Starlight polarization and CO observations towards the Lupus clouds

We performed an observational study of the dark ﬁlaments Lupus 1 and Lupus 4 using both polarimetric observations of 190 stars and a sample of 72 12 CO proﬁles towards these clouds. We have estimated lower limits to the distances of Lupus 1 and Lupus 4 ( * 140 and * 125 pc, respectively). The observational strategy of the survey allows us to compare the projected magnetic ﬁeld in an extended area around each cloud with the magnetic ﬁeld direction observed to prevail along the clouds. Lupus 4 could have collapsed along the magnetic ﬁeld lines, while in Lupus 1 the magnetic ﬁeld appears to be less ordered, having the major axis of the ﬁlaments parallel to the large-scale projected magnetic ﬁeld. These differences would imply that both ﬁlaments have different pattern evolutions. From the CO observations we have probed the velocity ﬁelds of the ﬁlaments and the spatial extension of the molecular gas with respect to the dust.


I N T R O D U C T I O N
The Lupus star-forming region is a group of dark clouds surrounding a great number of T Tauri stars (Krautter 1991). It is one of the largest T associations of the southern sky, stretching over at least 238 perpendicular to the Galactic plane, and represents one of the nearest low-mass star-forming regions.
The T Tauri stars are concentrated in four subgroups designated Lupus 1 to Lupus 4 (Schwartz 1977). The Lupus 1 and Lupus 2-4 subgroups are associated with small dark clouds which are embedded in a large complex of CO emission (Murphy, Cohen & May 1986). Besides T Tauri stars, other peculiar stellar objects have been found in these clouds, namely one Herbig Ae/Be star and a few Herbig-Haro objects. These findings are indicative of a very recent episode of star formation (10 5 -10 6 yr ago). Very recently Krautter et al. (1997), using ROSAT observations, have found a population of 'weak-line' T Tauri stars, which apparently are not correlated with the most obscured zones of the region. Krautter (1991) has reviewed the properties of the Lupus clouds and their related young objects. He claims that the mass spectrum of the T Tauri stars is rather unusual, suggesting that its initial mass function differs from those in other well known T associations like the Taurus-Auriga complex. He suggests as a possible explanation that the star formation in Lupus might be supported against gravitational collapse by the ambient magnetic field. Shu, Adams & Lizano (1987) proposed that magnetic field support leads to low star-formation efficiency and that, in the subcritical regime, if the magnetic forces exceed the gravitational ones, low-mass stars should preferentially be formed. From their polarization measurements, Strom, Strom & Edwards (1988) suggested that the Lupus 1 cloud has probably been forced to collapse along the magnetic field lines, which may have played a major role in controlling the properties of the protostellar cores.
Measurements of light polarization from stars that are either embedded in, or located beyond, a dark cloud can be used to map the geometry (projected on to the plane of the sky) of the magnetic field in the cloud, assuming that the polarization is mainly caused by non-spherical dust grains in the cloud aligned by the magnetic field (Davis & Greenstein 1951). Several efforts have been made to understand the patterns of the projected magnetic fields, as derived from starlight polarization measurements, that are likely to be associated with elongated dark clouds (Vrba, Strom & Strom 1976;Heyer et al. 1987;Goodman et al. 1990;Arnal, Morras & Rizzo 1993). In most cases, it looks as if the interstellar magnetic field plays an important role in determining the evolution of the molecular cloud.
As pointed out by Heiles et al. (1993), polarization maps of clouds presented in isolation and removed from their environments may be misleading for correlations or alignments found within a given cloud. However, these features might not be peculiar to the cloud itself. Instead, they might characterize not only the cloud but also the region where it is embedded. For example, individual clouds in Ophiuchus and Taurus which appear to be 'parallel' or 'perpendicular' to the magnetic field (Goodman et al. 1990;Heyer et al. 1987), when their magnetic field orientations are compared with large-scale polarization maps of their surroundings, it is found that the cloud orientation may be fortuitous, because of the bias of optical polarization which probes the visually absorbing gas located at the periphery of the cloud (Heiles et al. 1993). Thus, it is advisable to observe a sample of stars located at different angular distances from the cloud, and spread over a large area around it.
In this paper we present the results of an observational study of Lupus 1 and Lupus 4. Both clouds show signs of internal fragmentation into globule-like beads. A small globule seen in projection close to Lupus 4 was also observed. These clouds are included in the Catalogue of Dark Globular Filaments (GF) of Schneider & Elmegreen (1979) and are identified as GF19 (Lupus 1), GF17 (Lupus 4) and GF18 (the globule). These authors found that the opacity, measured in visual extinction units on the Lynds (1962) scale, ranges from 5 to 6 for Lupus 1 and between 4 and 6 for Lupus 4. Maps of the distribution of visual extinction in these clouds can be found in Andreazza & Vilas-Boas (1996). Formaldehyde, in the l 10 → l 11 rotational transition at 6 cm, was detected in absorption in these clouds by Sandqvist & Lindroos (1976). Similar observations toward Lupus 4 were also carried out by Goss et al. (1980). The observational material comes from polarimetric observations of both stars projected on the sky on the outskirts of the dark clouds and stars shining through the less obscured part of the clouds, together with 12 CO 115-GHz observations towards both filaments and their environments. The relationship between the magnetic field geometry and the dark cloud morphology may provide important clues about the most likely scenarios for the cloud evolution. The velocity field of both clouds was mapped by observing the 12 CO (J ¼ 1 → 0) transition. Since the gas and the dust are mixed, observations of polarization of stars with known distances could also be used to estimate a lower limit on the distances to the molecular clouds.

