The physical state of selected cold clumps
AAstronomy & Astrophysics manuscript no. AA201423428 c (cid:13)
ESO 2018April 16, 2018
The physical state of selected cold clumps (cid:63)
A. Parikka , , M. Juvela , V.-M. Pelkonen , , J. Malinen , and J. Harju Department of Physics, P.O.Box 64, FI-00014, University of Helsinki, Finland Institut d’Astrophysique Spatiale, Universit´e Paris Sud, 91405 Orsay Cedex, France Finnish Centre for Astronomy with ESO (FINCA), University of Turku, V¨ais¨al¨antie 20, 21500 Piikki¨o, FinlandReceived ¡date¿ / Accepted ¡date¿
ABSTRACT
Context.
The study of prestellar cores is essential to understanding the initial stages of star formation. With
Herschel more coldclumps have been detected than ever before. For this study we have selected 21 cold clumps from 20
Herschel fields observed as afollow-up on original
Planck detections. We have observed these clumps in CO (1-0), C O (1-0), and N H + (1-0) lines. Aims.
Our aim is to find out if these cold clumps are prestellar. We have examined to what extent independent analysis of the dust andthe molecular lines lead to similar conclusions about the masses of these objects.
Methods.
We calculate the clump masses and densities from the dust continuum and molecular line observations and compare theseto each other and to the virial and Bonnor-Ebert masses calculated for each clump. Finally we examine two of the fields with radiativetransfer models to estimate CO abundances.
Results.
When excitation temperatures could be estimated, the column densities derived from molecular line observations werecomparable to those from dust continuum data. The median column density estimates are 4.2 × cm − and 5.5 × cm − for theline and dust emission data, respectively. The calculated abundances, column densities, volume densities, and masses all have largeuncertainties and one must be careful when drawing conclusions. Abundance of CO was found in modeling the two clumps in thefield G131.65 + − . The abundance ratio of CO and C O was ∼
10. Molecular abundancescould only be estimated with modeling, relying on dust column density data.
Conclusions.
The results indicate that most cold clumps, even those with dust color temperatures close to 11 K, are not necessarilyprestellar.
Key words.
ISM: clouds – submillimeter: ISM, dust, molecular lines – stars: formation
1. Introduction
The details of the initial stages of star formation are still poorlyknown, despite many observational and theoretical studies. Thisis in part because star formation involves a very complex mixtureof e ff ects of gravity, turbulence, rotation, radiation, thermody-namics, and magnetic fields. Both large representative samplesof clumps and detailed studies of their properties and relationsto their cloud environment are needed. Herschel observations of star-forming clouds, e.g., thoseconducted within the Gould Belt Survey (Andr´e et al. 2010)and HOBYS program (Motte et al. 2010), have lead to the de-tections of hundreds of starless and protostellar condensations.This is particularly true for nearby star forming clouds wherethe
Herschel resolution is su ffi cient to resolve gravitationallybound cores. The objects appear preferentially within filaments,which are a conspicuous feature of all interstellar clouds (Andr´eet al. 2013). The filaments themselves contain both supercriticalgravitationally bound structures and subcritical filaments that arelikely to disperse with time (Arzoumanian et al. 2013).The Planck satellite has performed an all-sky survey tomap anisotropies of the cosmic microwave background (Tauberet al. 2010). At the same time
Planck is providing maps ofthermal dust emission from molecular clouds within the MilkyWay. From these data more than 10,000 compact cold sources (cid:63)
Herschel is an ESA space observatory with science instrumentsprovided by European-led Principal Investigator consortia and with im-portant participation from NASA. have been detected and a large fraction of these are believedto be prestellar cores or larger structures harboring prestellarcores (Planck Collaboration et al. 2011b). The
Planck and IRAS(100 µ m) surveys have been used to compose the Cold ClumpCatalogue of Planck
Objects (C3PO), from which 915 mostreliable detections were published as the Early Cold ClumpsCatalogue (353 −
857 GHz, 350 − µ m), ECC (see, e.g., PlanckCollaboration et al. 2011c,d,e). The sources were detected withthe method described in Montier et al. (2010) and the ini-tial results on the Planck detections were described in PlanckCollaboration et al. (2011e,b). The final version of C3PO is cur-rently in preparation and will provide a global view of the coldclump population over the whole sky.A number of Planck detections of cold clumps have beenconfirmed with the
Herschel
Space Observatory observations(100 − µ m) to be cold (T dust ∼
14 K or below) and also dense.For the present study, we selected clumps from fields coveredin the
Herschel open time key program Galactic Cold Cores(Juvela et al. 2010). The aim of this
Herschel program is to ex-amine a representative cross section of the source population ofcold clumps observed with
Planck and to determine the physi-cal properties of these clumps. The
Herschel results suggest thatmany of the sources are already past the prestellar phase (Juvelaet al. 2010, 2011). In this paper, we refer to sources that aregravitationally bound as prestellar objects. If the estimated massexceeds the virial mass, the object is expected to be gravitation-ally bound. Bonnor-Ebert (BE) spheres are used as an alternative a r X i v : . [ a s t r o - ph . GA ] M a r . Parikka et al.: The physical state of selected cold clumps model to recognize prestellar cores (Andr´e et al. 2010; K¨onyveset al. 2010).In an earlier study, a sample of 71 fields at distances rang-ing from ∼
100 pc to several kiloparsecs were examined. The
Herschel observations, together with AKARI and WISE in-frared data, were used to confirm the presence of cold dust. Inabout half of the observed fields, point sources were found inthe mid-infrared, indicating active star formation. However, thecold dust still dominated the submillimeter spectra in these ac-tive star formation areas (Juvela et al. 2012b).Some of the
