1: % SAMPLE1.TEX -- AASTeX sample paper with minimal markup.
2: \documentclass[12pt,preprint]{aastex}
3:
4: \usepackage{emulateapj5}
5:
6: \def\hii{H\,{\sc ii}}
7: \def\hi{H\,{\sc i}}
8: \def\kms{\relax \ifmmode {\,\rm km~s}^{-1}\else \,km~s$^{-1}$\fi}
9: \def\cm-3{\relax \ifmmode {\,\rm cm}^{-3}\else \,cm$^{-3}$\fi}
10: \def\Jb{\relax \ifmmode {\,\rm Jy\,beam}^{-1}\else \,Jy\,beam$^{-1}$\fi}
11: \def\mJb{\relax \ifmmode {\,\rm mJy\,beam}^{-1}\else \,mJy\,beam$^{-1}$\fi}
12: \def\deg{\hbox{$^\circ$}}
13: \def\arcmin{\hbox{$^\prime$}}
14: \def\arcsec{\hbox{$^{\prime\prime}$}}
15: \def\secd#1.#2{ #1\farcs#2 } % seconds over decimal point
16: \def\e{$\pm$}
17: \def\x{$\times$}
18:
19: \newcommand {\ha}{H$\alpha$}
20:
21:
22: \begin{document}
23:
24: \title{Interferometric Mapping of Magnetic Fields in Star-forming Regions I. W51 e1/e2 Molecular Cores}
25: \author{Shih-Ping Lai, Richard M. Crutcher,
26: Jos\'e M. Girart\altaffilmark{1}, and Ramprasad Rao\altaffilmark{2}}
27: \affil{Astronomy Department, University of Illinois, 1002 W. Green Street, Urbana, IL 61801; \\
28: slai@astro.uiuc.edu, crutcher@astro.uiuc.edu, jgirart@am.ub.es, ramp@oddjob.uchicago.edu}
29:
30: \altaffiltext{1}{Current address: Departament d'Astronomia i Meteorologia, Universitat de Barcelona, 08028 Barcelona, Catalunya, Spain}
31: \altaffiltext{2}{Current address: Department of Physics, University of Chicago}
32:
33: \begin{abstract}
34: We present the first interferometric polarization map of the W51 e1/e2
35: molecular cores obtained with the BIMA array at 1.3 mm wavelength
36: with approximately 3\arcsec\ resolution.
37: The polarization angle varies smoothly across the double
38: cores with an average position angle of 23\deg\e5\deg\ for W51 e1 and
39: 15\deg\e7\deg\ for W51 e2. The inferred magnetic field direction is
40: parallel to the minor axis of the double cores, which is consistent with
41: the theoretical picture that clouds collapse along the field lines.
42: However, the magnetic field may not determine the axis of angular momentum
43: of these two cores as the field directions of the two cores
44: significantly differ with the previously measured directions of
45: rotational axes.
46: The polarization percentage decreases toward regions with high intensity,
47: suggesting that the dust alignment efficiency decreases toward high density
48: regions. The field directions are highly ordered, and the small dispersion
49: of the polarization angles implies that magnetic fields are strong
50: ($\gtrsim$ 1 mG) and perhaps dominate turbulence in W51 e1/e2.
51:
52: \end{abstract}
53:
54: \keywords{ISM: magnetic fields -- ISM: individual: W51 -- polarization
55: -- star: formation -- techniques: interferometric }
56:
57: %\newpage
58:
59: \section{Introduction}
60:
61: Star formation in molecular clouds is one of the most fundamental
62: problems in astrophysics. It has become increasingly evident that
63: star formation cannot be fully understood without considering magnetic
64: fields (see recent reviews by Mouschovias \& Ciolek 1999, and Shu et
65: al.\ 1999). Magnetic fields provide explanations for
66: the support of clouds against self-gravity, the formation
67: and evolution of cloud cores, the origin of supersonic line widths,
68: the low specific angular momentum of cloud cores and stars,
69: and the formation of bipolar outflows.
70: Unfortunately, the magnetic field remains the most poorly measured
71: quantity in the star formation process.
72: In order to advance our understanding of star formation,
73: it is essential to improve our empirical knowledge of
74: magnetic fields in molecular clouds.
75:
76: The most promising technique to probe magnetic field morphology
77: in dense molecular cores is to measure the linear polarization
78: of the thermal emission from spinning, magnetically aligned dust grains
79: (Heiles et al.\ 1993). Interstellar dust grains are elongated with
80: their minor axes aligned with the magnetic field,
81: thus the magnetic field direction in the plane of sky is perpendicular
82: to the direction of polarization (Davis \& Greenstein 1951; Roberge 1996).
83: Millimeter interferometers detect the dust emission
84: in the low optical depth regime with high angular resolution,
85: thereby revealing the field morphology in the dense cores.
86: In the past few years, linear polarization observations with
87: the Berkeley-Illinois-Maryland Array (BIMA) of Orion-KL
88: and NGC\,1333 IRAS\,4A have demonstrated the capability
89: and reliability of polarization observations with this telescope
90: (Rao et al.\ 1998; Girart et al.\ 1999).
91: These results are not only consistent with the previous
92: single-dish observations, but also provide new information
93: at resolutions up to 3\arcsec. Therefore, it is reasonable for us
94: to carry out a large linear polarization survey of star-forming cores
95: with BIMA.
96:
97: W51 is a large molecular cloud/\hii\ region complex at a distance
98: of 7.0$\pm$1.5 kpc (Genzel et al.\ 1981). Associated with ultracompact
99: (UC) \hii\ regions (Scott 1978), strong infrared and submillimeter continuum
100: emission (Thronson \& Harper 1979; Jaffe, Becklin, \& Hildebrand 1984),
101: molecular gas (Ho, Genzel, \& Das 1983) and H$_2$O and OH masers
102: (Genzel et al.\ 1981; Gaume \& Mutel 1987),
103: W51 is one of the most active high-mass star formation regions
104: in our Galaxy. W51 e1 and W51 e2 are molecular cores located in the
105: eastern edge of W51. The mass of W51 e1 and e2 are $\sim$ 150 and
106: 110 $M_{\odot}$, derived from NH$_3$ observations (Zhang \& Ho 1997).
