Why Stellar Atmospheres are Important to Astronomy
With the exception of the Sun, to the human eye stars look like points of light on the sky. In fact, they are huge, intensely hot balls of gas; most are composed primarily of hydrogen, with a sprinkling of heavier elements. The stellar light that we see comes from the outer layers of the star's atmosphere. The chemical composition and temperature in these layers determines the star's spectrum, that is, the amount of light it emits as a function of color. Our knowledge of stars is built by using theories relate this light to stellar characteristics, such as stellar mass, chemical content, gravity, age, and temperature. Until now, the most detailed measurements to test the consistency of these atmospheric models came from only one star: the Sun. Teamwork, a network of small telescopes, and quickly-scheduled access to one of the world's largest telescopes has now yielded the first full set of such spectral measurements --- for a cool, giant star near the crowded central regions of the Milky Way, more than one billion times more distant than our own Sun.

How Microlensing can Resolve Stellar Atmospheres
Microlensing, nature's gravitational telescope, can be used to study stellar light and how it varies across the face of much more distant stars, in principle allowing stellar atmospheric models to be tested against a broader range of stellar types. About half of all stars come in pairs. When one of these binary star systems passes nearly directly between the observer and a background star, its gravitational field alters the path of light from the background star, magnifying its apparent brightness. As the lens and source move, the brightness as a function of time (light curve) changes. At certain positions called caustics (see figure below), the two lenses act cooperatively, producing a large magnification.

Left: Two lensing stars of equal mass (two black dots) create a caustic structure --- regions where the combined effect of both lenses is enormous (solid black line). Right: The light curves of sources passing over this caustic structure exhibit sharp peaks in brightness at the moment of crossing. The exact light curve shape depends on the source path (colored lines in the left-hand figure) and the mass and separation of the lenses. (Adapted from Paczynski 1996.)

When a background star crosses a caustic, a large peak is produced in its light curve, and the portions of the star directly over the caustic are magnified more than the rest of the star. Microlensing caustics can thus be used to magnify and resolve the the face of a very distant star.

The Unusual Microlensing Event EROS BLG-2000-5

International collaborations of astronomers watch millions of stars nightly to search for microlensing events. On 5 May 2000, Jean-Francois Glicenstein of the EROS group alerted astronomers to an apparently normal microlensing event in the direction of the Galactic Bulge. Since the event, named EROS BLG-2000-5, was rising in brightness, the PLANET team began to monitor it intensively, as part of its regular operations to search for brief anomalies that might be indicative of lensing Jupiter-mass planets. The MPS team was also watching this event, and on 8 June 2000 they issued an alert that the source star was undergoing an unusual brightening. PLANET observers immediately turned their full attention to this event, monitoring it nearly continously from five different observing sites for the weeks to come.

It was soon clear that EROS BLG-2000-5 was due to a binary lens. The sudden jump in the light curve on June 8 was tracked by PLANET. These data, together with real-time modeling, allowed the team to issue an electronic alert that included precise timing (to within 3 minutes) and characterization of the first caustic crossing. A second crossing could then be expected with a few days or weeks.

While the light curve of the event indicated that the source was in the ``caustic trough'' between the first and second crossings, the PLANET team applied for Director's Discretionary Time on the Very Large Telescope (VLT) at the European Southern Observatory (ESO) on Paranal in Chile to monitor the spectrum of the event during the second caustic crossing. Meanwhile, the team continued to monitor EROS BLG-2000-5 from their network of small, southern telescopes and to issue (second, third) public information about when the second crossing might be expected.

The Results from the Very Large Telescope

Within a few days, the request for VLT time was granted. Spectra would be taken with the FORS spectrograph on the first unit telescope, named Antu. The plan for the observations would be executed remotely, with service astronomers at ESO responsible for manipulating the telescope.

Spectra taken on 25 June 2000 when the event was highly magnified, but had not yet crossed the caustic for the second time, indicated a large flux at red wavelengths (9000 angstroms) compared to blue wavelengths (4000 angstroms). Absorption line strengths could be measured, such as the small dip at 6563 angstroms (the H-alpha line) caused by hydrogen atoms in the outer atmosphere absorbing light produced deeper in the star.

The FORS spectra indicated clearly that EROS-BLG-2000-5 was a cool, K giant star.

The FORS instrument takes images as well as spectra. In only 0.25 seconds of exposure time, the 8-meter diameter mirror of the VLT aided FORS in taking the image shown at right (2 arcminutes on a side). EROS-BLG-2000-5 is circled in red, and was already highly magnified at this point in time. The long opening slit of FORS was aligned with other stars in the field. In this way, changes in the spectrum of the EROS-BLG-2000-5 source star could be compared to the spectrum of other stars (which shouldn't be varying!), providing a useful check in the experiment.

The team continued to monitor the event. Modeling by PLANET team members at Ohio State University was somewhat uncertain, but clearly indicated that the second crossing would be long, lasting several days rather than the usual 10-20 hours.

So the observing plan was altered so that spectra could be taken on four nights (see figure at left) during the caustic crossing, when the light curve would be brightening, and then dimming dramatically as the source finally exited the caustic. Due to limb darkening, the cooler, outer regions of a star appear dimmer and redder than the center. Over the four nights of the crossing, the source would be moving across the caustic: first the cooler leading edge (limb) of the star would cross, then the hotter middle, and finally the trailing limb. Changes in the brightness and color across the face of a few microlensed stars had already been measured during caustic crossings. The PLANET team now expected to see changes in the strength of the absorption lines, which are determined by the temperatures in the stellar atmosphere, as well. This effect should be visible in the FORS spectra.

On the nights beginning 4, 5, 6 and 7 July 2000, the ESO service observers executed the last steps of the PLANET team plan; close contact was maintained via email and the Internet. Every night while the source star was nearly directly overhead in Chilean skies, the FORS spectrograph measured the spectral content of the light coming from EROS-BLG-2000-5. When the results were reduced, and analyzed by two different PLANET team members in two different ways, the results were the same: the strength of the H-alpha absorption line in EROS-BLG-2000-5 changed systematically throughout the second caustic crossing, the line strength of the control stars varied only slightly or not at all.

About 40 red spectra were taken in all. When the strength (or equivalent width, EW) of the H-alpha line was plotted as a function of time, the strength became stronger as the center of the star passed over the caustic, and then much weaker as the trailing limb exited the caustic. The measurements are shown as black dots in the figure at right; the red dots are the average strength of the absorption line on each of the five nights, one before the second crossing on 25 June, and one each for the four nights of the caustic crossing.

Stellar atmosphere specialist Peter Hauschildt of the University of Georgia prepared special models of K giant atmospheres that could be compared to these results. These models include the effects of hundreds of thousands of possible chemical transitions in the atmosphere of the cool giant star, each of which can affect, to a larger or smaller degree, the observed stellar spectrum. When PLANET team members simulated the effects of a simple caustic crossing passing over these theoretical stars, they obtained the solid lines shown in the plot above. The models (and thus the assumed chemistry of cool K giant star atmospheres) clearly match these observations in their general form.

What's Next?

But do the VLT observations of EROS BLG-2000-005 match the current stellar atmosphere models in more detail? Answering this question will require studying all regions of the FORS spectra, and developing a better model for how the source star crossed the caustic structure.

After the second crossing, the PLANET team continued to monitor EROS BLG-2000-005 and noticed that the light curve began to brighten again almost immediately. On 11 July 2000, PLANET issued a an electronic alert that the steep rise in brightness might indicate that the source star was approaching or crossing a sharp turn or "cusp" in the caustic structure. Indeed the light curve of the event looks very much like the yellow model curve shown in the first plot above. Notice that the third peak in brightness is caused by the source passing near the cusp on its exit. The simple models used for the simulations shown as solid lines in the figure above must therefore be modified.

PLANET continues to model its light curve data for to EROS BLG-2000-005 in order to improve understanding of the caustic crossing trajectory; it also continues to monitor the light curve into the 2001 observing season. These steps, together with new spectral observations of more caustic crossings, should lead to a more precise assessment of the validity of our understanding of the chemistry in the atmosphere of distant stars.

For More Information See the ESO Press Release.

To read the technical details about these measurements, download our technical paper at astro-ph/0011380, now published in the 2001 Astrophysical Journal Letters, 550, L173.