Magnetic Fields in Star-Forming Molecular Clouds IV. Polarimetry of the Filamentary NGC 2068 Cloud in Orion B

22 Pages·0.32 MB·English

by  B. C. Matthews

Magnetic Fields in Star-Forming Molecular Clouds. IV.
Polarimetry of the Filamentary NGC 2068 Cloud in Orion B
Brenda C. Matthews1 & Christine D. Wilson
McMaster University, 1280 Main Street West, Hamilton Ontario, Canada L8S 4M1
2 [email protected]
0 [email protected]
0
2

n
a
J ABSTRACT

1
We present submillimeter polarimetry at 850 µm toward the filamentary star-forming region
2
associated with the reflection nebulosity NGC 2068 in Orion B. These data were obtained using

1 the James Clerk Maxwell Telescope’s SCUBA polarimeter. The polarization pattern observed
v is not consistent with that expected for a field geometry defined by a single mean field direc-
8
tion. There are three distinct distributions of polarization angle, which could represent regions
4
of differing inclination and/or field geometry within the filamentary gas. In general, the po-
3
1 larization pattern does not correlate with the underlying total dust emission. The presence of
0 varyinginclinationsagainstthe plane ofthe sky is consistentwith the comparisonofthe 850µm
2 continuum emission to the optical emission from the Palomar Optical Sky Survey, which shows
0
thatthe westerndustemissionlies in the foregroundofthe opticalnebula while the easterndust
/
h emission originates in the background. Percentage polarizations are high, particularly toward
p the north-east region of the cloud. The mean polarization percentage in the region is 5.0% with
-
a standard deviation of 3.1%. Depolarization toward high intensities is identified in all parts of
o
r the filament.
t
s
a Subjectheadings: ISM:clouds,magneticfields,molecules—ISM:individual(NGC2068)—polarization
: — stars: formation — submillimeter
v
Xi 1. Introduction in molecular clouds,including OrionB,where the
efficiency is 1% (Carpenter 2000).
ar The study of polarized emission from molec- The Orio∼n B cloud, at a distance of 415
ular clouds is of great interest since, at long ∼
pc (Anthony-Twarog 1982), is the closest giant
wavelengths (λ > 25 µm), it effectively traces
molecular cloud. Within it, star formation is con-
the orientation on magnetic fields local to star-
centrated into five distinct regions: NGC 2071,
forming regions (Hildebrand 1988). Magnetic
NGC2068,LBS23(HH24),NGC2024andNGC
fields have been shownto be energeticallycompa-
2023, as determined from unbiased surveys for
rabletogravityandkineticmotionswithinmolec-
young stellar objects (Lada et al. 1991a) and
ular clouds (Myers& Goodman 1988;Crutcher et
dense CS gas (Lada et al. 1991b). Large scale
al. 1999;Basu2000)andare theorizedto provide
850µmdustemissionfromthese regionshasbeen
vitalsupport to clouds, preventingglobalcollapse
mapped by Mitchell et al. (2001) and Motte
(Mouschovias 1987; McKee et al. 1993, and ref-
et al. (2001). Maps of the polarized emission
erences therein). Such support is necessary to
from NGC 2071, LBS 23, and NGC 2024 have al-
explain the low star-forming efficiencies observed
ready appearedin the literature (see Matthews et
1Current Address: Department of Astronomy, 601 al. (2001a,hereafterPaperIII)).Inordertocom-
CampbellHall,UniversityofCaliforniaBerkeley,Berkeley, parefieldgeometriesacrossOrionB,wehavenow
CA94720-3411 mapped the polarized emission from NGC 2068.
1
ThesubmillimeteremissionfromNGC2068lies ical model of a magnetized cloud. It is impor-
southofthereflectionnebula(seeFigure1)which tant to recognize that the polarizations measured
is seen optically and contains an infrared cluster are vector sums along a particular line of sight
(Lada et al. 1991a). The overall structure of the through the cloud observed, weighted by column
dust emission is that of a “clumpy filament” in density. Fiege & Pudritz (2000a) present a model
which 18 distinct compact sources are identified. forafilamentarycloudinwhichahelicalmagnetic
Most of the submillimeter sources fall outside the field threads the filament and plays an important
boundary of the cluster identified in CS by Lada role in determining the radial density structure.
−2
et al. (1991b). Star formation is ongoing in this This model predicts an r density profile, which
region, as evidenced by detections of bipolar out- has been observed in several clouds, including
flows by Mitchell et al. (2001) around OriBN 51, the Integral-shaped Filament (Johnstone & Bally
from which evidence of outflow previously existed 1999)andseveralcloudsinCygnus(Lada,Alves&
(Edwards & Snell 1984; Gibb & Little 2000), and Lada1999;Alves,Lada&Lada1999). Thehelical
also around OriBN 35 and OriBN 36 peaks (see fieldgeometryalsopredictsdepolarizationtoward
their Fig. 1c or Figure 2 below), although the theaxisofafilamentduetocancellationeffectson
sources of these outflows are ambiguous due to either side of the axis. Fiege & Pudritz (2000b)
the close proximity of OriBN 33 and OriBN 37 to present predicted polarization patterns for cases
the positions of high velocity gas. OriBN 39 has in which the field is either poloidally or toroidally
a 2MASS infrared source associated with it, and dominated as well as for various filament incli-
thus should be a source of outflow. Evidence for nations. Qualitative extensions to these models
redshifted gas near OriBN 47 also suggests some have been shown to reproduce observed polariza-
outflow from this source. tion patterns in the filamentary clouds OMC-3
There are no prior observations of polarized (Matthews et al. 2001b, hereafter Paper II) and
emission from the NGC 2068 dust emitting re- NGC 2024 (see Paper III). The filamentary struc-
gion. However, Mannion & Scarrott (1984) mea- ture of NGC 2068 is therefore of particular inter-
sured linear polarization from scattering against estasafurthertestofaxiallysymmetricmagnetic
the reflection nebula north of the molecular con- field geometries.
