Eta Carinae's "Spectroscopic Events"


Once every 5.5 years, the spectrum of Eta Car changes for a few weeks or months, while its X-rays exhibit a maximum, a crash, and then a partial recovery. Today we know that this happens near periastron, the time when a companion star approaches closest to the primary star. Many strange phenomena then occur together, most of them poorly understood, controversial, and never seen in any other known astronomical object. They have helped to indicate Eta's longer-term change of state.

(These phenomena are sometimes called X-ray minima or periastron events.)

Observational History

The record was sparse before 1990. Eta Car had erupted twice, and was known to have a peculiar spectrum, but sometimes many years passed between useful observations []. Its spectrum usually showed hundreds of low-excitation emission lines plus a few high-excitation features. The latter, mainly due to helium and Fe III, required EUV photons from a very hot source with T > 30,000 K. Everything else in the spectrum seemed much cooler.

In 1948, 1964, and 1981, the high-excitation features nearly vanished, temporarily. Analyzing the 1981 case, Zanella, Wolf, and Stahl deduced that Eta had abruptly ejected some material, temporarily quashing the EUV source []. They also remarked that 1948-1965-1981 would fit a periodicity of 16.5 years. They were right, but the true period is 1/3 of 16.5 years, because earlier observers had failed to notice events in 1953, 1959, 1970, and 1975! Someone should have realized this when another occurred in 1986-1987, but no one did. The next event in 1992 received little attention -- until, several years later, Damineli used it to explicitly identify the 5.5-year period []. Later estimates gave P = 2023 ± 1 days = 5.54 years, see Appendix in [Meh11b].

For a while almost everyone interpreted the periodicity as a stellar pulsation and/or thermal instability. Then, in 1997, Corcoran and also Damineli proposed that Eta has a companion star with a 5.5-year orbit. This possibility had earlier been acknowledged but not strongly advocated [,]. Davidson then remarked that an O-type secondary star can account for the puzzling EUV radiation [,]. (This wasn't true in Damineli's model with a B-type companion star; but a hotter type seemed more likely anyway, for evolutionary reasons.) Later developments supported the binary hypothesis.

Spectroscopic events at t = 1998.0, 2003.5, 2009.1, and 2014.6 were observed much better, as the Hubble Space Telescope (HST) [[revolutionized this topic.]] The Space Telescope Imaging Spectrograph (STIS) could observe the central stellar wind without much contamination by Eta's nearby ejecta. It obtained a few spectra at t = 1998.0, intensive data in 2003, and a smaller amount in 2014. During the 2009 event, when STIS was not operational, observations were made with a larger ground-based telescope than those used for previous events.

What happens during a spectroscopic event?

For the complicated history of this topic in 2000-2018, see various authors' review articles in [] and later results in [,]. Here we sketch the best-bet scenario, though many details remain controversial.


Figure 1: colliding-winds-b.png
Figure 2: orbit-01-60d-b-soft.png
Figure 3: X-ray-lc-2009-b.png

XXXX memo: remember to mention correlation with 1843 event xxxx