next up previous contents index
Next: Spectroscopic Analysis of Up: StarsStellar Populations, Previous: HST Parallel Imaging

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

Is there a Dichromatic UV Laser in Eta Carinae?

S. Johansson, K. Davidson, D. Ebbets, G. Weigelt, B. Balick, A.Frank, F. Hamann, R.M. Humphreys, J. Morse, R.L. White

Department of Physics, University of Lund, Sölvegatan 14, S-223 62 Lund, Sweden Lund Observatory, University of Lund, Lund, Sweden University of Minnesota, Minneapolis, MN, USA Ball Aeroespace Corporation, Boulder, CO, USA Max Planck Institute for Radio Astronomy, Bonn, Germany University of Washington, Seattle, WA, USA University of California, San Diego, CA, USA University of Colorado, Boulder, CO, USA Space Telescope Science Institute, Baltimore, MD, USA



We propose that gas ejecta, located 0.2" from the central star in Carinae, act as a dichromatic UV laser. The proposal is based on intensity anomalies in two pairs of Fe II lines around 2510Å observed in recent spectra obtained with the HST/GHRS. The break-down of the branching ratios for spontaneous emission is explained as contributions from stimulated emission. An inverted population of two highly excited Fe II levels is due to selective photoexcitation by H I and N I.


Pre-COSTAR observations in 1991 aimed at three settings with the Faint Object Spectrograph (FOS) of the central core region of Carinae, which has been shown to consist of four objects (Weigelt & Ebergsberger 1986) separated by only a few tenths of an arcsecond. The objects are labeled A, B, C and D, where A is the central star and B, C, D most probably blobs of ejected material (Davidson et al. 1995). The 1991 observations gave spectra for two different settings (one failed) in the region 1200--5500Å, but the spectrum of the blobs could not be directly separated from the spectrum of the star. By making some assumptions about the blobs, the observed spectra were treated as linear combinations of the stellar spectrum (A) and the combined blob spectrum (BCD). From these linear combinations the stellar spectrum was separated from the blob spectrum, and the resulting spectra showed remarkable differences (see Figure 6 in Davidson et al. 1995). The stellar spectrum (component A) contains emission lines due to allowed Fe II transitions in the resonance region (2000--3000Å) and the Balmer series of hydrogen. The blob spectrum (components BCD) shows strong forbidden Fe II lines and a dominant emission feature at about 2508Å. The latter was identified as a blend of two highly-excited Fe II transitions, first explained in cool stars as fluorescence lines pumped by H Ly (Johansson & Jordan 1984). The same lines were discussed by Johansson & Hamann (1993) in a recent paper on UV fluorescence observed in IUE spectra, and the particular case of Eta Carinae was emphasized. To further study the 2508Å feature, new observations were needed at higher spatial and spectral resolution.


The central core region of Carinae was re-observed with HST in June 1995 using the Goddard High Resolution Spectrograph (GHRS) at medium resolution (grating G270M). The six different spectrograph settings, 1--6, of the four target objects ABCD are shown in Figure 1.

Figure: Positions of six 0.1" spaced GHRS 0.2" aperture observations of the core region of Carinae, covering the targets ABCD.

Figure: Spectra around the 2507 and 2509Å lines obtained at positions 2 and 4. (The intensity scale of the top panel has been changed in order to demonstrate the strength of the two lines in position 2.)

Three different wavelength regions about 40Å wide and centered at 2510, 2740 and 2845Å, respectively, were recorded. However, in this paper we discuss only the 2508Å feature. In Figure 2 is shown a comparison of the spectra around this feature recorded at target positions 2 (HRS-2) and 4 (HRS-4). GHRS resolves the feature observed in the FOS spectrum as two well-resolved, strong lines. The scale of the upper panel in Figure 2 has been changed in order to show the full intensity of the two lines.

A blow-up of the spectrum around 2508Å from positions 2 and 1 is shown in Figure 3. Besides the strong lines and , we have also marked the positions of two transitions and , which have connections to and , as

Figure: Spectra around the 2507 and 2509Å lines obtained at positions 2 (solid line) and 1 (crosses). The four lines discussed in the paper are marked

explained in the next section. The measured heliocentric vacuum wavelengths for and on HRS-2 are 2508.719 and 2507.170. The laboratory wavelengths are and , corresponding to a blueshift of 0.379 and 0.382Å (45 ). The same blueshift has been found for about 6--7 other high-excitation Fe II lines in the same spectral region.