Planck -detected clumps have already been stud-ied in molecular lines. Wu et al. (2012) surveyed 673 sourceswith single-point observations and mapped 10 clumps with 22identified potential cold cores. They found seven cores that arelikely to be gravitationally bound and, thus, on the verge of col-lapse. In a follow-up study, 71 of the sources observed only withsingle-point observations were examined and 90 % of the foundcores were starless (Meng et al. 2013). The clumps studied byWu et al. (2012) and Meng et al. (2013) did not include theclumps chosen for this paper.The definition of dense structures in the ISM is still an on-going process. The clumps are small and dense condensationsin the molecular clouds, often inside filaments. Cores are evensmaller and denser. Bergin & Tafalla (2007) defined the clumpsto have a mass of 50 −
500 M (cid:12) , size of 0.3 − − cm − . Cores were defined to have mass of 0.5 − (cid:12) ,size of 0.03 − − cm − . We use theword clump as a general term for the dense structures we inves-tigate here, however, some of these structures are likely to becores themselves or contain unresolved cores.Extensive surveys have been done in CO and CO overthe Galactic plane and of nearby star-forming clouds (see, e.g.,Combes 1991 for review). However, general surveys are not thebest method to research possible compact sources. For the firsttime, with
Planck and
Herschel , we can identify the coldestand densest clumps in the known molecular clouds complexesand focus on dedicated molecular line observations targeting themost interesting dense clumps in these regions.We combine
Herschel data with molecular line observationsto examine the physical properties of several cold clumps andtheir likelihood to evolve into star-forming cores. To investigatethe gravitational stability of the objects, molecular line obser-vations are essential to providing direct information of the ki-netic and turbulent forces and, thus, on the current stability ofthe clumps and the cores. The objects can become gravitation-ally unstable by cooling and accretion, possibly aided or hin-dered by turbulence and external forces (Bergin & Tafalla 2007;Ballesteros-Paredes et al. 2007).We investigate the physical state of a few clumps that wereoriginally selected based on the cold dust signature detected bythe
Planck satellite. We derive the CO column densities, vol-ume densities, and clump masses and estimate the gravitationalstability. The potential prestellar nature of the objects is inves-tigated by comparing their mass estimates with virial masses.We look for signs of CO abundance variations as further evi-dence of the nature of the sources. Analysis assuming a localthermodynamic equilibrium (LTE) will be complemented withmodeling. We selected one target for radiative transfer modelingwhere continuum data are used to constrain the models so thatdirect estimates of CO abundance are possible. The results alsoprovide insights into the nature of the larger
Planck cold clumppopulation.The paper is organized as follows. The observations are de-scribed in Sect. 2. Further, we go through the methods in Sect. 3. The results are presented in Sect. 4 and they are discussed inmore detail in Sect. 5. Finally, the conclusions are summarizedin Sect. 6.
2. Observations
We examine 21 clumps in 20 fields previously observed with
Herschel at wavelengths 100 - 500 µ m. The fields mappedwith Herschel were originally selected based on the
Planck survey, in which these clumps showed a significant excess ofcold dust emission (Planck Collaboration et al. 2011c). Ourselection was based on the
Herschel data from the
GalacticCold Cores Herschel key program (PI Juvela) that carried outfollow-up observations of 116 fields, each with one or more
Planck cold clumps. We chose the bright clumps, which basedon
Herschel data also contain very cold dust (color temperatureat T dust < ∼
14 K, 40 (cid:48)(cid:48) resolution). The fields are located in theMilky Way at galactic latitudes | b | = − ◦ , which ensures min-imal confusion from background emission. While they typicallyhave a general cometary or filamentary morphology, the submil-limeter data show that at small scales the clouds are fragmentedto several clumps. We mapped the selected clumps in the CO (1-0) spectral lineusing the 20-m radio telescope in Onsala Space Observatory,Sweden, in February and March 2012. Toward the CO peaks,we also observed C O (1-0) and N H + (1-0) lines. Part of thenorthern clump in field G131.65 + CO (1-0) in February 2011 (Planck Collaboration et al. 2011b). Theobserved clumps are summarized in Table 1 and the observedspectra are shown in the appendix A. We carried out the obser-vations in frequency switching mode, with frequency throws of ±
10 MHz for CO and C O and ± H + . For N H + in the 86 GHz band the half-power beam width, HPBW, is 44 (cid:48)(cid:48) and the main beam e ff ciency, η B , is 0.65. For CO and C O inthe 109 GHz band the HPBW is 35 (cid:48)(cid:48) and η B is 0.45.The telescope has a pointing accuracy of 3 (cid:48)(cid:48) rms. Thepointing was checked with SiO spectra toward bright sourcesR Cassiopeiae and U Orionis. The calibration was achievedthrough the chopper-wheel method, and the system gives an-tenna temperatures in units of T ∗ A (Kutner & Ulich 1981). Thebackend was a low-resolution digital autocorrelator spectrome-ter (ACS) with a bandwidth of 40 MHz divided into 1198 chan-nels. This gave a channel width of 25 kHz, corresponding to0.068 km s − at the frequency of the CO(1 −
0) line.On average the RMS noise for the observed CO lines was ∼ T MB ) scale.The corresponding values for the observed C O and N H + lines are ∼ ∼ / CLASS software package . The frequencyswitched spectra were folded, averaged, and the spectral base-lines were modeled and subtracted using third order polynomi-als, except for C O in sources G108.28 + + Institut de Radioastronomie Millim´etrique (IRAM):http: // iram.fr / IRAMFR / GILDAS /
2. Parikka et al.: The physical state of selected cold clumps
Table 1.