densations. Basedonthecentrosymmetricpattern This paper is the fourth in a series which seeks
observed, Mannion & Scarrott (1984) ruled out to compare the polarization patterns (and in-
thepresenceofalignedgrainswithinthereflection ferred magnetic field geometries)in different star-
nebula. They infer the presence of a foreground formingregions. Theobservationsanddatareduc-
assembly of grains illuminated from behind solely tion techniques are described in 2. The polariza-
§
by the star HD 38563N. The position of this star tion data are presented and analyzed in 3. The
′ §
is 5 north-east of the dust emission on which implications of these data for the local magnetic
∼
wereportinthiswork,concidentwiththenear-IR field geometry andthat ofOrionB as a whole are
cluster observed by Lada et al. (1991a). discussed in 4, and 5 summarizes our results.
§ §
Polarization data probe only two directions of
the magnetic field geometry – those on the plane 2. Observations and Data Reduction
of the sky. Additionally, they provide no infor-
Using the UK/Japan polarimeter with the
mation about the strength of the magnetic field.
SCUBA (Submillimeter Common User Bolome-
Where field geometries are simple and the direc-
ter Array) detector at the James Clerk Maxwell
tion of the magnetic field does not vary through
Telescope2 (JCMT), we have mapped polarized
the cloud depth, the polarized emission detected
thermal emission from dust at 850 µm toward a
is perpendicular to the mean magnetic field and
filamentary dust cloud associated with the star-
the latter can be inferred simply by rotating the
◦
polarizationvectorsby90 . Ifthefieldhasamore 2The JCMT is operated by the Joint Astronomy Centre
complex, non-uniform geometry, then interpreta- onbehalfoftheParticlePhysicsandAstronomyResearch
tion becomes more difficult. In such cases, it is Council of the UK, the Netherlands Organization for Sci-
best to compare directly the polarization maps entific Research, and the National Research Council of
Canada.
with polarization patterns predicted from a phys-
2
forming regionNGC 2068. The observations were which negative and positive values seem equally
taken from 11 to 16 October 1999,and additional distributed can be considered to contain zero flux
data were added on 18 February 2000. The po- and be used for sky subtraction.) At 850 µm, the
larizer and general reduction techniques are de- sky is highly variable on timescales of seconds.
scribed in Greaves et al. (2001) and Greaves et This variability must be measured and removed
al. (2000). More information on data reduction from the data. Chopping removes the effects of
and systematic errors can be found in Paper II. slow sky variability; however, fast variations re-
Six different SCUBA fields were required to main in the data, which require sky subtraction
map the entire NGC 2068 filament. Initially, four using array bolometers devoid of significant flux.
fieldswereused,butthenthenumberoffieldswas We used between one and three bolometers to
increasedandthepositionsshifted;hence10inde- determine the sky variability, using the existing
pendent field centers were used. The field centers scan maps of total intensity at 850 µm (Mitchell
andotherobservingparametersaresummarizedin etal. 2001)tohelpselectempty bolometers. The
Table 1. The data were obtained using a 16 point methods ofsky subtractionare discussedin detail
jiggle-map mode, in which the telescope is “jig- in Jenness, Lightfoot & Holland (1998). As part
gled” in order to completely sample the SCUBA of our analyses of other regions, we determined
field. Choppingtoareferencepositionwasdoneto that sky subtraction can be done effectively with
removeskyeffects. Themaximumchopthrowpos- a single bolometer (see Paper II) and that the
sible is 180′′. We typically used150′′ to ensurewe POLPACK reduction process is extremely robust
were not chopping onto polarized emission. The with respect to the selection of sky bolometers
extinctionoftheatmosphere,τ (225GHz),ranged (see Paper III). We have not added the mean flux
from0.03to0.09overtheobservations,but>85% removed by sky subtraction back into the NGC
were taken when tau(225 GHz) was 0.06 or 0.07. 2068maps, since the flux in the sky positions was
sufficiently close to zero (as determined from the
We have reduced the data using the Starlink
largescaleintensitymapofMitchelletal. (2001)).
software package POLPACK, designed to include
polarizationdataobtainedwithbolometricarrays. Once the instrumentalpolarizations(IPs) were
The data have been corrected for an error in the removed from each bolometer, all the data sets
SCUBA computer’s clock which placed incorrect were combined to produce a final map in three
LST times in the data headers from July 1999 to Stokes’ parameters, I, Q and U, where I is the
May2000. Thiserrordidnotaffectthetelescope’s total, unpolarized intensity and Q and U are two
acquisition or tracking but affects data reduction orthogonalcomponents oflinearly polarizedlight.
sincetheelevationandskyrotationarecalculated These three Stokes’ parameters were then com-
from the LST times in the data files. The evo- bined to yield the polarization percentage, p, and
lution of the magnitude of this error over time polarization position angle, θ, in the map accord-
has been tracked and can therefore be corrected ing to the relations:
retroactively by adjusting the times in the head-
Q2+U2 1
ers (see the JCMT website for details). The error p= ; θ = arctan(U/Q). (1)
p I 2
in timing after this adjustment is 10s.