Optical polarimetry
The polarimetric measurements were performed using the Vatican Polarimeter (VATPOL) mounted at the Cassegrain focus of the 2.15-m telescope at El Leoncito (CASLEO, Argentina). A thorough description of the VATPOL system was provided by Magalhaes, Benedetti & Roland (1984). The observations were carried out in two observing runs during 1991 June and 1992 June, respectively. Unfiltered, linear optical polarization measurements were obtained for 190 programme stars, which were selected on the basis of both their apparent brightness and their angular distance from Lupus 1 and Lupus 4. From this sample, 40 and 46 weak stars are seen projected on to the dust patches delineating Lupus 1 and Lupus 4, respectively. A total of 66 and 38 'field stars', that is to say stars located away from the dark clouds, were observed on the outskirts of Lupus 1 and Lupus 4, respectively. The position angle zero-point and instrumental polarization values were established by regularly observing a sample of highly polarized and nearby unpolarized standard stars, respectively. The instrumental polarization was found to be less than 0.03 per cent.

Carbon monoxide
Observations of the 2.6-mm line 12 CO J ¼ 1→ 0 were made with the 1.2-m Columbia Millimeter-Wave Telescope at Cerro Tololo, Chile, during 1985 January. At the CO frequency (115.2712 GHz) the radio telescope has an angular resolution (FWHM) of < 8.7 arcmin. A 256-channel filter bank of 100 kHz bandwidth per channel was used at the back end. The velocity resolution was 0.26 km s ¹1 , and the velocity coverage < 66 km s ¹1 . A detailed description of the instrument was given by Cohen (1983). The observational method, reduction and calibration of the profiles were similar to those used by Arnal et al. (1993). Typical final rms noise per profile was of the order of ,0:27 K.
A total of 41 points in Lupus 1 and 31 points in Lupus 4 were observed.

Optical polarimetry
The results of the observations of the 190 programme stars are given in Tables 1 and 2, respectively. In Table 1 the observational results corresponding to those stars seen in projection on to, or close to, the filamentary globules are given. A star sequence identification number is given in the first column. The observed percentage polarization P, its probable error eðPÞ, the position angle of polarization v (in degrees), and its error eðvÞ are given in the second, third, fourth and fifth columns, respectively. The position angles v are measured east of north and the uncertainty in v is given by the formula eðvÞ = 288 : 65 [eðPÞ=P] (Serkowski 1974). In Table 2, the results corresponding to the so-called 'field stars' are provided. There, the star HD and SAO number are given in the first and second columns, respectively. The corresponding P, eðPÞ, v and eðvÞ are given in the third, fourth, fifth and sixth columns, respectively. In the seventh column, when available, the visual magnitude is listed, as quoted in the SAO Catalogue. In the eighth column the spectral type according to the Michigan Catalogue (Houk & Cowley 1975) is listed. The numbers 1 or 4 quoted in the ninth column indicate the field to which the star belongs (Lupus 1 or Lupus 4).
Finding charts for the stars observed in the central parts of Lupus 1 and Lupus 4 are given in Figs 1(a) and (b), respectively. The dark clouds are clearly visible in the charts as well-defined dark lanes. The positions of the observed stars are indicated according to the number assigned in Table 1. In Figs 2(a) and (b) the polarization vectors of these stars have been drawn. The length of the polarization vector is proportional to P and it is oriented in the direction indicated by v. In Fig. 3 polarization vectors of the 104 'field stars' are plotted together. In Figs 2(b) and 3, polarizations lower than 0.5 per cent are indicated by an open circle.

