July/August 2000 Table of Contents
though we had good reason to believe planets orbit other stars like
the Sun, just five years did we uncover the first evidence to support
this belief. Now the discovery of such "extrasolar" worlds threatens
to become routine.
decades of false starts, the search for extrasolar planets is on
firm footing and charging full steam ahead. To those of us in the
business of finding planets, it is hard to believe that just five
years ago, Michel Mayor and Didier Queloz discovered the first extrasolar
planet. The heated debate over the reality of the existence of that
planet and the subsequent struggle to interpret such a strange world
still echoes in our daily reflections, even as we begin to publish
our new planet detections in bunches of six. The novelty of finding
an extrasolar planet may have worn off. But now, the real science
did it take so long to find planets?
don't emit their own light. Only stars with their tremendous gravitational
pressure have the power to ignite the thermonuclear reactions that
give off energy as light. In principle, planets could be seen by
reflected starlight. However, even a high-albedo, Jupiter-like planet
is down by a factor of more than a million in brightness relative
to its host star. With our largest telescopes and best detectors,
planets are absolutely invisible next to the bright spotlights of
the stars they orbit. Invisible, but not undetectable.
ago, astronomers reasoned that the presence of extrasolar planets
could be inferred by the gravitational tug they exert on the star:
as a planet orbits a star, it drags the star around a common center
of gravity. Even though the planets are invisible, astronomers knew
there should be a smoking gun. The star should wobble with a telltale
motion—a signature of the unseen planet(s).
the first attempts to look for these wobbles, astronomers measured
the positions of the stars with painstaking care, using the most
distant stars as their unmoving reference points (a technique called
astrometry). Confident that if they worked hard enough they could
beat this problem down with lots of data, these first planet hunters
spent fifty years on their quest, but ultimately failed. Given the
technology of the time (photographic plates, primitive computers,
and the inability to correct for atmospheric blurring), this technique
for measuring tiny wobbles in the position of the stars was just
the same time, another technique was being developed to study stars:
spectroscopy. Instead of measuring the changing positions of stars,
spectroscopy allows us to measure the speed of a star. Starlight
is sent through a spectrograph where it spreads out like a rainbow.
The absorption lines in the rainbow, or spectrum, of the star, tell
us what the star is made out of: typically, lots of hydrogen and
helium sprinkled with other elements from the periodic table. And
the shift of these absorption lines, due to the Doppler effect,
tells us how fast the star is moving along our line of sight (i.e.,
the radial velocity).
stars exhibit some Doppler shift because all stars are moving in
the gravitational field of our Galaxy. But the stars that move toward
us, then away from us, then toward us again—wobbling with periodic
motion—are gravitationally bound to another object. The magnitude
of this changing velocity, together with the period of the motion,
reveals the mass of an invisible companion to a visible star. Our
Sun moves with a speed of about 12 meters/sec due to the most massive
planet in our Solar System, Jupiter. The periodicity of the Sun's
motion is the same as the orbital period of Jupiter: one cycle takes
about twelve years.
the early 1980s, astronomers were measuring velocities with a precision
of about 1 km/sec. Then, high-resolution spectrometers, CCD detectors,
faster computers, and astronomers with the single-minded goal of
pushing this new technology as far as it could go, ushered in a
new era in high-precision Doppler measurements. The barometer of
velocity precision began to plummet: 500 meters/sec... 300 meters/sec...
100 meters/sec... 50 meters/sec... down to a mere 20 meters/sec
in 1995 when the first extrasolar planet was discovered. It is no
accident that the planet around 51 Pegasi (with a velocity amplitude
of 50 meters/sec) was discovered exactly when the Doppler technique
was finally cold enough and quiet enough to permit the measurement
of a minuscule periodic shift of the star's absorption lines.
first extrasolar planet was a surprise in every way. In an orbit
twenty times closer to its star than Earth is to the Sun, this planet
zipped around the star 51 Peg in just five days, and weighed in
with about half the mass of Jupiter. There was a brief moment of
stunned silence in the astronomical community. Then, the questions
started flying. The temperature would be sizzling-hot so close to
this were a gas-giant planet like our Jupiter, wouldn't the atmosphere
could such a planet form so close to a star?
it didn't form there, how did it migrate into its present position?
it migrated inward, how did the migration stop to park the planet
in its present orbit?
answers came more slowly. The gravity of a massive planet would
be sufficient to hold on to its atmosphere, even in such a hot environment.