±
After extinction correction, noisy bolometers The uncertainties in each of these quantities are
were identified and removed from the data sets. given by:
The data were sky subtracted using bolometers
with mean values close to zero, but not those dp=p−1 [dQ2Q2+dU2U2]; dθ =28.6◦/σp
which were significantly negative. (The area p (2)
mapped by each bolometer in a single jiggle map where σp is the signal-to-noise in p, or p/dp.
can be identified with reasonably accuracy, since A bias exists which tends to increase the p value,
little sky rotation occurs during the 16-point jig- even when Q and U are consistent with p = 0,
gle. Hence, by examining the flux levels of pixels because p is forced to be positive. The polariza-
within a bolometer, those in which a majority tion percentages were debiased according to the
of adjacent pixels are negative were considered expression:
significantly negative and not used. Those for p = p2 dp2. (3)
db
−
p
3
Future referencesto polarizationpercentage,orp, 3. NGC 2068 Polarization Data
refer to the debiased value.
Figure2showsthepolarizationdatatowardthe
In order to improve the σ , and hence dθ, the
′′ p NGC 2068 filamentary cloud. These data are dis-
datawerebinnedto15 sampling. Thisrebinning
playedoveraportionofthe Mitchelletal. (2001)
improves σ by a factor of five over the unbinned
p
′′ map of Orion B North. The polarization data
3 sampled data. Good polarization vectors were
′′ ′′
◦ are binned to 15 , slightly greater than the 14
selected to have σ >3 (which implies dθ <10 ),
p
beamwidth of the JCMT at 850 µm. Thus, each
dp < 1%, p > 1% and be coincident with posi-
vector canbe treated as an independent measure-
tions where the total unpolarized 850 µm flux ex-
mentofthelocalpolarizationat850µm. Thedata
ceeds20%ofthefaintestofthesixcompactpeaks,
are presented in tabular form in Appendix A. By
OriBN 35. The filtering of polarizations less than
eye, one can readily discern that three different
1%accountsfortheuncertaintiesintheIPvalues,
populations of position angles exist within Figure
as well as for any contamination due to sidelobe
2. Separationof the data into three regions based
polarization,asdiscussedinGreavesetal. (2001).
on boundaries in RA can generally separate these
Sidelobecontaminationreferstoapolarizedsignal
populations. Figure 3 shows the distributions in
measured within the main beam due to a source
position angle over the whole map, and for each
withinasidelobe. Ifthereisenoughpolarizedflux
of the three subsections. Region 1 encompasses
(p I)fromasourceinasidelobeposition,asignal
× the eastern part of the map, including all vectors
canbeproducedinthemainbeamdespitethefact
eastoftheJ2000positionof05h46m39s.8. Region2
that the power in the JCMT sidelobes at 850 µm
is west of Region 1, extending to RA 05h46m33s.8
is typically 1% that ofthe main beam. Greaves
≤ (J2000). Region 3 covers the western section of
et al. (2001) derive the minimum believable po-
the map, including all vectors west of 05h46m33s.8
larization percentage, p , based on: the ratio of
crit
(J2000).
powerin the sidelobe to the mainbeam, P /P ;
sl mb
NGC 2068 contains six compact dust conden-
theratioofunpolarized,totalfluxofthesourcein
sations (Mitchell et al. 2001), as well as a dozen
thesidelobetothatinthemainbeam,F /F;and
sl
more amorphous, fainter condensations. In the
the IP estimate at the sidelobe, p :
sl
OMC-3 filament of OrionA, the polarizationpat-
Psl Fsl tern was no different in the presence of embed-
p 2 p . (4)
crit ≥ × sl(cid:18)P (cid:19)(cid:18) F (cid:19) ded cores than elsewhere along the filament (Pa-
mb
per II). However,in Figure 2,polarizationvectors
We can estimate the extreme value of pcrit for near cores are smaller and more variable in direc-
the entire NGC 2068 region by calculating the tion,particularlynearthecoresOriBN35,47and
worst case scenario. This occurs for the field cen- especially 51. In the latter case, no polarization
h m s
tered at α(J2000) = 05 46 31.2 and δ(J2000) = is detected towardthe core. Towardeachof these
◦ ′ ′′
00 0025.5, in which the brightest source in our cores, there is evidence of outflow; they are not
− ′′
map, OriBN 51, was approximately 54 from the pre-protostellar in nature. We note however that
center of the array. The beam responses and across the outflow source OriBN 39 and the com-
IP values are obtained from maps of unpolarized pact core OriBN 38, the polarization vectors are
planets. Using polarization maps of Saturn from consistentwiththepatternssurroundingthecores.
