Before the discovery of the first exoplanets, astronomers thought they understood the architecture of planetary systems. Using our own solar system as a guide, it was expected that small rocky planets would orbit close in to their suns with orbital periods on the order of a few months to a few years with larger Jupiter-like planets found in more distant orbits beyond the “ice line” with orbital periods on the order of a decade and more. This view was upended in October 1995 with the discovery of 51 Pegasi b – a Jupiter-size exoplanet in a tight four-day orbit now characterized as a “hot Jupiter” (see “The Discovery of Extrasolar Planets”). In the two decades since this discovery, extrasolar giant planets have been found orbiting main sequence stars over a large range of distances making it apparent that the arrangement of planet types in our solar system is only one example along a continuum of possible planetary architectures (see “How Typical is Our Solar System?”).
Since this time, astronomers have continued to search for Jupiter-like exoplanets in distant Jupiter-like orbits to determine how common the arrangement of planets in our solar system actually is. The task is made all the more difficult by the years or even decades of astrometric or precision radial velocity (RV) measurements needed to detect the subtle reflex motion caused by such distantly orbiting bodies and characterize their properties. The interpretation of precision RV measurements, which have been used to find exoplanets of all types orbiting nearby stars, is complicated by long-term cycles in stellar activity such as the 11-year sunspot cycle observed on the Sun – a length of time comparable to the orbital period of a true Jupiter analog.
Despite the challenges, distantly orbiting “cold Jupiters” have been identified with more detections yet to come as the length of time covered by various databases increases and detection methods improve. On March 21, 2018, Fabo Feng, Mikko Tuomi and Hugh R.A. Jones (University of Hertfordshire, UK) submitted their analysis of a quarter century’s worth of precision RV data of the nearby Sun-like star Epsilon Indi A (also written as ε Indi A) for publication in Monthly Notices of the Royal Astronomical Society. Their work provides additional evidence for a distantly orbiting Jovian exoplanet in this system designated “ε Indi Ab” which is the closest exoplanet of this type currently known.
Background
The star ε Indi is a K5V type with a V magnitude of 4.83 located in the southern constellation of Indus – The Indian. Also known by various other catalog designations (e.g. HD 209100 and GJ 845), the common name of ε Indi was established in 1603 by the German celestial cartographer, Johann Bayer (1572-1625), in his famous star catalog called Uranometria. In 1847, German astronomer Henrich Louis d’Arrest (1822-1875) noted that the position of ε Indi had changed when comparing data in star catalogs going back to 1750 indicating that this star has a substantial proper motion. Current measurements indicate a proper motion of 4.7 arc seconds per year – the third fastest moving naked eye star in the sky after the much dimmer Groombridge 1830 and 61 Cygni.
The southern constellation Indus (The Indian) and its surroundings, as drawn in the famous Uranographia published 1801 of German astronomer Johann Elert Bode. (ESO/Uranographia)
Since high proper motion is suggestive of a nearby star, the parallax of ε Indi was measured for the first time in 1883 by Scottish astronomer David Gill (1843-1914) and American William L. Elkin (1854-1933) while working together at the Royal Observatory at the Cape of Good Hope in South Africa. Their distance measurement of about 15±2 light years was not too far off from today’s best measurement of 11.81 light years making ε Indi among the closest stars known. Because of its closeness, ε Indi was included in the first edition of the Gliese Catalogue of Nearby Stars in 1957 earning it the designation of GJ 845 after the creator of the catalog, German astronomer Wilhelm Gliese (1915-1993), and his long time collaborator on later editions, Hartmut Jahreiß.
3D map of stellar systems in the solar neighborhood within a radius of 12.5 light-years. The Sun is at the center and the ε Indi system lies near the bottom. (ESO/R.D. Scholz et al.)
The best current measurements of the properties of ε Indi give it an effective surface temperature of 4,630 K, a radius 0.73 times that of the Sun, a luminosity of 0.22 times and a mass estimated to be 0.76 times. With a metallicity of about 87% that of the Sun, ε Indi has a slightly lower concentration of elements heavier than helium compared to our system’s central star. Long term photometric observations suggest the period of rotation for ε Indi is about 35 days. Based on a comparison of the star’s properties with models of stellar evolution, it appears that ε Indi is about 1.4 billion years old, although some work suggests it may be older. Overall, ε Indi seems to be a slightly cooler, smaller, dimmer but younger version of the Sun.
ε Indi A (the bright star at far right) and its brown dwarf companion ε Indi B (circled). The upper image comes from one of the SuperCOSMOS Sky Surveys (SSS) I-band photographic plates. The lower image is a near-infrared image (Ks-band, 2.1 µm) from the Two Micron All Sky Survey (2MASS). Both images cover roughly 7 by 5 arcmin. Click on image to enlarge. (ESO/R.D.Scholz et al.)
