The Search for Planets Around Proxima Centauri

The α Centauri system, which is currently the closest known star system to the Sun (and also known by the ancient name, Rigil Kentaurus), contains three stars. At the heart of the system are a pair of Sun-like stars 4.37 light years from us which are locked in a moderately eccentric 80-year orbit with a mean separation of 23.4 AU – roughly the same as the distance between the Sun and Uranus in our solar system. About 15,000 AU (or about a quarter of a light year) from this pair of stars is α Centauri C better known as Proxima Centauri because, at a distance of 4.24 light years, it is the closest known star to our solar system (and 0.17 light years closer than α Centauri AB).

In 2012, the discovery of a planet orbiting the smaller of the Sun-like pair of stars, α Centauri B, was announced. This still unconfirmed planet was the subject of the first essay in this series about the search for planets in this nearby system (see “The Search For Planets Around Alpha Centauri”). A second essay covered what previous searches for planets orbiting either α Centauri A and B had found and what future searches could find (see “The Search For Planets Around Alpha Centauri – II”). In this essay, the results of the search for planets orbiting the third star of this system, Proxima Centauri, are summarized.

Background

Unlike α Centauri A and B which are about the same size and mass as the Sun, Proxima Centauri is a very small type M5V red dwarf star that displays all the characteristics of a magnetically active star including occasional intense flares. With an estimated mass of only 0.11 times that of the Sun and with 0.0017 times its luminosity, it is among the smallest main sequence stars known with a mass only about a third again more than the least massive normal star theoretically possible. Because of its low luminosity, Proxima Centauri has a V magnitude of only 11.0 and requires a small telescope to spot visually even though it is the closest known star. Despite its dimness, it still has a higher apparent brightness than most known red dwarfs and, as a result, it has been a target of study by those with an interest in these small stars.

Because of its close proximity to α Centauri AB and its shared apparent motion, it has been generally believed since its discovery in 1915 by Scottish astronomer Robert Innes that Proxima Centauri is bound to the pair of larger, Sun-like stars at the center of this system. But studies over the past two decades on the relative motions of these stars tends to suggest that Proxima Centauri might not be gravitationally bound to α Centauri AB. Instead these stars might be independent members of a stellar moving group that happen to be close to each other at the present time.

A view of Proxima Centauri acquired by the Hubble Space Telescope. (ESA/Hubble/NASA)

A study by Jeremy Wertheimer and Gregory Laughlin (University of California Observatories/Lick Observatory) published in 2006 using the best available information on the positions and motions of these stars shows that Proxima Centauri, with a velocity relative to α Centauri AB of 530±140 meters per second, is right on the cusp of being bound to α Centauri AB. The best measurements suggest that Proxima Centauri is near the apastron or most distant point in a highly eccentric orbit around α Centauri AB with an orbital period measured in hundreds of thousands to maybe even millions of years. However, given the measurement uncertainties, the possibility can not be excluded that Proxima Centauri might be unbound and simply passing by α Centauri AB. Only future improvements in the accuracy of the absolute radial velocity of Proxima Centauri will definitively resolve the issue.

Astrometric Method

One of the first sensitive methods to be employed in the search for unseen companions (i.e. planets and brown dwarfs) orbiting Proxima Centauri has been astrometry: the measurement of the position of a star on the celestial sphere. By precisely measuring the position of a star over time, not only can its parallax and proper motion be determined, but the reflex motion of that star as it is orbited by a smaller body can be characterized as well. While documented efforts of using astrometry to search for unseen companions orbiting Proxima Centauri date back at least a half a century, the most sensitive search to date using this technique was performed in the 1990s by a team led by G. Fritz Benedict (McDonald Observatory) using the Hubble Space Telescope (HST).

Benedict et al. used the HST to observe a handful of nearby red dwarf stars looking for the telltale wobble caused by orbiting unseen companions. These stars were ideal for this work since the low mass and relatively close proximity of these stars maximized the sensitivity for the astrometric detection of planets. But instead of using one of the imaging instruments to perform their measurements, Benedict et al. used one of HST’s Fine Guidance Sensors (FGS). Hubble’s three FGS units are white light Koester prism interferometers used to keep the telescope locked in position while other instruments collect data. They perform this task by precisely measuring the position of selected guide stars during an observing session. While not an imaging instrument, the FGS is capable of performing astrometry with a single measurement accuracy of about ±3 milliarc seconds (mas). Even better accuracy is possible by averaging the results of many individual measurements. With this capability in mind, FGS #3 was fitted with a broadband visible light filter specifically designed to make astrometric measurements.