14 October 1999, 16 October 1999 and 18 Febru- The effect ofthe coresonthe polarizationpattern
ary 2000, the ratios of power in the sidelobe ver- indicates that the filament is not dominating the
sus main beamare deducedto be 0.009,0.01,and polarized emission as was the case in OMC-3 in
0.004 respectively. The corresponding mean IP Orion A (see Matthews & Wilson (2000) and Pa-
values at this sidelobe position are 4.5%, 4.7%, per II).
and 4.3%. The ratio of the flux of Ori BN 51
to the center of the field is 7. Hence, using 3.1. Polarization Position Angle and Fila-
∼
equation (4), the p values are determined to ment Orientation
crit
be 0.57%, 0.66%, and 0.24%. Thus, by selecting
InRegion1,themeanpolarizationpositionan-
onlyvaluesofp>1%,thepolarizationscannotbe
◦
gle is 109 east of north, which is equivalent to
attributable to sidelobe contamination.
4
◦
71 in linear polarization (since vectors offset larizationis high. The meanis not biasedby only
− ◦
by 180 are indistinguishable). The amorphous afewvectorsofparticularlyhighpolarizationper-
cores near the north-east have the highest po- centage.
larizations, and the vector orientations there are One of the most interesting properties of dust
roughly perpendicular to the projected filamen- emissionpolarimetryhasbeenthemeasurementof
taryaxis,alignedinaroughlynorth-easttosouth- declining polarization percentage with increasing
west orientation. Interestingly, the vectors along unpolarized intensity. Figure 4 shows this same
the three cores OriBN 34, 35 and 39 align well trend in the NGC 2068 region and its subregions
with those to the north-east, but now appear to that has been observedin other regions. Over the
lieparalleltothestringofthreecores. InRegion2, entiredataset,the variationofpwithtotalinten-
whichincludes thecoresOriBN37,38andpartof sity follows a power law of the form: p = A Iγ,
36, the vectors also follow the string of cores. Fi- whereγ = 0.81 0.04andA=(1.6 0.4) ×10−2.
nally,in Region3,whichincludes allcoreswestof No systema−tic va±riation in polariza±tion p×ercent-
OriBN42,thepolarizationvectorsaredistributed age between regions is evident from examination
◦
about a mean of 74 , although the projected fil- of Figure 4. We do note, however, that the effect
ament direction varies considerably. The vectors is strongest in Region 3. In Region 1, the vectors
thus tend to follow the filament near OriBN 42 surroundingthecoresOriBN31and32areshown
to 48, and then lie perpendicular to the filamen- as filled triangles, while the rest of the region’s
tary dust emission along the whole eastern edge. vectors are plotted as open triangles. It is clear
Overall,thereisnostrongcorrelationbetweenthe that the highest polarizations are observed at the
filament orientation and the vector position an- extreme north-east of the NGC 2068 region.
gles.
As discussed in Paper II, a depolarization ef-
Table 2 summarizes the statistics of the three fect can be produced systematically by chopping
data subsets plus the whole data set for the NGC the telescope during observing. However, based
2068 region, as well as the results of single Gaus- on that analysis, the steep profiles shown in Fig-
sianfitstoeachofthe positionangledistributions ure 4 could not be explained by this effect unless
shown in Figure 3. The mean position angle is there were significant polarized flux in the chop
denoted µ, while the width of the distribution is position. For example, if the reference, or chop,
w(θ). The reduced chi-squaredvalue is also listed position had 25% of the peak flux in the source
and in each case is <1. The fit to Region 3 gives field and was polarized to twice the degree of the
◦ ◦
a meanof66 with awidth of18 . Region2gives source field, then a slope of 0.88 could be pro-
a mean of 93◦ with a width of 16◦, and Region ducedonalogpversuslogI p−lotin regions of low
◦ ◦
1 yields a mean of 123 and width of 24 . The total intensity. Even under this extreme scenario,
results of the fits are consistent with the statisti- in regions of high flux, we would not expect such
cal mean and standard deviation in the position asharpdeclineinpolarizationpercentagewithin-
angles. tensity. We thus conclude that the depolarization
effect in this regionis notproducedby systematic
3.2. Polarization Percentage
effects of chopping.
The results of a basic statistical analysis of the
4. Discussion
polarization percentage distributions in the NGC
2068 cloud and its subset of three regions are re-
4.1. Polarization Percentage in Orion B
portedinTable2. Themeanpolarizationpercent-
age in the whole filamentary cloud is 5.0% with Figure5showsthedistributionsofpolarization
a standard deviation of 3.1%. Similar values are percentageintheregionsofOrionBobservedwith
found for each separate region within NGC 2068. theSCUBApolarimeter. Thesedistributionshave
Nosystematicvariationinpolarizationpercentage been normalized to the total number of vectors
is obvious from examination of Table 2. Interest- measuredineachregion,andthecountsexpressed
ingly, Region 2 has the highest median value of p, asafractionofthe total. Theerrorbarsrepresent
anditistheonlyregionwherethemedianexceeds the √N statistical uncertainty in the number of
the mean, which indicates that in this region, po- counts, also normalized to the total in each map.