Lupus 1
In Fig. 4 the observed CO positions, marked by filled circles, along with a small but representative sample of CO profiles, are shown. The selected CO profiles are depicted inside rectangular boxes. A straight line joins every box with the actual observed position. Solid con tour lines corresponding to the visual absorption map of Andreazza & Vilas-Boas (1996) are also shown, to help the reader to pin down regions of high optical obscuration. Those positions observed at the H 2 CO 6-cm line by Sandqvist & Lindroos (1976) are indicated by open squares. Most of the profiles are single-peaked and exhibit velocity asymmetries. No clear correlation between the line asymmetries and the location of observed points has been found. A few of these profiles, three out of 41, show a clear double-peak structure. All profiles are broad (FWHM up to 3.2 km s ¹1 are observed, the average being 2.3 km s ¹1 ) and the peak temperature over the filament reaches , 8 K. Table 3(a) summarizes the observed line parameters: the first column gives a profile number; the second and third columns correspond to the equatorial coordinates (1950.0) of the observed point; the observed peak brightness temperature (T max ), the mean temperatureweighted radial velocity (V R ) and the line integral (W CO ) are listed in the fourth, fifth and sixth columns, respectively. Omitted parameters correspond to those points having T max , or W CO , values lower than 3 times the rms noise. The W CO along the filament has a clear decreasing trend from north-west to south-east.

Lupus 4
The main observational results for this cloud are summarized in Table 3(b). The meaning of the columns is similar to that in Table  3(a). We include for this region the CO observations of both Lupus 4 and the globule GF18. Due to a lack of observing time, no CO data were obtained outside the optical borders of the globule.
We present a montage of several of the observed CO profiles in Fig. 5. As in Fig. 4, we also have superimposed the visual absorption contours of Andreazza & Vilas-Boas (1996). All the CO profiles are single-peaked with mean FWHM values of 1.23 and 1.44 km s ¹1 for Lupus 4 and GF18, respectively. Peak temperatures are 7-8 K over Lupus 4 and 4-6 K over GF18. The radial velocities ranges between 4 and 5 km s ¹1 and 3 and 4 km s ¹1 for Lupus 4 and GF18, respectively.
The profiles in Lupus 4 are narrower and more similar to each other than those of Lupus 1. However, systematic profile asymmetries at positive velocities are present in almost all of them.

Cloud distances
Based on the spectroscopic information available for the 'field stars' (those listed in Table 2), an individual distance modulus, not corrected for extinction, was derived for every star of known spectral type. We could not obtain its counterpart for those stars observed in projection on to, or located at a small angular distance from, the clouds (those stars listed in Table 1), because photometric and spectroscopic information about them is lacking.
Since there is no available UBV photometry for most of the 'field stars', except for a few of the brightest and nearest ones, following Morras (1981) and Arnal et al. (1993), a lower limit to the visual extinction was estimated using the relationship P V # 3A V (Hiltner 1956).

Observations towards the Lupus clouds 499
q 1998 RAS, MNRAS 300, 497-510   Then, using the absolute magnitudes listed by Corbally & Garrison (1984), a distance modulus corrected for visual extinction was obtained for every individual star. Diagrams of P V versus this individual extinction-corrected distance modulus are shown in Figs 6(a) and (b) for stars in the fields of Lupus 1 and Lupus 4, respectively. There, a sudden increase in P V at a distance modulus < 5.7 mag (< 140 pc, Fig. 6a) and 5.5 mag (< 125 pc, Fig. 6b), respectively, is clearly noticed. We ascribed these jumps to the presence of aligned dust grains related to the molecular clouds where the dark filaments are immersed. No error bars were quoted in the individual stellar distance modulus due to a lack of true visual extinction values.
A star which appears to be in a discrepant position in Fig. 6(a) is HD 141978 (distance modulus < 4.4 mag). It was classified as a G5V star by Houk & Cowley (1975) and they assigned to their objective-prism spectra a quality factor of 2. The latter means that 'the spectra may be slightly over-or underexposed or slightly overlapped'. Thus, it is possible that the spectral type and/or the luminosity class could be slightly different: a small difference in the latter can mean that the distance modulus of this star will be greater (for a G5IV star, the distance modulus would roughly be 6.4 mag). Thus, its discordant position could be the result of a misclassification. Appenzeller, Mundt & Wolf (1978) have used the bright star HD 140748, which apparently illuminates matter belonging to the molecular cloud Barnard 228, and derived a distance of 125 pc to Lupus 1. Franco (1990) have studied the colour excess distribution from stars located in the neighbourhood of Lupus 4 and found a distance of 165 6 15 pc for this dark filament. Hughes, Hartigan & Clampitt (1993) estimated a distance of 140 6 20 pc to the Lupus clouds, assuming that all of the Lupus subgroups are at the same distance. All these values agree well with our values of 125-140 pc, and fall within the 130-170 pc range suggested by Murphy et al. (1986). Our results and the agreement among other works would suggest that the filamentary clouds Lupus 1 and Lupus 4 are embedded in a single diffuse cloud located at a distance range of 130-170 pc.
If these estimated distances are correct, from their observed major angular sizes and their average angular widths, as determined by Schneider & Elmegreen (1979), both dark clouds are 6-7 pc long, and 0.6-0.7 pc wide.