Summary of observations. The table lists the 21 targets, kinematic distances for the fields (Montillaud et al. 2015), temper-ature of the dust, coordinates for the CO (1-0) peak temperature (where the C O (1-0) and N H + (1-0) were observed), area insquare arc minutes selected for the CO (1-0) mapping, and number of observed points for C O (1-0) and N H + (1-0). Field d (pc) T dust α (J2000) δ (J2000) CO (sq (cid:48) ) C O (N) N H + (N)G86.97-4.06 700 11.5 ± +
43 18 47.1 4.55 1 1G92.04 + ± +
52 28 38.3 3.96 1 1G93.21 + ± +
56 54 42.6 6.79 2 1G94.15 + ± +
55 34 33.4 6.67 1 1G98.00 + ± +
60 08 55.1 3.79 1 1G105.57 + ± +
66 34 40.9 8.33 1 1G108.28 + ± +
72 53 43.2 8.01 7 1G110.80 + ± +
72 46 49.1 6.88 2 1G111.41-2.95 3000 13 ± +
57 36 44.5 7.77 1 1G131.65 + ± +
70 43 20.1 6.79 4 1G131.65 + ± +
70 36 09.5 8.89 7 1G132.12 + ± +
69 50 12.9 5.92 2 1G149.67 + ± +
55 13 40.6 4.89 1 1G154.08 + ± +
53 07 23.1 20.2 3 1G157.92-2.28 2500 10.9 ± +
45 24 22.5 8.33 2 1G159.34 + ± +
52 08 16.4 1 point 1 1G161.55-9.30 250 13 ± +
37 45 35.0 2.22 5 1G164.71-5.64 330 13 ± +
37 45 14.3 5.39 1 1G167.20-8.69 160 12 ± +
34 18 25.6 6.88 2 1G168.85-10.19 2610 13 ± +
31 43 01.2 1 point 1 0G173.43-5.44 150 13 ± +
31 20 46.3 1 point 1 0
Herschel
We observed the dust continuum data with the
Herschel
SPIREinstrument (Gri ffi n et al. 2010) between November 2009 andMay 2011. These observations consist of surface brightnessmaps of 250, 350, and 500 µ m. The raw and pipeline-reduceddata are available via the Herschel
Science Archive, the user-reduced maps are available via ESA site We reduced these data similar to (Juvela et al. 2012b) andonly a summary of these reduction steps is given here. We com-pleted the reduction using the
Herschel
Interactive ProcessingEnvironment HIPE v.10.0 and the o ffi cial pipeline with the it-erative destriper and extended emission calibration options. Theresulting maps are the product of direct projection onto the skyand averaging of the time-ordered data. The resolutions of themaps are 18 (cid:48)(cid:48) , 25 (cid:48)(cid:48) , and 37 (cid:48)(cid:48) at 250, 350, and 500 µ m, respec-tively.The accuracy of the gain calibration of Herschel data is bet-ter than 7 % in absolute terms and probably better than 2 % band-to-band . The surface brightness data do not have an absolutezero point and therefore, before temperatures or column densi-ties can be estimated, we must set a consistent zero point acrossall Herschel bands. One alternative is to compare
Herschel datawith
Planck and IRAS measurements (using IRAS data tied toDIRBE scale) and to make use of the zero points of those surveys(see, e.g., Juvela et al. 2012b). We opt for the more straightfor-ward procedure of selecting a reference area with a radius of 1.5 (cid:48) within the
Herschel map and measuring surface brightness val-ues relative to the average value found in the reference area. Thecentral coordinates of the reference areas can be seen in Table2. The derived dust temperature and column density estimatesignore the very di ff use medium to the extent that is visible inthe reference region. The reference regions have low extinction( A V ∼ http: // herschel.esac.esa.int / UserReducedData.shtml SPIRE Observer’s manual http: // herschel.esac.esa.int / Documentation.shtml
Table 2.
Center coordinates of the 1.5 (cid:48) radius reference areas.
Field α (J2000) δ (J2000)G86.97-4.06 21 16 40.8 +
43 34 45G92.04 + +
52 34 45G93.21 + +
56 44 49G94.15 + +
55 56 19G98.00 + +
59 51 19G105.57 + +
66 26 12G108.28 + +
72 51 20G110.80 + +
72 37 40G111.41-2.95 23 20 11.9 +
57 35 20G131.65 + +
70 45 20G132.12 + +
69 53 43G149.67 + +
55 29 55G154.08 + +
53 14 42G157.92-2.28 04 30 02.5 +
45 25 50G159.34 + +
51 54 33G161.55-9.30 04 15 46.2 +
37 40 56G164.71-5.64 04 42 47.6 +
38 22 45G167.20-8.69 04 35 10.1 +
34 11 09G168.85-10.19 04 36 06.8 +
31 50 54G173.43-5.44 05 07 55.2 +
31 04 55 dance of CO. Thus, even after background subtraction, the con-tinuum data may probe a volume larger than the actual molecularcloud. Therefore, the extrapolation of the relations like in Fig. 1does not necessarily go via the origin of that plot. This shouldnot be a significant problem, however, because we are interestedin the densest clumps that are far above the column density ofthe reference regions.The resolution of
Herschel column density maps is 40 (cid:48)(cid:48) . Thebeam size is 35 (cid:48)(cid:48) in CO observations and 40 (cid:48)(cid:48) in N H + obser-vations. Because of the relatively small di ff erences and the ex-tended nature of our sources, the data were compared directlywithout further convolution.
3. Parikka et al.: The physical state of selected cold clumps
3. Methods
For the calculation of the CO column densities, we used themethod described by Myers et al. (1983). Based on the data anal-ysis, we calculated the optical depth at the peak of C O line, τ ,the excitation temperature of the C O line, T , and the columndensity of the C O in the cloud, N . To derive the total columndensity of H , the column density of C O, N , is divided by thecommonly assumed abundance ratio [C O] / [H ] = − .The optical depths of CO and C O are related by (Myerset al. (1983), Eq. 2) τ = τ n ( J = n ( J = L L ∆ V ∆ V J ( T ) J ( T ) . (1)In the equation J ( T ) = T [exp( T / T ) − − , where T = O and 5.29 K for CO and T is the excitation temper-ature. The n (J =
1) and n (J =
1) are the number of moleculesat J = CO J = O, respectively. The L and L are the line-of-sight extent of the emitting gas and the ∆ V and ∆ V are the line widths of the CO and C O line,respectively. The equation assumes the same excitation temper-ature for both lines, the same beam filling (both lines originatein the same region), and the same velocity gradient. We also as-sume the terrestrial abundance ratio of 5.5 between CO andC O and we use the line width of C O for the calculations.For a more comprehensive explanation of the method, see Myerset al. (1983).We also derived the column density with the
Herschel dustcontinuum observations. The observed intensity can be stated as I ν = B ν ( T dust )(1 − e − τ ) ≈ B ( T dust ) × τ, (2)where I ν is the observed intensity at a frequency ν , B ν ( T dust ) isthe blackbody brightness of the object as a function of color tem-perature T dust , and τ is the source optical depth. The equationassumes a homogeneous source. In the far-infrared and at longerwavelengths the optical depths of the clouds are clearly belowone. This justifies the approximation made in Eq. 2. The opticaldepth can be written as τ = κ ν × N (H ) × µ, (3)where µ is the average particle mass per H molecule, 2.8 u.We assume a dust opacity of κ = . f / . × ) . cm / g,where f is the frequency (Beckwith et al. 1990). The constant2.0 is the dust opacity spectral index, β . While it may varyfrom source to source, the value 2.0 we used is consistent withmany observations of dense clumps although the average valuein molecular clouds is likely to be closer to ∼ N (H ) . From Eqs. 2 and 3 we can derive the molecularcolumn density N (H ) = I ν B ν ( T dust ) κ ν µ . (4)The derived column densities from molecular lines and dustcan be used to calculate the mass and density of a cloud or aclump as M c = N (H ) × π R × µ M (cid:12) , (5)where R is the radius of the clump. The average density, n , iscalculated using the hydrogen column density n = N (H )2 R , (6) where the clump diameter 2 R is the full-width at half maximum(FWHM) from the Herschel column density map.We derived the estimated virial mass from the formula(MacLaren et al. 1988) M vir = k σ RG , (7)where k depends on the density distribution. We choose k = . , which corresponds to density distribution ρ ( r ) ∝ r − . .The velocity dispersion is given by the equation σ = (cid:115) kT kin m + (cid:32) ∆ V ln (2) − kT kin m (cid:33) , (8)where m is the mean molecular mass (2.33 u assuming 10 %He), and m is the mass of the molecule used for observations.The temperature, T kin , is the kinetic temperature, which is as-sumed to be 10 K. The velocity dispersion provides an estimateof the nonthermal motions that provide further support againstgravity (Planck Collaboration et al. 2011b; Bertoldi & McKee1992). If the cloud’s mass is less than the virial mass, it is notgravitationally bound, and without external pressure for support,it will disperse.We can also examine the stability using the model of BEspheres. The BE mass assumes a static isothermal cloud with nomagnetic field (McKee & Ostriker 2007). We include the non-thermal component and use the e ff ective sound speed instead ofthe isothermal sound speed. As a di ff erence to the virial mass, theBE model includes the external pressure. The critical BE mass(Bonnor 1956) is M BE ≈ . R BE σ G , (9)where R BE is the BE radius, and G is the gravitational constant.
