next up previous contents index
Next: Eclipse Decomposition of Up: StarsStellar Populations, Previous: Is the Crab

Science with the Hubble Space Telescope -- II
Book Editors: P. Benvenuti, F. D. Macchetto, and E. J. Schreier
Electronic Editor: H. Payne

HST Studies of Carbon and K-M Giant/Supergiant Stars

Kenneth G. Carpenter
Laboratory for Astronomy and Solar Physics, Code 681 NASA-GSFC, Greenbelt, MD 20771

Richard D. Robinson
Computer Sciences Corporation @ LASP-NASA/GSFC, Code 681, Greenbelt, MD 20771



Results of a program to measure, using the Goddard High Resolution Spectrograph (GHRS) on the Hubble Space Telescope (HST), macroturbulent and flow velocities, the acceleration of winds, and the amount of hot (transition-region) plasma in the outer atmospheres of Carbon and K-M Giant/Supergiant stars are presented. We have measured in several such stars the acceleration of the stellar winds in the chromospheres (from initial velocities of 3--9 to upper velocities of 15--25 ) and the chromospheric macroturbulence ( 25--35 ). Numerous new emission features have been detected and identified, including (e.g., in the non-coronal giant Tau) weak C IV emission indicative of hot transition-region plasma, many new fluorescent lines of Fe II, and fluorescent molecular hydrogen emission and Ca II recombination lines seen for the first time in a giant star. The UV spectrum of two carbon stars have been studied with unprecedented resolution and reveal extraordinarily complicated Mg II lines nearly smothered by circumstellar absorptions.

Keywords: globular clusters,peanut clusters,bosons,bozos


UV spectra of K-M giant and supergiant stars and of carbon stars have been acquired with the Goddard High Resolution Spectrograph (GHRS) on the Hubble Space Telescope (HST). These spectra have been used to measure chromospheric flow and turbulence velocities, study the acceleration of their stellar winds, acquire constraints on the outer atmospheric structure of such stars, and provide data needed to understand the radiative line transfer in these atmospheres. We have observed the normal, oxygen-rich giant stars Dra (K5 III hybrid), Tau (K5 III), Cru (M3.4 III), Gem (M3 IIIab), and 30 Her (M6 III) and the supergiant Ori (M2 Iab), as well as the carbon stars TX Psc (N0; C6,2) and TW Hor (N0; C7,2). The high resolution and wavelength accuracy of these data have allowed the direct measurement of the acceleration of the stellar winds and the macroturbulence in the chromospheres of several of these stars. The high signal-to-noise and large dynamic range of these spectra have allowed the detection and identification of numerous new emission features, including weak C IV emission indicative of hot transition-region plasma in the non-coronal giant Tau (Carpenter, Robinson, & Judge 1994), many new fluorescent lines of Fe II, and the first detection of molecular hydrogen and of Ca II recombination lines in the UV spectrum of a giant star (McMurray et al. 1996). The UV spectrum of two carbon stars have been studied with H. Johnson et al. (see, e.g., Johnson et al. 1995) with unprecedented resolution and reveal extraordinarily complicated Mg II lines strongly mutilated by overlying circumstellar absorptions. Details and some examples of these results are given below. Although, there are far too many data to be shown in this paper, further examples (some indicated herein) can be found in Carpenter (1996).


We have used the profiles and widths of optically-thin emission lines to characterize the macroturbulence in the chromospheres of these stars. Our primary diagnostic is the C II] (UV 0.01) multiplet of semi-forbidden lines near 2325Å. These lines show no evidence for opacity broadening, but are still much broader than one would expect on the basis of thermal microturbulence (about 6 in these chromospheres). We have used both the GHRS G270M and Echelle-B gratings in these observations. A sample of these data are shown in Carpenter (1996).

