X-ray Observations of Eta Carinae

This quicktime movie displays visually the variation of the X-ray emission from the central source of Eta Car, based on our RXTE observations from January 1997 through September 1999, using an image of the core from a CHANDRA/ACIS image obtained in Sep 1999. The false-color image shows regions of bright X-ray intensity as reddish-white, and regions of fainter intensity as bluish-black. At the start, the source shows brief periodic flashes of X-rays. In mid 1997, the brightness of the central source increases dramatically and the variability of the source becomes much more extreme: then suddenly, in December 1997, the X-ray emission almost entirely disappears. It reappears in about 3 months, and regains (almost) its previous brightness and the X-ray flaring resumes. For comparison see the radio movie created by Stephen White and Bob Duncan.

  1. Introduction
  2. Detailed Monitoring of Eta Car with the Rossi X-ray Timing Explorer: X-raying the Beating Heart of the Homunculus
  3. Long-Term X-ray Variability
  4. HRI Observations - Spatially resolved Variability
  5. ASCA Observations of the X-ray Spectrum
  6. Upcoming X-ray Observations
  7. For More Information


Extremely massive stars are key astronomical objects, as they play a role in chemical enrichment and galactic evolution. They mark the end of their stellar lives as supernovae explosions in which a single supernova can equal the entire radiant output of a galaxy of a billion stars. Recently the extreme members of this class have been suggested by Paczynski (1998) to produce the "hypernovae" which might explain the bursts of gamma radiation which have been an astronomical mystery for 30 years. The energy emitted in a "hypernova" is astounding; perhaps the equivalent to the radiant energy output of an entire universe of galaxies. Such extraordinary explosions require stellar precursors of unusually large mass, and so should be relatively rare. Alarmingly, the Milky Way possesses one possible member of this putative class, the massive, luminous, and relatively nearby star, Eta Carinae. Eta Carinae is both an extremely massive star and an extremely unstable one. It's notorious for erupting in the mid-19th century, and is surrounded by the ejecta from this eruption. The Hubble Space Telescope has an impressive image of Eta Car and this ejecta .

The entire field is beautiful at X-ray energies, as you can see from this ROSAT High Resolution Imager image. For comparison, here's the same field in the optical.

Recent observations in the optical and near IR by Augusto Damineli and co-workers suggest that certain emission lines vary periodically with a cycle of 5.52 years. The exact cause of this periodicity is not really known, but the most likely explanation is that the period reflects the orbital motion of one star around another star: Eta Car may be two stars, not one!.

If Eta Car really is a binary system, then our understanding of the evolution of the star and the nature of its instabilities will need to be radically revised. So this is a crucial question: Is the star really a binary, or could single-star models be found to explain the spectral variations?

X-ray observations can help resolve this important issue.

Eta Car is a strong X-ray source. X-rays are produced as the ejecta expands into the circumstellar medium near the star at speeds of 100-1000 km/s. There is also a mysterious point-like source of hard emission centered on the star itself. Recent X-ray observations with the ROSAT PSPC and HRI show that the X-ray emission from this point-like source varies by a factor of 3 on timescales of months (Corcoran et al., 1995, ApJ, 445, L121, or see the press release).

The ROSAT PSPC provides energy-filtered X-ray imaging of Eta Car, too. Notice how the X-ray emission is extended at lower energies, but more point-like at higher energies.

Detailed Monitoring of Eta Car with the Rossi X-ray Timing Explorer: X-raying the Beating Heart of the Homunculus.

The ROSAT observations show that the X-ray source varies, but give little information about the timescale of the variability, since sampling is so poor. Since early 1996 we (M. Corcoran, J. Swank, K. Ishibashi, K. Davidson, R. Petre, and others, see Nature 390, 587) have been monitoring the X-ray flux from Eta Car using RXTE. RXTE provides good time sampling (roughly one observation every two weeks with limited intervals of daily observations) of the hard (2-10 keV) X-ray flux from Eta Car, since Eta Car happily lies in the RXTE continuous viewing zone. This sampling is the key to understanding how the hard X-ray flux from Eta Car varies, and for correlating the changes in the X-ray emission with those changes observed at other energies.

Our RXTE analysis of the 2-10 keV X-ray spectrum (Ishibashi et al., ApJ 1999, in press; the postscript version of the paper, and figures, and the tabular data is available.) is the first such detailed analysis for this star (or for any massive star for that matter). For the first time we are able to document the variety of changes in Eta Car's X-ray emission properties:

So what's going on? If Eta Car really is a binary, the hard X-ray emission could naturally arise at the shocked interface created by the collision of the wind from the primary component with the wind (or surface) of the secondary. Such "Colliding Wind" emission has been observed in a number of other massive binary systems. In this "colliding wind" scenario, the periodic variation in the X-ray flux could be produced by periodic variations in the strength of the wind-collision shock, and/or by periodic changes in the amount of absorbing material between the shock and us. Such changes can easily be accomplished in the type of eccentric binary model suggested by the emission line variations. However, numerical models predict that at periastron passage of the system (i.e. the time when the 2 stars are at closest approach) the intrinsic X-ray flux in the 2-10 keV band should be at a maximum. But our RXTE (and recent ASCA) observations show that the intrinsic flux is at a minimum. This could suggest that the binary model is wrong, or that the simplyfying assumptions that make the numerical modelling tractable are in fact too simple>. Our best guess is that the assumptions are too simple, and that the star really is a binary - the X-ray "low state" looks too much like an occultation!