4. Results
First we compare the column densities calculated from themolecular line observations to the column densities calculatedfrom the dust continuum observations. The sizes of the clumpswere estimated as a FWHM from the dust column density mapsand the physical sizes in pc are in Table 3. Twelve of the clumpsare larger than the size criteria (0.03 - 0.2 pc) given by Bergin& Tafalla (2007) for cores, and, thus, are larger structures thatcould contain further substructures.We did not get a result for the excitation temperature in all ofthe cases. This is most likely because of the greater ratio between CO and C O abundances than the 5.5 that was assumed in thecalculations. Thus, we initially used an estimate of T ex =
10 K(see, e.g., Dobashi et al. 1994; Alves et al. 1999). The resultsfrom the calculations showed, however, that when T ex could beestimated, it was typically 5 K (see Table 3). Therefore, we showin Table 3 column densities for a fixed value of T ex = T ex could be derived are plot-ted with square symbols, the solid line showing the corre-sponding linear least-squares fit. The column densities calcu-lated with T ex = T ex =
4. Parikka et al.: The physical state of selected cold clumps
Table 3.
Clump sizes (FWHM from dust column density), column densities derived from dust and line data, and estimated C Oexcitation temperatures. N (H ) N (H ) N (H ), T ex = cm − ] T ex [K] lines [10 cm − ] lines [10 cm − ]G86.97-4.06 0.4 0.17 12 ± ± + ±
20 4 ± ±
40 37 ± + ± ± ± ± + ± ± ± ± + ± ± ± ± + ± ± + ± ± + ± ± ± ± ± ± + ± ± + ± ± + ± ± + ± ± + ± ± ± ± ± ± + ± ± ± ± ± ± ± ± ± ± ± ± ± ± Fig. 1.
Comparison of the column densities derived from molec-ular line and dust continuum observations. The squares are thecolumn densities, where the excitation temperature could becalculated, and the circles are the column densities calculatedwith an assumed excitation temperature of 5 K. The colors in-dicate the dust color temperature. Blue is below 11 K, yellowis 11 - 13 K, and red is warmer than 13 K. The solid lineis the linear least-squares fit to the column densities derivedfrom the calculated excitation temperature, and the dashed linefits the T ex = N (H ) dust = N (H ) line .values are usually within the margin of errors. The col-umn densities calculated from dust are typically higher butalso often within the margin of error. For example, forG159.34 + N (H ) dust = ± × cm − and N (H ) = ± × cm − . In some cases the column densitiesare quite di ff erent, however. This is the case for G86.97-4.06, where N (H ) dust = ± × cm − , while the value derivedfrom the line data is much lower, N (H ) = ± × cm − .Uncertainty of the background subtraction could contribute tosome of these di ff erences. For example, in G86.97-4.06 thereference region is estimated to have a column density of ∼ cm − , which corresponds to a di ff use part of the cloud withlittle CO. However, the presence of a small column density ofgas without CO molecules cannot explain the di ff erence in thecolumn density estimates of several times 10 cm − .The dust color temperatures of the clumps are shown in Fig.1 in di ff erent colors: blue for clumps below 11 K, yellow for therange of 11 −
13 K, and red above 13 K. The column densitiesderived from dust show an anticorrelation with the dust colortemperature as denser regions are cooler. This anticorrelation onthe dust color temperature was not seen as clearly in the columndensities derived from molecular line observations, as there areseveral column densities ∼ . × cm − independent of thedust color temperature.The maps of T mb ( CO) are shown in Figs. 2 and 3. Thecontours correspond to the column densities derived from dustcontinuum data (subtracting the local di ff use background). Thestatistical error in the column density maps of the dust is at level2 × cm − . All the contours are at more than 10 σ level and,thus, statistically significant. The S / N of CO maps is lower. Ina few cases, we get over 5 σ detection for the whole map, namelyfor fields G105.57 + + + + σ in part of the field.In all the maps, we see a peak in the main beam temperatureof CO close to the dust column density peak. For few fields,e.g., G98.00 + (cid:48) . In most casesthe shift is smaller than the resolution of the observations. Insome cases, as in G149.67 + ∼ −
5. Parikka et al.: The physical state of selected cold clumps
Fig. 2.