Figure: Fits to C II Lines assuming isotropic and anisotropic macroturbulence

The Echelle data, in particular, show that the lines are broadened at the base, relative to the single Gaussian profile expected for simple isotropic macroturbulence. Following a suggestion by David Gray, we have found that the observed profiles are much better fit assuming an anisotropic macroturbulence, in which the macroturbulent velocities are confined to the radial and/or tangential directions, as one might expect at the edges and tops of convective cells. The best fit is produced assuming pure radial macroturbulence although tangential motions produce similar profiles and some contribution from such motions cannot be completely ruled out on the basis of these data. Figure 1 illustrates the best fits assuming isotropic macroturbulence (dots) and an anisotropic, radial-tangential macroturbulence (dashes) to the observed C II 2325Å line in Ori and Tau. Table 1 lists our estimates of the mean macroturbulence derived from this fitting process (along with other results to be discussed later), assuming a pure radial distribution of the macroturbulent velocities.

Table 1: Measured Macroturbulence, Mean Flow Velocities and Wind Acceleration for Cool Giants and Supergiants (in km s).

Plasma Flows and Wind Acceleration

A second very important diagnostic of the velocity fields in these stars are the large variety of Fe II lines seen in the mid-UV spectral region. The Fe II profiles are more complicated in appearance than the simple emission lines seen in C II, in that many of the lines have one or more absorption self-reversals superposed on the emission component. The Fe II UV 1 2586Å line in three stars is shown in Carpenter (1996), where it can be seen that Cru shows two self-absorption components (a strong, blue-shifted one and a weaker red-shifted one) like all K-M giant stars, Ori shows a single blue-shifted absorption, and Vel shows multiple self-absorptions, all blue-shifted. The blue-shifted components are indicative of formation in an outflowing stellar wind and the variation of their shift with intrinsic line strength provides a way to measure the acceleration of the stellar wind with height. The red-shifted self-absorptions may indicate a weaker downflow of some material or, more likely, be the result of a subtle radiative transfer effect and a turbulence field which changes with height (Ensman & Johnson 1995).

We have measured mean flow velocities of the C II and Fe II ions in these chromospheres by the offset of the mean wavelength of their emission profiles, and, for Fe II, of the self-absorptions, from the laboratory wavelength values, adjusted for the stellar radial velocity. These flows are tabulated in Table 1. The mean flows of the emission components of both ions correlate well with each other and generally show either a slight inflow or slight outflow of 2--4 . The mean flows of the blue-shifted Fe II self-absorptions indicate means outflows of several to as much as 10 , while the mean apparent velocities of the weaker ``inflows'' are seen at 8 to 14 . We also find that the observed blue-shift of the strong Fe II reversals increases with increasing line strength, indicating that the outflow, i.e., the stellar wind, is accelerating with increasing distance from the stellar photosphere (since the self-reversals of the stronger, more opaque, lines are formed higher up in the atmosphere). We characterize the relative line strengths by a relative line center opacity (), computed for a temperature of 6000 K, a hydrogen column density of cm, a microturbulence of 6 , an electron density of cm, and a solar Fe II/H abundance. Figure 2 shows the observed shift of the stronger Fe II self-absorption versus this relative log() in the Cru rest frame.

Figure: The increase of Fe II self-absorption blue-shift with increasing line strength (height in chromosphere), reflecting the acceleration of the wind in Cru.

We are first able to detect the wind at about 7 (where the Fe II absorption first becomes thick enough to observe) and can follow it higher into the atmosphere, as it accelerates up to about 15 . None of the Fe II lines are opaque enough to allow us to sample higher and perhaps faster moving regions. In order to sample higher regions in giant stars, we must use other, more opaque, lines. One good diagnostic of these regions is the O I UV 2 multiplet, three lines near 1304Å which are also self-reversed, but even stronger than the strongest of the observed Fe II lines (the 2756Å line). Figure 3 compares the observed profiles of these lines to the Fe II 2756Å and 2737Å lines in the K-giant Tau. In this star, the use of the O I lines allows us to follow the wind up to about 25 .

Figure: The acceleration of the wind in Tau as seen in the increasing blue-shift of the Fe II and O I self-reversals with increasing line strength.

