Herschel Observations of Nearby Interstellar Filaments Philippe André CEA - Lab. AIM Paris-Saclay

IC5146 Outline: • « Universality » of the filamentary structure of the ISM • The key role of filaments in the core/star formation process • Implications for the IMF, open issues, and conclusions

With: D. Arzoumanian, V. Könyves, P. Palmeirim, A. Menshchikov, N. Schneider, A. Roy, N. Peretto, P. Didelon, J. Di Francesco, S. Bontemps, F. Motte, D. Ward-Thompson, J. Kirk, M. Griffin, S. Pezzuto, S. Molinari, J.Ph. Bernard, Polaris Y. Shimajiri, B. Merin, N. Cox, A. Zavagno, L. Testi & Herschel the Herschel Gould Belt Survey KP Consortium 250/350/500 µm Herschel has revealed IC5146 a “universal” filamentary structure in nearby clouds 5 pc Pipe

2 pc

Arzoumanian + 2011 Peretto + 2012 Taurus Aquila

Polaris 2 pc

5 pc

Palmeirim + 2013 Kirk + 2013 4.5pc ~ 1deg Ward-Thompson + 2010 Miville-Deschênes + 2010 Könyves + 2010 The observed filaments are reminiscent of numerical simulations of cloud evolution with large-scale flows Turbulence Gravity

Padoan et al. 2001 Heitsch et al. 2008 Gomez & Vazquez-Semadeni 2014

Turbulence + Gravity + B fields

Z.Y. Li et al. 2010 Chen & Ostriker 2014 Ntormousi et al. Structure of the cold ISM prior to star formation Gould Belt Survey PACS/SPIRE // mode 70/160/250/350/500 µm

Polaris flare translucent cloud: non star forming

d ~ 150 pc

~ 2200 M☉ (CO+HI) unbound: Mvir/Mtot ~ 10 Heithausen & Thaddeus ‘90

~ 13 deg2 field Miville -Deschênes et al. 2010 Ward - Thompson et al. 2010 ~ 5 pc Men ’shchikov et al. 2010

Herschel/SPIRE 250 µm image André et al. 2010

Tracing the underlying filamentary networks Different techniques: Projection on curvelets (Starck+2003), DisPerSE (Sousbie2011), getfilaments (Men’shchikov+2013) … ’Disorganized’ network Examples of getfilaments results

Polaris ’Hub-filament’ network - see Myers (2009) Oph/L1688

Gravity-dominated ? Turbulence-dominated ? See also ‘nests’ vs. ‘ridges’ (Hill+2011 – Vela C) Evidence of much fainter filaments + high degree of universality with Herschel Taurus B211 filament: M/L ~ 50 Musca Musca filament: M /pc P. Palmeirim et al. 2013 M/L ~ 30 M /pc ⊙

N. Cox et al in⊙ prep.

1 pc

SPIRE Herschel SPIRE 250 µm Gould Belt 250µm Evidence of much fainter filaments + high degree of universality with Herschel B field Musca filament: Taurus B211 filament: M/L ~ 50 M/L ~ 30 M /pc M /pc P. Palmeirim et al. 2013

N. Cox et al in⊙ prep. ⊙

1 pc

Polarization vectors overlaid on Herschel images Optical Polarization Pereyra & Heyer+2008 Magelhaes 2004 Heiles 2000 Very common pattern: main filament or “ridge” + network of perpendicular striations or “sub-filaments”  DR21 in Cygnus X: Serpens-South filament: Suggestive of accretion flows into the main filaments M/L ~ 4000 M /pc M/L ~ 250 M /pc Hennemann, Motte et al. 2012 Könyves+2010, Maury+2011, Sugitani+2011, Sugitani+2011 H. Kirk+2013 Also Schneider+2010, Csengeri+2011⊙ ⊙

4 pc 4 pc

1 pc

1 pc

Herschel Herschel HOBYS Gould Belt col. density 3 color cylinder Depth along 0.1los pc ~ = isothermal NB: p for 4 Ν Diameter flatinnerof plateau: ρ ρ & Ward Whitworth (cf. Plummer Plummer Taurus B211/3filament

(r) = (r) = (r) = Resolvingthe structure of filaments with SPIRE 250 SPIRE ρ ρ Ν 2R

c c ( Ostriker

c -like / [1 + (r/ R / [1 + (r/R -like profile density 2): = (p

/ [1 + (r/R flat

density ~ 0.1 pc pc 0.1 ~

1964) µ - Thompson 2001) m

flat flat flat

(D. Li &Goldsmith profile ) )

2 2 ) equilibrium 2 ] ] p/2

] (p-1)/2 Palmeirim+2013 Arzoumanian+2011 1 pc 1

:

‘12) - NNH2 [cm [cm- 2H2 2]] Tdust [K] Beam Beam Beam Beam profile Temperature log plot -log log profile density Column R

flat

Radius [pc] Radius [pc] RadiusRadius [pc] [pc] Radius [pc]

Radius [pc]

- South - - South South - South

- North - North East East - North - North Herschel fit fit fit Best Best Best fit Best

West West West West East East

γ = 0.97 = γ with T

∝

ρ

γ

− 1

Filaments have a characteristic inner width ~ 0.1 pc Network of filaments in IC5146 Statistical distribution of widths Herschel 500/250 µm for > 270 nearby filaments

0.1 pc IC5146 Orion B Aquila Polaris Ophiuchus Taurus Distribution of Pipe Jeans lengths Musca ~ 5 pc

Number of filaments per bin bin of filaments per Number Example of a filament radial 0.1 profile Inner radius Filament width (FWHM) [pc] ~ 0.05pc

] D. Arzoumanian et al. 2011 + PhD thesis 2 - Outer radius [see also Alves de Oliveira+2014 for Chamaeleon;

[cm 0.5pc Some variations along each filament: Ysard+2014] H2 N beam background  Strong constraint on the formation Radius [pc] and evolution of filaments Column density profiles of low- vs. high-density filaments

M ~ 2 M /pc subcritical Taurus 1021 line  FWHMdec=0.08+-0.02 pc

Contrast ~1.2

Gaussian fit

Palmeirim + 2013 Mline ~ 1 M/pc subcritical Mline ~ 50 M/pc supercritical FWHMdec=0.09+-0.01 pc FWHMdec=0.09+-0.02 pc Power law Contrast likely shaped ~15 Contrast by gravity ~1.3

Gaussian fit Gaussian fit Variation of filament angular diameter with cloud distance

Median angular width of the filaments in each region

Musca Aquila Common physical inner width ~ 0.1 pc Taurus IC5146 Ophiuchus Pipe Polaris Orion B

Cloud distance [pc] Deconvolved filament width width (FWHM) filament Deconvolved [pc] D. Arzoumanian’ s PhD thesis

Filaments due to large-scale supersonic turbulence ?

Filament width ~ 0.1 pc: ~ sonic scale of interstellar turbulence ?

Linewidth-Size relation in clouds Simulations of turbulent fragmentation (Larson 1981)

0.5 σV(L) L

∝

c 0.2 s km/s

0.1 pc Log (Size) [pc] Log (Velocity Dispersion) Dispersion) [km/s] Log (Velocity Sonic scale

 Corresponds to the typical thickness λ Padoan, Juvela et al. 2001 of shock-compressed layers in HD λ ~ L/M(L)2 ~ 0.1 pc  Filaments from a combination of MHD turbulent compression and shear; width set by the dissipation scale of MHD wavescompression ? (Hennebelle ratio 2013) (isothermal HD shock)

Velocity dispersion of filaments vs. column density

18 IRAM 30m C O, Mass per unit length, mline + N2H observations [M /pc] 15 [km/s] [km/s]

C18O(1-0) tot ⊙ Non Self- σ self-gravitating gravitating

1/2

σtot~ NH2 + σ c N2H (1-0) NT s 0.2 ≾ Total velocity dispersion, dispersion, velocity Total

Arzoumanian et al. 2013 -2 Column density, NH2 [cm ] Low-density filaments have subsonic levels of internal turbulence: σturb < cs (Hacar & Tafalla 2011; Arzoumanian et al. 2013) +15 ~ 75 - 5 % of prestellar cores form in filaments, N > 7x1021 cm-2 above a column density threshold H2 ~ -2 <=>

Aquila curvelet N map (cm ) Unstable H2 1021 1022 Av >~ 7 2 Σthreshold ~ 150 M☉/pc

1 M Examples of Herschel line prestellar cores (▵ ) /M Blow-up N map (cm-2) H2 line,crit

0.1 Unbound

André et al. 2010, Könyves et al. 2010 + in prep

Quantifying the connection between filaments & cores

Aquila 0.1 pc filament “footprints” associated with DisPerSE & getfilaments skeleton : Prestellar cores found with getsources

Könyves et al., in prep. Mass budget in the Aquila cloud complex Column Density Probability Density Function for Aquila

H2 Lognormal PDF Power-law tail logN

∆ above Av ~ 5 N/ ∆ Turbulence ? Gravity (See Schneider+2013 ~ 85% of and Schneider+2014 the mass ~ 15% for other, similar of the mass column density PDFs from Herschel)

Av ~ 7 Number of pixels per bin: per of pixels Number -2 Column density, NH2 (cm ) Könyves et al. in prep