Maps of CO main beam temperature ( T mb [K], resolution 35 (cid:48)(cid:48) ). The contours show the column density derived from dustcontinuum observations (resolution 40 (cid:48)(cid:48) ), with contour steps 10 % of the peak value. The name of the field and the maximum valueof N(H ) derived from dust emission are marked in each frame.Significant di ff erences in the peak positions of dust and lineemission are often observed because of molecular depletion,other chemical gradients, or because of temperature di ff erences.Marsh et al. (2014) found shifts with a median distance of 59 (cid:48)(cid:48) for H CO + and dust (250 µ m). The dust and gas have beenfound to be coupled in the interior of cores with densities above3 × cm − . At lower densities the temperature di ff erence be- tween the two components depends on the strength of the ex-ternal UV field and not on density (Galli et al. 2002). Also, thedepletion can increase the gas temperature in low and moderatedensities (Goldsmith 2001). If we compare the densities calcu-lated from the molecular line observations in Table 4, the clumpsare less dense and, thus, dust and gas are not likely coupledin these clumps. However, the asymmetric displacement of the
6. Parikka et al.: The physical state of selected cold clumps
Fig. 3.
Continued from Fig. 2. CO emission in the densest clumps is more likely to be a signof variations of CO abundance.
The masses and densities along with the virial and BE massesare listed in Table 4. The uncertainties for densities and masseswere determined with the Monte Carlo method. The uncertain-ties are big in many cases, up to 53 %, and we must be care-ful when drawing conclusions from the density and masses. Themass errors do not include the error for distance ( ∼
30 %, in thecase of G168.85-10.19 the error is closer to ∼
50 %), which fur-ther increases the uncertainty. In addition, in most cases the ex-citation temperature could not be calculated and the excitationtemperature of 5 K was assumed. In the cases where the excita-tion temperatures could be calculated, the derived densities andmasses are within the margin of error of those derived from thedust emission data. The field G92.04 + ff erent structural scales. Followingthe mass criteria of Bergin & Tafalla (2007), 11 of them are moremassive than typical cores ( > ocal (cid:12) ). However, most of themshow a single kinematic component and are therefore consistentwith the idea of clumps as velocity coherent objects. If both M line and M dust are larger than the virial and BE mass, we considerthem to be gravitationally bound, and thus (potentially) on theverge of collapse. If M line or M dust exceeds BE mass, the objectis potentially bound, but if the estimated mass is smaller thanboth the virial and BE mass, we consider the object not to beprestellar.Two clumps, G92.04 + + + T ex = + T ex = + + + N H + To estimate the N H + column densities, we performed fits of thehyperfine structure of the J = T ex = ff ect of the uncertainty of T ex . Apart from G86.97-4.06,shown in Fig. 4, the N H + spectra are shown in Appendix B.For most of the studied clumps, N H + line was not de-tected, suggesting that these clumps are not dense enough forsignificant CO depletion and the associated rise of N H + abun-dance (Bergin & Langer 1997; Charnley 1997). The line was de-tected in six clumps: G86.97-4.06, G92.04 + + + + + T mb ∼ − ∼
7. Parikka et al.: The physical state of selected cold clumps
Table 4.
The used line width of C O, densities, masses, virial masses, and BE masses for all the targets using molecular lineobservations and mass value calculated from the dust observations. The values from molecular line observations are derived with T ex calculated from C O observations and with a fixed value of T ex = ∆ V C O n [10 cm − ] M line [M (cid:12) ]Field [km s − ] n [10 cm − ] ( T ex = line [M (cid:12) ] ( T ex = dust [M (cid:12) ] M vir [M (cid:12) ] M BE [M (cid:12) ]G86.97-4.06 0.74 - 7 ± ± ± ± ± + ±
20 32 ± ±
70 100 ±
20 130 ±
40 110 ±
30 70 ± + ± ± ± ± ± ± ± + ± ± ±
30 39 ± ± ± ± + ± ± ± ± ± ± ± + ± ± ± ± ± + ± ± ± ± ± + ± ± ± ± ± ± ± ± ±
20 120 ±
50 500 ±
100 290 ± + ± ± ± ±
10 30 ± + ± ± ± ± ± + ± ± ±
30 110 ±
30 70 ± + ± ± ± ± ± + ± ± ± ± ± ± ± ± ± ±
10 40 ±
10 26 ± + ± ±
60 130 ±
30 100 ±
30 60 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
10 130 ±
40 50 ±
30 30 ± ± ± ± ± ± Table 5.
The main beam temperature, velocity, noise and column density of N H + in the fields it was detected. Field T mb [K] V [km s − ] ∆ V [km s − ] rms [K] N [10 cm − ]G86.97-4.06 0.34 ± ± ± ± + ± ± ± ± + ± ± ± ± + ± ± ± ± + ± ± ± ± + ± ± ± ± tation temperature is much lower than the kinetic temperature.In the studied clumps, the excitation temperature of the C O inmost cases was ∼ ff erence between theexcitation and kinetic temperatures.The calculated N H + column densities are only a few times10 cm − . These can be compared with the N H + survey ofnearby, low-mass cores conducted by Caselli et al. (2002). In thatstudy, for the low-mass cores, M vir ∼
10 M (cid:12) , the average N H + column density was ∼ × cm − , and they found starless coreshad a marginally lower column density than protostellar cores.Our column densities correspond to a lower end of those val-ues. Conversely, Pirogov et al. (2003) observed a sample of 35massive molecular cloud cores (mean virial mass over 600 solarmasses) and found an average column density of 29 × cm − .Our low column density values are consistent with the low massof the clumps, but also suggest that the densities (and CO deple-tion) are not significant enough for a large N H + abundance. The observations were compared to radiative transfer models ofCO lines and of dust continuum emission. The modeling is ex-plained in more detail in Appendix C and we present only themain results. The modeling was limited to the two positions inthe field G131.65 + CO abundances that areclose to the value of 10 − . Because the abundances dependgreatly on the column density estimated from continuum data,the values are uncertain. The abundance ratio for CO and C Owas found to be ∼
10, instead of the 5.5 assumed in Sect. 3,independent of the di ff erent assumptions used in the modeling(see Table C.1). Harjunp¨a¨a & Mattila (1996) and Anderson et al.(1999) have also found similar abundance ratios in dark cloudsand cores.
