociated with embedded protostellar cores (Morgan et al. 2008).Furthermore, a significant proportion of these sources have beenfound to be associated with water masers (Cesaroni et al. 1988;Henning et al. 1992; Wouterloot et al. 1993; Xiang & Turner 1995;Claussen et al. 1996; Valdettaro et al. 2005, 2008) and kinematicline profiles indicative of molecular outflows (see Morgan et al.2009; and references therein). In Fig. 1 we present an example of a BRC that appears to be agood candidate for triggered star formation. In this figure we havecombined two Spitzer IRAC bands and a MIPS image to create athree colour image of SFO 13 (IRAC 4.5 and 8.0 µ m and MIPS24 µ m bands are coloured in green, blue and red respectively). Wehave over-plotted contours of the NVSS radio continuum emission(white) and the SCUBA 850 µ m emission (yellow) which traceionised gas and thermal dust emission respectively.The multiple wavelength data that comprise Fig. 1 show manydi ff erent features of this small molecular cloud on the edge of anHII region. The 4.5 µ m emission in green likely indicates shockedH gas (e.g. Teixeira et al. 2008), possibly associated with thejets / outflows of young stars. Probing the interface between theneutral molecular material of the cloud itself and the ionized gasof the HII region, the 8.0 µ m emission, traced in blue, indicatesthe presence of polycyclic aromatic hydrocarbons (PAHs), com-plex molecules which absorb far UV photons and re-emit in the IRregime (Leger & Puget 1984). Dominant lines of these PAHs are at7.7 and 8.6 µ m which are within the passband of the Spitzer 8.0 µ mband. This reveals the photon-dominated region (PDR), which ispreceded by the ionization front at the head of the cloud. It is herethat the cloud is actively supporting the presence of an establishedIBL (as described in e.g. Lefloch & Lazare ff (1994)), emitting free-free emission which is traced by the 20 cm NVSS contours in white.The 24 µ m emission, shown in red in Fig. 1, traces small grains ofheated dust (e.g. Koenig et al. 2008). This emission peaks in a cen-trally condensed region at the head of the cloud, coincident withthe peak of 850 µ m contours in yellow, which show submillimetre-emission associated with warm dust within the cloud and surround-ing the protostellar core at the head of the cloud.In order to explore the possible origins of these protostars, it isessential to explore the physical properties of the protostellar coresthemselves and their environs. In this paper we present a set of am-monia observations made towards 42 BRCs in an e ff ort to confirmthe presence of protostellar cores and to probe their physical condi-tions. Analysis of the line properties of ammonia rotational transi-tions at ∼
23 GHz allows for determinations of kinetic temperature,column density and velocity dispersion. These properties will allowthe definitive identification of protostellar cores within our sampleand allow us to determine the global properties of molecular mate-rial within BRCs.The structure of the paper is as follows: in Sect. 2 we describethe observational set up, data reduction procedures and derive thephysical parameters. In Sect. 3 we present the observational results,in Sect. 4 we discuss these results with respect to triggered starformation. In Sect. 5 we present a summary and highlight our mainfindings.
Observations of the NH (1,1) and (2,2), (3,3) and (4,4) rotationaltransitions were made over three sessions from the 11 th of Februaryto the 20 th of March 2005 on the Green Bank telescope operated bythe National Radio Astronomy Observatory . The National Radio Astronomy Observatory is a facility of the NationalScience Foundation operated under cooperative agreement by AssociatedUniversities, Inc. c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds Table 1.
Observational setupDate Centre Frequencies (GHz) RMS a (K) T asys (K)11 th Feb 2005 23.710, 23.877 0.05 41.012 th Feb 2005 23.710, 23.877 0.04 31.220 th Mar 2005 23.695, 23.724, 23.871, 24.140 0.07 38.4 a -RMS and Tsys values represent mean values of measurements takenfrom single two minute integrations at intervals over the entire observingperiod. Observations were performed in ‘nod’ mode in order to re-move sky contributions and a noise diode was observed in switch-ing mode throughout the observations in order to achieve abso-lute flux calibration on the T ∗ A scale to an accuracy of ∼ ∼
650 km s − ) bandpass with the (3,3) transition in anotherspectral window (the (4,4) transition was not observed in these ses-sions). The first two sessions were observed with a spectral channelwidth of 6.1 kHz ( ∼ − ). By the time of the third observingsession, hardware problems had been resolved and all four transi-tions were observed at the optimum spectral resolution of ∼ ∼ − ) in 12.5 MHz bandwidth ( ∼
160 km s − ) spectralwindows. All transitions were observed simultaneously using theconfigurations presented in Table 1.Weather conditions were stable across all three observing ses-sions, typical pointing o ff sets were ∼ ′′ in azimuth and eleva-tion with a beamsize of ∼ ′′ . Sources were observed for periodsranging between four and 66 minutes, dependent upon the signalstrength of observed lines. The median observing time was 20 min-utes per object.In total, observations were made towards 42 BRCs, includingSFO 11E and 11NE, identified by Thompson et al. (2004) but notincluded in the original SFO catalogue. Of the 44 sources listed inthe northern SFO catalogue, four (SFO 2, 3, 27 and 40) were notobserved due to time and hour angle limitations. The data of twosources (SFO 11 and 32) were later found to be corrupt, resultingin a total source sample size of 40 out of a potential 46. In Table 2we present a summary of the BRCs observed and the pointing cen-tres used for the observations. Source positions were largely deter-mined from examination of SCUBA observed dust emission maps,presented in Morgan et al. (2008) and Thompson et al. (2004). Forsources not observed in those works, the coordinates of the IRASsource associated with the relevant object were used. Data were reduced using the GBTIDL data analysis package. Badscans were removed and channels outside the region of interestwere discarded. In reduction, use was made of the GBTIDL pro-cedural ability to smooth the ‘o ff ’. This involves the smoothing of http: // gbtidl.nrao.edu / Table 2.