Non-coronal vs. Hybrid Star Chromospheres

The difference in hybrid vs. non-coronal stellar atmospheres is well-illustrated by the differences in the spectra of Dra and Tau around 1550Å (see Figure 4).

Figure: The region near C IV (UV 1) in a non-coronal and a hybrid giant.

We have obtained two deep exposures of this region for the purpose of measuring the amount of flux (or better upper limits) arising from hot (transition region) material in these stars. The non-coronal stars (K2 and later on the giant branch) like Tau typically show evidence for slow, massive winds, but little or no evidence of transition region or coronal material, while the hybrid-chromosphere stars, although later than the dividing-line spectral type of K2, look more like the earlier-type coronal giants with evidence of hot material and faster winds. The GHRS data we obtained were surprising in that we detected a C IV surface flux from the non-coronal giant comparable in strength to that in the hybrid star, suggesting a non-trivial amount of transition-region material in the non-coronal star. However, the higher density, cooler temperatures of the non-coronal atmosphere is spectacularly evident in the spectrum of Tau in the form of a myriad of narrow fluorescent emission lines from Fe II and H, and recombination lines of Ca II. These lines are weak or absent in the hybrid star, but dominate the spectrum of the non-coronal star (see McMurray et al. 1996).

The absorptions superposed on the Mg II h & k emission profiles are, like the Fe II lines, useful probes of the stellar wind, but they also frequently contain absorptions originating in circumstellar shells and the interstellar medium. Figure 5-A compares the profiles of the Mg II k-line in two K5 III stars, the normal giant Tau and the hybrid-chromosphere star Dra.

Figure: Mg II profiles in Tau, Dra, Ori, Cru, and TX Psc.

The profile from the latter star differs from the former in three ways: 1) it has a strong ISM absorption (marked ``ISM'' on the plot) to red of line center (in Tau the ISM absorption is masked by the chromospheric self-reversal), 2) the wind absorption extends to much higher velocities (up to 70 vs. 30 ), and 3) it lacks the sharp, weak absorption at zero velocity (relative to the star) seen in Tau (marked ``S'' on the plot). The visibility of the ISM absorption (item 1) is controlled by chance, i.e., whether or not the stellar and ISM radial velocities happen to be similar or not. The higher speed winds (item 2) are characteristic in general of hybrid stars. The feature noted in item 3 however, has, to date, only been seen in Tau and we believe it is formed by a shock formed as the star passes through the interstellar medium and not a feature intrinsic to this or other non-coronal giants.

Circumstellar Shells

Figure 5-B&C compare the k-line profiles in three stars, the M-giant Cru, the M-supergiant Ori, and the carbon star TX Psc. The carbon star has a profile intermediate in width between the M giant and supergiant (indicating an intermediate luminosity), although the superposed circumstellar absorptions resemble more closely those in the supergiant. The h-line in the carbon star is even more mutilated by overlying circumstellar absorptions than the k-line and by features which do not appear at all in the M-supergiant (Figure 5-D). The circumstellar absorptions identified as Mn I UV 1 and Fe I UV 3 in Ori show outflow velocities consistent with previous measures of CS shell velocities (Bernat, 1977).


We are gratefully acknowledge our collaborations with H. Johnson et al. which have produced the excellent observational data on the carbon stars and the collaborative program with Alex Brown which obtained the data on Dra. We also acknowledge helpful conversations with Phil Judge, Graham Harper, and Carole Jordan.


Bernat, A. P. 1977, ApJ, 213, 756

Carpenter, K. G. 1996, in Ninth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, eds. R. Pallavicini & A. K. Dupree, ASP Conference Series, in press

Carpenter, K. G., Robinson, R. D., & Judge, P. G. 1994 BAAS, 26, 1380

Ensman, L. & Johnson, H. R. 1995, BAAS, 27, 839.

Johnson, H. R. et al. 1995, ApJ, 443, 281

McMurray, A. et al. 1996, in Ninth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun, eds. R. Pallavicini & A. K. Dupree, ASP Conference Series, in press

next up previous contents index
Next: Eclipse Decomposition of Up: StarsStellar Populations, Previous: Is the Crab