5. Discussion
In about one third of the clumps, where excitation temperaturecould be estimated, the column densities are similar to those de-rived from dust (see Fig. 1 and Table 3). The column densitiescalculated with T ex =
10 K, were lower by a factor of three. Using T ex = ∼ cm − ). Higher columndensities tend to have larger uncertainties. For example, the col-umn density for clump G92.04 + ± × cm − from dustand 9 ± × cm − from molecular lines is still inside the mar-gin of errors. The cases where T ex =
8. Parikka et al.: The physical state of selected cold clumps −0.10.00.10.20.30.4 T M B ( K ) G86.97−4.06 T MB =0.34V =5.50∆V =0.43 −5 0 5 10 15v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.06
Fig. 4.
The N H + spectrum observed toward G86.97-4.06. Thered dashed line is the fit to the hyperfine spectra and the lowersubframe shows the fit residuals. The main beam temperatureof the 23-12 component, the fitted radial velocity, the FWHMline width (km s − ), and the residual rms noise (as main beamtemperature) are given in the frames. Fig. 5.
FWHM (pc) vs. distance with the best fit.mostly lower column densities than those where T ex was solved.One explanation for this could be that CO has been depleted inthese areas and, thus, the column density and the volume densitycalculated from CO observations are too low. The most likelyexplanation for the di ff erence to the dust estimates is, however,the excitation temperature that is likely to vary from clump toclump.The distances vary from 150 pc to 3 kpc across the sample,but the physical sizes (see Table 3 and Fig. 5) of the objectsare, with one exception, ∼ + + T ex values,the mass and column density estimates indicate that the selectedclumps might not be very dense, and subsequently may not beactual prestellar cores. The lack of N H + emission also supportsthis conclusion, with upper limits T mb (N H + ) = − CO and C O , which was actually higher thanthe assumed upper limit of 5.5. When extinction A V is small,the abundance ratio can range much above the terrestrial value(Minchin et al. 1995). The abundance ratio we found in ourmodels was ∼
10, compared to the used terrestrial value of 5.5.We found the abundance for CO, however, to be close to thecanonical 10 − , provided that we trust the dust-derived columndensities. The di ff erence in the estimated abundance ratio canbe caused by a real di ff erences in the abundances or excitationtemperatures or by optical depth e ff ects. The modeling, how-ever, roughly takes the e ff ect of optical depth and the expecteddi ff erence in the excitation temperature of the isotopomers intoaccount. This di ff erence is caused by the di ff erence in the pho-ton trapping resulting from the di ff erence in the optical depth ofthe lines. The assumption of the homogeneous source in the LTEanalysis is inaccurate and, in reality, the excitation temperatureof C O is probably below the excitation temperature of CO(and not equal as assumed) and drops precipitously toward thecloud surface. If the CO is optically thick, the observed in-tensity originates in a di ff erent part of the cloud than the C O.Our calculations did not show CO to be optically thick, butwe calculated the optical thickness using the assumption that theexcitation temperatures are equal. Because of the smaller opticaldepth, more of the C O line photons escape, leading to a smallerexcitation temperature. This is also clearly significant when thekinetic temperature varies within the cloud (Juvela et al. 2012a).Based on the line data, 16 of the studied clumps are sub-critical and, even considering the uncertainties of the mass esti-mates, only five sources could be gravitationally bound. In threecases, G92.04 + + + + + + > σ , although G92.04 + T ex = σ -level.However, only four clumps have masses below the virial andBE masses on a level > σ . In the coldest supercritical clumps,G92.04 + T dust = + T dust = + T dust = H + line,which would support the idea that these are being denser andpossibly prestellar. The N H + emission seemed in general tobe connected to the coldest clumps of ∼ −
11 K, although itwas not observable in all the clumps of these temperatures. Forthe clumps where N H + was detected, we get a FWHM linewidth of (cid:46) − , values typically found in prestellar cores(Johnstone et al. 2010).If we rely on the mass estimates derived from dust ob-servations instead of CO lines, the results are similar, onlythree clumps are below the virial mass and BE mass limits.
9. Parikka et al.: The physical state of selected cold clumps
Four of the clumps were larger by more than the error mar-gins and could be classified as prestellar. The four clumps wereG92.04 + + +
6. Conclusions and summary
We investigated 21 clumps, that show low dust color tempera-tures T ∼ −
15 K and are therefore potential places of star for-mation. To study the conditions in the clouds, we used molecu-lar line and dust continuum observations. Our comparison of thetwo tracers shows that, even though the gas and dust emission aremainly morphologically compatible, the dust peak is sometimesshifted up to 30 (cid:48)(cid:48) relative to the CO maximum. The data wereexamined with standard LTE analysis and with radiative transfermodeling to understand better the physical and chemical state ofthe clumps.The column density calculations from molecular lines weremostly within the uncertainties to those calculated from dust,when the excitation temperature could be calculated. With an as-sumed fixed excitation temperature value of T ex = CO was found to be close to the typicallyassumed value of 10 − . However, the abundance ratio of COand C O was ∼
10, higher than the terrestrial value 5.5.The calculated masses had big uncertainties and one mustbe careful when drawing conclusions. When the masses werecompared to virial and BE masses, only five clumps had ahigh enough mass to be gravitationally bound. In three of theseclumps, we also found N H + , so the clumps found in thefields G92.04 + + + + + T dust ∼ + T dust =
12 K. The highest temperatures in the examined clumpswere T dust (cid:38)
13 K.The stability of the clumps requires further study.Observations of higher CO isotopomer transitions and furtherdensity and temperature tracers are needed to better quantify thegas component in these cold clumps.
Appendix A: CO and C O spectra
A sample spectrum of CO of all the fields are shown in Fig.A.1 and A.2. The spectra are from the CO peak position ofeach field. The C O spectrum is shown from the same position;the baseline has been moved to -2.0 K.