Atomic Physics

The lines and are identified as two Fe II transitions labeled and , where and (Johansson 1978). From LS selection rules the intra-parental transition is supposed to give a strong line, whereas is an extraordinary intercombination line and an unlikely transition. However, the upper levels of and have about the same excitation energy, and are, therefore, strongly mixed (Johansson 1978). The level mixing results in two states having the same decay channels, for which the relative intensities of lines to a common level is constant and determined by the strength of the mixing. A similar case of level mixing in Fe II has recently been studied (Johansson et al. 1995). The upper levels in the transitions and can also decay to the J=9/2 level of through the transitions and , as shown in Figure 4. These transitions result in the satellite lines and , respectively (see Figure 4). To get the predicted stellar wavelengths of and , we have corrected the laboratory wavelengths for a

Figure: Simplified level diagram of Fe II illustrating the transitions and that correspond to the lines in Figure 3.

blueshift of 0.380Å and obtain Å and . The intensity ratio of two lines coming from the same upper level have been determined from a laboratory emission spectrum of Fe II. We get and . As the lines are close in wavelength, these intensity ratios reflect the ratios between the Einstein coefficients for spontaneous radiation, A. We should point out that is blended in the laboratory spectrum. Due to the particular level mixing in this case and the general complex structure of Fe II, theoretical transition probabilities have large uncertainties and have not been used.


The satellite lines and are very faint in the stellar spectrum, which result in much larger values of the intensity ratios and . At target position HRS-2, and , and at HRS-1, . is difficult to measure as is hidden in the noise. There are two possibilities to get an increase in the intensity ratios:
A) The strong lines are enhanced or
B) the satellite lines are reduced in intensity.

A) The strong lines can be enhanced due to blends with other lines or because of stimulated emission. Accidental blending of both and requires that other lines coincide in wavelength with and within a few mÅ. The blending lines also need to be selectively excited like and . We have not found any candidates for blending that fulfill these requirements. The stimulated emission requires a combination of a strong radiation field at these particular wavelengths as well as an inverted population.

B) The satellite lines can be reduced due to self absorption or absorption by other transitions coincident in wavelength. The two lower levels in the two pairs of transitions belong to the same term, , and they have similar decay channels down to in the infrared. Therefore, there is no build-up in the J=7/2 level that should cause self-absorption. The line lies close to a broad absorption line of Fe II at 2505.80Å. Considering the widths of and the wavelength difference between the two lines, we estimate that the Fe II absorption has a very little effect on . We find no absorption feature in the spectrum in the vicinity of and no potential candidate that could affect it by absorption.


From this discussion we draw the conclusion that and are enhanced due to stimulated emission. The low probability for stimulated emission at UV wavelengths is probably compensated by a strong inverted population. The upper levels are close in energy and can be populated through photoexcitation by the same pumping line. Different possibilities were discussed by Johansson & Hamann (1993) for different stars, but the photoexcitation by an accidental resonance in wavelength between two Fe II lines and HLy was supposed to dominate and HLy should be the major pump. However, the great intensity of the two fluorescence lines in Carinae initiated the search for another pump that could build up the population of the upper levels. This pump has been found in NI lines around 1130Å. A support for N I pumping is the presence of other strong nitrogen lines in the spectrum of Car due to a high nitrogen abundance in the star. A more detailed discussion of the pumping will be included in a forthcoming paper on these two laser lines in Eta Carinae.


Davidson, K., Ebbets, D., Weigelt, G., Humphreys, R.M., Hajian, A.R., Walborn, N.R., & Rosa, M. 1995, AJ, 109, 1784

Johansson, S. 1978, Physica Scripta, 18, 217

Johansson, S. & Jordan, C. 1984, MNRAS, 210, 239

Johansson, S. & Hamann, F.W. 1993, Physica Scripta, T47, 157

Johansson, S., Brage, T., Leckrone, D.S., Nave, G., & Wahlgren, G.M. 1995, ApJ 446, 361

Weigelt, G. & Ebersberger, J. 1986, A&A, 163, L5

next up previous contents index
Next: Spectroscopic Analysis of Up: StarsStellar Populations, Previous: HST Parallel Imaging