The other important result from the RXTE data is the mysterious, 85-day "flaring". We still don't have a firm idea on the cause of this period, but there are at least 3 likely possible causes:

  1. It may be the orbital period of a close binary system. Is Eta Car really a triple system?
  2. The period could represent the stellar rotation period. Perhaps streams of fast-moving material from the star collide with localized slow-moving gas once per rotation period, similar to Dermot Mullan's ``corotating interaction regions'' scenario. Or perhaps the second star interacts with the rotating streams to produce large amounts of shocked gas every 85 days.
  3. The X-ray period could represent a stellar pulsation period. A photospheric pulsation could presumably modulate the density in the stellar wind, which could manifest itself as a periodic variation in the volume emission measure (and resultant flux) of the X-ray zone.

ROSAT HRI Observations of Eta Car: Variations of the Core and the X-ray Nebula

Here's a mosaic of a WFPC1 image of Eta Car with an overlay of smoothed contours from the 1994 ROSAT HRI. Note how extended the X-ray "shell" is compared to nebulosity visible in the optical. The ROSAT HRI, while providing no spectral information, does provide the best spatially-resolved imaging of the X-ray emission from Eta Car currently available. In addition, we can compare HRI images taken at different epochs in order to look for X-ray spatial variations. This contour map shows the exposure-corrected, aspect corrected HRI image of Eta Car in Jun 1992 (red contours) compared to the exposure and aspect corrected image in Jun 1994. The spatial scale in the image is 2 arcsec/pixel, or roughly 0.02 pc/pixel at the distance of Eta Car (roughly 2600 pc). The point source about 1 arcmin to the NE of Eta Car is the O3 star HDE 303308 which serves as a reference for the exposure and aspect correction. If we subtract the earlier image from the latter image we immediately see the brightening of the core of Eta Car, which is unresolved even at the spatial resolution of the HRI. We also see evidence of 2 rings of emission in the outer part of the nebula, one to the SE and one to the NW of the core. These rings are more apparent if we compare the difference image and a map of the signal-to-noise of the difference image . The image on the left is the difference image, while the image on the right is the S/N ratio of the difference, generated by taking the square root of the square of the difference map divided by the square of the error of the difference map. Brighter features in the S/N map indicate features with higher statistical significance. The brightening of the core is detected with a high statistical significance, < 99%. Though the "rings" surrounding the core are detected at a lower significance, ~ 68%, the similarity of the morphology of the X-ray "rings" to the "bubbles" seen in HST images of the homunculus suggest that the X-ray rings are in fact real (although the X-ray "rings" are larger than the "bubbles" in the Hubble images). The X-ray "rings" seen in the difference maps could represent the expansion of the X-ray nebula in the interval 1992-1994, though the velocity of the material would need to be near 10,000 km/s. Such a velocity is higher than any yet measured for material near Eta Car. Here's an overlay of the smoothed difference contours on the WFPC1 mosaic.

The X-ray Spectrum of Eta Car

We've recently obtained a deep (100 ksec) exposure of Eta Car with the ASCA SIS and GIS instruments. This is probably the best spectrum of Eta Car yet obtained. The ASCA observation took place on Jul 29 1996. This is close in time to the "maximum" of the XTE lightcurve which occurred on Jul 22 1996.

You can download the submitted version of the paper, entitled "The ASCA X-ray Spectrum of Eta Car" as a gzipped postscript file.

Our spectral modeling is used to derive the "unfolded" SIS Eta Car spectrum. Note how similar the 1-10 keV spectrum is to the X-ray spectrum of WR 140, a colliding wind system. The WR 140 spectrum shown here is from an ASCA SIS PV observation, when the system was near periastron passage. Both WR 140 and Eta Car show highly absorbed, very high-temperature thermal components. Eta Car also has considerable emission at energies below 1 keV, but this mainly originates in the extended emission associated with the homunuculus.

The region of the spectrum near the Fe K-alpha line at 6.7 keV is complex. Here's a plot of the photon spectrum in the Fe K-alpha region along with the best-fit RS thermal equilibrium model discussed above. For an appreciation of the actual resolution of the SIS, the observed SIS0 count spectrum and RS model is also shown. The strongest lines in this region of the spectrum are:

The 7.5 keV RS model (with slightly reduced Fe abundances) does a good job of reproducing the observed line strengths, and implies that the emitting plasma is in thermal equilibrium.


The campaign of observations (covering radio through UV and X-ray to Gamma-ray observations) of the upcoming X-ray eclipse/"shell event" is available at http://lheawww.gsfc.nasa.gov/users/corcoran/eta_car/2003.5/

For More Information

Page Author: Dr. Mike Corcoran.

Last modified December 1, 1999

Key Resource