The masses
It must be pointed out that we did not attempt to obtain the total molecular mass of these clouds, because the molecular gas distribution is likely to have been undersampled by our observational scheme and no information from the 13 CO data has been obtained. Thus, we have only used our 12 CO observations in order to estimate the molecular mass of each condensation. To do this, we selected the CO profiles located inside the condensations defined by Andreazza & Vilas-Boas (1996), namely A to F in Lupus 1 and A to D in Lupus 4. Then, we computed the total CO intensity of every condensation as where A cond is the angular size of the condensation, as given by Andreazza & Vilas-Boas (1996) in arcmin 2 , A CO is the CO beam, (8.7 arcmin) 2 , and W mean CO is the average CO intensity. After this, we used the relation suggested by Strong et al. (1988) to estimate the H 2 masses of the condensations.
It is very hard to determine the formal errors involved in those masses, because several sources of uncertainties are present. First, the CO/H 2 conversion factor of Strong et al. (1988), based on diffuse galactic g-ray maps, H I surveys and CO surveys, probably has an uncertainty of & 20 per cent. Although the 12 CO line J ¼ 1 → 0 is optically thick, this conversion factor is frequently used in the literature due to the fact that it remains at approximately the same value in different regions of the Galaxy, except towards the Galactic Centre (Ramana Murthy & Wolfendale 1993 and references therein). Secondly, the value of 3.2 assumed by Andreazza & Vilas-Boas (1996) for R V may be another source of uncertainty: several works Rydgren 1984 andArnal et al. 1993, among others) have shown that R V increases toward the densest parts of the dark clouds.
Bearing in mind the factors mentioned above, the agreement between both estimates of the molecular masses of every individual condensation can be considered reasonable.