5
The distribution for NGC 2068 is comparable to ing dark lane. (North of the H II region lies an-
those measured in the other star-forming regions otherdustcondensationnotyetobservedwiththe
ofOrionBNorth– NGC 2071andLBS23N–and polarimeter.) Hence,itislikelythatthedustemis-
with measurements toward the OMC-3 filament sionmaylieontheouteredgeofthereflectionneb-
in Orion A (Paper II). The NGC 2024 region is ula with the western material in the foreground
dominated by low percentage polarizations, with and the eastern material behind. Thus, it is clear
the majority between 1-2%. This result could im- that this filament does not lie in the plane of the
ply NGC 2024 has weaker magnetic fields, poorer sky, and that the inclination on the sky is likely
grainrotationoralignment,adifferentgraincom- variable.
position, or some combination of these factors. An obvious question is whether or not all the
The fact that NGC 2071, NGC 2068 and LBS dust emission arises from material which is spa-
23N show distributions which are reasonably flat tially related. It is noteworthy that the core
from 1-6%could indicate similar grainproperties, OriBN 34 does not appear to line up well with
field strengths, and degrees of grainaligment,but the filamentary material of Figure 2. In fact, our
this cannot be proven with these data. The den- comparisonoftheopticalanddustemissionshows
sities toward the NGC 2024 cores have been esti- that the location of OriBN 34 is optically dark.
8 −3
matedtobeabnormallyhigh(10 cm Mezgeret This could be a coincidence, or OriBN 34 could
al. (1988)), but Schulz et al. (1991) suggest the becloserthanthedustemissionappearingnextto
valuesaremoretypicalofcores(106cm−3). Polar- it on the SCUBA image. OriBN 39 also appears
izationdataatotherwavelengths(suchas350µm) dark and could be in the foreground. Mitchell et
couldhelpconstrainthedustpropertiesaccording al. (2001) present 13CO J = 2 1 toward NGC
to models (see Hildebrand et al. (2000) and ref- 2068andapartialmapofthenor−th-easternregion
erencestherein), andobservationsofdense molec- in C18O J =2 1 (see their Figs. 6 and 7). The
ular tracers like OH for Zeeman splitting could 13COcontoursa−recloselycorrelatedtothe850µm
provide more detailed information about the field dustemission. ThesameistruefortheC18Oemis-
geometries in these four regions. sion where data exist. These maps are integrated
Toward all four regions of Orion B observed over velocity ranges from 5 km s−1 < v < 15
LSR
in polarized emission, the depolarization effect is km s−1 and 7 km s−1 < v < 13 km s−1, re-
LSR
detected (see Fig. 5 of Paper III and Figure 4). spectively. Thus, the emission is confined to the
Paper III shows that NGC 2024 exhibits a simi- Orion B cloud. The fact that OriBN 34 and 39
lar depolarization signature to the northern core show similar polarization position angles as the
NGC 2071. However,LBS 23N has a significantly eastern area but may be in the foreground would
steeper slope of 0.95, which is more consistent argue against these regions being completely spa-
−
with that of Region 3 in our data set. This trend tially distinct (unless the cores are not contribut-
has been observed in many other regions as well, ing asmuchpolarizedemissionas the background
including massive cores such as OMC-1 (Schleun- diffuse material. If the material is spatially sepa-
ingetal. 1997)andprotostellarandstarlesscores rated,thenthesimilarorientationsofthepolariza-
(Girartetal. 1999;Ward-Thompsonetal. 2000). tionpositionanglescouldsuggestthatthe field, if
not defined by a single meanfield direction, could
4.2. Evidence for Varying Inclinations in be organized on spatial scales at least as large as
NGC 2068 the separation between them.
The filamentary dust emission of NGC 2068 The 2.4 µm emission from the NGC 2068 re-
′
likely arises from dust at different depths in the flection nebula spans approximately 6 in spatial
extent(Sellgren1984). Thisisconsistentwiththe
OrionBcloud. ThecomparisonbetweenthePalo-
mar Observatory Sky Survey optical data and physical size of the nebula in optical emission as
the dust emission has been made by Mitchell et shown in Figure 1, and at a distance of 415 pc
corresponds to 0.7 pc. If the nebula is as deep as
al. (2001). We produce a similar image in Fig-
ure 1, which shows that while the western dust it is wide, then the source of dust emission at the
north-east could be more than a parsec displaced
emissionlinesupwellwiththeopticallydarkdust
′
lane,theeasterndustemissionhasnocorrespond- spatially from the cores at the west (taking 7 as
6
an estimate of their separation in projection). region must be modeled separately.
4.3. Field Geometry 4.3.2. More Complex Geometries
4.3.1. A Single Mean Field Direction As discussed above, there is evidence that the
dust arises from a connected gas structure. How-
The vectors of NGC 2068 do not support a
ever, the spatial separation between the filament
single field orientation in NGC 2068. If one as-
edges could be > 1 pc. The regions of different
sumes that the magnetic field geometry through-
position angles could indicate regions of different
outacloud’sdepthisreasonablywelldefinedbya
field geometry or inclinations. We propose two
meanfielddirection,thenthedirectionofthefield
model geometries which could potentially explain
can be obtained by rotating emission polarization
the variable position angles observed.
◦
databy90 . DoingthisintheNGC2068polariza-
Since the cores are mainly aligned along a fila-
tion map will clearly not produce aligned vectors,
mentary structure, a helical field geometry is ap-
any more than the polarization data themselves
pealing. It can explain the confinement of gas in
arealigned. Therefore,wecanruleoutauniform,
the filament, the fragmentation to cores and the
unidirectional field across all of NGC 2068. This
elongationoftheasymmetricconcentrationsalong
wasalsothecaseinOMC-3inOrionA(seePaper
theaxisofthefilament,whichiscertainlythecase
II); however, in that region, the filamentary axis
for OriBN 31, 32, 33, 41 and 49. Fiege & Pu-
was easy to define, and a strong correlation be-
dritz (2000b) show possible polarization patterns
tweentheaxisandpolarizationpositionanglewas
fordifferenthelicalfieldconditions(i.e.poloidally-
observed along 75% of the filament. The vectors
dominated versus toroidally-dominated). These
ofNGC2068donotshowanobviousalignmentof
models are developed for straight filaments, and
polarization position angle with the filament ori-
Fiege & Pudritz (2000b) find that for such fila-
entation.
ments, the polarization vectors align either paral-
We can compare the polarization position an-
lel or perpendicular to the projectedfilament axis
glesacrossthoseregionsobservedthusfarinOrion
regardlessof the inclination of the filament. How-
B. In the three regions of Orion B North, no ev-
ever, Paper II presents a model for a helical field
idence for similar polarization orientations exists.
threadingabentfilament. Inthiscase,thevectors
Toward NGC 2071, the polarization vectors align
mayadoptanyorientationrelativetothefilament
generallywiththeprominentoutflowfromtheIRS
due to the asymmetries in the filament. The rel-
◦
1 source at a position angle of 40 east of north.
ative alignmentdepends on the filament’s inclina-
Within the LBS 23N string of cores to the south
tion and rotation on the plane of the sky.
of NGC 2068, the vectors are generally aligned
◦ Thus, as inclination and angle in the plane of
north-south(positionangle0 )althoughthe scat-
the sky vary, a helically-threaded filament should
ter in this faint region is considerable. Neither
produce different polarization position angles rel-
of these orientations is dominant in NGC 2068.
ative to the projected filament orientation. For
Thus, there is no support for a mean field direc-
instance, if the polarization positions are aligned
tioninNGC 2068oracrossthe threestar-forming
with the filament, a toroidally-dominated heli-
regions of Orion B North.
cal field could wrap the filament locally. Con-
The one region observed in Orion B South is
versely, where the vectors appear perpendicular
NGC2024,andthepolarizationpatternfromthat
tothefilament,ahelicalfieldwouldbepoloidally-
region has been modeled as arising either from a
dominated, or the filament and field could be sig-
helicalfieldgeometryorfromtheexpansionofthe
nificantly bent. At the north-western edge near
ionization front due to the associated H II region
thecoresOriBN33and41,thevectorsareneither
in Paper III. The latter geometry is favored since
parallel nor perpendicular to the filament. This
it is most compatible with the total physics and
could also indicate that the filament is bent along
geometry of NGC 2024. It is clear that the polar-
its length in this area. Modeling such a complex
ization patterns across these star-forming regions
filamentary structure will be difficult, since more
can be strongly correlated with the dust and gas
thanonefilament/fieldgeometrycanproducesim-
structures of a particular region, and that each
ilar polarization patterns.
7
One additional test for the presence of a heli- reflection nebula could produce a similar effect in
calfield geometryis the radialprofile ofthe total, NGC2068,althoughthepatternproducedismuch
unpolarized emission. Using the 850 µm map of more complex. In NGC 2024,the ridge is entirely
Mitchelletal. (2001),aradialprofilecanbebuilt located on the far-side of the HII region,whereas
upbytakingseveralslicesacrossthefilament. We in NGC 2068 the comparison of optical and 850
have attempted to confine the cuts to the regions µm dust emission in Figure 1 indicates that the
between cores along NGC 2068. Figure 6 shows dust emission arises from both the near and far-
the profiles through several such slices, taken be- side of the nebula.
tweentheOriBN44and48,between32and39and A recent publication on the formation of qui-
between35and36. Forthesecondcut,bothsides escent cores through turbulent flows (Padoan et
oftheprofilewereused;fortheothertwo,oneside al. 2001) provides a third possible interpretation
of the slice was discarded due to the presence of of the polarization pattern. This work discusses
extended emission from the filament near OriBN the potential formation of cores due to the pres-
47(forthe cutbetweenOriBN44and48)andex- ence of super-sonic turbulent flows within molec-
tended emission from core OriBN 34 (for the cut ular clouds. In this model, cores form by accre-
between OriBN35 and 36). The fluxes have been tionalongfilamentarystructures,withthebright-
normalizedto the maximumthroughthe cuts (re- estcoresformingatthelociofintersectingshocks.