. Below AV ~ 7: ~ 20 % of the mass in the form of filaments, < 1% in prestellar cores

. Above AV ~ 7: > 50 % of the mass in the form of filaments, ~ 15 % in prestellar cores Strong evidence column a of density Number of cores per bin: ∆N/∆ΝH2 Distribution of background column densities Parent filament column density, column filament Parent N for the formationthe for prestellar of cores Σ for the Aquila prestellar cores

~ 150 M ~ ⊙ pc -2

H2 H2 [10 21

cm -2 ]

“threshold” Johnstone+2004 Onishi+1998 also: See in prepKönyves etal. André+2014 PPVI

Σ A arefound above with cores identified prestellar the of 90% ~ In Aquila,

v ~ 150 M ~ ~ 7 7 ~

Herschel Herschel  ⊙

pc

-

2

Distribution of mass in the parent cloud and background-dependent completeness imply that this threshold is very significant !

) Completeness curves vs. background A

H2 v

Av ~ 5

~ 85% of ~ 15% of the mass, the mass ~ 95% of

the area Av ~ 20

A ~ 7 Completeness Aquila v

-2 Column density, NH2 [cm ] Cumultive mass fraction(> N True core mass, M (M☉)

Könyves et al. in prep

(See also Lada+2010 for similar mass Completeness when AV because “cirrus fraction plots based on extinction) noise” fluctuations (cf. Gautier et al. 1992)

Real “threshold” or rising probability of core/star formation with increasing AV ?

Probability of finding a prestellar core below a given extinction ) )

V Sharper transition than Predictions of models with

) / Area (< A no SF threshold V and εff ~ 1%

Krumholz+2012

Hennebelle & Chabrier 2011 Av ~ 7 Herschel GBS (NB: Similar Aquila conclusion by

Number of pre* cores (< A (< cores of pre* Number Evans+2014 using Background extinction, A [mag] YSOs) V Könyves et al. in prep

Σ or M/L threshold above which Interpretation of the star formation threshold interstellar filaments are gravitationally unstable : Prestellar cores ▵ -2

Aquila curvelet N map (cm ) Unstable H2  The gravitational instability 1021 1022 of filaments is controlled by the mass per unit length M

line 1 M (Ostriker(cf. Ostriker’64, 1964, Inutsuka & Miyama ’97): Inutsuka & Miyama 1997): line • unstable if Mline > Mline, crit /M • unstable if Mline > Mline, crit

line,crit • unbound if Mline < Mline, crit unbound if M < M • line2 line, crit • Mline, crit = 2 cs /G ~ 16 M☉/pc 2ρΣ thresholdthreshold •for M Tline, ~ 10crit K = 2 cs n/GH2 ~ threshold 16 M☉/pc 23 0.1 for T ~ 10 K ~ 16016002 × M 10M☉4☉ /pccm/pc- 3 Unbound

 Simple estimate: Mline NH2 x Width (~ 0.1 pc) ∝ Unstable filaments highlighted André et al. 2010 in white in the NH2 map Toward a new paradigm for ~ M star formation ? See PPVI chapter (André, Di Francesco, Ward-Thompson, Inutsuka, Pudritz, Pineda 2014 - astro-ph/1312.6232)

1) Large-scale MHD supersonic 2) Gravity fragments the densest ‘turbulence’ generates filaments filaments into prestellar cores 2 M / L > Mline,crit = 2 cs /G

Protostellar Cores

50’ 50’ ~ 2 pc ~ 2 pc

Ward-Thompson + 2010 Palmeirim + 2013 Miville-Deschênes + 2010 J. Kirk + 2013 Polaris – Herschel/SPIRE 250 µm Taurus B211/3 – Herschel 250 µm Filament fragmentation may account for the peak of the prestellar CMF and the “base” of the IMF

Core Mass Function (CMF) in Aquila Complex logM ∆

N/ IMF ∆ Jeans mass: CMF × 2 MJeans ~ 0.5 M☉ (T/10 K) × (Σ /160 M pc-2)- × ε crit ☉ 1 system IMF 0.6 M

⊙

Number of cores per mass bin: bin: mass per of cores Number Mass (M☉) Könyves et al. 2010 + in prep. André+2014 PPVI

 CMF peaks at ~ 0.6 M ≈ Jeans mass in marginally critical filaments  Close link of the prestellar CMF with the stellar IMF: M★ ~ × ⊙ 0.3 Mcore

 Characteristic stellar mass may result from filament fragmentation Can filament fragmentation account for the Salpeter power-law of the IMF ?