107: With infall motions identified, these molecular cores are
108: in the process of forming OB associations (Ho \& Young 1996).
109: Subarcsecond VLA observations reveal that W51 e1 contains
110: three UC \hii\ regions, e1, e3, e4 (Gaume, Johnston, \& Wilson 1993),
111: and a possibly dust source, e8 (Zhang \& Ho 1997).
112: In the vicinity of the UC \hii\ region e2,
113: there is a compact water maser concentration known as W51 Main
114: (Genzel et al.\ 1981).
115: Previous single-dish measurements with resolution $\sim$30\arcsec\
116: show very low polarization in W51 e1/e2
117: consistent with non-detection (Kane et al.\ 1993; Dotson at al.\ 2000);
118: however, this could be caused by averaging over the large beams.
119: Our observations with 10 times better resolution
120: will allow us to explore the field structure in the W51 e1/e2 cores.
121:
122: In this paper, we present the first interferometric polarization
123: map of the W51 e1/e2 cores at 1.3 millimeter wavelength.
124: Other sources in our survey will be reported in later papers.
125:
126: \section{Observations and Data Reduction}
127:
128: Several observations were carried out from 1999 August to 2000 April
129: using nine BIMA antennas with 1-mm SIS receivers and quarter-wave plates.
130: The digital correlator was setup to observe the continuum emission
131: with a 750 MHz window centered at 226.9 GHz in the lower sideband
132: and a 700 MHz window centered at 230.9 GHz in the upper sideband.
133: Strong CO $J$=2--1 line emission was isolated in
134: an additional 50 MHz window in the upper sideband.
135: The primary beam is $\sim$ 50\arcsec\ at 1.3 mm wavelength.
136: Data were obtained in the B, C and D array configurations
137: with projected baseline ranges from 6--170 kilowavelengths,
138: resulting in a linear polarization map of the W51 e1/e2 cores
139: with a long integration time, $\sim 25$~hours.
140:
141: The BIMA polarimeter and the calibration procedure are described
142: in detail by Rao et al.\ (in preparation).
143: A pair of quarter-wave plates was placed in front of the linearly
144: polarized receiver of each antenna in order to enable
145: the detection of the left ($L$) and right ($R$) circularly
146: polarized radiation. Four cross correlations,
147: $LL,\ RR,\ LR$, and $RL$, must be measured at each baseline
148: in order to obtain the four Stokes parameters, $I$, $Q$, $U$, and $V$
149: (Thompson, Moran, \& Swenson 1986).
150: Because there was only one receiver at each BIMA antenna,
151: the quarter-wave plates were switched on a time scale shorter
152: than the $uv$-cell transit time to achieved quasi-simultaneous
153: dual polarization observations.
154: Walsh function switching patterns (Thompson, Moran, \& Swenson 1986;
155: Harmuth 1969) were used to maximize the efficiency of obtaining four cross
156: correlations (Rao 1999).
157: Sixteen switch patterns were needed to complete a cycle.
158: The integration time of each pattern was set to 11.5 seconds
159: and the plate switch took an additional 2--3 seconds;
160: therefore, one cycle was about 4 minutes.
161:
162: Calibration and data reduction were carried out using
163: the MIRIAD software (Sault, Teuben, \& Wright 1995).
164: The instrumental polarization response (``leakage")
165: was calibrated by observing the strongly polarized quasars
166: 3C273 or 3C279 over a wide hour angle range, typically
167: more than 5 hours. The observed polarization, including
168: contributions from the instrumental polarization
169: and the polarization of the calibrator, varied with hour angle
170: due to the fact that the polarization vector of the calibrator
171: rotates with respect to the linear feed horn.
172: On the other hand, the instrumental polarization is constant with time.
173: Therefore, the leakages can be solved by assuming that the
174: polarization of the calibrator was constant in the frame of sky
175: over the whole track.
176: Task GPCAL of MIRIAD was used to solve simultaneously for the leakages
177: and antenna gains from the polarization calibrator.
178: The average leakage of each antenna was 5.9\%\e0.4\%
179: for our observations. The leakage correction of each antenna was
180: applied to visibility data before further processing.
181:
182: Channels with line emission were carefully flagged out in the visibility
183: data of the continuum bands.
184: As the continuum emission of the the W51 e1/e2 cores was stronger than
185: the phase calibrator (QSO 1751+096), self-calibration was performed
186: for refining antenna gain solutions. The Stokes $I$ image was made
187: with Briggs' robust weighting of 0.5 (Briggs 1995; Sault \& Killeen 1998)
188: to acquire a smaller beam size (2\farcs7\x2\farcs0, PA=1\deg)
189: for better determination of the self-calibration model.
190: The model was used to calibrate the visibility data in order to
191: obtain new gain solutions with shorter time intervals.
192: The above iterations were repeated until the Stokes $I$ image
193: was not sufficiently improved.
194: The Stokes $Q$ and $U$ images were then made with natural weighting
195: to obtain the highest S/N ratio. The resulting synthesized beam
196: was 3\farcs2$\times$2\farcs3 with PA=2\deg.
197: Maps of Stokes $Q$ and $U$ were deconvolved and binned to approximately
198: half-beamwidth per pixel (1\farcs5\x1\farcs2) to reduce oversampling
199: in our statistics. We then combined the maps to obtain the observed
200: linearly polarized intensity,
201: \begin{equation}
202: I_{p,obs}= \sqrt{Q^2+U^2},
203: \end{equation}
204: \noindent and the polarization position angle,
205: \begin{equation}
206: \phi = 0.5~\tan^{-1}{(U/Q)}.
