The Palomar High-precision Astrometric Search for Exoplanet Systems


Description (from an SPIE newsroom article):


The Palomar High-precision Astrometric Search for Exoplanet Systems (PHASES) is a unique approach to finding Jupiter-like planets around other stars.  The development of multiple tools for studying new planetary systems is important because each method provides astronomers with different information to answer questions about their sizes, the diversity of environments in which they are found, and how they form.  As they learn more, astronomers get closer to answering questions about where else life might exist, how common the phenomenon of life might be, how the Earth came to be, and its role in the universe.

The majority of star systems are comprised of more than one star; the Sun is one of the exceptions, rather than the norm.  However, the question of whether planets are as abundant in these ``binary'' systems as they are around single stars remains.  There is reason to question how planets could form when an extra star is present, especially if the stars are close together.  Would the second star disrupt the planet-forming dust and gases before the planets are built?  Are the planets' orbits dynamically stable, or will the stars rapidly eject planets?

Two versatile configurations exist for which planets in binary systems can have stable orbits over long periods of time (Holman and Wiegert, 1999).  The popular movie ``Star Wars'' depicts one configuration:  the Skywalkers' home planet Tatooine orbits two stars that are close together.  These can be called ``Circumbinary'' planets (see our sister project “TATOOINE”).  Another configuration is where the planet orbits just one star, with the second star orbiting distantly.  The (now-disproven) ``Nemesis'' theory posed that the Sun had such a distant stellar companion, periodically perturbing distant comets, causing periods of excess bombardment on Earth and mass extinctions in the geologic record (Harrison 1977).  Though the theory has passed, the name remains to describe such configurations as ``Nemesis'' systems.  The PHASES program is an effort to find Nemesis systems.

The PHASES search uses a different method than most extrasolar planet searches:  astrometry.  When a planet orbits a star, Newton's Third Law of Motion---for every action there is an equal and opposite reaction---predicts that the star will move in response.  The two bodies each orbit the center-of-mass of the planetary system.  Because the star is much more massive than the planet, this center-of-mass is located much more closely to the star, and its motion is smaller, but not zero.  While astronomers cannot yet see the planets themselves, they can measure these small motions of the stars.  The very successful radial velocity (RV) method measures motions in a direction along the line connecting the Earth and target star.  Astrometry measures the side-to-side motions of the stars in the two perpendicular directions---the positions of stars on the sky.

From the nearest stars, the Sun would appear to move by an amount smaller than a tenth the resolution of the Hubble Space Telescope (HST).  PHASES uses multiple small telescopes combined into a system called an interferometer, which can operate with a resolving power greater than 10 times that of HST.

Astrometry requires reference stars to calibrate measurements of a target's positions.  Ground-based astrometry is made difficult by the atmosphere, as turbulent winds and temperature variations introduce apparent changes in stars' separations.  However, this effect is small when the stars are close together in the sky.  One good way to find references is to look for pairs of stars orbiting each other.  Thus, this is an excellent method for searching out planets in Nemesis systems.

The Palomar Testbed Interferometer was built with the ability to measure the separations of pairs of stars to extremely high precisions (Colavita et al. 1999).  It is a unique facility with the ability to use simultaneously two cameras, one that rapidly monitors changes in the atmosphere for calibration, and another that measures star positions with high accuracy.  This double measurement configuration---called ``phase-referencing'' (Lane and Colavita 2003, thus the PHASES acronym)---is necessary to measure the separations well enough to detect planets.

PHASES studies pairs of stars, looking for wobbles superimposed on the motion of the stars as they orbit each other.  These perturbations would be caused by objects orbiting either star (the Nemesis setup), but not both simultaneously (the Circumbinary case).

The PHASES measurements have successfully identified wobbles in superimposed on the orbits of several systems.  It turns out that the faint companions in those systems are not small enough to be planets, but instead are fainter, lower-mass stars.  This is an important auxiliary science product, as the orbital configurations tell astronomers about the environments in which multiple stars form.

Planets as small as twice Jupiter's mass can already be ruled out by the PHASES observations in several systems.  As the study continues and the technique is improved, better limits can be placed on planetary companions.  If PHASES finds that many giant planets exist in the target systems, this result will show that planet formation can occur regardless of disruptions by the presence of extra stars.  Otherwise, astronomers will have learned that the pool of stars potentially hosting planets is much smaller.