It was long thought that ε Indi was a single star similar to the Sun-like nearby stars ε Eridani and τ Ceti (see “Habitable Planet Reality Check: Tau Ceti”) making it a target for the search of a solar system analog which might include a Earth-size planets orbiting in its habitable zone (HZ) as well as a frequent SETI target. But in January 2003, Scholz et al. announced the discovery of a brown dwarf moving with ε Indi at a projected distance of 1,459 AU. With the orbital period probably on the order of a couple of tens to a couple of hundreds of thousands of years, it will be some time before any orbital motion would become apparent. Subsequent high-resolution infrared imagery acquired by McCaughrean et al. later that year using the NAOS/CONICA adaptive optics imaging system on the European Southern Observatory’s (ESO’s) 8.2-meter VLT in Cerro Paranal, Chile revealed that ε Indi B, as it was now called, was actually a pair of brown dwarfs with an apparent separation of just 0.732 arc seconds making this the closest brown dwarf binary then known (surpassed in 2013 by the discovery of Luhman 16 just 6.5 light years away). The discovery also made ε Indi the third closest triple star system known after α Centauri (see the α Centauri page) and Luyten 789-6.
Artist conception of the ε Indi B binary brown dwarf. Click on image to enlarge. (Gemini Observatory)
The pair of brown dwarfs distantly orbiting the Sun-like ε Indi A have been designated ε Indi Ba and Bb. The brighter component, ε Indi Ba, is a type T1 brown dwarf with a temperature in the 1,352 to 1,385 K range and an estimated mass of 47±10 times that of Jupiter or MJ. The dimmer ε Indi Bb is spectral type T6 with a cooler temperature in the 976 to 1,011 K range and an estimated mass of 28±7 MJ. With a projected separation of about 2.65 AU and an orbital period expected to be around 15 years, detailed astrometric measurements promise to provide accurate dynamical masses for this pair of substellar objects. Combined with their relative closeness and brightness, ε Indi Ba/Bb promise to provide much information on the properties and early evolution of brown dwarfs
A Jupiter-like Gas Giant
Because of the Sun-like nature of ε Indi and its relative closeness, it has been a high-priority target for southern hemisphere exoplanet surveys. The best published survey results as of the turn of the century were by Endl et al. in 2002. This international collaboration observed ε Indi as part of the CES Survey of 37 late-type, nearby star which ran from November 1992 to April 1998. Using the Coude Echelle Spectrometer and Long Camera (CES LC) on ESO’s 1.4-meter Coude Auxiliary Telescope (CAT) in La Silla, Chile, Endl et al. obtained a long series of precision RV measurements in the hopes of detecting exoplanets orbiting their nearby targets.
ESO’s 1.4-meter Coude Auxiliary Telescope (CAT) was used between November 1992 and May 1998 to gather precision RV measurements of ε Indi and other stars as part of the CES Survey. (ESO)
In the case of ε Indi (which was still thought to be a single star at the time), Endl et al. secured 73 RV measurements over a span of 5.2 years. While the team failed to find any statistically significant periodic signals in the data indicative of an exoplanet in a short-period orbit, they did note a long term trend in the RV amounting to an increase of 21 meters per second during their survey. The measurements the increase in RV was 4.4 meters per second per year with an RMS scatter of 11.6 meters per second in their data set. This trend could be explained by the presence of a companion in an orbit with a period greater than 20 years. Endl et al. suggested that the trend could be caused by an exoplanet with a semimajor axis of ~6.5 AU (equivalent to an orbital period of ~19 years) with an MPsini of 1.6 MJ. Because the inclination, i, of an exoplanet’s orbit to the plane of the sky can not be determined directly from RV measurements alone, only the minimum mass or MPsini can be determined. The actual mass would almost surely be higher. When the ε Indi B brown dwarf binary was found in 2003, it was quickly realized that they could not be responsible for the observed RV trend bolstering the case for a Jupiter-like gas giant distantly orbiting ε Indi A.
The RV measurements from the original CES Survey showing the clear increase over time. Click on image to enlarge. (Endl et al.)
As the ESO team continued to gather precision RV data to provide a definitive detection, others employed ever improving direct imaging techniques to spot this possible exoplanet. The most sensitive such search to date was made by Janson et al. who made observations of ε Indi A on the nights of July 3, October 31 and November 2, 2008 using ESO’s NAOS/VLT operating at the infrared wavelength of about 4 μm where a young, massive gas giant would be expected to be comparatively bright. Janson et al. failed to find anything near ε Indi A in their images placing strict upper limits on the mass of the suspected exoplanet. Combined with the trend derived from the latest RV measurements (which was now pegged at 2.6 meters per second per year) and considering the the projected separation could be smaller than the semimajor axis at the time of their observations, Janson et al. concluded that the possible companion of ε Indi A must have a mass in the 5 to 20 MJ range (depending on the assumptions made about the age of the system), a semimajor axis somewhere in the 10 to 20 AU range and an inclination, i, of greater than 20°.