207: \end{equation}
208: \noindent As the polarized intensity is an intrinsically positive
209: quantity, Eq.\ (1) will tend to overestimate the polarized intensity.
210: Therefore, a bias correction for the polarized intensity must
211: be performed (Leahy 1989),
212: \begin{equation}
213: I_p = \sqrt{I_{p,obs}^2-\sigma_{I_p}^2},
214: \end{equation}
215: \noindent where the rms noise value of polarized intensity, $\sigma_{I_p}$,
216: was taken as the average value of the rms noise in the $Q$ and $U$ maps,
217: as the noise levels in these two maps are comparable.
218: The polarization percentage was calculated from
219: \begin{equation}
220: p=\frac{I_p}{I}.
221: \end{equation}
222: The measurement uncertainty in the position angle,
223: \begin{equation}
224: \sigma_{\phi} = 0.5~\tan^{-1}{(\sigma_{I_p}/I_p)},
225: \end{equation}
226: \noindent depends on both $\sigma_{I_p}$ and $I_p$, thus it varies across
227: the map. When weighted with $I_p$, the average measurement uncertainty
228: in the position angle for our observations was 5\fdg6\e1\fdg7.
229:
230:
231: \section{Results and Analysis}
232:
233: Figure 1 displays a contour map of the 1.3-mm continuum from the W51 e1/e2
234: cores overlaid with polarization vectors.
235: Polarization is called detected at positions where the linearly polarized
236: intensity is greater than 3$\sigma_{I_p}$ (1$\sigma_{I_p}$ = 4.7 \mJb)
237: and the total intensity is greater than 5$\sigma_I$
238: (1$\sigma_I$ = 27 \mJb, which is dominated
239: by incomplete deconvolution rather than thermal noise).
240: Under these criteria, the polarized emission extends over an area of
241: $\sim$12 beam sizes.
242: Table 1 lists the polarization measurements in the W51 e1/e2 cores
243: at selected positions, separated by approximately the synthesized beamsize.
244: The distributions of the polarization angle and the polarization percentage
245: are shown in Figure 2 and 3, respectively.
246:
247:
248: \subsection{Continuum and Polarized Emission}
249:
250: The south and north continuum condensations of the double
251: cores in Figure 1 correspond to W51 e1 and W51 e2, respectively.
252: W51 e1 is extended (FWHM=5\farcs0\x3\farcs0, PA=11\deg)
253: and its integrated flux is 8.2 Jy.
254: W51 e2 is also resolved (FWHM=4\farcs4\x3\farcs3, PA=$-$8\deg)
255: with an integrated flux of 13.6 Jy.
256: We will refer to the intensity maxima of these two cores as
257: ``e1 1mm peak'' and ``e2 1mm peak''.
258: Analysis of the continuum flux at multiple wavelengths has shown that
259: W51 e1 and W51 e2 contain both free-free emission and dust emission
260: (Rudolph et al.\ 1990). The expected flux of the free-free emission
261: at 1.3 mm wavelength for these two cores can be estimated
262: from their SEDs (Rudolph et al.\ 1990), which are $\sim$ 0.5 Jy
263: for both cores. Therefore, the average fractions
264: of the free-free emission in W51 e1 and e2 are $\sim$6\% and $\sim$4\%.
265:
266: The polarized emission does not show two distinct peaks associated with
267: W51 e1 and e2, but it mainly arises from an extended region across
268: the two cores. The peak of polarized intensity is at 1\farcs2 south of
269: e1 1mm peak (Table 1). There is also a compact region of polarized
270: emission $\sim$4\arcsec\ northwest of e2 1mm peak
271: (for convenience, we name this position ``e2 pol NW"; cf.\ Fig.\ 1).
272: The polarized flux drops to zero between e2 1mm peak and
273: e2 pol NW, and e2 pol NW shows very different polarization angle.
274: Therefore, the polarization in e2 pol NW may sample a different region,
275: and is excluded in our statistical analysis.
276:
277: Most of the polarized flux is associated with W51 e1.
278: The elongated shape of W51 e1 is due to the fact
279: that it is composed of emission from nearby UC \hii\ regions
280: (e1, e3 and e4), a newly discovered continuum source
281: (e8: Zhang \& Ho 1997), and the dust emission in the W51 e1 core.
282: The nearest source to e1 1mm peak is e8, which indicates that e8 is
283: stronger than the UC \hii\ region e1 at 1.3 mm. Because the UC \hii\
284: region e1 is stronger than e8 at 1.3 cm (Zhang \& Ho 1997) and e8 is not
285: detected at 2 cm (Gaume et al.\ 1993), the spectral index of e8 is very
286: different from free-free emission. This fact is consistent with Zhang
287: \& Ho's suggestion that e8 is a dust-dominated continuum source. Hence,
288: e8 is the most likely site for active star formation in the W51 e1 core.
289:
290: The position of the UC \hii\ region e2 (accurate to 0\farcs2;
291: Gaume et al.\ 1993) is offset $\sim$1\arcsec\ to northwest of e2 1mm peak .
292: It is known that the molecular gas in W51 e2 is not evenly
293: distributed around the \hii\ region (Zhang \& Ho 1997).
294: If we consider that molecular gas is in general associated with
295: dust grains, then, although the offset is small, it is evident that
296: the free-free emission and the dust emission in W51 e2 do not coincide.
297: Since the polarized intensity only originates from the dust emission
298: and the fraction of the free-free emission could be significantly higher
299: than the average value at the position of the UC \hii\ region,
300: this offset may be part of the reason for the absence of polarization
301: between e2 1mm peak and e2 pol NW.
