≡ Menu

Closing in on Centauri A and B with Astrometry

When it comes to finding planets around Centauri A and B, the method that most intrigues me is astrometry. At the recent Breakthrough Discuss sessions, Rachel Akeson (Caltech/IPAC) made the case for using the technique with data from the Atacama Large Millimeter Array (ALMA). My interest is piqued by the fact that so few of the more than 4300 known exoplanets have been discovered using astrometry, although astronomers were able in 2002 to characterize the previously known Gliese 876 using the method. Before that, numerous reported detections of planets around other stars, some going back to the 18th Century, have proven to be incorrect.

But we’re entering a new era. ESA’s Gaia mission, launched in 2013, is likely to return a large horde of planets using astrometry as it creates a three-dimensional map of star movement in the Milky Way. Dr. Akeson’s case for using ALMA to make detections on the ground is robust, despite the challenges the method presents. She points out that if we viewed the Solar System from a distance of 10 parsecs, Jupiter’s impact on the movement of our star would be 500 microarcseconds (µas), which works out to 1.4 X 10-7 degrees. Keep that figure in mind.

Astrometry is about measuring the movement of a star’s position on the sky, and it has been put to good use for a long time in identifying binary star systems. As the star’s position changes over time, the gravitational pull of an orbiting planet should likewise be revealed. This is something like the well established radial velocity technique for planet detection, but it operates within the plane of the sky instead of along the line of sight (radial velocity measures the Doppler shift in the star’s light as it is pulled alternately toward and then away from Earth along the line of sight). The beauty of astrometry is that it complements radial velocity by being best suited to planets with a wide separation from their star.

Image: Astrometry is the method that detects the motion of a star by making precise measurements of its position on the sky. This technique can also be used to identify planets around a star by measuring tiny changes in the star’s position as it wobbles around the center of mass of the planetary system. Credit: ESA.

Back to Gaia for a moment. It’s a mission that has given astrometry a dazzling upgrade, offering 10 to 20 µas performance for a large sample of observed stars, making the upcoming release of its exoplanet catalog an event that should vastly enlarge our statistical understanding of planetary systems. Remember that figure from Jupiter as seen from 10 parsecs — the movement of the Sun is 500 µas. Gaia should identify tens of thousands of exoplanets out to 1600 light years from the Sun, all of this by tracking minute changes in the stars’ position. But for the nearest star system, Gaia is far less effective because of Alpha Centauri’s brightness.

Enter ALMA. 5000 meters up on the Chajnantor Plateau in Chile’s Atacama desert, the site offers 66 antennas with baseline lengths of 150 meters to 16 kilometers. At the longest baseline, the resolution is 12 milliarcseconds (one thousandth of an arcsecond, abbreviated mas). We’ve looked often in these pages at ALMA studies of protoplanetary disks (Akeson refers to the famous image of HL Tau, shown below, with the roughly one million year old disk dividing into clearly visible rings and gaps).

With ALMA, the proximity of Centauri A and B becomes an advantage. The relative positions of the two stars can be measured, as opposed to having to compare them to reference stars in other fields, meaning that the precision of the measurement is greatly increased. To be sure, this remains a tough measurement — in 2015 observations, position uncertainties were in the range of 10 milliarcseconds, a figure that needs to be reduced to hundred of microarcseconds.

Can ALMA handle this level of precision? Akeson’s team requested three pairs of observations in 2018-2019, with two achieved and the third incomplete, a dataset that will be complemented with future observations to identify deviations in the orbital motions of these stars. The need for high resolution has meant that only a few of the configurations, using the longest baselines between ALMA antennas, can be used, which puts limits on the observing time available.

Image: ALMA image of the protoplanetary disc around HL Tauri – This is the sharpest image ever taken by ALMA — sharper than is routinely achieved in visible light with the NASA/ESA Hubble Space Telescope. It shows the protoplanetary disc surrounding the young star HL Tauri. These new ALMA observations reveal substructures within the disc that have never been seen before and even show the possible positions of planets forming in the dark patches within the system. Credit: ALMA (ESO/NAOJ/NRAO.

The highest resolution the team has yet achieved is in the August 2019 dataset. Here the absolute position of the stars was limited to 3 milliarcseconds (atmospheric noise is a factor here), and their relative separation showed uncertainties of 300 µas. Out of all this the plan is to use ALMA’s astrometrical data along with archival data (this includes radial velocity) to tighten up constraints on the orbits of both stars. The data allowed the team to improve our calibration of the stars’ masses by 2-3 percent. From a paper (citation below) describing the ALMA work, illustrating how data from numerous sources play into this effort:

The combination of historical measurements of α Cen A-B position angle and separation, more recent PRV [precision radial velocity] measurements from HARPS, and absolute astrometry from Hipparcos, and ALMA yields improved α Cen A and B orbital elements, system proper motion and parallax, and component masses.