The ambient magnetic field
The polarization data can also be used to examine the difference between the polarization angles of those background stars, the angular distance of which from the dark clouds is small, and the polarization angles of those stars located further away (the 'field stars'). To this end, the polarization angle data, in the direction of stars with interstellar polarization greater than 0.5 per cent, were grouped into bins 108 wide, for stars observed in or close to Lupus 1 and Lupus 4 (Figs 7a-7b), and 208 wide for the so called 'field stars' (Figs 7c-7d). The mean values of the E-vector polarization angles v and its standard deviations are given in Table 4. Myers & Goodman (1991) determine mean values for v of 508 and 228 for Lupus 1 and Lupus 4, respectively, in good agreement with our 'close stars' values -the first and second rows of Table 4.
The observed dispersion in the polarization angle distribution is always greater than the instrumental uncertainty, suggesting that this dispersion may contain some information about the degree of variation of the magnetic field direction. A model explaining such dispersion in dark clouds, as arising from a three-dimensional magnetic field having uniform and non-uniform parts, can be found in Myers & Goodman (1991). Unfortunately, we can not apply this model to our observations, because information about the line-of-sight component of the magnetic field is not available.
In the case of Lupus 4 we note -after a visual inspection of Figs 2(b), 3 and 7, and comparing the results of Table 4 -that the projected magnetic field mean angle is very similar in both sets of data ('field stars' and 'close stars'), which suggests an alignment of the overall magnetic field roughly perpendicular to the optical filament. This observational finding allows us to suggest that Lupus 4 could have been formed as the result of matter collapse along the magnetic field lines.
The picture in Lupus 1, however, appears more complicated. Apparently, the role played by the magnetic field in controlling the properties of the dark filament is not as simple as Strom et al. (1988) suggested. Our data show that the magnetic field direction deduced from stars close to Lupus 1 is significantly different from the magnetic field direction determined from the 'field stars' data. The position vectors of the 'field stars' are remarkably well aligned with each other and with the projected long axis of the cloud (see Fig. 3 Andreazza & Vilas-Boas (1996). For the CO profiles, the plotted magnitudes are T ¬ A in K (ordinates) and V LSR in km s ¹1 . Dark points correspond to the CO observations reported in this paper, while open squares indicate the H 2 CO positions observed by Sandqvist & Lindroos (1976).
Downloaded from https://academic.oup.com/mnras/article/300/2/497/1037573 by guest on 23 November 2021 characterized by its mean value, i.e. the distribution of individual v ranges mainly from 208 to 1008. This high dispersion could be explained by: (i) an intrinsically more complex v distribution, such as a bimodal distribution like the one suggested by Goodman et al. (1990) for the Perseus complex; or (ii) the fact that the magnetic field did not play a major role in determining the cloud structure at small scales. Andreazza & Vilas-Boas (1996) pointed out that the distribution of the magnetic field orientation does not have any apparent relationship with either the major axis of the condensations or the filaments. We tested these conclusions by observing the relative orientation of the mean polarization angles of the stars located in projection very close to the condensations in both filaments. Based on the results obtained, the magnetic field in Lupus 1 appears to be highly variable from one condensation to another and even inside the same condensation: the polarization angles are 178 (stars 4 and 6) or 1008 (stars 5 and 7) for the condensation A, 558 for condensation D, and 238 (stars 23 and 27) or 758 (stars 24, 25, 26, 28 and 31) for the joined condensations E-F. In contrast, the polarization vectors in Lupus 4 appear highly collimated: average values of 258, 228, 238and 288 are obtained for condensations A to D. These facts, along with the features found from our CO profiles (line asymmetries, broad profiles and double-peak structures) could indicate that Lupus 1 has a more intricate cloud structure than Lupus 4.

Molecular gas motion
In order to illustrate both the velocity field and the distribution of the molecular material in Lupus 1, we have analysed three subsets of Observations towards the Lupus clouds 507 q 1998 RAS, MNRAS 300, 497-510  Sandqvist & Lindroos (1976), shown in Fig. 8 as open circles, agree with the trend shown by our CO measurements. Fig. 9 shows the changes in V R along the Lupus 4 filament. The radial velocity reaches a minimum of < 4 km s ¹1 towards the central part of the filament. It could be indicating the presence of a small expansion or contraction in the cloud. Fig. 9 also includes the H 2 CO measurements of Sandqvist & Lindroos (1976) and Goss et al. (1980). The agreement between both data sets is quite good.
All measured velocities in GF18 are in the narrow range from 3.5 to 4 km s ¹1 , slightly lower than the corresponding ones for Lupus 4 itself. In addition, a systematic asymmetry in all CO profiles was noticed: V R is always greater than the peak velocity, suggesting the presence of a low-intensity component at higher velocities (see Fig. 5).
A crude one-dimensional model aimed at studying the dynamical stability of the clouds was constructed for both filaments. We have considered each filament as a bunch of aligned condensations moving at different velocities along the line of sight. This simpleminded model shows that Lupus 1 would not be recognizable as a single cloud after 3 or 4 Myr. In constrast, Lupus 4 would keep its actual morphology for at least 5 Myr.  Sandqvist & Lindroos (1976), and the triangles indicate the H 2 CO observations of Goss et al. (1980). In the filament, contour lines are from the visual absorption map of Andreazza & Vilas-Boas (1996), while the dashed line in the globule represents the optically defined outer boundary of the dark cloud.    interstellar medium. The mean radial velocities derived from both the CO and the H 2 CO data are consistent with the outward peculiar motion of the H i, dust clouds and bright young stars outlining the local system called Gould's Belt (Lindblad et al. 1973;Olano 1982;Pöppel 1997).

C O N C L U S I O N S
From the starlight polarization data we estimated a lower distance limit of 140 and 125 pc for Lupus 1 and Lupus 4, respectively. The magnetic field towards Lupus 4 has a well-defined direction, which coincides with the magnetic field direction inferred from the 'field stars' data. This field is almost perpendicular to the major axis of the cloud.
Close to Lupus 1 the magnetic field direction seems to change from place to place without exhibiting a clear pattern. The major axis of this cloud is parallel to the magnetic field direction derived from the 'field stars'.
The radial velocity derived from molecular line data, and the distance determination of these clouds, seems to indicate that they belong to Gould's Belt.