Appendix B: N H + The spectra with N H + detections, besides the field G86.97-4.06in Fig. 4, are shown in Fig. B.1. Appendix C: Radiative transfer models
In Sect. 4 the dust emission was analyzed assuming a constanttemperature along the line-of-sight (LOS), but this may under-estimate the true column density of externally heated, opticallythick clumps (Malinen et al. 2011; Juvela et al. 2013). Similarly,the CO(1–0) and C O(1–0) lines were analyzed using the LTEassumption, without the possibility of independent estimates oftheir relative abundances. To examine these questions, we car-ried out radiative transfer modeling where the density distribu-tion was first derived from dust continuum observations and wasthen used as a basis for modeling the lines. The modeling waslimited to the two positions in the field G131.65 + C.1. Modeling of dust surface brightness
We extracted 5 (cid:48) × (cid:48) continuum surface brightness maps centeredon the selected positions and resampled the data on 3 (cid:48)(cid:48) pixels.The maps are 100 ×
100 pixels in size and corresponding modelclouds were constructed using a cartesian grid of 100 ×
100 cells.In the line-of-sight direction the density distribution was as-sumed to follow a Plummer-like (Whitworth & Ward-Thompson2001; Plummer 1911) function ρ ( r ) ∼ | + ( r / R flat ) | ] , with a centralflat part with R flat equal to 0.03 pc (not well resolved with theemployed discretization). Corresponding to the apparent clumpsizes in the plane of the sky, the FWHM of the density distri-bution was set to ∼
17 pixels, which corresponds to ∼ (cid:48)(cid:48) or alinear scale of 0.25 pc at the distance of 1070 pc. This may over-estimate the size because of the e ff ects of beam convolution andradial temperature gradients in the clumps. For this reason, andto check the general sensitivity to the LOS extent, we also cal-culated another set of models with FWHM at half of the valuegiven above. Note that the clump size can also be significantlylarger along LOS than in the plane of the sky. There is evensome bias in this direction because the emission from elongatedclumps and filaments becomes stronger when they are alignedalong LOS.We performed the calculations iteratively, and on each itera-tion we solved the dust temperature distributions and calculatingmodel predictions of the surface brightness that we then con-volved to the resolution of the observations. The calculationswere carried out with a Monte Carlo radiative transfer program(Juvela 2005). The initial external radiation field corresponded tothat of Mathis et al. (1983). The modeling was performed withtwo dust models. The first, in the following MWD, representsdust in normal di ff use interstellar medium (Li & Draine 2001).The other one, in the following OH, was taken from Ossenkopf& Henning (1994) and corresponds to dust that has coagulatedand accreted thin ice mantles over a period of 10 years at a den-sity 10 cm − . In our calculations, the dust opacities κ (250 µ m)were 0.045 cm g − for MWD and 0.22 cm g − for OH, thevalue of Sect. 4 falling between the two. The value of κ is crucialbecause it a ff ects the column density, an important parameter ofthe subsequent line modeling.In the plane of the sky, the column density corresponding toeach map pixel was adjusted by comparing the observed 350 µ msurface brightness with the model prediction. The external radi-ation field was adjusted so that finally the 160 µ m and 500 µ mpredictions also agreed with the observations to within 10 %.
10. Parikka et al.: The physical state of selected cold clumps
Fig. A.1.
The observed CO and C O spectra at the CO peak position. The x-axis shows the velocity and the y-axis shows themain beam temperature. For plotting, the C O spectra have been shifted by 2 K.
11. Parikka et al.: The physical state of selected cold clumps
Fig. A.2.
Continued from Fig. A.1.The observations could be fitted well with both dust models,with an external radiation field 70 - 80 % of the Mathis et al.(1983) values. Because we used background-subtracted surfacebrightness data, the model represents only the inner parts of thecloud without the di ff use envelope. Therefore, the radiation fieldin the models is somewhat weaker that the full radiation fieldoutside the cloud. The column densities are listed in Table C.1.The values obtained with the OH dust model are close to thecolumn densities estimated in Sect. 4. With MWD, the column densities are higher and in the northern point by more than theratio of dust opacities, ∼
5. For a given dust opacity and radiationfield, there is a maximum surface brightness that can be pro-duced with any column density. Thus, for a too small value of κ ,the column density of the model might be grossly overestimated.When estimated with the NICER method (using 2MASS stars, aspatial resolution of two arcminutes, and a value of R V = ff erence in the resolution into account, this is still consis-
12. Parikka et al.: The physical state of selected cold clumps tent with the Sect. 4 estimates and the result from the modelswith OH dust. However, the A V measurement appears to ruleout the values obtained with MWD dust (assuming the A V is notseverely underestimated either because of very clumpy columndensity structure or the presence of foreground stars) and givesome support to the idea of dust opacity higher than that of dif-fuse medium. C.2. Modeling of the CO(1–0) and C O(1–0) lines
We modeled the CO(1–0) and C O(1–0) lines separately, tak-ing the density distribution directly from the continuum models.We only modeled the spectra toward the center of the clumps.The fractional abundance and kinetic temperature were assumedto be constant but in the non-LTE models the excitation temper-ature does vary and gives more weight to the dense regions. Thevelocity field was initialized by giving each cell a random veloc-ity vector with σ = . − and assuming a turbulent linewidth with Doppler velocity ∆ v D = √ σ = . − . Thusthe initial ”microturbulence” within the cells is slightly largerthan the ”macroturbulence” between the cells. In the actual cal-culations, to match the observed line widths, both velocity com-ponents are scaled by the same number, typically by ∼ ff erentassumptions of the LOS cloud extent and the two di ff erent dustmodels. Furthermore, some models were also calculated witha kinetic temperature of T kin = . T kin = . T kin = . T kin = . CO values are on theorder of the canonical value of 10 − and the estimates are simi-lar for both assumed values of T kin . However, because the abun-dances directly depend on the assumed column density, and thusthe values of dust κ , the absolute values are very uncertain. Therelative abundance between CO and C O should be more re-liable, as suggested by the similarity between the OH and MWDcases and the two cases of LOS density distribution. The abun-dance ratio is ∼
10 and thus larger than the value of 5.5 that wasassumed in Sect. 3.
References
Alves, J., Lada, C. J., & Lada, E. A. 1999, ApJ, 515, 265Anderson, I. M., Caselli, P., Haikala, L. K., & Harju, J. 1999, A&AAndr´e, P., Di Francesco, J., Ward-Thompson, D., et al. 2013, eprintarXiv:1312.6232Andr´e, P., Men’shchikov, A., Bontemps, S., et al. 2010, A&A, 518, L102Arzoumanian, D., Andr´e, P., Peretto, N., & K¨onyves, V. 2013, A&A, 553, A119Ballesteros-Paredes, J., Klessen, R. S., Mac Low, M.-M., & Vazquez-Semadeni,E. 2007, Protostars Planets VBeckwith, S. V. W., Sargent, A. I., Chini, R. S., & Guesten, R. 1990, AJ, 99, 924Bergin, E. A. & Langer, W. D. 1997, ApJ, 486, 316Bergin, E. A. & Tafalla, M. 2007, ARA&A, 45, 339Bertoldi, F. & McKee, C. F. 1992, ApJ, 395, 140Bonnor, W. B. 1956, MNRAS, 116Boulanger, F., Abergel, A., Bernard, J.-P., et al. 1996, A&ACaselli, P., Walmsley, C. M., Zucconi, A., et al. 2002, ApJ, 565, 344Charnley, S. B. 1997, MNRAS, 2911 T M B ( C O )( K ) [ CO(North)]=3.2e−07 −10 −9 −8 −7 −6
V (km s −1 ) −0.10.00.10.20.30.40.5 T M B ( C O )( K ) [C O(North)]=2.8e−08 [ CO(South)]=1.2e−06 −10 −9 −8 −7 −6
V (km s −1 ) [C O(South)]=8.3e−08
Fig. C.1.