But I also noticed, from the same paper, this intriguing bit about Proxima Centauri, whose status as bound or unbound to the central binary has remained problematic. The ALMA work seems to resolve the issue:

The significance of the gravitational link between AB and Proxima has therefore increased over the last three years from 4.4σ in (Kervella et al. 2017b), to 5.5σ in Kervella et al. (2019) to 8.3σ in the present work (< 10−15 false alarm probability). Proxima becomes a yet more valuable check on lower main sequence stellar modeling…

Leading to this:

Our accurate determination of the α Cen AB system parallax, RV [radial velocity], and proper motion, when compared to those values for Proxima Centauri, confirms that the three stars constitute a bound system.

All this is obviously painstaking work, and in the way of astronomical observation, it allowed the scientists to now use the tightened orbital information as a baseline for their continued search for deviations caused by planets. Akeson argues that the highest resolution configuration of the ALMA antennas will allow a resolution 2.5 times higher than the August 2019 data, which would take ALMA into the 100 microarcsecond range. As the paper notes:

Our estimates of ALMA’s relative astrometric precision suggest that we will ultimately be sensitive to planets of a few 10s of Earth mass in orbits from 1-3 AU, where stable orbits are thought to exist.

We may be closing in on a potential planet like Kevin Wagner’s tentatively identified C1. Here we can see the synergy between detection methods, with direct imaging (the subject of multiple proposals) exploring the same space.

But both these methods play off radial velocity studies as well. Remember that astrometric data is best suited to planets in wider orbits than radial velocity. While the latter continue and the ALMA work pushes ahead, we have imaging possibilities through the James Webb Space Telescope to add to the mix. The observational ring around Centauri A and B, then, is tightening as astrometry yields more refined orbital parameters for both stars. We can hope for exoplanet discovery here that can be confirmed in short order through multiple approaches.

For more on the ALMA work on Alpha Centauri, see Akeson et al., “Precision Millimeter Astrometry of the α Centauri AB System,” in process at the Astrophysical Journal (preprint).

{ 10 comments… add one }
  • Alex Tolley April 28, 2021, 11:50

    As the star’s position changes over time, the gravitational pull of an orbiting planet should likewise be revealed. This is something like the well established radial velocity technique for planet detection, but it operates within the plane of the sky instead of along the line of sight (radial velocity measures the Doppler shift in the star’s light as it is pulled alternately toward and then away from Earth along the line of sight).

    While astrometry is most complete in the plane view, shouldn’t it still work in the line of sight – with the star moving in 1 dimension (to and fro) rather than 2? Orientation of the orbital plane between the 2 extremes should allow for some interpolation of the true orbit[s] of the planet[s]. If the major planet has a very elliptical orbit, I assume astrometry, if accurate enough, can determine the orbit.

    Remember that astrometric data is best suited to planets in wider orbits than radial velocity.

    The wider the orbit, the greater the orbital period, making observations more difficult as they must be made over a longer period of time. Jupiter has a nearly 12-year orbital period, and a Jupiter-size planet at the orbit of Neptune has an orbital period of 165 years. So we still seem to be forced to look for low-hanging fruit observations – large planets around small stars with orbital periods short enough to be able to infer the orbits from the stellar movements over some fraction of an orbit.

  • Geoffrey Hillend April 28, 2021, 16:26

    Yes we would be able to see astrometric movement in a one dimensional plane, the star would move left to right instead of in a circle. That’s the correct way to think about it because a two dimensional flat plane is one dimensional seen edge on. A one dimensional astrometry might be more difficult to detect? A wider orbit would take longer to observe. We should be able to see the astrometry of Alpha Centauri A and B if there is any. Thinking about this more carefully, the stars astrometry caused by an exoplanet should be independent from any astrometry caused by the three body star system and it’s center of mass as it moves through space.

  • James T. Early April 28, 2021, 18:01

    Twenty years ago JPL had a Space Interferomety Mission (SIM) which would use two spacecraft to achieve high resolution which might detect planets by astrometry. I believe the mission was cancelled for budgetary reasons rather than technical problems. Would this mission be possible now with our lower launch costs? Was the resolution adequate for planet detection?

    • Paul Gilster April 28, 2021, 20:09

      Yes, the resolution of SIM would have made planet detection possible. One of the goals of the mission was to detect Earth-like planets around nearby stars. A pity this one never flew, but you’re right, budgetary issues shot it down after a great deal of money had already been spent.