The planet probably didn't form in its present position; more likely,
it migrated inward when its primordial orbit was destabilized by
gravitational interactions with other planetesimals or with the
material in the circumstellar disk. The theorists, who had been
constructing planet formation models, based on our Solar System,
struggled to interpret this new piece of evidence. If we learned
anything from 51 Peg, it was to expect the unexpected.
the beginning of 1999, the velocity precision had dropped to a few
meters per second and almost twenty extrasolar planets had been
discovered with the Doppler technique. All of these planets had
Jupiter-like masses and all were relatively close to their host
stars, presumably having migrated in from a more distant origin.
Just when discoveries started to seem ho-hum in the planet-hunting
business, there were three big breakthroughs: the discovery of a
system with three planets, an observation of a planet transit, and
the detection of two, Saturn-mass planets.
A Multiple-Planet System
planet, similar to that discovered orbiting 51 Peg, had already
been found around the star Upsilon Andromedae back in 1996. At that
time, planet-hunters Geoff Marcy (University of California, Berkeley)
and Paul Butler (Carnegie Institution of Washington), who reported
this discovery, realized that there was more to this system than
just one planet. In addition to a four-day period derived from the
star's radial velocity data, they observed a longer-period trend
that suggested the presence of an additional planet. From Kepler's
laws of planetary motion, the farther a planet is from its host
star, the longer the planet's orbital period. So, a waiting game
ensued as everyone watched to see when the putative second planet
would complete one, full, orbital period.
March 1999, the trend in the velocities for Upsilon And showed that
the second planet had executed more than one, full, 3.5-year orbit.
However, when the mathematical model was constructed to describe
these two planets, a third planet's orbital signature was discovered
in the data. Snuggled between the known inner planet and the anticipated
outer planet was an unexpected interloper—a planet with an orbital
period of 242 days.
discovery of a triple-planet system was an amazing technical feat,
but the truly stunning breakthrough came when theorists demonstrated
that this system was both dynamically stable and dynamically full.
If another planet were magically dropped into this system, the orbits
of the detected planets would become chaotic and at least one planet
would be lost—scattered by the gravitational tugs of the other planets.
Indeed, this is the same situation in the Solar System. Again, the
questions came. Do planets form as single, precious jewels in a
protoplanetary disk, or are they born with many siblings? How does
nature create dynamically full but stable planetary systems?
answers are still being worked out on computers throughout the world,
but one hypothesis is that enormous numbers of planetesimals form
in the protoplanetary disk. As the fledgling planets accrete disk
material and grow, a gravitational tug-of-war develops. In the end,
only the planets in gravitationally stable niches survive. If correct,
this suggests that stars commonly form with several planets. Astronomers
are now engaged in another waiting game, watching a few more stars
with planets in known short-period orbits that, like Upsilon And,
appear to have one or more additional planets in longer-period orbits.
A planetary transit. Here
is an artist's conception of an extrasolar planet passing in front
of its parent star. Such a transit of the star HD 209458 was observed
by two research groups last year. Illustration courtesy of NASA.
A Planet Observed in Transit
curiosity is the nature of the scientist. Even with the discovery
of so many extrasolar planets, astronomers yearned for more information.
The Doppler technique only provides the mass of the planet. In fact,
because the inclination of a planet's orbit about a star is hardly
ever known well, only a lower limit for the planet's mass may be
determined. In a statistical sense, the true mass is expected to
be usually within a factor of two of the Doppler-determined mass.
But, it was unclear whether these extrasolar planets were megalithic
Earths or gas giants like Jupiter. The gas-giant scenario was generally
favored because astronomers reasoned that as soon as a rocky core
formed in a protoplanetary disk, it would gravitationally sweep
up an enormous gaseous atmosphere.
was only a matter of time until the next breakthrough settled this
question. Astronomers had anticipated that as the number of extrasolar
planets increased, one would eventually be found in an edge-on orbit.