Example of line mass fluctuations along the Theoretical arguments long axis of a marginally critical filament (Inutsuka 2001) suggest that Filament’s crest H2 Column density map this is possible provided turbulence has generated the appropriate power spectrum of initial density fluctuations

Column density fluctuations along z axis

Mean line mass ≈ 18 M/pc

 Statistical analysis of the line-mass fluctuations for a sample of 80 subcritical Critical line mass ~ 16 M/pc for Roy et al. 2014 or marginally supercritical an isothermal cylinder at T = 10 K Herschel filaments Statistical properties of line-mass fluctuations Implication Power spectrum of line-mass fluctuations Evolution toward a Salpeter-like Power spectrum index

/pc] core mass function when initial 2 α  power spectrum index α ~ -1.5 P(s) [M

s [arcmin-1] Nyquist spatial frequency Beam-corrected Uncorrected power spectrum index power α = -1.7±0.3 spectrum index α ± = -1.9 0.1 Statistics based on 80

filaments Log (dN/dM) Nb. of filaments per per bin Nb. of filaments

Log (M/M ) Roy+2014 Power spectrum index, α min Inutsuka 2001 Summary: A filamentary paradigm for star formation in GMCs ?

 Observational facts: Most SF occurs in dense gas above AV ~ 7; > 50% of this dense gas is in the form of filaments; > 75% of prestellar cores are within dense filaments.

 Herschel results suggest star formation occurs in 2 main steps: 1) ~ 0.1 pc-wide filaments form first in the cold ISM, probably as a result of the dissipation of large-scale MHD turbulence; 2) The densest filaments grow and fragment into prestellar cores via gravitational instability above a critical (column) density -2 × threshold Σth ~ 150 M pc  AV ~ 7  nH2 ~ 2 104 cm-3 ⊙  Filament fragmentation appears to produce the peak of the prestellar CMF and may account for the « base » of the IMF, possibly more (?)

Origin of the characteristic width of filaments ?

Non self-gravitating Self-gravitating Paradox:

Dense filaments should radially 0.1 contract !

Thermal Jeans length Filament width [pc] width (FWHM) Filament

-2 Central column density NH2 [cm ]

D. Arzoumanian et al. 2011 + PhD thesis Key: Evidence of accretion of background material (striations) onto self-gravitating filaments

Example of the B211/3 filament in the Taurus cloud (Mline ~ 54 M☉/pc) Palmeirim et al. 2013 (see also H. Kirk, Myers ea 2013 for another example: Serpens-South)

Herschel CO map 250 µm

1 pc VLSR~6 km/s

B211/3 filament Estimate of the mass accretion rate: CO observations from Goldsmith et al. 2008

Mline ~ 25-50 M☉/pc/Myr

Accretion-driven MHD turbulence can prevent the radial contraction of dense filaments Model of accreting Filament width vs. Column density

filaments Unbound Self-gravitating

0.1

Thermal Jeans length Balance between accretion- driven turbulence (Klessen & Filament width [pc] width (FWHM) Filament Hennebelle 2010) and dissipation of MHD turbulence due to ion- -2 Central column density NH2 [cm ] neutral friction

D. Arzoumanian et al. 2011 + PhD thesis « Dynamical » equilibrium + Hennebelle & André 2013 (see also Heitsch 2013) with ~ 0.1 pc ‘Fibers’: A possible manifestation of accretion-driven turbulence ?

Hacar et al. (2013)’s C18O « fibers » overlaid on Filtered 250 µm image showing the fine Herschel 250 µm image (Palmeirim et al. 2013) structure of the Taurus B211/3 filament

2deg

Obtained with getfilaments (Men’shchikov 18’’ resol. 2013) ArTéMiS : A powerful tool to study massive star-forming filaments (‘ridges’) beyond the Gould Belt APEX ×3.4 higher resolution OSO 12m than SPIRE First 350 µm observations with ArTéMiS at × 3-(10) faster than APEX in July/Sep 2013 : NGC 6334 (d ~ 1.7 kpc) SABOCA 8’’ resol. ~ 0.06 pc PACS “Blue” Focal plane

2048 pixels ~ ~ 10 pc

20 ’ 16x16 pixels

ArTéMiS : 2304 pixels @ 450 µm ESO Photo Release 2304 pixels @ 350 µm ArTéMiS 350 µm 1152 pixels @ 200 µm + VISTA near-IR

Resolving the NGC 6334 main filament (d ~1.7 kpc) with ArTéMiS + Herschel ArTéMiS + SPIRE (350 µm res.: ~ 8’’) Radial intensity profile of NGC6334 filament

beam 8’’

Beam ~ 5~

Intensity [MJy/sr] Intensity 10 ’ pc

Radius [pc]  Deconvolved FWHM width and diameter of flat inner plateau: ~ 0.1-0.2 pc  Linear resolution: < 0.07 pc See Russeil et al. 2013, A&A, for the Herschel/HOBYS view of NGC6334