spectively 0.07 Jy beam−1, 0.32 Jy beam−1, and ThecoreOriBN47liesattheintersectionofthree
0.28 Jy beam−1); this position is assumed to be filamentary segments. As predicted by Padoan et
the axis of the filament. The NGC 2068 filament al. (2001), this core exhibits depolarization, al-
is so narrow and faint, it is difficult to interpret though at the distance of Orion B, it is difficult
the profiles. To guide the eye, we have drawn on toachieveagoodsamplingofpolarizationvectors
Figure 6 lines corresponding to slopes of -0.5 and across this core. Thus, we do not see the large
-1. Near the axis, the flux falls off with a distri- changesinpositionanglepredictednearcores(see
bution consistent with a powerlaw index of 0.5. their Fig. 3) and routinely observed in Bok glob-
−
Toward lower fluxes, the index could be closer to ules andstarlesscoresincloserregionsofstarfor-
1; however, at these low flux levels, interpreta- mation (e.g. see Ward-Thompson et al. (2000)).
−
tionoftheindexbecomesdifficult. Thefluxprofile However,thepolarizationpatternalongthelarger
of r−0.5 corresponds to a density profile of r−1.5 scale filamentary structure is well sampled in our
foranisothermalfilament. Thehelicalfieldgeom- map. In the turbulent flows model, the polariza-
etrypredictsadensityprofilewithindex 2foran tion vectors along the filamentary structures are
−
isothermal equation of state. For other equations seentoalignwellwiththefilaments’axes. InNGC
ofstate(e.g.thelogotrope,McLaughlin&Pudritz 2068, only the filament segment east of OriBN 48
(1996)), a shallower range of density profiles can exhibitsthisbehavior. Thesegmentstothenorth-
begenerated. Unmagnetized,isothermalfilaments westandsouth-westshowvectorsorientedroughly
predict much steeper profiles, with indices of 4, perpendicular to the filamentary structure, which
−
which are clearly not consistent even with these does not agree well with the turbulent flows pic-
poor profiles. ture. A simulation of turbulent flows along fila-
Asecondpossiblefieldgeometryissuggestedby mentsofhigherdensitymayagreebetter withour
the position of the molecular filament so near the observations in NGC 2068. If a threshold in ex-
periphery of the reflection nebula of NGC 2068. tinction exists beyond which grains are not effec-
PaperIIIpresentsamodelofthepolarizationdata tively aligned (Padoan et al. 2001), then denser
towardtheNGC2024denseridgeofcoresinwhich turbulent flows could exhibit less correlation be-
thefield,compressedbytheexpansionoftheNGC tween the filamentary axis andthe inferredpolar-
2024HIIregion,isthenmouldedaroundthedense ization.
ridge as the ionizationfront approachesthe cores.
This picture accounts for both the polarization 5. Summary
pattern at 850 µm and the measurements of the
We have detected polarized emission from
line-of-sightfield directionand strengthmeasured
aligned dust grains at 850 µm with the SCUBA
by Crutcher et al. (1999). The expansion of the
8
polarimeter. These data reveal strong degrees of ficult,sincetheremaybemanydegeneratechoices
polarization (up to 17%) toward the NGC 2068 offilament/fieldconfigurationswhichcanproduce
region, with higher polarization percentages de- the observed polarization pattern.
tected toward regions of fainter total intensity. These are the first observations of polarized
This depolarization effect has been observed in emissiontowardthisfilament. Thestrengthofthe
most regions of polarized emission. Significant polarization percentages detected suggests that
depolarization has been detected toward several polarized emission should be detectable at other
compact cores along the filament. Within the wavelengths. Forinstance,observationsat350µm
cores of LBS 23N and some cores along the NGC with the Hertz polarimeter would provide some
2024 dense ridge, significant depolarization was information on the polarization spectrum for this
also measured (see Paper III) We note that this region. Thiscouldleadtoconstraintsonthegrain
was not the case in the OMC-3 filament of Orion population along the filament. Additionally, Zee-
A, where the polarization patterns behaved con- man data on B field would provide significant
los
sistently in the presence or absence of embedded constrainttothethree-dimensionalfieldstructure.
cores (see Paper I and Paper II). This was at-
tributed to the fact that filamentary dust dom-
The authors thank the members of the Cana-
inated the polarization pattern in Orion A, but
dian Consortium on Star Formation for the 850
that this is not the case in Orion B.
µmscanmapoftheOrionBNorthcloud. Thanks
The polarization pattern is inconsistent with to J. Greaves, T. Jenness, and G. Moriarty-
that expected for a uniform field geometry, for Schieven at the JCMT for assistance during and
which vectors should be aligned with one another after observing. The research of BCM and CDW
acrosstheentireregion. Thepolarizationposition has been supported through grants from the Nat-
angles are roughly organized into three distinct ural Sciences and Engineering Research Council
distributions which do not align with the orienta- of Canada. BCM acknowledges funding from On-
tions of the filamentary total intensity emission. tario Graduate Scholarships during the period of
This is also different than the pattern observedin this research.