As a planet in an edge-on orbit passes in front of the host star,
it dims the starlight by an amount that is proportional to the cross-sectional
area of the planet. With this measured diminution, it is then straightforward
to estimate the density of the planet and determine whether the
planet is primarily gaseous or rocky. The Doppler technique predicted
exactly when the planet would pass in front of the star but could
not determine whether the orbit was edge-on.
made careful measurements of the brightnesses of the stars with
very close planets (where the probability of catching a transit
was highest) before, during, and after the predicted transit times.
In the autumn of 1999, two groups independently observed a 2% dimming
in the brightness of the star HD 209458 at precisely the predicted
transit time. Their observations determined an absolute mass for
the planet and verified that this is, indeed, a gas-giant planet,
silencing the few remaining skeptics who had argued that the extrasolar
planets might be brown dwarfs or even low-mass stars with extreme
Finding a Saturn-mass world.
From the Doppler-shifted absorption lines in the spectrum of star
HD 46375, astronomers were able to calculate the star's radial
velocity as a function of time...
biggest planets are the easiest to find because they exert a strong
tug on the host star. But, despite this ease of detection, only
a few planets have been found that are more than five times the
mass of Jupiter. None are more than ten times the mass of Jupiter.
This observed upper limit in the mass of the detected planets tells
us that there is a real, physical limit to how big planets can grow
in the planetary nursery.
is also a lower limit to the Doppler-detectable planet mass. It
is difficult to detect planets with less than one half the mass
of Jupiter. Recently, this limit has been pushed down somewhat,
but it is not likely that Earth-mass planets will be detected with
the Doppler technique. This is because lower-mass planets exert
smaller tugs on the star with correspondingly smaller Doppler shifts
that are harder to measure.
March 2000, using one of the giant Keck telescopes in Hawai'i, the
low-mass threshold, stubbornly held since 1995, was broken. Two
planets with masses less than Saturn were found orbiting the stars
HD 46375 and HD 16141. It is possible that the Doppler technique
will eventually find Neptune-mass planets if they exist in close
orbits. The technical feat involved in detecting such low-mass companions
is impressive. But as usual, it is the broader interpretation that
seems most exciting. Simply put, there are more little planets than
big ones. Even though the Doppler technique may never enable us
to detect planets with Earth-like masses, there is a clear trend
that suggests that low-mass planets may be predominant in nature.
This is a theme that rings commonly in nature: there are more low-mass
stars than high-mass stars, more grains of sand on the beach than
...And with this information they
esitmate HD 46375's planetary companion to have a mass of about
80 percent that of Saturn. Both illustrations this page courtesy
of Geoff Marcy (Univ. of CA - Berkeley)
appears to be a strong correlation between the metal content—to
astronomers, every element heavier than helium is considered a metal—of
the host star and the presence of a gas-giant planet. This has some
first stars in the Galaxy were composed only of hydrogen and helium.
It took a few stellar generations to enrich the interstellar medium
with heavier elements that could condense into grains and form the
cores of planets. If the first generation of stars could not form
planets, when was the interstellar enrichment sufficient for planet
formation to turn on? If the most ancient stars did not form planets,
would this reduce the likelihood of ancient civilizations in the
Galaxy? It remains unclear whether this correlation with metallicity
will extend to terrestrial-mass planets.
of May 2000, 42 extrasolar planets had been discovered by Doppler
surveys. These alien worlds seem different from the planets orbiting
the Sun. For example, in the Solar System, most of the planets travel
in circular orbits, which provide a reasonably stable temperature
on the surface of the planet. This was probably an important factor
in the evolution of life on Earth. However, all of the extrasolar
planets at distances beyond two-tenths of an Earth-Sun distance
from their host stars reside in elongated, elliptical orbits. These
planets spend most of the time in the cold, distant reaches of their
orbits and then quickly plunge in close around their stars. Such
huge swings in the ambient temperature must provide an extra challenge
for biological evolution.
orbital periods of the extrasolar planets range from just three
days to a few years. A true analog of our Solar System would have
a Jupiter in a twelve-year orbit. It has been suggested that Jupiter
plays a protective role for biological evolution on Earth. A big
planet like Jupiter may serve as a gravitational sink, swallowing
up many of the incoming comets (à la Comet Shoemaker/Levy
9) that might otherwise pummel Earth and cause mass extinctions.