302: It is interesting that the location of W51 Main is right in
303: the polarization gap between e2 1mm peak and e2 pol NW,
304: and a stream of water masers in W51 Main has been detected
305: with proper motions headed toward position angle 200\deg\
306: (Lepp\"anen, Liljestr\"om, \& Diamond 1998),
307: which seems well matched with the direction of the polarization gap.
308:
309:
310: \subsection{Polarization Angle Distribution}
311:
312: The polarization angle distribution in the W51 e1/e2 cores
313: appears to be well matched to the morphology of the cores.
314: Except that e2 pol NW has a very different position angle
315: at 58\deg$\pm$8\deg,
316: the rest of the polarization vectors are uniformly distributed
317: along the major axis of the double cores with an average position
318: angle of 21\deg$\pm$6\deg.
319: Polarization vectors seem to be approximately parallel to local core edges:
320: in W51 e1, polarization angles decrease westward and increase eastward
321: from the central ridge to the edge; in W51 e2, position angles
322: are close to 0\deg\ at the northeast edge; even for e2 pol NW,
323: the difference
324: between the polarization angle and the edge is less than 20\deg.
325: The polarization angles also follow the ridge connecting W51 e1 and W51 e2.
326: Excluding e2 pol NW, we calculate the mean polarization angles for W51 e1
327: and W51 e2 weighted by the measurement uncertainty, which are
328: 23\deg$\pm$5\deg\ for W51 e1 and 15\deg$\pm$7\deg\ for W51 e2.
329: Compared to the measurement uncertainty,
330: the variation of polarization angles corresponding to
331: the core morphology is only slightly suggested by our data.
332:
333: Figure 2 shows the distribution of the polarization
334: angles in W51 e1 and W51 e2. The observed polarization angle dispersion
335: $\delta\phi_{obs}$, defined to be the standard deviation of
336: the position angles, is 4\fdg8\e2\fdg0 in W51 e1 and 6\fdg6\e3\fdg0
337: in W51 e2 (excluding e2 pol NW). Interestingly, these values are
338: equal to the measurement uncertainty of the position angles $\sigma_{\phi}$,
339: 4\fdg9\e1\fdg4 in W51 e1 and 6\fdg7\e3\fdg4 in W51 e2.
340: The observed dispersion is made up of contributions from the measurement
341: uncertainty and the intrinsic dispersion $\delta\phi$,
342: $\delta\phi_{obs}^2 = \delta\phi^2+\sigma_{\phi}^2$,
343: which indicates that the intrinsic dispersions in W51 e1 and e2
344: are very small.
345: Therefore, if we take $\sigma_{\phi}$ as small as possible
346: ($\sigma_{\phi}-1\sigma_{\sigma_{\phi}}$) and $\delta\phi_{obs}$
347: as large as possible ($\delta\phi_{obs}+1\sigma_{\delta\phi_{obs}}$),
348: we can obtain the upper limit of the intrinsic dispersion,
349: which is 5\fdg8 in W51 e1 and 9\fdg0 in W51 e2.
350: Our results for the dispersion analysis are summarized in Table 2.
351:
352:
353: \subsection{Polarization Percentage Distribution}
354:
355: The average polarization percentage of the W51 e1/e2 cores is
356: 1.8\%\e0.4\%, including positions for which polarization vectors
357: are not plotted in Fig.\ 1 because $I_p$ is too small.
358: W51 e2 has a lower polarization percentage (1.1\%\e0.3\%)
359: than that of W51 e1 (3.2\%\e0.6\%). To better compare with previous
360: observations, we convolved our map to 30\arcsec\ resolution with
361: a Gaussian beam and obtained 1.3\%$\pm$0.2\% polarization with a position
362: angle of 25\deg$\pm$3\deg. The polarization detections we have obtained
363: are clearly different from the single-dish results: at 1.3 mm (HPBW=30\arcsec),
364: Kane et al.\ (1993) measured a polarization percentage of 0.50\%\e0.21\%
365: at an angle of $-$31\deg\e11\deg; at 100 $\mu$m (HPBW=35\arcsec),
366: Dotson at al.\ (2000) obtained zero polarization (0.54\%$\pm$0.58\%).
367: However, all results could be still consistent,
368: because the low S/N ratio of the these single-dish measurements can
369: lead to large uncertainty in the polarization angle.
370: Our observations demonstrate that high resolution is important
371: for successful detection and mapping of magnetic fields in molecular cores.
372:
373: Figure 3 shows plots of the polarization percentage
374: versus Stokes $I$. These plots show that the polarization
375: percentage decreases as the total intensity increases in both W51 e1
376: and W51 e2. Such a decrease in the polarization toward high intensity
377: regions ({\it the depolarization}) has been commonly seen in polarization
378: observations, e.g., L1755 (Lazarian, Goodman, \& Myers 1997),
379: OMC-1 (Schleuning 1998), and OMC-3 (Matthews \& Wilson 2000).
380: Lazarian, Goodman, \& Myers (1997) have shown that the depolarization
381: can be explained as a consequence of the dust alignment efficiency
382: decreasing toward the inner parts of the dark clouds,
383: as all known mechanisms of grain alignment fail under the typical
384: physical conditions of dark cloud interiors.
385: On the other hand, beam smearing over small-scale field structure
386: can also produce low polarization percentage (Rao et al.\ 1998);
387: this can only be verified by observations with higher resolution.
388:
389: Due to the apparent close anti-correlation of the polarization
390: percentage and the total intensity in Figure 3,
391: it is interesting to perform a least-squares linear fit
392: on these two quantities.