The upper limits of the masses found by Janson et al. for three different presumed ages for any exoplanet orbiting ε Indi A as a function of projected separation (which can be smaller than the actual semimajor axis). The solid line are the values suggested by the RV trend with i=0° while the the dashed line represent i=60°. Click on image to enlarge. (Janson et al.)
By 2013, the ESO team performing RV surveys searching for exoplanets had a substantially expanded data set to characterize the possible substellar companion of ε Indi A. In the best published results until recently, Zechmeister et al. presented a new analysis for ε Indi A based on not only the original 78 CES Survey RV measurements but new ones with significantly improved measurement accuracy. Between November 1999 and May 2006, Zechmeister et al. secured 54 new measurements taken over 5.8 years using CES with the VLC (Very Long Camera) now on ESO’s 3.6-meter telescope at La Silla. From November 2003 to December 2009, an additional 457 measurements were acquired over a period of 5.9 years using the then-new HARPS (High Accuracy Radial velocity Planet Search) spectrograph also on the 3.6-meter telescope which has been used to spot so many nearby exoplanets in recent years (see the HARPS page).
The ESO 3.6m telescope equipped first with CES VLC and then the HARPS spectrograph was used to gather precision RV measurements of ε Indi A. (ESO/H.H.Heyer)
As before, Zechmeister et al. found no evidence for any statistically significant periodic signals in their precision RV data set that would suggest the presence of exoplanets in short period orbits. They did, however, confirm the 2.4 meter per second per year increase in RV noted earlier in the combined CES LC and VLC data set. This was much larger than the 0.009 meter per second per year rate expected from the reflex motion from the ε Indi B brown dwarf binary. The data suggested an orbital period greater than 30 years corresponding to a semimajor axis in excess of 9 AU for a planet with a MPsini of at least 0.97 MJ. These new findings further confirmed the limits based on the work by Janson et al..
The CES LC, CES VLC and HARPS data sets analyzed by Zechmeister et al. showing the clear increase in the RV of ε Indi A over time. Click on image to enlarge. (Zechmeister et al.)
The latest RV analysis results from Feng et al. started with the data set used by Zechmeister et al. and added 4,149 new RV measurements derived from ESO’s archived HARPS spectra including 3,636 high cadence measurements taken over two weeks as part of a campaign to monitor oscillations in ε Indi A. Feng et al. also analyzed various subsets of these newer data: 518 measurements which excluded the high cadence measurements with a low signal to noise ratio and a conservative set of 465 point which excludes HARPS RV measurements made after 2015 which have an as yet imprecisely characterized offset after an upgrade to the instrument’s optical fiber feed. Using the HARPS spectra, Feng et al. also calculated a series of standard stellar activity indicators to see how they may correlate with periodic signals in the RV measurements. A strong correlation would favor a non-planetary explanation for any observed RV variations.
A detailed analysis of the various data sets (and subsets) did indeed reveal three signals with periods of 11, 17.8 and 278 days. Unfortunately these periods seem to be the result of stellar activity and how it is modulated by the 35-day rotation period of ε Indi A. There is also a period of 2500 days noted in the measures of stellar activity corresponding to long term cycles in magnetic activity. The results suggest that there are no exoplanets present around ε Indi A which would produce RV variations with a semiamplitude greater than about 1 meter per second. For the habitable zone (HZ) of ε Indi A which is conservatively defined by Kopparapu et al. (2013, 2014) to range from 0.47 to 0.87 AU (corresponding to orbital periods of ~136 to ~339 days), this result eliminates the possibility of any exoplanets with an MPsini of about 7 to 9 times the mass of the Earth. While this effectively excludes the possible presence of Neptune-mass exoplanets or larger orbiting inside of the HZ, Earth-size exoplanets could still be present and remain undetected by the surveys to date.
The latest analysis of the RV data for ε Indi A combining the various data set over the decades showing that the increase in RV is beginning to flatten out. The data are fitted to an orbit with a period of 52.62 years. Click on image to enlarge. (Feng et al.)