    • Michael Fidler April 28, 2021, 21:02

      I’m wondering if surface based telescopes could be combined with space based to form a much larger quantum telescope interferometer?

      Quantum Astronomy Could Create Telescopes Hundreds of Kilometers Wide.


      While looking for this article I came across a very interesting article that could be seeing implementation very quickly instead of years from now. It may give angular resolutions of 10 μas in a few hours of observation of two bright stars.

      Quantum-Assisted Optical Interferometers: Instrument Requirements.

      It has been recently suggested that optical interferometers may not require a phase-stable optical link between the stations if instead sources of quantum-mechanically entangled pairs could be provided to them, enabling extra-long baselines and benefiting numerous topics in astrophysics and cosmology. We developed a new variation of this idea, proposing that photons from two different sources could be interfered at two decoupled stations, requiring only a slow classical connection between them. We show that this approach could allow high-precision measurements of the relative astrometry of the two sources, with a simple estimate giving angular resolution of 10 μas in a few hours’ observation of two bright stars. We also give requirements on the instrument for these observations, in particular on its temporal and spectral resolution. Finally, we discuss possible technologies for the instrument implementation and first proof-of-principle experiments.


      To top it all off we have now come to a Alice in Wonderland imaginary world in Quantum Mechanics. Schrödinger’s cat looks to be turning into the Cheshire Cat!

      Physicists Prove That the Imaginary Part of Quantum Mechanics Really Exists!

      “Now, we have theoretically and experimentally proved that there are quantum states that can only be distinguished when the calculations are performed with the indispensable participation of complex numbers,” explains Dr. Streltsov.
      Complex numbers are made up of two components, real and imaginary. They have the form a + bi, where the numbers a and b are real. The bi component is responsible for the specific features of complex numbers. The key role here is played by the imaginary number i, i.e. the square root of -1.”


      Lest we forget Maxwell’s equations for the wave nature of light is made up in part of complex numbers.

      Maybe Scalar waves really exist… }

      “Other gestures to the Cheshire Cat’s tropes of disappearance and mystique have been seen in scientific literature coming from the field of Physics. “The Cheshire Cat” is a phenomenon in quantum mechanics in which a particle and its property behave as if they are separated,[35] or when a particle separates from one of its physical properties.[36] To test this idea, researchers used an interferometer where neutron beams passed through silicon crystal. The crystal physically separated the neutrons and allowed them to go to two paths. Researchers reported “the system behaves as if the neutrons go through one beam path, while their magnetic moment travels along the other.”[36]”

      Cheshire Cat.


  • Spaceman April 28, 2021, 23:18


    You mentioned the “upcoming release” of Gaia’s exoplanet catalogue. Is this exoplanet catalogue going to be released in sometime in 2021 or later?

    • Paul Gilster April 29, 2021, 9:24

      According to ESA, Gaia Data Release 3 (Gaia DR3) is planned for the first half of 2022. The ‘full release for the nominal mission’ is still marked TBD — I assume this would contain the final catalog.

  • wdk April 29, 2021, 14:04

    Have been spending much time of late studying detection methods.
    Noticed that Gaia has 2 to 200 micro-arcsecon precision depending on
    luminosity of source, but owing to its survey techniques it re-visits
    particular stars about once every 70 days. That would not be bad for the
    case of Alpha Centauri A and B. Their period is about 80 years, but what
    you would be looking for in an HZ planet is something cycling on order of 400 days for A. Were the it a single star, this would be simpler.
    A single case, you would have an oscillation but about a dozen or so
    angles reading to determine the HZ planet’s orbital signal. Neptune masses are largely ruled out already what with the previous study of ground based imaging, but they still have their so-called Candidate 1 in the article within the dynamic limit stability limits. The more massive the planet the better, but for Earth and sun it is only a ratio of 330,000 to 1. So, if the planet swings out an arc-second….And then A and B
    are in binary motion… Wonder what the current separation of A and B is? Best if tried near true anomaly of about 180 degrees.

  • Geoffrey Hillend April 29, 2021, 16:17

    The Space Interferometry Mission used a nulling technique needed to use at least two telescopes because they combined the light waves of two separate images of the star or mixed the light waves together which cancel out the star light of the star so we can see only the exoplanet reflected starlight and get direct imaging. I liked the idea, but it was very expensive. https://en.wikipedia.org/wiki/Terrestrial_Planet_Finder
    It’s not made obsolete by a coronagraph since their sensitivity is getting better or can be improved.

  • Afterthought April 29, 2021, 23:53

    Very exciting! Thank you for your dedication and clarity.

Leave a Comment