393: We obtain $\log_{10}(p)=(-1.52\pm0.01)-(0.63\pm0.07)\times\log_{10}(I)$
394: with a correlation coefficient of $-$0.91 in W51 e1,
395: and $\log_{10}(p)=(-1.73\pm0.01)-(0.99\pm0.04)\times\log_{10}(I)$
396: with a correlation coefficient of $-$0.97 in W51 e2.
397: However, due to the difficulty of excluding the free-free
398: emission at every position, direct comparison between
399: our results and the dust alignment mechanisms cannot be done.
400: The tight correlations hint that the depolarization in the W51 e1/e2 cores
401: is a gradual and global effect. Therefore, it is unlikely that
402: the depolarization is entirely caused by a sudden change of the alignment
403: mechanism in a local region, as was reported in Orion-KL (Rao et al.\ 1998).
404:
405:
406: \section{Discussion}
407:
408: \subsection{Magnetic Field Morphology}
409:
410: The magnetic field direction inferred from the paramagnetic relaxation
411: of grain alignment is perpendicular to the direction of the linear
412: polarization of the dust emission (Davis \& Greenstein 1951).
413: Therefore, the average directions of the magnetic fields
414: are 113\deg$\pm$5\deg\ in W51 e1 and 105\deg$\pm$7\deg\ in W51 e2
415: (excluding e2 pol NW).
416: The field directions are approximately parallel to the minor axis
417: of the two cores, suggesting that matter collapsed along the field
418: lines to form the parent cloud of the double cores.
419: The uniform field structure in the W51 e1/e2 cores also suggests that
420: ionized gas in the UC \hii\ regions does not have much interaction
421: with the magnetic field, or the interaction only exists
422: on a scale smaller than our synthesized beam.
423:
424: Comparison between the field morphology and the rotational
425: axes of W51 e1 and W51 e2 provides interesting implications
426: for the relation between magnetic fields and core formation.
427: Rotational motions have been observed toward these two cores
428: by Zhang, Ho, \& Ohashi (1998) from analysis of the velocity
429: gradient of CH$_3$CN emission,
430: and the derived rotational axis is at 66\deg$\pm$27\deg\
431: for e1 and at 20\deg$\pm$17\deg\ for e2. Clearly,
432: the rotational axes and the magnetic field directions
433: of these two cores do not align with each other.
434: Although it has been suggested that protostellar disks
435: may drag the field lines as they rotate (Holland et al.\ 1996),
436: it may not be the case for these two cores.
437: The field directions vary smoothly along the two cores.
438: Thus, it is unlikely that two independent rotating cores that
439: twist the field lines separately would produce a matched field
440: morphology. Therefore, the smooth magnetic field structure is
441: most likely associated with the common envelope of e1 and e2.
442: Zhang, Ho, \& Ohashi (1998) caution that the uncertainty in their
443: analysis may be large. However, if the difference
444: between the directions of rotational axes and magnetic fields
445: is indeed significant, the magnetic field of the parent cloud
446: seems to have had no effect in determining the axis of angular
447: momentum of these two cores.
448:
449:
450: \subsection{Estimation of the Magnetic Field Strength}
451:
452: Owing to the lack of understanding of the detailed mechanism
453: of grain alignment, the magnetic field strength cannot be
454: directly derived from linear polarization measurements
455: of dust emission (Lazarian, Goodman, \& Myers 1997). However,
456: Chandrasekhar \& Fermi (1953, hereafter CF) have proposed a method
457: to estimate the field strength from the dispersion of polarization
458: angles ($\delta\phi$) by assuming that the variation in polarization
459: angles results from the perturbation of Alfv\'en waves on the field
460: lines. In this case, the field strength projected in the plane of
461: the sky ($B_p$) is given by
462: \begin{equation}
463: B_p = Q \sqrt{4\pi\bar{\rho}}~~\frac{\delta v_{los}}{\delta\phi},
464: \end{equation}
465: \noindent where $\bar{\rho}$ is the average density, $\delta v_{los}$
466: is the rms line-of-sight velocity, and $Q$ is 1 for CF's derivation.
467: Detailed studies show that the CF formula tends to overestimate
468: the field strength. $Q$ can be reduced by several factors,
469: such as the inhomogeneity of clouds and the line-of-sight averaging
470: (Zweibel 1990; Myers \& Goodman 1991). Ostriker, Stone, \& Gammie
471: (2001) calculate the value of $Q$ from their simulations
472: of turbulent clouds, and suggest that
473: the CF formula with $Q\approx0.5$ can account for
474: the complex magnetic field and density structure, and
475: provide accurate estimates of the plane-of-sky field strength
476: under strong field cases when $\delta\phi \leq 25\deg$.
477: Since the angle dispersion we measured in the W51 e1/e2 is small
478: ($\delta\phi<9\deg$; \S3.2), it is interesting to work out
479: the field strength estimates with the modified CF formula.
480:
481: However, the two other quantities, $\bar{\rho}$ and $\delta v_{los}$,
482: needed to complete this exercise have uncertainties not smaller
483: than that of $\delta\phi$. The density of the W51 e1/e2 cores varies
484: with the physical components and scales, and we should use the average
485: density obtained from the region that the observed polarization
486: is associated with.
487: Zhang \& Ho (1997) measure the NH$_3$ emission in the infall regions of
488: the cores and derive $n_{H_2}\sim3\times10^6\cm-3$,
489: which is probably too high for our purpose because the polarized intensity
490: in the W51 e1/e2 seems to originate in a more extended region.
491: From spectral decomposition of the whole W51\,A region
492: (including W51\,e1/e2 and W51\,IRS 1), Sievers et al.\ (1991) show that
493: W51\,A consists of a 20 K cold dust component with
494: $n_H\sim5\times10^4\cm-3$ ($n_H=2n_{H_2}$) and a 60 K
495: warm dust component with $n_H\sim10^6\cm-3$.
496: This provides a range for the density
497: as W51 e1/e2 may contain a mixture of both components.