One of Hubble’s Fine Guidance Sensors being exchanged by astronaut Gregory J. Harbaugh during Servicing Mission 2 in February 1997. (NASA/ESA)

In the penultimate summary of the work on Proxima Centauri by Benedict et al. published in 1999, the team presented the analysis of astrometric data from 152 shorter exposures of FGS #3 acquired between March 1992 and October 1997 and 15 longer exposures from July 1995 to July 1996. The data over this 5.6-year time span were analyzed to measure the parallax of Proxima Centauri to a precision of ±0.3 mas and its proper motion to ±0.1 mas/year – about an order of magnitude more accurately than was possible using data from ESA’s Hipparcos astrometry mission which is still considered the “gold standard” of astrometry even two decades after it completed its mission. After these effects were taken into account, any small perturbations in the motion of Proxima Centauri could reveal the presence of an unseen companion.

Subtle instrumental artifacts and the irregular spacing of the data with large gaps in time complicated the analysis of the measurements. There was also a “false alarm” in 1993 when the team thought it had detected a Jupiter-mass planet in an 80-day orbit that turned out to be the related to activity on the surface of Proxima Centauri modulated by its 84-day period of rotation (see “The Hubble Space Telescope and the Search for Faint Extrasolar Companions”). Once all of these issues had been addressed, the team had failed to detect any planets orbiting Proxima Centauri with orbital periods in the 50 to 2000-day range (corresponding to an orbital distance range of about 0.13 to 1.49 AU). A detailed statistical analysis and a series of Monte Carlo simulations where artificial signals were added to the data to gauge their sensitivity showed that perturbations with an amplitude of 1 mas would have been detected at a 1% false-positive level with a 5% miss rate.

Plot showing the detection limits (at a 95% confidence level) as a function of companion period for Proxima Centauri based on the astrometric work by Benedict et al. using the Hubble Space Telescope. Click on image to enlarge. (Adapted from Benedict et al.)

This null result from the work of Benedict et al. excludes the presence of unseen companions more massive than Jupiter (or 1 M_J) with orbital periods between 50 and 1000 days to a 95% certainty level or better. For orbital periods greater than 400 days, the detection limit is less than the mass of Saturn or 0.3 M_J. These detection limits for planets with orbital periods in the 50 to 2000-day range still represent the best results using astrometry.

While measurements made by Benedict et al. using HST are the most accurate astrometric results to date, they only cover orbital periods out to about 2000 days owing to the limited time span of the available data. It turns out that less accurate measurements obtained over a longer period of time can prove to be useful in detecting large planets in wider orbits. This was the approach used by a team led by John C. Lurie (University of Washington – Seattle) in their astrometric analysis of the motions of  Proxima Centauri that was recently published.

A view of the four SMARTS consortium telescopes at the Cerro Tololo Inter-American Observatory in Chile. (CTIO/SMARTS)

Lurie et al. observed Proxima Centauri as part of the RECONS (Research Consortium On Nearby Stars) long-term astrometric and photometric survey program using the CTIO/SMARTS (Cerro Tololo Inter-American Observatory/Small and Moderate Aperture Research Telescope System) 0.9-meter telescope in Chile. For their analysis, they used data from 205 observations acquired over 14 seasons from mid-2000 to early-2013 with an average single-measurement uncertainty of ±4.83 mas. The analysis of their data resulted in an absolute parallax with an uncertainty of ±1.0 mas and a proper motion with an uncertainty of ±0.6 mas/year – only a factor of three and six, respectively, worse than the measurement uncertainties of the HST results from the 1990s despite the use of a more modest ground-based telescope. Just like the earlier HST result, Lurie et al. found no significant perturbations in their data of the motion of Proxima Centauri that would indicate the presence of a planet.

The detection thresholds at a 90% confidence level for five of the stars studied by Lurie et al.. Proxima Centauri is the bottom curve in the group. Click on image to enlarge. (Lurie et al.)

A detailed statistical analysis and a series of Monte Carlo simulations where artificial signals were added to the data to gauge their sensitivity showed that Proxima Centauri does not possess a brown dwarf of a mass greater than 12 M_J with an orbital period in the 2 to 12-year range to a 99% confidence level. These results also exclude the presence of planets with masses greater than 2 M_J in orbits with periods of 2 to 5 years – significantly higher upper limits than the results from Benedict et al. for this same range. But because of the longer time span of the data analyzed by Lurie et al., they were able to set upper limits for planets in much longer period orbits. To a 90% confidence level, they found that there were no planets detected with masses greater than about 1.0 M_J and 0.6 M_J with periods of 5 and 12.6 years, respectively. The combination of results from Benedict et al. and Lurie et al. effectively eliminates the possibility that Proxima Centauri has any companions with a mass comparable to or greater than Jupiter in orbits with periods ranging from 0.14 to 12.6 years (corresponding to orbital radii of about 0.13 and 2.6 AU, respectively).

Radial Velocity Method

The other means of measuring the reflex motion of a star orbited by an unseen companion is precision radial velocity measurements. In fact, before NASA’s Kepler mission, the majority of extrasolar planets were discovered using this technique. Since this method is more sensitive to planets in smaller orbits, it nicely complements the astrometric method which is more sensitive to planets in larger orbits. One of the drawbacks of the radial velocity method not shared by astrometry is that only the minimum mass of a planet or M_Psini (where i is the inclination of the orbit to the plane of the sky) can be calculated. Since the inclination can not be determined from radial velocity measurements alone, it must be found by other means or, failing that, only the probability that the actual mass of a planet falls within some range of interest can be calculated.

View of the European Southern Observatory’s Very Large Telescope array at Cerro Paranal in Chile. One its telescopes, UT2, has been used to measure the radial velocity of Proxima Centauri in search of planets. (J.L. Dauvergne & G. Hüdepohl/ESO)

While published results from radial velocity searches for unseen companions of Proxima Centauri go back over a decade and a half, the most accurate radial velocity results to date are based on the work of Michael Endl (McDonald Observatory) and Martin Kürster (Max Planck Institute) published in 2008. Endl and Kürster were involved in an ongoing program to survey 40 M-dwarf stars using the UVES spectrograph on the 8.2-meter telescope designated UT2 – one of four such telescopes that make up the VLT or Very Large Telescope at the European Southern Observatory on Cerro Paranal in Chile. For their analysis of Proxima Centauri, they used 229 individual spectra obtained over the course of 76 nights between March 2000 (only a year after first light for UT2) and March 2007. The differential radial velocity measurements they obtained had an RMS scatter of 3.11 meters/second – a factor of 17 better than earlier published radial velocity results.

As with the astrometric results, no signal indicating the presence of an orbiting extrasolar planet was detected. Based on an analysis of the data and the ability of their data analysis software to detect simulated planet signals, it was found that there were no planets detected orbiting Proxima Centauri with M_Psini values of two times that of the Earth (or 2 M_E) with an orbital period of 2 days and up to 20 M_E for planets with orbital periods of nearly 2000 days (corresponding to orbital radii of 0.015 to 1.49 AU, respectively). Planets with M_Psini values in the 2 to 3 M_E range were excluded from the conservatively defined habitable zone of Proxima Centauri which spans orbital periods of 3.6 to 13.8 days and distances of 0.022 to 0.054 AU. Because of the high probability of spurious signals with periods of about one year, orbital periods from 300 to 400 days were excluded from the analysis by Endl and Kürster (although we know from the astrometric work of Benedict et al. that the presence of planets more massive than Saturn have been excluded from this zone).

Minimum mass upper limits as a function of period for planets in circular orbits around Proxima Centauri based on work by Endl and Kürster. All test signals with amplitudes corresponding to masses on and above the solid line were recovered with a >99% significance. The dashed, dashed-dotted and long-dashed lines show the mass range where they successfully recovered 75, 50 and 25% of the test signals. The shaded area labelled “HZ” displays the approximate location of the conservatively defined habitable zone. The “1-year” window was excluded from their analysis because of spurious signals. Click on image to enlarge. (Endl and Kürster).

While we do not know how these M_Psini values translate into actual mass values owing to the unknown value of i, we can calculate the probability that a randomly oriented orbit has a value of i less than a certain amount. A 95% detection probability corresponds to i of less than 18.2°. This inclination yields an actual mass value that is 3.20 times the M_Psini. Using this scaling factor, we can state that there is a 95% chance that there are no planets more massive than 6 M_E with orbital periods of two days and no planets more massive than 64 M_E (equivalent to 0.2 M_J) with a period approaching 2000 days – in rough agreement with the upper limits based on the astrometric results of Benedict et al..

The radial velocity survey results effectively eliminate the possible presence any Saturn-size planets (95 M_E or 0.3 M_J) with orbital periods less than about 2000 days or Neptune-size planets (17 M_E or 0.05 M_J) with orbital periods less than about 40 days to a 95% confidence level or better. Inside of the habitable zone, there are no planets with masses greater than the 6 to 10 M_E at the same confidence level. This lack of any planets detected inside the habitable zone is actually good news. While the existence of planets about as small as mini-Neptunes has been effectively excluded, still smaller planets can be present and have escaped detection – smaller planets that have a much higher likelihood of having a rocky composition and therefore being potentially habitable.

