Vistas in Astronomy, Vol. 29, pp. 253-280, 1986 0083-6656/86 $0.00 + .50 Printed in Great Britain. All rights reserved. Copyright © 1986 Pergamon Journals Ltd., POLARIMETRY OF COOL GIANTS AND SUPERGIANTS Hugo E. Schwarz Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, Surrey, RH5 6NT, U.K. ABSTRACT The polarimetry of cool, bright stars is reviewed, with the emphasis on observational work. The progress made in spectral and spatial resolution, wavelength coverage and sensitivity is discusseg Polarigenic models are considered and some theoretical ideas are discussed. Several interesting classes of object are treated in some detail and some suggestions for future work are made. i. INTRODUCTION Historically both the intensity and direction of arrival of the light that reaches us from the stars have been measured and analysed as a function of time and wavelength. Until recently, all these observations were made in the visible wavelength range; that is roughly between 4000A and 8000A. In this review I will consider mainly this wavelength band but also some infra-red measurements. All the starlight collected contains not only intensity and directional information but also possesses a state of polarization. The latter has been ignored for most of the time during which astronomical observations have been made. The first (theoretical) step towards the concept of intrinsic stellar polarization was taken by Chandrasekhar (1946, 1947) who published the solution of the radiative transfer equation in an illuminated plane parallel atmosphere with Rayleigh scattering. This model predicts polarization effects in certain binary systems and in early-type single stars if a significant departure from LTE, that is a radiative asymmetry, is present. This is in fact the effect that Hall (1949) and Hiltner (1949) were looking for when they serendipitously discovered interstellar polarization (ISP). Through the subsequent study of the wavelength dependence of the ISP, many of the properties of the interstellar medium have been determined. For instance much has been learned about the size distribution and chemical composition of interstellar dust. Another well-studied feature of the interstellar medium is the structure of the Galactic magnetic field. Aligned polarization vectors trace the field lines through the paramagnetic grain alignment mechanism of Davis and Greenstein (1951). In the decades after Hall and Hiltner's discovery it was found that some stars showed time variability in their polarization and a wavelength dependence unlike that generally observed for the ISP. It was realised that these stars were intrinsically polarized. The study of J~vA 29:3-B 253 254 H.E. Schwarz this intrinsic polarization which only certain groups of stars show, forms an important part of astronomical polarimetry and has contributed greatly to the rapid growth of this field in the past twenty years or so *. Since the degree of intrinsic polarization of starlight is usually of order one percent or less, the original method of measurement, using photographic plates, could only be applied to a small number of highly polarized stars. The lower limit to the degree of polarization, p, which can be detected photographically is 5 to 10%, while modern photoelectric devices can measure p to accuracies of about 10-3%. The advantage of the photograph in being capable of recording many images simultaneously is now rivalled by the latest two-dimensional detectors such as the CCD, IPCS and the electronograph. Polarization maps of extended sources and complete spectra of point sources can now be obtained in one observation. However, most high precision measurements of individual stars are made using photomultiplier tubes; their advent has accelerated the development of more accurate photometric and polarimetrie instrumentation. Advances in polarimetric optics have also contributed to this increase in measurement accuracy. For instance Serkowski (1974) and Tinbergen (1974) have described the various optimum combinations of optical elements such as half- and quarter-wave plates, superaehromats and efficient, wide band polarizers. Schemes to compensate automatically for sky background light polarization have been proposed by Serkowski (1974) and Metz (1979). The method used by Metz also removes the wavelength dependent rotation of the polarization angle, present in superachomatic half wave plates. An era has now arrived in which polarimetry can be performed at high spectral and spatial resolution, using low noise detectors with high detective quantum efficiency. The models explaining the observed phenomena are more constrained and have to be better and more complicated than ever before. 