Table 1 (click here) presents the stellar parameters for our sample. Earlier photometric and polarimetric work in the Chamaeleon region has analyzed the stellar content of both member and field stars (Whittet et al. 1987), and a search of the SIMBAD database and reading of literature has produced the notes included in Table 1 (click here). Important in the discussion of any polarimetry is the possibility of intrinsic polarization from the source. We chose to observe stars which were optically determined to be field stars, which are expected to have little or no intrinsic polarization. Three of the stars in our sample are IRAS point sources, which is probably a result of the late spectral type and bright magnitude. Recent pointed ROSAT observations (Feigelson et al. 1993; Zinnecker et al. 1996), however, have identified four new YSOs which are associated with our ``field'' stars, these are indicated in Table 1 (click here). Two of them are both IRAS point sources and X-ray sources. The polarimetry and the IR emission of these stars will probably be influenced by circumstellar material.
Figure 1 (click here) shows an IRAS 100 
image of the Chamaeleon I
cloud, with the locations of our program stars indicated.
Figure 1: Location of program stars plotted on IRAS 
m
map of the Cha I cloud
Measurements of optical linear polarization of starlight can be used
to trace the geometry of the interstellar magnetic field component,
, in the plane of the sky. The observed polarization
is produced by the differential extinction by non-spherical dust
grains associated with the interstellar clouds along the line of
sight. It is assumed in fact that elongated paramagnetic dust grains
are aligned by the local interstellar magnetic field, via the
Davis-Greenstein mechanism with their short axes
parallel to the field direction. Strong evidence that extinction by dust grains is responsible for the
observed polarization comes from the existence of a correlation between
the degree of polarization and the amount of reddening (Spitzer 1978).
In Fig. 2 (click here) we present a plot for the Chamaeleon I cloud, with the
observed polarization vectors for our program stars, superimposed onto
a plot of IRAS 100 
m contours, with equatorial coordinates
labeled. The magnetic field vectors would be expected to follow the
polarization vectors, and they are approximately parallel to the
galactic plane. The dispersion in the values of the polarization angle
is very small, typically 2 degrees or less. The complete results of the
polarization measurements are presented in Table 2 (click here) for our sample of
stars in the five passbands.
Figure 2: Average polarization vectors for polarimetric program stars.
The 100 
m contours of IRAS are also included, as are coordinate grids for
equatorial coordinates and galactic plane parallel. The length of the symbol is
proportional to the percentage polarization
Figure 3:
a) Least square fits to Serkowski's law for individual stars.
The Serkowski polarization model was fitted with K = 1.15, and free
parameters 
and 
We have adopted the parameterisation of the wavelength dependence of the linear polarization, referred to as Serkowski's model (Coyne et al. 1974),
where 
is the wavelength at which the maximum
polarization, 
, is observed, and 
. The value
of 
for our sample ranged from 0.35 to 0.9 
m and
typically was found to be near 
= 0.55 
m. Larger
values of 
are expected in direction of dense clouds,
indicating an increase in the mean grain size. A discussion of the
theoretical arguments behind the Serkowski polarization model may be
found in Spitzer (1978).
For selected stars we have fitted the percentage polarization with a
Serkowski polarization model to derive values of 
and
P
. The data points, and their best fits are presented in
Fig. 3 (click here). The Serkowski model was fitted only to those stars which
had accurate polarization measurements in at least four passbands. We
excluded measurements from our fit which had polarization
uncertainties 
.
To check the effect of fitting the Serkowski model with four points,
we also did all the fits using both 5 and 4 points and, typically, the
values of 
derived in the two ways do not differ by
more than 100 Å. For 19 stars we have fitted the Serkowski
polarization model with 5 points, while 9 stars were done with 4
points, since the late spectral types of the stars prevented accurate
measurements of U band polarization. Seven stars were not fitted at all,
since both the U and B polarization were found to be inaccurate, or
the star was found to be unpolarized.
In Fig. 4 (click here) we present plots of the percentage polarization for each
of the five bands against the color excess 
. It is clear that
the polarization rises for 
at a different slope than
for the largest values of 
. A similar change in slope
has been observed by Whittet et al. (1991), and it is believed to
reflect a decrease in the efficiency of polarization at the highest
extinctions due to either more spherical grains, or de-alignment of
the grains in the densest parts of the sight lines. Figure 5 (click here) shows
the maximum polarization 
of each star, plotted against the
color excess 
. Also shown in Fig. 5 (click here) are separate fits for
the sight lines where 4 or 5 points were used to fit the data to the
Serkowski model. We see a gradually increasing function, with a fair
amount of dispersion in the measured values of 
.