Examples of the modeled CO (upper frames) andC O spectra (lower frames) for the northern (left frames) andsouthern (right frames) positions in the field G131.65 + Combes, F. 1991, ARA&A, 29, 195Dobashi, K., Bernard, J.-P., Yonekura, Y., & Fukui, Y. 1994, ApJS, 95, 419Galli, D., Walmsley, M., & Gonc¸alves, J. 2002, A&A, 394, 275Goldsmith, P. F. 2001, ApJ, 557, 736Gri ffi n, M. J., Abergel, A., Abreu, A., et al. 2010, A&A, 518, L3Harjunp¨a¨a, P. & Mattila, K. 1996, A&AHildebrand, R. H. 1983, QJRAS, 24Johnstone, D., Rosolowsky, E., Tafalla, M., & Kirk, H. 2010, ApJ, 711, 655Juvela, M. 1997, A&A, 322, 943Juvela, M. 2005, A&A, 440, 531Juvela, M., Harju, J., Ysard, N., & Lunttila, T. 2012a, A&A, 538, A133Juvela, M., Malinen, J., & Lunttila, T. 2013, A&A, 553, A113Juvela, M., Ristorcelli, I., Montier, L. A., et al. 2010, A&A, 518, L93Juvela, M., Ristorcelli, I., Pagani, L., et al. 2012b, A&A, 541, A12Juvela, M., Ristorcelli, I., Pelkonen, V.-M., et al. 2011, A&A, 527, A111K¨onyves, V., Andr´e, P., Men’shchikov, A., et al. 2010, A&A, 518, L106Kutner, M. L. & Ulich, B. L. 1981, ApJ, 250, 341Li, A. & Draine, B. T. 2001, ApJ, 554, 778MacLaren, I., Richardson, K. M., & Wolfendale, A. W. 1988, ApJ, 333, 821Malinen, J., Juvela, M., Collins, D. C., Lunttila, T., & Padoan, P. 2011, A&A,530, A101Marsh, K. A., Gri ffi n, M. J., Palmeirim, P., et al. 2014, MNRAS, 439, 3683Mathis, J. S., Mezger, P. G., & Panagia, N. 1983, A&A, 128, 212McKee, C. F. & Ostriker, E. C. 2007, ARA&A, 45, 565Meng, F., Wu, Y., & Liu, T. 2013, ApJS, 209, 37Minchin, N. R., White, G. J., & Ward-Thompson, D. 1995, A&AMontier, L. A., Pelkonen, V.-M., Juvela, M., Ristorcelli, I., & Marshall, D. J.2010, A&A, 522, A83Montillaud, J., Juvela, M., Rivera-Ingraham, A., et al. 2015, A&A, submitted(not yet accepted)Motte, F., Zavagno, A., Bontemps, S., et al. 2010, A&A, 518, L77Myers, P. C., Linke, R. A., & Benson, P. J. 1983, ApJ, 264, 517Ossenkopf, V. & Henning, T. 1994, A&A, 291, 943Pirogov, L., Zinchenko, I., Caselli, P., Johansson, L. E. B., & Myers, P. C. 2003,A&A, 405, 639Planck Collaboration, Abergel, A., Ade, P. A. R., et al. 2011a, A&A, 536, A25Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2011b, A&A, 536, A23Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2011c, eprintarXiv:1101.2035Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2011e, A&A, 536, A22Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2011d, A&A, 536, A7Plummer, H. C. 1911, MNRAS, 71Tauber, J. A., Norgaard-Nielsen, H. U., Ade, P. A. R., et al. 2010, A&A, 520, A2Whitworth, A. P. & Ward-Thompson, D. 2001, ApJ, 547, 317Wu, Y., Liu, T., Meng, F., et al. 2012, ApJ, 756, 76
13. Parikka et al.: The physical state of selected cold clumps
Table C.1.
Result from radiative transfer modeling. The columns are (1) observed position, (2) column density estimated in Sect. 4,(3) width of the LOS density profile, (4) dust model, (5) column density of the continuum model, (6) assumed kinetic temperature,(7) CO abundance of the model, and (8) relative abundance [ CO] / [C O].
Position N(H ) LOS width Dust N(H ) T kin [ CO] [ CO] / [C O](10 cm − ) (10 cm − ) (K) (10 − )G131.65 + + + + + + + + + + + + −0.10−0.050.000.050.100.150.200.25 T M B ( K ) G92.04 +3.93 T MB =0.17V =−1.80∆V =0.51 −10 −5 0 5 10v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.04 −0.10.00.10.20.3 T M B ( K ) G93.21 +9.55 T MB =0.23V =−1.73∆V =0.41 −10 −5 0 5 10v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.07 −0.10.00.10.20.3 T M B ( K ) G98.00 +8.75 T MB =0.23V =4.98∆V =0.54 −5 0 5 10 15v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.06 −0.10.00.10.20.3 T M B ( K ) G105.57 +10.39 T MB =0.31V =−10.10∆V =0.46 −20 −15 −10 −5 0v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.05−0.10.00.10.20.30.4 T M B ( K ) G132.12 +8.95 T MB =0.32V =−12.41∆V =0.51 −20 −15 −10 −5v (km/s)−0.10−0.050.000.050.100.15 R e s i du a l s ( K ) Rms =0.04
Fig. B.1.
The N H + spectra observed toward G92.04 + + + + + − ), and theresidual rms noise (as main beam temperature) are given in the frames.), and theresidual rms noise (as main beam temperature) are given in the frames.