The results of the analysis by Feng et al. also indicates that the long term RV trend appears to be flattening out and possibly even reversing. The analysis of the combined long term RV data sets spanning a quarter of a century by Feng et al. confirm that ε Indi A is apparently orbited by a gas giant in a distant orbit. Their best fit to the available data suggest an essentially circular orbit with a period of 52.62 +27.70/-4.12 years corresponding to a semimajor axis of 12.82 +4.18/-0.71 AU – broadly consistent with the earlier limits by Janson et al.. The semiamplitude of 24.67 +14.28/-3.50 meters per second suggests an exoplanet with a MPsini of 2.71 +2.19/-0.44 MJ. Although not strictly considered a “Jupiter analog” because of its comparatively large orbit, this represents the closest known Jovian exoplanet.
NASA’s James Web Space Telescope should be capable of directly imaging ε Indi Ab providing independent confirmation of this find as well as new data about its properties. (NASA)
As always, more data will be needed to confirm the presence of this putative “ε Indi Ab” and pin down its properties. Combined with more studies of the brown dwarfs ε Indi Ba and Bb, Feng et al. consider this system to be a natural laboratory for a side by side comparison of the formation and evolution of brown dwarfs and Jovian exoplanets. As new precision RV measurements are being made, Feng et al. point out that NASA’s James Web Space Telescope (JWST) should be able to spot this gas giant due to its expected brightness in the infrared and its predicted maximum separation of 3.3 arc seconds. The expected periodic “wobble” of ten or more milliarc seconds in the motion of ε Indi A in response to this orbiting exoplanet should also be detectable not only by JWST, but in long term astrometric surveys including data from ESA’s ongoing Gaia mission. These data should allow independent confirmation of the planetary nature of “ε Indi Ab” as well as allow its properties to be determined including its actual mass. These data along with those from a new generation of instruments which will be capable of making even more precise RV measurements promises to allow astronomers not only to refine our knowledge of this possible nearby Jovian exoplanet, but probe other parts of the system for small exoplanets including inside the HZ.
Follow Drew Ex Machina on Facebook.
Related Reading
“A New Definition for Gas Giants”, Drew Ex Machina, June 30, 2016 [Post]
“How Typical is Our Solar System?”, Drew Ex Machina, August 15, 2015 [Post]
“Where Are the Jupiter Analogs?”, Centauri Dreams, December 18, 2015 [Post]
General References
F. Feng, M. Tuomi and H.R.A Jones, “Detection of the closest Jovian exoplanet in the Epsilon Indi triple system”, arXiv 1803.08163 (Submitted for publication in Monthly Notices of the Royal Astronomical Society, March 21, 2018 [Preprint]
M. Janson et al., “Imaging search for the unseen companion to ɛ Ind A – improving the detection limits with 4 μm observations”, Monthly Notices of the Royal Astronomical Society, Vol. 399, No. 1, pp. 377-384, October 2009
R.K. Kopparapu et al., “Habitable zones around main-sequence stars: new estimates”, The Astrophysical Journal, Vol. 765, No. 2, Article ID. 131, March 10, 2013
Ravi Kumar Kopparapu et al., “Habitable zones around main-sequence stars: dependence on planetary mass”, The Astrophysical Journal Letters, Vol. 787, No. 2, Article ID. L29, June 1, 2014
M.J. McCaughrean et al., “ε Indi Ba, Bb: The nearest binary dwarf”, Astronomy & Astrophysics, Vol. 413, pp. 1029-1036, January 2004
R.D. Scholz et al., “ε Indi B: A new benchmark T dwarf”, Astronomy & Astrophysics, Vol. 398, pp. L29-33, January 2003
M. Zechmeister et al., “The planet search programme at the ESO CES and HARPS. IV. The search for Jupiter analogues around solar-like stars”, Astronomy & Astrophysics, Vol. 552, Article ID A78,, April 2013
Rather confusingly, in the 2018 discovery confirmation paper, Epsilon Indi A is referred to as a K2V star, whereas all the previous descriptions of that star I have seen before give its stellar type as K4 or K5V. Also, it gives the age of Epsilon Indi as 1.4 billion years, which is consistent with what I have read before, but it quotes a paper where the star isn’t even mentioned, well at least I can’t find Epsilon Indi in the paper via Vizier, to back that figure up. However, Bonfanti et al 2015 does give the age for Epsilon Eridani as 1.4 billion years, but I suppose that may just be a coincidence.
“Epsilon Indi A is a good candidate for searching Earth analogs. Epsilon Indi A (HIP 108870, HR 8387, HD 209100, GJ 845) is a young (1.4 Gyr, Bonfanti et al. 2015) nearby K2V star (3.62pc according to van Leeuwen 2007) with a mass of 0.762±0.038 M
(Demory et al. 2009), and a luminosity of 0.22L.”
Perhaps they used the algorithms described in the Bonfanti et al. 2015 paper to calculate the age and it’s merely coinciding with the age of Epsilon Eridani, but still I can’t get my head around the description of Epsilon Indi being K2V.