Summary

The best planet survey results to date have failed to detect any planets or brown dwarfs orbiting Proxima Centauri. These results effectively eliminate the possibility of any Jupiter-size or larger planets with orbital periods ranging from as short as two days to in excess of 12 years to a confidence level of 90% to 95% or better. The results also effectively eliminate the possible presence any Saturn-size planets with orbital periods less than about 2000 days or Neptune-size planets with orbital periods less than about 40 days to a 95% confidence level or better. No planets with masses greater than 6 to 10 M_E are likely to be present in the habitable zone of Proxima Centauri leaving the possibility that rocky, potentially habitable sub-Earth-size to super-Earth-size planets can exist in this zone and still escape detection by published searches to date with a high probability.

While there have been some who have been alarmed by the lack of any planet detections to date, it is not unexpected given what we have learned about the planetary systems of other M-dwarf stars. A recent statistical analysis of the Kepler database for M-dwarf stars performed by Courtney Dressing and David Charbonneau (Harvard-Smithsonian Center for Astrophysics) has shown that planets with radii greater than about 2.5 times that of the Earth (corresponding to 0.64 times the radius of Neptune) and orbital periods less than 200 days are rare. Planets larger than Neptune are exceptionally rare in M-dwarf systems. And since the “typical” M-dwarf in the analysis by Dressing and Charbonneau is over four times more massive than the diminutive Proxima Centauri (with the corresponding planets also tending to be larger), the lack of any planetary detections to date is even less surprising (see “Occurrence of Potentially Habitable Planets Around Red Dwarfs”).

At this stage, the best thing to do for those anxious about the discovery of planets orbiting Proxima Centauri is just to sit tight and wait. Radial velocity data with ever-improving quality continues to be gathered for this star. While astronomers’ ability to correct their data for stellar activity will ultimately limit what can be achieved using this method, refinements in instruments and procedures should improve the detection threshold. ESA’s Gaia mission, which is currently gathering data from the L2 Lagrange point of the Earth-Sun system, promises to generate astrometric measurements for Proxima Centauri well over an order of magnitude more precise than any obtained to date. Combining Gaia’s data with earlier measurements promises a significant improvement in the detection of planets in long-period orbits as well especially if Gaia manages to operate longer than its nominal five-year mission. As the planet detection capabilities for the Proxima Centauri system continue to improve, it is only a matter of time before planets are found orbiting our nearest stellar neighbor.

Follow Drew Ex Machina on Facebook.

Related Reading

“The Search For Planets Around Alpha Centauri”, Drew Ex Machina, August 11, 2014 [Post]

“The Search For Planets Around Alpha Centauri – II ”, Drew Ex Machina, September 25, 2014 [Post]

“Happy Anniversary Alpha Centauri Bb?”, Centauri Dreams, October 16, 2014 [Post]

“Occurrence of Potentially Habitable Planets Around Red Dwarfs”, Drew Ex Machina, January 12, 2015 [Post]

“The Hubble Space Telescope and the Search for Faint Extrasolar Companions”,  SETIQuest, Volume 3, Number 2, pp. 1-9, Second Quarter 1997 [Article]

“Publication Watch: A Possible Companion to Proxima Centauri”, SETIQuest, Volume 4, Number 2, p. 19, Second Quarter 1998 [Article]

“Publication Watch: Wide Field Planetary Camera 2 Observation of Proxima Centauri: No Evidence of the Possible Substellar Companion”, SETIQuest, Volume 4, Number 3, p. 20, Third Quarter 1998 [Article]

General References

G. Fritz Benedict et al., “Interferometric Astrometry of Proxima Centauri and Barnard’s Star Using HUBBLE SPACE TELESCOPE Fine Guidance Sensor 3: Detection Limits for Substellar Companions”, The Astronomical Journal, Vol. 118, No. 2, pp. 1086-1100, August 1999

Courtney D. Dressing and David Charbonneau, “The Occurrence of Potentially Habitable Planets Orbiting M Dwarfs Estimated from the Full Kepler Dataset and an Empirical Measurement of the Detection Sensitivity”, arVix 1501.01623 (submitted to The Astrophysical Journal), January 7, 2015 [Preprint]

M. Endl and M. Kürster, “Toward detection of terrestrial planets in the habitable zone of our closest neighbor: Proxima Centauri”, Astronomy and Astrophysics, Vol. 488, No. 3, pp.1149-1153, September 2008

John C. Lurie et al., “The Solar Neighborhood. XXXIV. A Search for Planets Orbiting Nearby M Dwarfs Using Astrometry”, The Astronomical Journal, Vol. 148, No. 5, Article id. 91, November 2014

Jeremy G. Wertheimer and Gregory Laughlin, “Are Proxima and α Centauri Gravitationally Bound?”, The Astronomical Journal, Vol. 132, No. 5, pp. 1995-1997, November 2006