2. EARLY YEARS There are several groups of stars that have been identified as showing intrinsic polarization: Be stars, RV Tauri and T Tauri stars, R CrB stars, FK Gom stars, novae, white dwarfs, cool long period variables and irregular variable stars, symbiotics, 6 Scu stars and Ap stars, IE stars and carbon stars. In this review I will discuss only the cool stars among these groups. The presence of intrinsic (as opposed to interstellar) polarization is usually inferred from temporal variations or a peculiar wavelength dependence of the observed degree and angle of polarization. The particular form of these variations can then be used to investigate the mechanism responsible for the intrinsic polarization and hence to study the physics of star atmospheres and the cire%tmstellar environment. Most of the earlier observations were made using broad passbands or with just one wide-band filter giving only crude wavelength resolution or none at all. Some of this work on cool stars is presented in papers by Capps and Dyck (1972); Co]me and Kruszewski (1968); Coyne and Shawl (1973); Dyek (1968); Dyck et el. (1971a); Dyck and Jennings (1971); Kruszewski et *The number of abstracts in Astron. Astrophys. Abs. pertaining to polarimetry has increased by about 300% relative to the total number of abstracts over the period 1969 to 1984. Polarimetry of Cool Giants and Supergiants 255 al. (1968); Serkowski (1971a) and zappala (1967). Important papers are also DombrovskiJ (1970); Dyek, et al. (1971b,c); Dyek and Sandford (1971); Serkowski (1966a, 1966b, 1969a, 1969b, 1970, 1971b); Shakhovskoj (1964); Shawl (1969, 1972), and Shawl and Zellner (1970). These broad-band observations generally show that the polarization of most cool, bright stars decreases with wavelength and that variations with time of both the degree and angle of polarization are present. The form of the wavelength dependence can also vary with time: regularly for some stars, seemingly at random for others, with time scales from days to several years. Outbursts are sometimes also observed, for instance in symbiotics. The groups of cool stars generally show an infra-red excess at ll~m which has been ascribed to eircumstellar silicate grains by Woolf and Ney (1969). Evidence of a correlation between this IR excess and the degree of polarization has been presented by Dyck et al. (19715). Their graph of the degree of polarization, p, against the flux ratio at ll~m and 3.5~m is reproduced in Figure i. l • M STARS • CARBON STARS • S STARS Figure i. Average degree of polarization versus the flux ratio at ii #m and 3.5 ~m (from Dyck et al. 19715). el, • f ] [ IOx I0 -2 20 X I0"2 30xJO -z 40xlO "z FX (l'ly.m) / F),.(5.5p.m) Scattering in an asymmetrical grain distribution surrounding the star has been suggested as a possible mechanism to produce the observed polarizations. Correlations of the degree of polarization with fundamental stellar parameters form an important area of study and some striking correlations have been found, especially with light curves and spectral type. For instance Dyek (1968) obtained observations of Mira-type variables which showed a decrease in the polarization with increasing visuallight. Kruszewski et al. (1968), whose diagram is reproduced in Figure 2, reported a maximum in the degree of polarization at minimum light for V CVn, a cool, irregular supergiant. However, these correlations are not found in all stars. Zappala (1967) reported that he found no simple correlation between the polarization and brightness of a sample of Mira-type variables. 256 H.E. Schwarz i ~ L i I I I I I I 6.5 @@ 7.0 v 7.5 ,,% ~o 8.o 8.5 8 o oo 6 oo N 8o Figure 2. V band light curve, degree P% 4 and angle of polarization for V CVn °%° (from Kruszewski et al., 1968). 2 110 o0OO 8 ° ;... ~ , 100 I I i i I i I I I 2430200 2430300 J.D. Large variations and strong correlations have been found for some individual stars and Shawl (1974) combined all the then available data on Mira in a folded (six cycles) diagram, reproduced in Figure 3. i I I i r [ E %% 4 2 1 Figure 3. U band polarization versus A phase for Mira (from Shawl, 1974). 0 T OOA•% I O Z~ O O O I I r E ; i i O0 02 04 06 08 I0 12 PHASE Clearly, a periodic modulation of the degree of polarization is present which is correlated with the stars light curve in the sense that maximum light coincides with (near) maximum polarization. These variations were ascribed to changes in a cloud surrounding the star, made up of dust and grains in a mixture which varies its composition with the phase of Mira. Note that the sense of the correlation between lightcurve and degree of polarization for Miras is opposite to that found by Dyck (1968). Kruszewski et al. (1968) observed several cool stars and found changes with time in both the degree and angle of polarization which for some well-observed stars were correlated with their light curves. They also found that carbon stars, unlike M-type stars, have a flat wavelength dependence in the yellow-blue region. M-types tend to have a polarization which decreases monotonically towards longer wavelengths over the visible range.