the OMC-3 part of the Integral-shaped Filament
in Paper II where the vectors aligned with the REFERENCES
filament orientation along 75% of the filament’s
Alves, J., Lada, C.J., & Lada, E.A. 1999, ApJ,
length. Taken in comparison with the polariza-
515, 265
tion data for three other star-forming regions in
Orion B (Paper III), there is also no evidence for Anthony-Twarog,B. J. 1982,AJ, 87, 1213
a single field orientationacross the Orion B cloud
as a whole. Basu, S. 2000, ApJ, 540, 103
Comparison of optical and 850 µm dust emis- Carpenter, J. M. 2000, AJ, 120, 3139
sionestablishesthatthe dustfilamentis notlying
in the plane of the sky, but must be twisted such Crutcher, R. M., 1999,ApJ, 520, 706
that its western edge is in the foreground of the
Crutcher, R.M., Roberts, D.A., Troland, T.H., &
optical reflection nebula, while the eastern edge
Goss, W.M. 1999, ApJ, 515, 275
lies behind. These changes in inclination can help
explain the polarization pattern observed. Paper Edwards, S. & Snell, R.L. 1984, ApJ, 281, 237
II shows that a bent filament threaded by a heli-
Fiege,J.D.,& Pudritz,R.E.2000a,MNRAS,311,
calfieldgeometrycanproducepolarizationvectors
85
offset from the filament orientations by varying
degrees. Material lying on the edge of an expand- Fiege,J.D., & Pudritz,R.E.2000b,ApJ,544,830
ing reflection nebula could also exhibit different
magnetic field orientations at different locations Gibb, A.G., & Little, L.T. 2000, MNRAS, 313,
depending on how effectively the field has been 663
compressed by the expansion. The complexity of
Girart, J.M., Crutcher, R.M., & Rao, R. 1999,
this filamentary structure will make modeling dif-
ApJ, 525, 109
9
Greaves, J.S., Holland, W.S., Jenness, T., Mitchell, G.F., Johnstone, D., Moriarty-Schieven,
Moriarty-Schieven, G., Chrysostomou, A., G.,Fich,M.,&Tothill,N.F.H.2001,ApJ,556,
Berry, D.S., Murray, A.G., Nartallo, R., Ade, 215
P.A.R., Gannaway, F., Haynes, C.V., Tamura,
Motte, F., Andr´e, P., Ward-Thompson, D., &
M., Momose, M., & Morino, J.-I. 2001, MN-
Bontemps, S. 2001,A&A, 372, 41
RAS, submitted
Mouschovias,T. C. 1987,in Physical Processes in
Greaves, J.S., Jenness, T., Chrysostomou, A.C.,
Interstellar Clouds, 453
Holland, W.S., & Berry, D.S. 2000, in Imaging
at Radio through Submillimeter Wavelengths,
Myers, P.C., & Goodman, A.A. 1988, ApJ, 326,
eds. J.G. Mangum & S.J.E. Radford, ASP-CS
L27
217, 18
Padoan, P., Goodman, A.A., Draine, B., Juvela,
Hildebrand, R.H. 1988, QJRAS, 29, 327 M., Nordlund, ˚A., & Ro¨gnvaldsson, O¨. 2001,
ApJ, 559, 1005
Hildebrand,R. H., Davidson,J.A., Dotson,J.L.,
Novak, G., & Vaillancourt, J. E. 2000, PASP,
Schleuning,D.A.,Dowell,C.D.,Hildebrand,R.H.,
112, 1215
Platt, S.R., & Novak, G. 1997,PASP, 109,307
Jenness, R., Lightfoot, J.F., & Holland, W.S.
Schulz, A., Gu¨sten, R., Zylka, R., & Serabyn, E.
1998,Proc. SPIE, 3357, 548
1991,A&A, 246, 570
Johnstone, D., & Bally, J. 1999,ApJ, 510, L49
Sellgren, K. 1984, ApJ, 277, 623
Lada, C.J., Alves, J., & Lada, E.A. 1999, ApJ,
Ward-Thompson,D., Kirk, J.M., Crutcher, R.M.,
512, 250
Greaves,J.S.,Holland,W.S.,&Andr´e,P.2000,
ApJ, 537, 135
Lada,E.A., Bally,J., &Stark,A.A. 1991a,ApJ,
368, 432
Lada,E.A.,DePoy,D.L.,Evans,N.J.,&Gatley,
I. 1991b, ApJ, 371, 171
Mannion, M.D., & Scarrott, S.M. 1984, MNRAS,
208, 905
Matthews,B.C.,&Wilson,C.D.2000,ApJ,531,
868 (Paper I)
Matthews, B.C., Fiege, J.D., & Moriarty-
Schieven, G. 2001a, submitted to ApJ(Paper
III)
Matthews, B.C., Wilson, C.D., & Fiege, J.D.
2001b, ApJ, 562, 400 (Paper II)
McKee, C. F., Zweibel, E. G., & Goodman, A. A.
1993,in Protostars and Planets III, 327
McLaughlin, D. E., & Pudritz, R. E. 1996, ApJ,
469, 194
Mezger,P.G.,Chini,R.,Kreysa,E.,Wink,J.W.,
& Salter, C. J. 1988, A&A, 191, 44
This2-columnpreprintwaspreparedwiththeAASLATEX
macrosv5.0.
10