Astrometry is more sensitive to planets with long orbital periods, like those found in Earth's solar system.  Future astrometric planet searches, such as the Space Interferometry Mission (SIM), will be used to study systems more similar to our own.  The PHASES program is developing many of the tools and procedures that will be used for SIM.


Recent Results:


Detection Limits (see the full paper here or here):

The 13 Pegasi (HD 207652) Mass-Period companion phase space shows PHASES observations can rule out tertiary objects as small as two Jupiter masses.  Any companions with masses greater than the lines plotted are ruled out by PHASES observations (calculation assumes circular, face-on orbits).  The most massive planets (M = 13 Jupiter Masses) are ruled out for orbital periods ranging from 10-1000 days.



Triple Star Systems (see papers here and here or here and here):

We have used the astrometric measurements the mutual inclinations of hierarchical triple star systems.  In such systems, one of the two components of a relatively wide pair of stars is itself a much more narrowly separated binary.  We measure the positions of the stars of the wide system, and see a center-of-light of the narrow pair.  This is the same signature one would expect to see from an exoplanet, just larger (and in these cases, shorter period).

The apparent orbit of the kappa Pegasi system, which shows a six-day perturbation because one component is in fact two stars.  The center-of-light motion of that pair is measured.

Mutual inclination measurements are particularly valuable, as the dynamical relaxation process undergone by multiples after formation is expected to leave a statistical ``fingerprint'' in the distribution of inclinations (Sterzik & Tokovinin, 2002).  The resultant distribution in mutual inclinations depends upon the distributions of angular momenta in star forming regions.  These measurements thus provide a way by which one can investigate statistically properties of typical star-forming regions in which the local populations of stars originated.  The structure of star forming regions is as yet poorly understood, and observational constraints are important for setting initial conditions in simulations of star formation.  These simulations in turn predict the typical encounters a star might undergo while in the star forming region.  Such encounters can influence the evolution and stability of star systems and the disks surrounding young stars in which planets are thought to form.  Initial observational results suggest evidence for non-random structure in typical star forming regions, though more measurements are required as only a few (six) mutual inclination measurements have yet been made.

While bridging RV and visual observations for binary systems is a challenge already, doing so for triple star systems requires an extra order of magnitude in resolving power for the visual orbit, for reasons as follow.  For complete description of the triple system, RV orbits for both the wide pair and narrow subsystem are required.  It should be noted that the absolute spatial orientation of a binary system has two degenerate solutions without complementary RV data; thus the mutual inclination of a hierarchical triple is also degenerate in any system for which RV orbits cannot be determined for both subsystems.  Thus, the ``wide'' pair must have an orbital period (and corresponding separation) as short as the two-component binaries previously studied by combined RV and interferometric methods.

            To study the mutual inclinations of the orbits in such a system, both RV and visual orbits for both systems in a triple are required.  If only one component of the orbit is available for either system, one is left with a degeneracy in the longitude of the ascending node of that system, which propagates as a degeneracy in the mutual inclination.  Thus, determinations of the mutual inclinations of the two orbits in hierarchical triple stellar systems are rare, with previously only six unambiguous determinations available in the literature (see Muterspaugh et al 2005 for a complete list). The PHASES team has made two of the six mutual inclination measurements that have been reported.

Cumulative distribution of the six systems for which unambiguous mutual inclinations have been measured (set 6U).  Also shown is the distribution from Sterzik & Tokovinin 2002, who included both possible mutual inclination angles for each of 22 systems for which only ambiguous mutual inclination determination was possible (set ST).  By including both possible angles, distribution ST may “diluted” and partially biased towards randomly oriented orbits.  The theoretical distribution for random orientations is also shown (set Random).  A two-sided Kolmogorov-Smirnov (K-S) test shows 6U and Random do not agree (at 96% probabilty); however, better statistics are required to determine the correct underlying distribution.  The proposed study will double the statistics of the 6U set by observing another six systems.  Figure is from Muterspaugh et al. 2005.


You might also want to visit some other team members’ websites:

Dr. Ben Lane
Dr. Bernard Burke
Dr. Shri Kulkarni