Figure 4:
Polarization percentage for each of the five passbands for the
polarimetric sample. Included are the best fits to the data (dashed line) and
``maximum" polarization, 
, (solid line)
from Spitzer (1978). The different slope between high
and low extinction sightlines is apparent
Figure 5:
Maximum polarization, 
, plotted against 
for the
polarimetric sample. The filled circles indicate sightlines which had 5
measurements for fitting to the Serkowski model, while the open circles
are sightlines which only had measurements in 4 filters.
The dotted line shows the best fit to all of the points, and the solid
line the best fit to sightlines which included 5 measurements.
The dashed line represents the ``maximum" polarization efficiency from
Spitzer (1978)
From the parameterisation of the polarization wavelength dependence
it is also possible to determine 
, and we examined the behaviour
of 
as a function of extinction. The values of
increase with 
and a plot is presented in
Fig. 6 (click here) for our sample. The increase in 
at large
is consistent with larger dust grains occurring in denser
sight lines with high extinction, perhaps as the result of ices
accumulating on the grains.
Figure 6:
Wavelength of maximum polarization, 
versus 
.
The dotted line represents the least square fits to all data points, while
the solid line gives the least square fits for only values of
derived from more than three points fits to the Serkowski's
law (filled circles)
In Fig. 7 (click here) we present a plot of the distribution of 
,
which appears to be bimodal, with a smaller population of points
having larger values of 
. It appears that most of the
large values of 
also coincide with the peak 100 
m
emission from the cloud, based on an examination of the location of
the program stars on the IRAS map of Fig. 1 (click here).
Figure 7: Histogram of 
values, showing a bimodal population
with the larger values of 
concentrated mostly in the center
of the cloud
Variations in 
are generally attributed to variations
in the mean grain size toward the region. Some complications can
arise, however, in regions with multiple clouds along the line of
sight, which may have different values of 
, or
different orientations of the mean magnetic field. In the former case,
significant deviations from the Serkowski model may be found, while in
the latter case, the electric field vector orientation may be a
function of wavelength. Clarke and Al-Roubaie (1984) have modeled the
effects of various grain size distributions, and specifically the
rotation of the polarization vector 
with the
angle 
between two clouds along the sight line. Additionally a
linear relation exists between the width parameter K and
. The physical significance of the parameter K from
the Serkowski model has been examined by Whittet et al. (1991), but is
presently uncertain whether K varies significantly with direction
within the Galaxy.
We have assumed a value of 1.15 for the Serkowski K parameter, and
also that most of the polarization results from a single component.
The spectroscopic results of Sect. 3.3 suggest that there are two
components of the ISM in the line of sight to the Cha I
association. The second component, which appears in several of the Ca
II spectra, is much weaker, however, and therefore our assumption of a
single dominant component is valid. Our observed rotation of the
polarization angles for a given sight line varied by less than 
for nearly all sight lines, which is also consistent with a single
dominant component of polarizing interstellar medium (Clarke &
Al-Roubaie 1984).
Infrared observations of nearby molecular clouds and atomic cirrus
have also been used to deduce an increased small grain content from
increased R(12,100), and decreased 
near the edges of
clouds, which undergo rapid fluctuations of the number of small grains
attributed variously to condensations from gas, removal from larger
grains, and from slight redistributions between small (a < 30 Å)
and mid-sized (30 < a < 100 Å) grains (Désert et al. 1990;
Bernard & Boulanger 1993).
Table 3: Infrared data from IRAS. The 100
IR fluxes, I(100),
are in mJy Sr
We have analyzed the values of R(12,100) and other IRAS colours, and
present the results for our program star sight lines in Table 3 (click here). The
IRAS colours were also examined for systematic trends with some of our
other polarimetric and spectroscopically observed quantities. Figure 8 (click here)
presents a plot of percentage polarization against R(12,100). A clear
decrease in polarization is seen as R(12,100) increases, suggesting
that the processes which disrupt the grains in the cloud to produce
the smaller particles, also are capable of either realigning or
destroying the larger grains responsible for optical polarization. The
fact that in Fig. 9 (click here) (top) we see little or no correlation between
and R(12,100) suggests that the process of creating
the small grain population minimally affects the grain size
distribution of the polarizing grains, and therefore the small grains
producing the enhanced values of R(12,100) are probably an independent
population from the polarizing grains.
Figure 8: Plot of percentage polarization against R(12,100) infrared color
Figure 9:
Infrared color indices versus the wavelength of maximum polarization,
.
A strong correlation is seen with 
, and a very weak trend
is seen between R(12,100) and 
, suggesting that the
small grains responsible for R(12,100) enhancements do not appreciably
affect the grain size distribution
Bernard & Boulanger (1993) have modeled the various contributions
of PAH and other small molecules to the small grain content, and
Désert et al. (1990) have determined expected IR colours with
mixtures of PAH's and small grains. Boulanger et al. (1994)
have determined that variations in the far UV rise of the extinction
curves for stars in Chamaeleon I and II were not correlated with 12
and 
m emission, which they interpreted as proof that the
R(12,100) fluctuations were caused by slight variations in the size
distributions of mid sized and small grains having the same
composition as the larger grains. Our results are consistent with this
model, since only a slight trend is seen between R(12,100) and
which is plotted in Fig. 9 (click here). We do see a strong
correlation between 
and 
, however, and a
slight anticorrelation between R(60,100) and 
. These
results suggest that the larger grains associated with increased
values of 
also have reduced grain temperatures and
larger 
m fluxes.
We have also made plots of the IRAS fluxes and colours against the
maximum percentage polarization 
, and these are presented in
Figs. 10 (click here)a and 10b. A definite correlation between 
and
is seen, as is an anticorrelation between R(12,100) and
. These results strongly suggest that the larger grains which
have increased values of 
and 
also are
responsible for large amounts of polarization. Colour-colour plots of
R(12,100) against 
(Fig. 11 (click here)) reinforce the trends shown
in Figs. 10 (click here)a and 10b, as does the plot of R(60,100) vs. R(12,100)
(Fig. 11 (click here), bottom), which suggests that the larger values of
R(60,100), usually associated with heated grains, also give rise to
larger numbers of small particles which are responsible for the
increase of R(12,100).
Figure 10: a)
Plots of IRAS fluxes against 
, the maximum polarization
percentage from Serkowski fit. Strong correlations are seen in the
60 and 100 
m fluxes
Figure 10: b)
Plots of IRAS colors against 
, the maximum polarization
percentage from Serkowski fit. Strong anti-correlations are seen in the
R(60,100) and R(12,100) plots, suggesting that sightlines with grain heating
also have substantial dealignment of the polarizing grains
Figure 11: Color-color IRAS plot for the points on the Cha I cloud toward stars
in our polarimetric sample. (Top): R(12,100) values are seen to highest where
is minimized, which is consistent with an interpretation of
R(12,100) enhancements as heating at the edge of the Cha I cloud.
(Bottom): R(60,100) values are seen to correlate with R(12,100), as is
seen in other diffuse clouds
The transitions from large to small grains are generally attributed to
variations in either the ambient UV radiation field or convective
turbulent motions within the interstellar cloud. Both processes would
be expected to affect the depletion of ices from the surfaces of
grains, and to release highly depleted elements such as Ca into the
gas phase. The observation of Ca II in absorption from background
stars should therefore provide a useful probe of grain destruction or
ablation from radiation or kinetic processes. We present in Fig. 12 (click here)
the absorption lines of Ca II for our sample, in order of Right
Ascension of the star. Line profile models for all the spectra have
been computed, and the values of 
, b, and N(Ca II) are
presented in Table 4 (click here). The Ca II absorption is quite uniform across the
cloud in velocity, although in some sight lines a weak second
component of absorption appears. The second component is strongest in
the spectra at either edge of the Cha I cloud, suggesting that this
component may arise from either a neighboring cloud, or higher
velocity gas which is impacting the Cha I region.
Table 4: Spectroscopic results
Figure 12:
a)
Spectra for all of our spectroscopic program stars in Ca II
absorption, in order of right ascension. The Cha I cloud Ca II absorption
appears at a nearly constant velocity of 
km s
, with
a second blended component visible in sightlines at either end of the cloud.
The strongest Ca II absorption appears at the East end of the cloud
High resolution spectra of CH absorption are presented for the sample
of stars in Fig. 13 (click here). The fitted line profiles are included, and the
values of 
, b, and N(CH) are also presented in Table 4. The
molecular content of clouds is generally found to correlate with
m emission, since the larger grains which emit 
m are
effective in shielding the molecules from the dissociating
interstellar radiation field. However, if transient processes exist
in molecular clouds which heat small grains and can produce molecular
CH, then it should be possible to see trends between N(CH) with
and R(60,100). Recent observations have found significant
amounts of CH in the warm envelopes of molecular clouds, suggesting
that both CH and CH
may exist as a transient phase of molecular
gas in some molecular clouds (Crane et al. 1995).
Figure 13:
a)
Spectra for all of our spectroscopic program stars in CH II absorption,
in order of right ascension. The Cha I cloud CH II absorption appears in
a single strong component at the same velocity of the dominant Ca II
absorption line. The strongest CH absorption appears at either end of the
cloud
Figures 14 (click here)a-d present stacked spectra for single sight lines, and it
is clear that the CH/CH
ratio is highly variable in the Cha I
cloud. In the sections below, we discuss the data for each of the four
Cha I sight lines, detailing the relationships between IRAS and
spectroscopic results.
Figure 14 (click here)a shows a stacked plot of the spectra for the HD 99759
sight line, which appears toward the Eastern edge of the Cha I cloud,
where the IRAS 
m flux is 
mJy Sr
.
The observed Ca II absorption profile has been fitted by two
components giving a total column density N(Ca II) = 
cm
.
The molecular content for this sight line is high with
N(CH) = 3.0 
cm
and N(CH
cm
.
The CH
profile was fitted by a single component and
appears asymmetric, which may be the result of two blended components.
Figure 14:
a) Stacked spectra for the HD 99759 sight line.
The absorption lines are discussed in Sect. 3.3.1.
Figure 14:
b)
Stacked spectra for the HD 97048 sight line. The absorption
lines are discussed in Sect. 3.3.2.
Figure 14:
c) Stacked spectra for the HD 96675 sight line. The absorption
lines are discussed in Sect. 3.3.3.
Figure 14:
d) Stacked spectra for the HD 97300 sight line. The absorption
lines are discussed in Sect. 3.3.4.
The extinction for the sight line is only 
, and the
column density of CH is one of the highest per unit of extinction, as
seen in Fig. 15 (click here). In low extinction sight lines, like the one toward
HD 99759, the dissociation rate of CH from the interstellar radiation
field must be balanced by an increased production of CH to account for
the large molecular column densities. A production mechanism which
could be attributed to shocks which collide with the Cha I cloud would
be most pronounced near the edges of the dark cloud region, such as
the material toward HD 99759. The large value of N(CH
(CH) = 0.63
is consistent with this hypothesis, since CH
has been
thought to be produced in shocks (Allen 1994).
Figure 15: CH column density, N(CH), is plotted against 
for spectroscopic
program stars, showing enhanced N(CH)/
content for sightlines
toward HD 94414, HD 97300, HD 96675, and HD 99759
Figure 14 (click here)b shows a stacked plot of the spectra for the HD 97048 sight line,
which appears toward the Southern central part of the Cha I cloud, where the
IRAS 
m flux is 
mJy Sr
.
HD 97048 is thought to be embedded
in the Cha I cloud, and the large IRAS fluxes therefore come from
circumstellar material being heated by the star.
HD 97048 is associated with the reflection nebula Ced 111, and has been found
to be surrounded by additional IRAS sources, which suggests that HD 97048 is a
center for low-mass star formation. Recent reviews on the HD 97048 and HD 97300
sight lines include Assendorp et al. (1990), and Steenman
&
Thé
(1989).
The latter
have found that the two stars have anomalous extinction, with 
,
which would result in an optical extinction for HD 97048 of 
,
which is less than half of the estimated maximum extinction of 
for the Cha I cloud. The HD 97048 sight line would therefore be
on the near side of the Cha I cloud, within a reflection nebula
which consists of a mixture of radiation processed gas and dust. The CH
column density for HD 97048 is slightly higher than the HD 99759,
at N(CH) = 3.4 
cm
, which is surprising considering the large
radiation field which must be present from the embedded source HD 97048.
The CH
column density for this sight line is extremely low, with
N(CH
cm
.
It is interesting to note that two sight lines from the same cloud, HD
97048 and HD 99759, have nearly identical colour excesses of 
, and yet have extremely different ratios of N(CH
)/N(CH),
with value of N(CH
(CH) < 0.043 for HD 97048 and
N(CH
(CH)
for HD 99759, respectively. This difference
may be due to the fact that HD 99759 is preferentially sampling the
edge of the cloud, where our results strongly suggest 
production is enhanced.
The Ca II column density toward HD 97048 is large, with
N(Ca II) = 6.7 
cm
, from a single component of absorption. The
HD 97048 Ca II absorption is more symmetric than the HD 99759 Ca II
absorption, which is also consistent with the enhanced CH
production toward HD 99759 resulting from evaporation or shock
processing of gas at the edge of the Cha I dark cloud, which might
introduce the extra component of Ca II absorption.
Figure 14 (click here)c shows a stacked plot of the spectra for the HD 96675 sight
line, which appears toward a Northern section of the Cha I cloud,
where the IRAS 100 
m flux is 
mJy Sr
. The
Ca II column density toward HD 96675 is moderate, with N(Ca II) =
cm
, distributed over two components as for HD
99759. The extinction is again similar to that of the previous two
sight lines, with 
, and with a substantial CH column
density of N(CH) = 3.1 
cm
, one of the highest per unit
of extinction, as seen in Fig. 15 (click here). The CH
column density is
very weak, with N(CH
cm
. It turns out
that the HD 96675 sight line has a large molecular content, yet a
small ratio of N(CH
(CH) and N(Ca II)/N(CH). The high value
of 
may hint at a different value of 
for
this sight line, or may reflect a substantial population of large
grains which would shield the CH from dissociation, but perhaps
provide a less favourable environment for the production of
CH
. The value of the total to selective extinction ratio
calculated from the wavelength dependence of the polarization is
, although the value of 
suggested from the
polarimetry is 
Å, which is average for the
polarimetry sample. The IRAS color index R(12,100)=0.04 is very small,
suggesting a reduced population of heated small grains, and the value
of R(60,100)=0.21 corresponds to one of the lowest grain temperatures
of the sample.
Figure 14 (click here)d shows a stacked plot of the spectra for the HD 97300 sight
line, which appears slightly South of HD 96675 but still
in the central part of the cloud. Like HD 97048, HD 97300 is believed
to be embedded in the Cha I dark cloud, and therefore the IRAS colours
are from heated circumstellar material. The IRAS 100 
m flux is
reported to be 
=282 mJy Sr
. The CH
column
density for this sight line is higher than for HD 97048, with
N(CH
cm
, and N(CH
(CH) = 0.08,
which is at least twice as high as the HD 97048 sight line, but still
relatively low compared with other molecular absorption sight lines.
The Ca II column density is substantially higher than for the HD 97048
sight line, with N(Ca II) = 
cm
. The column
densities of CH
and Ca II therefore seem to be correlated, at
least for our limited spectroscopic sample, which again suggests that
the CH
appears in the warmer outer envelopes of molecular
clouds.
and IRAS colours
Figure 15 (click here) is a plot of N(CH) vs. 
and we see a definite
increase in N(CH) with increased extinction. Also notable is the very
large value of N(CH)/
for the stars HD 94414, HD 97300, HD
96675, and HD 99759. Variations in N(CH)/
have been observed
previously in high galactic latitude molecular clouds and may trace
shock formation of molecules (Penprase 1993;
Penprase et al. 1990).
One of the more interesting results from the comparison of
spectroscopic, polarimetric and IRAS data involved the differing
behaviour of N(Ca II) and N(CH) with IRAS colours and 
.
Figure 16 (click here) presents a plot of the column densities N(CH) and N(Ca II)
against R(60,100), which is commonly considered to be a good indicator
of grain temperature. Values of N(CH) are seen to rise steadily with
R(60,100), suggesting strongly that the warmer grains are somehow
producing additional CH. At the same time, Ca II appears to decrease
with grain temperature, which is surprising since much of the gas
phase Ca II is expected to result from the disruption (and therefore
heating) of grains at the edges of clouds.
Figure 16: Plots of N(Ca II) (top) and N(CH) (bottom) against R(60,100), commonly
considered as a diagnostic of dust temperature. An increasing CH column
density is seen toward the warmer sightlines, while Ca II content decreases
Figure 17: Plots of normalized densities N(CH)/
and N(Ca II)/
against 
, derived from the polarimetry. The 
,
values are considered an indicator of grain sizes, and therefore it appears
that CH is produced more rapidly in regions with smaller mean grain sizes
Another interesting relation is found between N(CH)/
and
, which is shown in Fig. 17 (click here).
N(CH)/
decreases with 
, while N(Ca II)/
is
unchanged. One interpretation of this result is that the CH production
may be more favourable on smaller grains, which have decreased values
of 
. Further observations are needed to test this
possibility.