498: Because the column density of the warm dust is $\sim$5 times more
499: than that of the cold dust and W51\,e1/e2 is the densest region
500: in W51\,A (Sievers et al.\ 1991), W51\,e1/e2 is more likely
501: to have a density close to $10^6\cm-3$.
502: It is also difficult to decompose $\delta v_{los}$
503: that is associated with Alfv\'enic motion
504: from the molecular linewidths which are broadened by the complicated
505: dynamics in W51 e1/e2, such as infall (3.5 \kms: Zhang et al.\ 1998),
506: rotation ($\sim$4 \kms: Zhang \& Ho 1997), and possible outflow
507: activities. The linewidth of an optically thin NH$_3$
508: line measured in the envelope of e1 and e2 by Young, Keto, \& Ho (1998),
509: $\Delta v$=1.25 \kms, may contain little contamination from
510: dynamical motions. Therefore, we adopt $\delta v_{los} =
511: \Delta v / \sqrt{8\ln2} = 0.53 \kms$.
512:
513: Given the large uncertainty of the input parameters,
514: we can only calculate the lower limits of $B_p$
515: using the upper limits of the angle dispersions obtained in \S3.2.
516: The derived $B_p$ for the density limits of W51 e1/e2
517: are listed in Table 2. Our results show that the lowest value
518: of $B_p$ in W51 e1/e2 is just comparable to the typical magnetic
519: field strengths measured in molecular cores with $n_{H_2}=10^{5-6}$
520: \cm-3 using the Zeeman effect ($\sim500\mu$G: Crutcher 1999).
521: Since this typical field strength provides crucial contribution
522: in cloud evolution (Crutcher 1999), our results show that
523: the magnetic energy is an important component in the energetics
524: of the W51 e1/e2 envelope.
525: It can be shown that the ratio of the turbulent to magnetic energy
526: is simply proportional to the square of the angle dispersion
527: (Lai et al.\ 2001, in preparation).
528: The small angle dispersions we measured here imply that
529: the magnetic field dominates the turbulent motion
530: in the regions that emit polarized flux in the W51 e1/e2 molecular cores.
531:
532:
533: \section{Conclusions}
534:
535: We have measured linear polarization of the thermal dust emission
536: at 1.3 mm toward the W51 e1/e2 cores with the BIMA array.
537: The conclusions of our observations are the following:
538:
539: \begin{itemize}
540: \item The magnetic field is parallel to the minor axis of
541: the W51 e1/e2 cores, consistent with the theoretical picture that
542: cloud cores form due to gravitational contraction along field lines.
543: However, the significant difference between the directions
544: of magnetic field and rotational axes hints that the magnetic field
545: of the parent cloud does not determine the axis of the
546: angular momentum of these two cores.
547:
548: \item The polarization percentage decreases toward the high
549: intensity regions. The tight anti-correlation between $\log~p$
550: and $\log~I$ in both W51 e1 and W51 e2
551: implies that depolarization is a global effect
552: and may be caused by the decreasing dust alignment efficiency
553: toward higher density regions.
554:
555: \item The small dispersion of the polarization angles in W51 e1/e2
556: suggests that the magnetic field is strong ($\gtrsim$ 1 mG) and
557: dominates turbulence in the regions where the polarized flux arises.
558: \end{itemize}
559:
560:
561: \acknowledgments
562:
563: This research was supported by NSF grants AST 99-81363 and AST 98-20651.
564: We would like to thank the staff at Hat Creek, especially Rick Forster
565: and Mark Warnock for assistance with the polarimeter control system.
566:
567:
568: %\clearpage
569:
570: \begin{thebibliography}{}
571: \bibitem{} Briggs, D.\ S.\, 1995, PhD dissertation (available at the website http://www.aoc.nrao.edu/ftp/dissertations/dbriggs/diss.html)
572: \bibitem{} Chandrasekhar, S., \& Fermi, E.\ 1953, \apj, 118, 113 (CF)
573: \bibitem{} Crutcher, R.\ M.\ 1999, \apj, 520, 706
574: \bibitem{} Davis, L., \& Greenstein, J.\ L.\ 1951, \apj, 114, 209
575: \bibitem{} Dotson, J.\ L., Davidson, J., Dowell, C.\ D., Schleuning, D.\ A., \& Hildebrand, R.\ H.\ 2000, \apjs, 128, 335
576: \bibitem{} Gaume, R.\ A., \& Mutel, R.\ L.\ 1987, \apjs, 65, 193
577: \bibitem[Gaume, Johnston \& Wilson(1993)]{1993ApJ...417..645G} Gaume, R.\ A., Johnston, K.\ J., \& Wilson, T.\ L.\ 1993, \apj, 417, 645