Summary of source names, observed positions and observationdates. IRAS α δ
Obs. DateSource Source (J2000) (J2000)SFO 01 23568 + / / + / / + / / + / / + / / · · · · · · · · · / / + / / + / / + / / + / / · · · / / · · · / / + / / + / / + / / · · · · · · · · · / / + / / − − / / + / / + / / · · · · · · · · · / / − − / / − − / / − − / / − − / / + / / · · · · · · · · · / / + / / + / / − − / / − − / / − − / / − − / / + / / + / / + / / · · · · · · · · · / / + / / · · · · · · · · · / / + / / + / / + / / + / / · · · · · · · · · / / + / / + / / + / / + / / + / / the reference spectra before the usual spectral subtraction and cal-ibration are performed. This results in higher signal to noise ratiosfor the final calibrated spectrum with no loss of spectral resolution.After careful testing of this procedure, the default smoothing ra-tio of 16 channels was used for all sources except SFO 31. Thissource showed emission in the reference spectrum which resultedin negative emission seen in the final calibrated spectrum. Throughextensive smoothing of the reference spectrum (by 250 channels)this ‘o ff ’ emission was greatly minimised. Temperature scale cor-rections for atmospheric opacity were made using the zenith valuesprovided from local weather models. c (cid:13)000
Obs. DateSource Source (J2000) (J2000)SFO 01 23568 + / / + / / + / / + / / + / / · · · · · · · · · / / + / / + / / + / / + / / · · · / / · · · / / + / / + / / + / / · · · · · · · · · / / + / / − − / / + / / + / / · · · · · · · · · / / − − / / − − / / − − / / − − / / + / / · · · · · · · · · / / + / / + / / − − / / − − / / − − / / − − / / + / / + / / + / / · · · · · · · · · / / + / / · · · · · · · · · / / + / / + / / + / / + / / · · · · · · · · · / / + / / + / / + / / + / / + / / the reference spectra before the usual spectral subtraction and cal-ibration are performed. This results in higher signal to noise ratiosfor the final calibrated spectrum with no loss of spectral resolution.After careful testing of this procedure, the default smoothing ra-tio of 16 channels was used for all sources except SFO 31. Thissource showed emission in the reference spectrum which resultedin negative emission seen in the final calibrated spectrum. Throughextensive smoothing of the reference spectrum (by 250 channels)this ‘o ff ’ emission was greatly minimised. Temperature scale cor-rections for atmospheric opacity were made using the zenith valuesprovided from local weather models. c (cid:13)000 , 1–10 L. K. Morgan et al.
Table 3.
Sources not detected in our survey with RMS values of the obser-vations. Source RMS (mK)SFO 04 14.8SFO 06 9.6SFO 08 22.3SFO 10 12.8SFO 15 15.3SFO 19 33.8SFO 21 20.6SFO 22 19.4SFO 28 19.6
As data from di ff erent observing sessions were taken in dif-ferent configurations, in certain cases, individual objects may havebeen observed in multiple configurations. Spectra for these ob-jects were smoothed to a common resolution before combination.All (1,1) and (2,2) spectra were ultimately smoothed to a resolu-tion of 7.4 kHz ( ∼ − ) while (3,3) and (4,4) spectra weresmoothed to a resolution of 50.2 kHz (due to the expected lowersignal to noise ratio). A polynomial baseline was subtracted fromall spectra for normalisation. Several sources were detected in the(3,3) and (4,4) transitions. However, analyses of these data con-tribute little to the scope of this paper and so the spectra and relatedparameters are presented in appendix A.The corrected antenna temperature T ∗ A , hyperfine linewidth ∆ v , and central velocity V LSR values for each source in the (1,1)and (2,2) inversion transitions were determined by fitting with theNH3(1,1) METHOD process in the CLASS data analysis pack-age . In some cases (SFO 09, 11E, 17, 26, 35 and 42) the mainquadrupole of ammonia was detected but the hyperfines were notseen at a level required for CLASS to fit them. For these sources,and the (3,3) and (4,4) transitions presented in Appendix A, themain quadrupoles were fitted by a single Gaussian using the ‘fit-gauss’ procedure in GBTIDL. The optical depth ( τ ) associated with ammonia emission in oursources may be determined through the ratio of main to satellite an-tenna temperatures of the (1,1) inversion transition (Ho & Townes1983): ∆ T ∗ A ( J , K ) m ∆ T ∗ A ( J , K ) s = − e − τ ( J , K ) m − e − a τ ( J , K ) m (1)where m and s subscripts indicate quantities associated with themain and satellite quadrupole lines, respectively, and a is the ratioof intensities of satellite to main lines (0.278 and 0.221 for innerand outer satellites, respectively). Optical depth values were de-termined through use of the NH3(1,1) METHOD process in theCLASS data analysis package.Given the optical depth associated with the (1,1) transition,corrected antenna temperatures can be used to calculate the rota-tional temperature ( T r ) associated with the (2,2) and (1,1) transi-tions (Ho & Townes 1983): T r = − T ln (cid:26) − . τ m(1 , ln (cid:20) − ∆ T ∗ Am(2 , ∆ T ∗ Am(1 , (1 − e − τ m(1 , ) (cid:21)(cid:27) (2) http: // / IRAMFR / GILDAS where T = E (2 , − E (1 , k B ≈ . T k < T , a relationship between rotational and kinetic tem-perature may be calculated by consideration of (1,1), (2,2), and(2,1) states only (Walmsley & Ungerechts 1983; Swift et al. 2005): T r = T k + T k T ln h + . (cid:16) − . T k (cid:17)i , (3)although kinetic temperatures calculated by this method associatedwith T r ≈ T may be overestimated. Calculated T k values rangefrom 11.9 to 27.3K, suggesting that this approximation is reason-able for these source objects.In order to calculate the column density of our sources, it isnecessary to calculate the excitation temperature of each source.This may be found via T ∗ A = η mb η f [ J ( T ex ) − J ( T bg )][1 − e − τ m ] , (4)where the quantity J ν ( T ) is defined as h ν/ k( e h ν/ k T − , η mb is the mainbeam e ffi ciency of the GBT (0.89 ) and η f is the filling factor of thesource in the relevant observation. The filling factor of each sourceis somewhat hard to determine; values of T ex derived assuming avalue of η f equal to unity are presented in Table 4 along with val-ues of η f determined by assuming LTE in our sources and letting T ex = T k . This assumption is supported by the rough equivalence of T r , T k and the values of T dust determined by Morgan et al. (2008).A typical value of a filling factor in a similar study was determinedby Rosolowsky et al. (2008) in their study of dense cores in Perseusto be ∼ η f appear reasonable, ranging from0.03 to 0.35 with a mean of 0.12.If excitation conditions are homogeneous along the beam andall hyperfine lines have the same excitation temperature, then thecolumn densities at a given (J,K = J) transition can be written as(Rosolowsky et al. 2008) N (1 , = π / ν √ ln 2c A (cid:20) − e − h ν T ex (cid:21) − ∆ v τ, (5)where A is the Einstein spontaneous emission coe ffi cient. Our de-termination of the filling factors associated with each source indi-cate that the T ex of our sources may be significantly underestimated.This is likely due to beam dilution or clumping within the beam ofour observations, in order to account for this e ff ect we have as-sumed LTE in our sources and set T ex = T k in our determinations ofcolumn density.Overall column density is obtained from N(1,1) and / or N(2,2)and the partition function: NZ = N ( i , i ) Z ( i ) −→ N = N ( i , i ) ZZ ( i ) ; Z = X i Z i (6)where the partition function, Z J = (2J +
1) S(J) exp − h [B J (J + + (C − B) J ]k T k . (7)The values of the rotational constants B and C are 298117 and186726 MHz respectively and the function S(J) is 2 for J = T ex and column density are listed foreach source in Table 4. from ‘The Proposer’s Guide for the Green Bank Telescope’c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds Table 4.
Observationally measured and derived physical source parameters.NH (1,1) NH (2,2) T ∗ A ∆ v T ∗ A ∆ v V LSR τ T ex η f T k N NH T dust N bH Source (K) (km s − ) (K) (km s − ) (km s − ) (K) (K) (10 cm − ) (K) (10 cm − ) Triggered Sources
SFO 01 0.52 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a ± · · · · · · -40.3 · · · · · · · · · · · · · · · · · · · · · SFO 12 0.33 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a ± ± · · · · · · · · · · · · · · · · · · · · · SFO 36 0.38 ± ± ± ± ± ± · · · · · · SFO 37 0.23 ± ± ± ± ± ± ± ± ± ± ± ± · · · · · · SFO 41 0.12 ± ± ± ± ± ± a ± · · · · · · -2.61 · · · · · · · · · · · · · · · · · · SFO 43 0.22 ± ± ± ± ± ± ± ± ± ± ± ± · · · · · · Non-Triggered Sources
SFO 09 a ± · · · · · · -48.3 · · · · · · · · · · · · · · · · · · · · · SFO 11NE 0.39 ± ± ± ± ± ± · · · · · · SFO 16 2.35 ± ± ± ± ± ± a ± · · · · · · · · · · · · · · · · · · · · · · · · · · · SFO 18 1.48 ± ± ± ± ± ± ± ± ± ± ± ± · · · · · · SFO 23 0.52 ± ± ± ± ± ± ± ± ± ± ± ± · · · · · · SFO 26 a ± · · · · · · · · · · · · · · · · · · · · · · · · SFO 29 0.26 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± / or the (2,2) main line was not detected. b-Determined from the observations of Morgan et al. (2008). We detected ammonia (1,1) emission towards 31 of the 40 suc-cessfully observed sources. We present spectral plots of both theNH (1,1) and (2,2) lines towards all of these in Fig. 5. In theseplots we show the baseline corrected data (black histogram) andthe model fits (red line) to the data. The measured properties of themain NH (1,1) and NH (2,2) lines are given in Table 4. In thistable we have separated the sample into those clouds consideredto be good triggered star formation candidates and those in whichthe star formation is unlikely to have been induced. The determi-nation of which sources are triggered candidates and which not isdescribed in Morgan et al. (2009) and is based upon the detectionof a strong IBL in radio and PDR in IR emission.The hyperfine structure of the NH (1,1) transition is clearlyresolved towards 25 BRCs, as is the main quadrupole of theNH (2,2) transition, allowing estimates of the optical depth, kine-matic and excitation temperatures and column densities to beobtained (Several sources also show hyperfine structure in theNH (2,2) transition 07, 14, 16, 18, 30, 38, 44). Towards the remain-ing six sources only the main NH (1,1) quadrupole line is visibleabove the noise and of these the NH (2,2) transition is only visiblein three. Our results can thus be separated into three groups, strong detections, weak detections and non-detections, each of which con-tain 25, 6 and 9 clouds, respectively. Non-detections with respectiveRMS noise levels are listed in Table 3.The overall detection rate is 75%, with both transitions beingstrongly detected towards roughly ∼
60% of the sources. Ammoniais a high density tracer needing a critical density of n crit = cm − before becoming thermally excited (Swade 1989). Due to the re-quirement for this density before ammonia becomes thermally ex-cited, ammonia is better able to probe the properties of the denseprotostellar core surrounding the accreting protostar than CO ordust continuum emission, which tend to probe the whole columnof the gas along the line of sight.As previously mentioned, the ammonia observations werecentred on the position of either the dust cores revealed by theSCUBA observations (Morgan et al. 2008), or towards the IRASpoint source thought to be protostellar in nature. It is somewhatsurprising then to find 25% of sources observed resulted in non-detections, which would seem to rule out the presence of a densecore and subsequently call into question the association of theseclouds with protostars. Moreover, the weaker emission detected to-wards another six clouds may indicate that conditions are not con-ducive for star formation. We will investigate these weak and non-detections in more detail in Sect. 4.1. c (cid:13) , 1–10 L. K. Morgan et al.
Figure 2.
Histograms showing the distributions of the physical parameters derived from the ammonia detections. Filled red histograms represent our entiresample while blue represents our triggered sample only.
Table 5.
Summary of measured and derived parameters.Parameter Min Max Mean Median Std. dev.Excitation Temp. (K) 2.5 4.3 3.2 3.1 0.5Kinetic Temp. (K) 11.9 27.3 19.5 19.9 4.6 τ ) (cm − )] 13.9 15.2 14.4 14.4 0.3 ∆ v (km s − ) 0.48 1.93 1.07 1.10 0.38 In Fig. 2 we present a set of four histograms showing the dis-tribution of ammonia (1,1) linewidth, optical depth, kinetic tem-perature and column density. In these plots we show the distribu-tion for all of the detections (filled red) and the parameters associ-ated with BRCs identified as triggered candidates by Morgan et al.(2009) (filled blue). The ranges and averages of the properties ofall detected sources are presented in Table 5.The histograms of Fig. 2 show that the temperatures of oursources do not show any large variation dependent upon their trig- gered status. It would appear that the fundamental observationalproperties of ammonia emission are somewhat insensitive to en-vironmental factors. Temperatures and linewidths of protostellarsources, as observed in ammonia, appear fairly constant acrossmany sample bases. Some examples drawn from the literatureinclude Molinari et al. (1996) ( ∆ v = − , T k = ∆ v = − , T r = ∆ v = − , T k = We used the results of our infrared, submillimetre, molecular lineand radio analyses (Morgan et al. 2004, 2008, 2009) to identify 26BRCs that show strong evidence they are 1) undergoing recentlyinitiated star formation and 2) are being subjected to intense lev-els of ionizing radiation. These clouds are therefore considered to c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds be excellent candidates in which the observed star formation mayhave been triggered via RDI. Although there is evidence that starformation is present in many of the remaining 18 clouds of the SFOcatalogue there is little or no evidence that the ionisation from thenearby OB stars is having a significant dynamical impact on theseclouds. In this section therefore we will only concern ourselveswith the BRCs that remain in the refined triggered BRC cataloguepresented in Morgan et al. (2009). In Morgan et al. (2009) we identified 26 clouds from the SFOcatalogue in which there is a strong likelihood that observed starformation is the result of the RDI triggering process. Twenty-oneof these have been observed in ammonia here. Four of the trig-gered candidates are non-detections (SFO 4, 6, 10, 15) and a fur-ther two fall into the weak category (SFO 35 and 42). These sixclouds were either not detected or only marginally detected byMorgan et al. (2008) at submillimetre wavelengths. SFO 6 wasnot detected in that study and the remaining five sources weremarginally detected at 850 µ m but not detected at 450 µ m. Thesemarginal / non-detections suggest that these clouds are not host toprotostars, though they may still be at an early point in their evolu-tion with respect to the observed ionization fronts and may, in thefuture, develop protostellar cores. These sources shall be discardedfrom further consideration of triggered protostellar souces.Ten of the sources in our observational sample have been pre-viously associated with water maser activity (see Table 6). Theseaccount for 40% of our ‘strongly detected’ sources, a figure con-sistent with the 40% detection rate for Class 0 protostars foundby Furuya et al. (2001). All ten of the sources associated with wa-ter masers are also associated with outflow activity (see Table 6),supporting the general association of jets and outflows with wa-ter masers (Codella et al. 2004 and references therein). Of theseten sources, only two are found within our non-triggered sam-ple. While this finding is perhaps not statistically significant, giventhe small sample size, it does indicate that high radiation environ-ments are unlikely to suppress maser emission, as suggested byValdettaro et al. (2008). The presence of molecular outflows andwater masers towards so many triggered candidates is a strong in-dication that star formation is currently taking place within them.Observed linewidths within sources host to outflows and masersare typically greater than in other sources (median linewidths are1.4 km s − for maser sources, compared to 1.1 km s − for othersources), indicating that the high density gas surrounding the pro-tostellar cores traced by ammonia emission is dynamically linkedto the collisional processes associated with the maser emission. ff erenttracers In this section we will compare the derived ammonia and dust prop-erties to identify correlations and anti-correlations in the data andto check for inconsistencies between the di ff erent tracers. In Sec-tion 2.3 we derived the kinetic temperatures and NH column den-sities which are the most readily available quantities we have athand to compare with quantities derived from the dust emission. Inthe next section we will describe how the dust temperatures weredetermined and estimate the H column densities. Table 6.
Outflow and water maser activity in our sourcesSource ∆ v (km s − ) Outflow MaserSFO 05 1.27 1 3SFO 07 1.54 1 4SFO 14 1.62 2 5SFO 18 ⋆ ⋆ ⋆ BRCs not considered to be triggered candidates.References (NB, only most recent reference cited): 1-Morgan et al. (2009),2-Wu et al. (2004), 3-Xiang & Turner (1995), 4-Wouterloot et al. (1993), 5-Henning et al. (1992), 6-Claussen et al. (1996), 7-Valdettaro et al. (2005),8-Cesaroni et al. (1988), 9-Valdettaro et al. (2008)
In an earlier paper (Morgan et al. 2008) we presented observationsof submillimetre emission which tracing the distribution of warmdust. Spectral energy distributions (SEDs) were determined by fit-ting greybody functions to the measured submillimetre fluxes andmid- and far-infrared fluxes. Fits were presented for all sources forwhich good quality data was available.The H column density associated with each SFO object de-tected by Morgan et al. (2008) may be calculated using N (H ) = S ν / [ Ω µ m H κ ν B ν ( T d )] (8)where S ν is the 850 µ m flux density, Ω is the solid angle associatedwith the 30 ′′ aperture used to observe each core in Morgan et al.(2008), µ = H is the mass of ahydrogen atom, κ ν is the dust opacity per unit mass at 850 µ m (0.02cm g − , following Morgan et al. (2008)) and B ν ( T d ) is the Planckfunction, evaluated at dust temperature T d . Resulting values of H column density are presented in the final column of Table 4. In Fig. 3 we present two scatter plots comparing the temperatures(upper panel) and column densities (lower panel) derived from thetwo tracers of ammonia and submillimetre emission. We show thelinear-square fit to the data as a solid line. Additionally, in the tem-perature plot, we include a dashed line indicating the position ofthe data if both temperatures were equal. There is quite a lot ofscatter in the distributions seen in both plots. However, there is ageneral correlation between the kinetic and dust temperatures andthe H and NH column densities. The dust temperatures are gen-erally slightly higher than the observed kinetic temperatures. Theratio of dust temperature to T k ranges from 0.9 to 1.6 with a meanof 1.2, this slightly higher dust temperature may be attributed to thefact that submillimetre emission is associated with a wide range ofdensities, covering the star-forming core itself as well as the warmenvelope surrounding the core (and the interface between the two).As ammonia emission requires a critical density of ∼ cm − , itis likely to trace the inner, more dense regions of the protostellarcore. It should be noted that the dust temperatures were derivedusing fluxes from the IRAS with a significantly larger beam thanthe present observations. These observations are therefore likely to c (cid:13) , 1–10 L. K. Morgan et al.
Figure 3.
Upper panel: Dust temperature vs. kinetic temperature, a line ofbest fit is shown as a solid line while the dotted line indicates the line ofequality. Lower panel: NH vs. H column density, a line of best fit is shownas a solid line while the dotted line indicates the line of constant fractionalabundance. incorporate more of the hotter dust at the edges of the BRCs. Wetherefore conclude that the two tracers are probing material at sim-ilar temperatures (di ff erence between median averages is < and NH column densities the scatter is significant.The fractional abundance of NH may be found through simplecomparison of the derived H and NH column densities . Lookingat individual sources we find the fractional abundances rangefrom a few times 10 − to a few times 10 − . The mean fractionalabundance of NH to H is 2.6 × − , this is the value usedin Fig. 3 to illustrate a line of constant fractional abundance.The scatter in the plot of NH vs. H column density reflects Note that the beam size of the SCUBA observations was ∼ ′′ , sig-nificantly smaller than the ∼ ′′ beamsize of the ammonia observationspresented here. However, the fluxes taken from Morgan et al. (2008) weresummed over a 30 ′′ aperture, thus making a direct comparison valid. Thedisparity between the pointing positions of each set of observations is typi-cally small ( ∼ ′′ ) and only exceeds a half-beamwidth for one source, SFO29. the variation of fractional abundance from source to source.Overall, the determined values of fractional abundance are typicalacross a wide range of protostellar environments, from low-massstarless cores (Tafalla et al. 2006; Crapsi et al. 2007) and low tointermediate mass dense cores (Hotzel et al. 2001; Friesen et al.2009), to complex, PDR-associated regions (Larsson et al. 2003)and high-mass star forming regions (Kuiper et al. 1995; Pillai et al.2006).The fractional abundances found here reflect a more generaltrend in the properties of ammonia in star forming regions. Thephysical properties of our sources, as determined from our ammo-nia observations, are typical in most star forming environments,with only very hot cores showing any significant variation in col-umn density or temperature (c.f. Longmore et al. 2007; Pillai et al.2007). The implication of our analysis is that ammonia, once ex-cited beyond its critical threshold, is insensitive to environmen-tal circumstances, i.e. resistant to depletion in cold, dense coresand likely shielded from photoionisation in high-radiation environ-ments. The contribution of turbulent motions to the observed linewidthof each source was estimated by removing linewidth contributionsfrom other broadening agents. Thermal contributions to each linewere calculated via d V therm = p B T k ln2 / m NH , where T k is thekinetic temperature of the source in question and m NH is the meanmolecular mass of an ammonia molecule (17.03 amu).After removing the thermal broadening contribution to our ob-served linewidths we may evaluate the contribution of turbulentmotions in our clouds. An analysis of the turbulent velocity dis-persions ( σ = p < ∆ v > / (8ln2)) of our sources, separated basedupon their triggered status (see Sect. 1), reveals some interestingdi ff erences between the two samples. A histogram of the velocitydispersions of the potentially triggered and non-triggered samplesshows that our non-triggered sources have typically lower velocitydispersions than our triggered candidates (Fig. 4). A Kolmogorov-Smirnov test of the σ of each of the two samples indicates thatthe two distributions are drawn from separate populations with aprobability of 99.8%. An interesting point to note in the distribu-tions of the σ of each sample is that the non-triggered sources arelargely subsonic, based upon an estimate of the sound speed in oursources of ∼ − (e.g. Thompson et al. 2004). In contrast, thetriggered sample typically exhibit supersonic velocities.Several explanations of the di ff erence in σ between our twosamples exist, the sources in our triggered sample are likely un-dergoing the progression of shocks through the host clouds. Asthis sample was selected on the basis of the presence of an ion-isation front, the presence of shocks in the observed materialwould thus be consistent with the models of triggering put forwardby Bertoldi (1989); Lefloch & Lazare ff (1994); Miao et al. (2006);Gritschneder et al. (2009); Miao et al. (2009). An alternative expla-nation would be that the increased turbulent motions within ourtriggered sample are due to increased systematic motions within therelevant clouds (due to larger numbers of outflows, for example).Alternatively, these clouds may simply be larger in extent and / ormass. However, we found no trend toward higher dust mass forthese clouds from the submillimetre observations of Morgan et al.(2008). The source luminosities from those observations show thatthe most luminous sources of the entire sample are those of thetriggered sub-sample, as suggested by Morgan et al. (2008). The c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds Figure 4.
Histogram of turbulent velocity dispersion for triggered and non-triggered Sources, Group I (triggered) sources are shown in blue and GroupII sources are shown in red. Binsize is 0.1 km s − . median luminosity of the triggered sample is 201 L ⊙ while the non-triggered sample median average is almost an order of magnitudelower at 28 L ⊙ . While this may go some way to explain the highervelocity dispersions within the triggered sample, no direct correla-tion between source luminosity and linewidth can be drawn. Sucha correlation might be expected if the higher turbulent velocity dis-persions of the triggered sample were due to outflow momentum(e.g. Cabrit & Bertout 1992; Bontemps et al. 1996). The ambigu-ity in the true nature of the increased velocity dispersions in theseclouds cannot be resolved with single pointing data and must be ad-dressed with maps of the molecular emission associated with eachregion. We present observations made with the GBT of the ammonia (1,1)and (2,2) inversion transitions towards 40 bright-rimmed cloudstaken from the sample compiled by Sugitani et al. (1991). We de-tected the (1,1) transition towards 31 of the sources observed, withthe (2,2) transitions seen towards 28 of these. We use these emis-sion lines to derive the optical depths, kinetic temperatures and NH column densities towards the embedded protostars located withinthese clouds.Across the entire sample, the ammonia kinetic temperaturesof our sources correlate reasonably well with the dust temperaturesseen towards the same cores by Morgan et al. (2008). Dust temper-atures are typically slightly higher than the observed kinetic tem-peratures, likely due to the slightly di ff erent material traced by thetwo sets of observations. A correlation between H and NH col-umn densities is also seen, although there is somewhat more scatterin the plotted points. This likely reflects variation in the fractionalabundance of ammonia from source to source. A mean fractionalabundance of NH to H is 2.6 × − , typical for most observa-tions of protostellar regions.Using a combination of mid-infrared, submillimetre and ra-dio images, and CO molecular line data presented in earlier works(Morgan et al. 2004, 2008, 2009), we have refined the originalsample of 44 BRCs. Our e ff orts have identified 26 bright-rimmedclouds in which the data are consistent with the hypothesis that anyobserved star formation is likely to have been triggered. Within thisrefined sample we have detected strong ammonia emission towards15. In combination with the submillimetre continuum and CO line emission results of Morgan et al. (2008) and Morgan et al. (2009)respectively, in addition to outflow and maser detections from theliterature, our ammonia detections leave little doubt of the star-forming nature of these sources. Having already established thelikelihood that these sources represent photoionisation-triggeringprocesses in progress, these 15 sources are some of the best exam-ples yet known of the RDI process.An investigation of our samples, separated based upon trig-gered status, indicates a bimodality within observations of the tur-bulent velocity dispersion. Those sources which have been identi-fied as likely triggered in nature show typically supersonic turbulentvelocity dispersions. While non-triggered sources are more oftenfound to be subsonic. This disparity may stem from the presence ofshocks, traversing the clouds in the triggered sample. It is also pos-sible that the higher observed velocity dispersion in the triggeredsample is simply due to the higher occurence of outflows found inthat sample. It is tempting to draw conclusions of di ff ering physicalprocesses occuring within our triggered and non-triggered samples.However, we are not able to determine the true cause of this findingwithout mapping of molecular emission in these clouds. ACKNOWLEDGMENTS
The authors would like to thank an anonymous referee for a carefulexamination of this work which has resulted in considerable im-provements. We would like to thank the helpful sta ff of the GreenBank Telescope and Bill Saxton for his assistance in creating Fig.1.LKM is supported by a STFC postdoctoral grant (ST / G001847 / / California Institute of Technology, funded by the NationalAeronautics and Space Administration and the National ScienceFoundation.
REFERENCES
Bertoldi F., 1989, ApJ, 346, 735Bontemps S., Andre P., Terebey S., Cabrit S., 1996, A&A, 311,858Cabrit S., Bertout C., 1992, A&A, 261, 274Cesaroni R., Palagi F., Felli M., Catarzi M., Comoretto G., diFrancos Giovanardi C., Palla F., 1988, A&AS, 76, 445Claussen M. J., Wilking B. A., Benson P. J., Wootten A., MyersP. C., Terebey S., 1996, ApJS, 106, 111Codella C., Lorenzani A., Gallego A. T., Cesaroni R., MoscadelliL., 2004, A&A, 417, 615Crapsi A., Caselli P., Walmsley M. C., Tafalla M., 2007, A&A,470, 221De Vries C. H., Narayanan G., Snell R. L., 2002, ApJ, 577, 798Friesen R. K., Di Francesco J., Shirley Y. L., Myers P. C., 2009,ApJ, 697, 1457Furuya R. S., Kitamura Y., Wootten H. A., Claussen M. J.,Kawabe R., 2001, ApJ, 559, L143Gritschneder M., Naab T., Walch S., Burkert A., Heitsch F., 2009,ApJ, 694, L26 c (cid:13) , 1–10 L. K. Morgan et al.
Henning T., Cesaroni R., Walmsley M., Pfau W., 1992, A&AS,93, 525Ho P. T. P., Townes C. H., 1983, ARA&A, 21, 239Hotzel S., Harju J., Lemke D., Mattila K., Walmsley C. M., 2001,A&A, 372, 302Jijina J., Myers P. C., Adams F. C., 1999, ApJS, 125, 161Koenig X. P., Allen L. E., Gutermuth R. A., Hora J. L., BruntC. M., Muzerolle J., 2008, ApJ, 688, 1142Kuiper T. B. H., Peters III W. L., Forster J. R., Gardner F. F.,Whiteoak J. B., 1995, ApJ, 446, 692Larsson B., Liseau R., Bergman P., Bernath P., Black J. H.,Booth R. S., Buat V., Curry C. L., Encrenaz P., Falgarone E.,and 26 coauthors 2003, A&A, 402, L69Lefloch B., Lazare ff B., 1994, A&A, 289, 559Leger A., Puget J. L., 1984, A&A, 137, L5Longmore S. N., Burton M. G., Barnes P. J., Wong T., PurcellC. R., Ott J., 2007, MNRAS, 379, 535Miao J., White G. J., Nelson R., Thompson M., Morgan L., 2006,MNRAS, 369, 143Miao J., White G. J., Thompson M. A., Nelson R. P., 2009, ApJ,692, 382Molinari S., Brand J., Cesaroni R., Palla F., 1996, A&A, 308, 573Morgan L. K., Thompson M. A., Urquhart J. S., White G. J., 2008,A&A, 477, 557Morgan L. K., Thompson M. A., Urquhart J. S., White G. J., MiaoJ., 2004, A&A, 426, 535Morgan L. K., Urquhart J. S., Thompson M. A., 2009, MNRAS,400, 1726Pillai T., Wyrowski F., Carey S. J., Menten K. M., 2006, A&A,450, 569Pillai T., Wyrowski F., Hatchell J., Gibb A. G., Thompson M. A.,2007, A&A, 467, 207Rosolowsky E. W., Pineda J. E., Foster J. B., Borkin M. A., Kau ff -mann J., Caselli P., Myers P. C., Goodman A. A., 2008, ApJS,175, 509Sugitani K., Fukui Y., Ogura K., 1991, ApJS, 77, 59Swade D. A., 1989, ApJ, 345, 828Swift J. J., Welch W. J., Di Francesco J., 2005, ApJ, 620, 823Tafalla M., Santiago-Garc´ıa J., Myers P. C., Caselli P., WalmsleyC. M., Crapsi A., 2006, A&A, 455, 577Teixeira P. S., McCoey C., Fich M., Lada C. J., 2008, MNRAS,384, 71Thompson M. A., Urquhart J. S., White G. J., 2004, A&A, 415,627Thompson M. A., White G. J., Morgan L. K., Miao J., FridlundC. V. M., Huldtgren-White M., 2004, A&A, 414, 1017Urquhart J. S., Morgan L. K., Thompson M. A., 2009, A&A, 497,789Urquhart J. S., Thompson M. A., Morgan L. K., Pestalozzi M. R.,White G. J., Muna D. N., 2007, A&A, 467, 1125Urquhart J. S., Thompson M. A., Morgan L. K., White G. J., 2004,A&A, 428, 723Urquhart J. S., Thompson M. A., Morgan L. K., White G. J., 2006,A&A, 450, 625Valdettaro R., Migenes V., Trinidad M. A., Brand J., Palla F.,2008, ApJ, 675, 1352Valdettaro R., Palla F., Brand J., Cesaroni R., 2005, A&A, 443,535Walmsley C. M., Ungerechts H., 1983, A&A, 122, 164White G. J., Lefloch B., Fridlund C. V. M., Aspin C. A., DahmenG., Minchin N. R., Huldtgren M., 1997, A&A, 323, 931Wouterloot J. G. A., Brand J., Fiegle K., 1993, A&AS, 98, 589 Table A1.