To find a Jupiter in a twelve-year orbit, Doppler surveys need a
twelve-year baseline of high precision data. Most programs are about
Doppler surveys are only the first reconnaissance for planet hunting.
Many clever ideas are now on the table that will build on current
discoveries to search for lower-mass, Earth-like planets. Most of
these ideas require the stillness of space to achieve the highest
Earth's atmosphere, transit telescopes will stare at crowded star
fields, running a time series of high-precision, photometric observations
and analyses to look for a slight decrease in the brightness of
an individual star as a planet passes in front of it. This work
is already being tested on the ground, and space-born, transit
telescopes, which will provide the higher precision needed to
find terrestrial planets, are likely to launch as early as 2002.
is a fascinating technique that is already being tested to find
planets. This technique relies on a background star to provide
a source of light. Then, foreground stars act as gravitational
lenses when they serendipitously pass in front of the source star
and cause the background starlight to brighten and bend through
an annular region called the Einstein radius. If there is a planet
near the Einstein radius, it acts as a secondary lens and a bright,
narrow peak will be superimposed on the first light curve. Like
transit observations, microlensing will achieve the higher precision
needed to detect Earth-mass planets from space. As the most populous
stars in the Galaxy, low-mass, M-class dwarfs will be common lensing
stars, so microlensing observations may provide important statistical
information on the rate of occurrence of planets around these
Space Interferometry Mission (see "Interferometry
in Space") is being designed to find planets with masses that
are just a few times that of Earth around nearby stars. With a
probable launch date of 2006, SIM is also being designed to serve
as a testbed for the technique of nulling interferometry that
will be used by the Terrestrial Planet Finder, planned for launch
in about 2011. Nulling interferometry with TPF will allow direct
imaging and spectroscopy of Earth-like planets around nearby stars.
only seen the tip of the iceberg in the discoveries of extrasolar
planets. The next generation of planet hunting should be the most
productive. Whatever we find, we should expect the unexpected.
FISCHER is a postdoctoral fellow at the University of
California, Berkeley, working on the Marcy-Butler planet-search
project for the past two years. She monitors the radial velocities
of some 300 stars from Lick Observatory. She also has a special
interest in education and public outreach and is a volunteer with
ASP's Project ASTRO. She can be reached via email at email@example.com.
atmosphere of Earth hampers ground-based searches for planets around
Sun-like stars. While Doppler searches have been successful in finding
gas-giant planets similar to Saturn or Jupiter, the undisputed prize,
rocky Earth-like planets, will be easiest to find from space. NASA
is expected to launch the Space Interferometry Mission (SIM) within
the next decade to find smaller planets, more like the one we call
will consist of two light collectors separated by a ten-meter baseline
and, as the name implies, will employ interferometry to measure
tiny wobbles in the positions of nearby stars. This high-tech version
of good, old-fashioned astrometry demands that the baseline between
the light collectors be measured with a precision of about 50 picometers
(5´ 10-11 meter), which is about the diameter of a hydrogen atom.
Such an engineering feat will undoubtedly have spin-off applications
in other fields.
An artist's rendering of NASA's planned SIM spacecraft.
Illustration courtesy of JPL.
an eye to the future, NASA is also considering a design for a post-SIM
mission, the Terrestrial Planet Finder (TPF), which will use an
array of light collectors to carry out nulling interferometry—a
technique that is expected to provide the first real glimpse of
planets orbiting nearby stars. Normally in interferometry, it is
desirable to fine-tune the pathlength of the light from the collectors
so that the re-combined light constructively interferes and the
star appears bright. With nulling interferometry, however, the goal
is to arrange the optical elements so that the combined light destructively
interferes, canceling out the light from the central star and allowing
us to see the much fainter planets orbiting the stars. For the first
time, we may have a picture of a pale blue dot.
and TPF are engineering marvels that will not be as cheap as some
other planned missions that look for changes in brightness from
transiting planets or for microlensing events. But they uniquely
offer us the undeniable appeal of spying on those stars that are
our closest neighbors.