578: \bibitem{} Genzel, R.\ et al.\ 1981, \apj, 247, 1039
579: \bibitem{} Girart, J.\ M., Rao, R., \& Crutcher, R.\ M.\ 1999, \apj, 525, 109
580: \bibitem{} Harmuth, H.\ F.\ 1969, {\it IEEE Spectrum}, 6, No.\ 11, 82
581: \bibitem{} Heiles, C., Goodman, A.\ A., McKee, C.\ F., \& Zweibel, E.\ G.\ 1993, in Protostars and Planets III, ed. M.\ Matthews, \& E.\ Levy
582: (Tuscon: University of Arizona Press), 279
583: \bibitem{} Ho, P.\ T.\ P., Das, A., \& Genzel, R.\ 1983, \apj, 266, 596
584: \bibitem{} Ho, P.\ T.\ P., \& Young, L.\ M.\ 1996, \apj, 472, 742
585: \bibitem{} Holland, W.\ S., Greaves, J.\ S., Ward-Thompson, D., \& Andre, P.\ 1996, \aap, 309, 267
586: \bibitem[Jaffe, Becklin, \& Hildebrand(1984)]{1984ApJ...279L..51J} Jaffe, D.\ T., Becklin, E.\ E., \& Hildebrand, R.\ H.\ 1984, \apjl, 279, L51
587: \bibitem{} Kane, B.\ D., Clemens, D.\ P., Barvainis, R., \& Leach, R.\ W.\ 1993, \apj, 411, 708
588: \bibitem{} Lai, S.-P., Crutcher, R.\ M., Girart, J.\ M., \& Rao, R.\ 2001, submitted to \apj
589: \bibitem{} Lazarian, A., Goodman, A.\ A., \& Myers, P.\ C.\ 1997, ApJ, 490, 273
590: \bibitem{} Leahy, P., VLA Scientific Memoranda No.\ 161
591: \bibitem{} Lepp{\"a}nen, K., Liljestr{\"o}m, T., \& Diamond, P.\ 1998, \apj, 507, 909
592: \bibitem{} Matthews, B.\ C.\ \& Wilson, C.\ D.\ 2000, \apj, 531, 868
593: \bibitem{} Mouschovias, T.\ Ch.\ 1976, \apj, 207, 141
594: \bibitem{} Mouschovias, T.\ Ch., \& Ciolek, G.\ E.\ 1999, in The Origin of Stars and Planetary Systems, eds. C.\ J.\ Lada \& N.\ D.\ Kylafis
595: (Kluwer Academic Press), 305
596: \bibitem{} Myers, P.\ C., \& Goodman, A.\ A.\ 1991, \apj, 373, 509
597: \bibitem{} Ostriker, E.\ C., Stone, J.\ M., \& Gammie, C.\ F.\ 2001, \apj, 546, 980
598: \bibitem{} Rao, R., Crutcher, R.\ M., Plambeck, R.\ L., \& Wright, M.\ C.\ H.\ 1998, \apjl, 502, L75
599: \bibitem{} Rao, R.\ 1999, Ph.D.\ Dissertation, University of Illinois at Urbana-Champaign
600: \bibitem{} Rao, R., Crutcher, R.\ M., Girart, J.\ M., Lai, S.-P., Wright, M.\ C.\ H., Plambeck, R.\ L., \& Lugten, J.\ B.\ 2001, in preparation
601: \bibitem{} Roberge, W.\ G.\ 1996, ASP Conf.\ Ser.\ 97: Polarimetry of the Interstellar Medium, 401
602: \bibitem{} Rudolph, A., Welch, W.\ J., Palmer, P., \& Dubrulle, B.\ 1990, \apj, 363, 528
603: \bibitem{} Sault, R.\ J., \& Killeen, N.\ E.\ B.\ 1998, Miriad Users Guide (available at the website http://www.atnf.csiro.au/computing/ software/miriad)
604: \bibitem{} Sault, R.\ J., Teuben, P.\ J., \& Wright, M.\ C.\ H.\ 1995, ASP Conf.\ Ser.\ 77: Astronomical Data Analysis Software and Systems IV, 4, 433
605: \bibitem{} Schleuning, D.\ A.\ 1998, ApJ, 493, 811
606: \bibitem[Scott(1978)]{1978MNRAS.183..435S} Scott, P.\ F.\ 1978, \mnras, 183, 435
607: \bibitem{} Shu, F.\ H., Allen, A., Shang, H., Ostriker, E.\ C., \& Li, Z.\ 1999, in The Origin of Stars and Planetary Systems, eds. C.\ J.\ Lada \& N.\ D.\ Kylafis (Kluwer Academic Press),193
608: \bibitem[Sievers et al.(1991)]{1991A&A...251..231S} Sievers, A.\ W., Mezger, P.\ G., Kreysa, E., Haslam, C.\ G.\ T., Lemke, R., \& Gordon, M.\ A.\ 1991, \aap, 251, 231
609: \bibitem{} Thompson, A.\ R., Moran, J.\ M., \& Swenson, G.\ W.\ 1986, Interferometry and Synthesis in Radio Astronomy (New York: Wiley)
610: \bibitem[Thronson \& Harper(1979)]{1979ApJ...230..133T} Thronson, H.\ A.\ \& Harper, D.\ A.\ 1979, \apj, 230, 133
611: \bibitem{} Welch, W.\ J., et al.\ 1996, \pasp, 108, 93
612: \bibitem{} Young, L.\ M., Keto, E., \& Ho, P.\ T.\ P.\ 1998, \apj, 507, 270
613: \bibitem{} Zhang, Q., \& Ho, P.\ T.\ P.\ 1997, \apj, 488, 241
614: \bibitem{} Zhang, Q., Ho, P.\ T.\ P., \& Ohashi, N.\ 1998, \apj, 494, 636
615: \bibitem{} Zweibel, E.\ G.\ 1990, \apj, 362, 545
616:
617: \end{thebibliography}
618:
619:
620:
621: \begin{table}
622: \caption{Polarization Measurements in the W51 e1/e2 regions}
623: \vspace*{0.3cm}
624: \begin{tabular}{lcccl}
625: \hline
626: \hline
627: Position & Stokes $I$ & Polarization & Polarization & Note \\
628: (\arcsec,\arcsec)\tablenotemark{a} & (\Jb) & Percentage(\%) & Angle (\deg) &\\
629: \hline
630:
631: (-6.3,~3.3) & 0.25$\pm$0.03 & ~6.4$\pm$2.0 & 58$\pm$~8 & e2 pol NW \\
632: (-1.0,~2.0) & 0.35$\pm$0.03 & ~5.6$\pm$1.4 & ~1$\pm$~7 & e2 \\