Observationally measured parameters of (3,3) and (4,4) detec-tions. T ∗ A ∆ v Source (K) (km s − ) (3,3) Sources SFO 01 0.10 ± ± ± ± ± ± ± ± (4,4) Sources SFO 14 0.04 ± Wu Y., Wei Y., Zhao M., Shi Y., Yu W., Qin S., Huang M., 2004,A&A, 426, 503Wu Y., Zhang Q., Yu W., Miller M., Mao R., Sun K., Wang Y.,2006, A&A, 450, 607Xiang D., Turner B. E., 1995, ApJS, 99, 121
APPENDIX A: (3,3) AND (4,4) DETECTIONS
Several sources were detected in the (3,3) and (4,4) rotationaltransitions of ammonia. Spectra of these sources are presented inFigs. A1 and A2 and relevant parameters are listed in Table A1,RMS values of these observations are presented in Table A2.Those sources detected in the (3,3) line of ammonia were ex-amined for signs of masing. If masing is occuring in this transi-tion then the brightness temperature of that source would be largerthan the brightness temperature in the (1,1) transition and com-parable to the (1,1) kinetic temperature (e.g. Kuiper et al. 1995).However, none of our sources show such symptoms. The (4,4)line of ammonia may often be observed toward warmer cores (e.g.Longmore et al. 2007), only one of our sources provided a detec-tion in this line, SFO 14. This core is indeed one of the warmersources in our sample at T k ∼ / L A TEX file prepared by theauthor. c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds Figure 5.
Spectral lines are presented with corrected antenna temperature, T ∗ A , plotted against doppler shifted velocity, V LSR . Multiple lines are plotted on thesame axis range with the (1,1) transition spectra on the bottom and (2,2) spectra on top. Fitted profiles are overlaid in red, determien.c (cid:13) , 1–10 L. K. Morgan et al.
Figure 5. (cont.) Spectral lines are presented with corrected antenna temperature, T ∗ A , plotted against doppler shifted velocity, V LSR . Multiple lines are plottedon the same axis range with the (1,1) transition spectra on the bottom and (2,2) spectra on top.
Figure A1. NH (3,3) transition spectral lines. T ∗ A , plotted against doppler shifted velocity, V LSR , with GBTIDL fitted Gaussian profiles overlaid in red.c (cid:13) , 1–10 mmonia observations of bright-rimmed clouds Figure A2. NH (4,4) transition spectral line observed towards SFO 14. T ∗ A , plotted against doppler shifted velocity, V LSR , with GBTIDL fitted Gaussianprofiles overlaid in red.
Table A2.
RMS values of our (3,3) and (4,4) transition observations.Source (3,3) RMS (mK) (4,4) RMS (mK)SFO 01 13.3 4.7SFO 04 13.1 · · ·
SFO 05 12.0 · · ·
SFO 06 4.4 4.2SFO 07 14.8 4.6SFO 08 10.1 9.3SFO 09 14.3 · · ·
SFO 10 12.6 · · ·
SFO 11E 10.1 9.9SFO 11NE 10.3 5.3SFO 12 4.4 · · ·
SFO 13 11.4 · · ·
SFO 15 13.7 · · ·
SFO 16 4.4 4.7SFO 17 11.0 11.7SFO 18 5.5 4.0SFO 19 16.0 14.1SFO 20 7.3 6.6SFO 21 9.6 8.5SFO 22 9.7 8.7SFO 23 4.5 3.9SFO 24 7.2 · · ·
SFO 25 5.7 · · ·
SFO 26 10.0 9.4SFO 28 7.9 8.2SFO 29 7.3 6.2SFO 30 7.9 · · ·
SFO 31 3.8 · · ·
SFO 33 6.7 5.0SFO 34 6.4 5.2SFO 35 9.4 9.8SFO 36 4.5 · · ·
SFO 37 5.2 · · ·
SFO 38 28.0 5.2SFO 39 4.4 · · ·
SFO 41 7.7 4.6SFO 42 13.4 10.5SFO 43 4.8 · · ·
SFO 44 5.4 · · ·· · ·
Source not observed in this transition.c (cid:13)000