633: (-3.6,~0.6) & 3.22$\pm$0.03 & ~0.7$\pm$0.2 & 19$\pm$~6 & e2 1mm peak \\
634: (-1.6,~0.3) & 0.91$\pm$0.03 & ~1.5$\pm$0.5 & 10$\pm$10 & e2 \\
635: (-1.6,-2.9) & 0.32$\pm$0.03 & ~7.9$\pm$1.6 & 16$\pm$~5 & e2 south \\
636: (-3.9,-2.9) & 0.74$\pm$0.03 & ~3.1$\pm$0.6 & ~9$\pm$~6 & e2 south \\
637: (-2.0,-6.3) & 0.37$\pm$0.03 & ~6.5$\pm$1.4 & 19$\pm$~6 & e1 \\
638: (-4.2,-6.3) & 1.79$\pm$0.03 & ~2.1$\pm$0.3 & 29$\pm$~4 & e1 1mm peak \\
639: (-4.2,-7.5) & 1.45$\pm$0.03 & ~3.0$\pm$0.3 & 28$\pm$~3 & Peak of polarized flux \\
640: (-6.5,-7.5) & 0.30$\pm$0.03 & ~5.8$\pm$1.6 & ~7$\pm$~8 & e1 \\
641: (-2.0,-9.5) & 0.20$\pm$0.03 & 10.0$\pm$2.7 & 26$\pm$~7 & e1 \\
642: (-4.2,-9.5) & 0.63$\pm$0.03 & ~4.8$\pm$0.8 & 28$\pm$~4 & e1 \\
643: \hline
644: e1 average & -- & ~3.2\e0.6 & 23\e~5 \\
645: e2 average & -- & ~1.1\e0.3 & 15\e~7\tablenotemark{b} \\
646: total average & -- & ~1.8\e0.4 & 21\e~6\tablenotemark{b} \\
647: \hline
648: \hline
649: \tablenotetext{a}{Offsets are measured with respect to
650: the phase center: $\alpha_{2000}$=19$^h$23$^m$44\fs2,
651: $\delta_{2000}$=14\deg30\arcmin33\farcs4.}
652: \tablenotetext{b}{e2 pol NW is excluded.}
653: \end{tabular}
654: \end{table}
655:
656: %\clearpage
657:
658: \begin{table}
659: \caption{Estimates of magnetic field strengths in the plane of the sky}
660: \vspace*{0.3cm}
661: \begin{tabular}{lccccc}
662: \hline
663: \hline
664: Region & $\sigma_{\phi} (^{\circ})$\tablenotemark{a}
665: & $\delta\phi_{obs} (^{\circ})$\tablenotemark{b}
666: & $\delta\phi$ (\deg)\tablenotemark{c}
667: & $n_H$ (\cm-3)\tablenotemark{d}
668: & $B_p$ (mG)\tablenotemark{e} \\
669: \hline
670: e1 &4.9$\pm$1.4 & 4.8\e2.0 & $<5.8$ &
671: \begin{tabular}{@{}c@{}}
672: 5\x10$^4$ \\
673: 10$^6$
674: \end{tabular} &
675: \begin{tabular}{@{}c@{}}
676: $>$ 0.3 \\
677: $>$ 1.3
678: \end{tabular} \\
679: \hline
680: e2 & 6.7$\pm$3.4 & 6.6\e3.0 & $<9.0$ &
681: \begin{tabular}{@{}c@{}}
682: 5\x10$^4$ \\
683: 10$^6$
684: \end{tabular} &
685: \begin{tabular}{@{}c@{}}
686: $>$ 0.2 \\
687: $>$ 0.8
688: \end{tabular} \\
689: \hline
690: \hline
691: \tablenotetext{a}{The measurement uncertainty of the polarization angle}
692: \tablenotetext{b}{The observed polarization angle dispersion}
693: \tablenotetext{c}{The intrinsic polarization angle dispersion}
694: \tablenotetext{d}{The number density of atomic hydrogen}
695: \tablenotetext{e}{The plane-of-sky magnetic field strength}
696: \end{tabular}
697: \end{table}
698:
699:
700: \begin{figure}
701: \plotone{f1.eps}
702: \figcaption{Polarization map of the W51 e1/e2 cores.
703: The contours represent Stokes $I$ at $-5, -3, 3, 5, 10, 20, 30, ...,
704: 110~\sigma$ levels. The 1 $\sigma$ noise level of Stokes $I$ is 27 \mJb.
705: The line segments are polarization vectors, and
706: their lengths are proportional to the polarization percentage
707: with a scale of 3\% per arcsec length.
708: The dashed line that crosses the saddle point of the continuum
709: is used to separate the data of W51 e1 and W51 e2 for our statistics.
710: The star symbols mark the positions of e2, e4, e8, e1 and e3 from north
711: to south. The triangle marks the position of H$_2$O masers near e2 (W51 Main).
712: e2 pol NW refers to the two vectors northwest of W51 Main.
713: The ellipse indicates the synthesized beam of the Stokes $Q$ and $U$ images,
714: which is 3\farcs2$\times$2\farcs3 with PA=2\deg.
715: Positions with high intensity but without polarization detections
716: may have higher fraction of free-free emission or/and
717: have an environment inappropriate for the magnetic alignment
718: of the dust grains.}
719: \end{figure}
720:
721: \begin{figure}
722: \epsscale{0.65}
723: \plotone{f2.eps}
724: \figcaption{Distribution of polarization angle in W51 e1/e2.
725: The average and the dispersion of the polarization angle in e1 and e2 are
726: labeled, and the double arrow indicates that e2 pol NW with
727: $\phi=50\deg-60\deg$ is excluded in the calculation.}
728: \end{figure}
729:
730: \begin{figure}
731: \plotone{f3.eps}
732: \figcaption{ The variation of the polarization percentage with Stokes $I$
733: in W51 e1/e2. The dashed lines are the results of the least-squares linear
734: fit.}
735: \end{figure}
736:
737: \end{document}
738:
739: