We’re getting first results from the Gemini Planet Imager Exoplanet Survey (GPIES), a four-year look at 531 young, nearby stars that relies on the instrument’s capabilities at direct imaging. Data from the first 300 stars have been published in The Astronomical Journal, representing the most sensitive, and certainly the largest direct imaging survey for giant planets yet attempted. The results of the statistical analysis are telling: They suggest that planets slightly more massive than Jupiter in outer orbits around stars the size of the Sun are rare.
The Gemini Planet Imager (GPI), located at the Gemini South Telescope in Chile, can achieve high contrast at small angular separations, making it possible to see exoplanets directly, as opposed to the indirect methods that have dominated the field, such as transits and radial velocity analysis. As successful as the latter have been, they are most effective with planets closer to their stars, whereas an instrument like the GPI can find planets in regions outside the orbit of Jupiter. The GPI can directly image exoplanets a millionth as bright as the host star.
Bruce Macintosh (Stanford University), principal investigator for the GPI, calls this effort “…the most sensitive direct imaging survey for giant planets published to date.” The question it raises is significant: Just how representative is our Solar System in having gas giants like Jupiter and Saturn in outer orbits around a G-class star? We’re only beginning to learn the answer, but surveys like this one are on track to tell us. The answer may have astrobiological consequences, as Franck Marchis (SETI Institute), a co-author of the just published report, explains:
“We suspect that in our solar system Jupiter and Saturn sculpted the final architecture that influences the properties of terrestrial planets such as Mars and Earth, including basic elements for life such as the delivery of water, and the impact rates. A planetary system with only terrestrial planets and no giant planets will probably be very different to ours, and this could have consequences on the possibility for the existence of life elsewhere in our galaxy.”
Image: Close-up Picture of Gemini Planet Imager currently located at Gemini South Observatory in Cerro Pachon. Photo by J. Chilcote.
Out of the 300 stars in the thus far released Gemini survey data, 123 are more than one-and-a-half times more massive than the Sun. What the data show is that the hosts of the planets thus far detected are all among the higher mass stars. This despite the fact that given the differential between stellar light and that of a planet, a giant planet orbiting a fainter star more like the Sun is actually easier to see. This relationship of mass to giant planet frequency has been discussed in the literature and is now strengthened by the results of the GPI survey.
The results of the Gemini survey pick up on the theme that other planetary systems tend to be different from our own, despite the assumption that gas giants in outer orbits and rocky worlds on inner orbits would be a fairly standard pattern. Both the GPIES and other exoplanet surveys point to the rarity of giant planets around stars as small as the Sun. Worlds several times more massive than Jupiter and above (the GPIES is not sensitive enough to pick up planets of as low a mass as Jupiter itself) tend to be hosted by stars more massive than the Sun. Our own wide-orbit Jupiter, then, may be a statistical outlier, although that is yet to be determined.
From the paper’s conclusion:
From the first 300 stars observed out of the planned 600-star survey, reaching contrasts of 106 within 1?? radius, GPIES is one of the largest and deepest direct imaging surveys for exoplanets conducted to date. Our analysis of the data shows that there is a clear stellar mass dependence on planet occurrence rate, with stars >1.5 M? [i.e. 1.5 times Solar mass] more likely to host giant planets (5–13 MJup ) at wide separations (semimajor axes 10–100 au) than lower-mass stars.
The paper reports the imaging of six planets and three brown dwarfs, with a sensitivity to planets of several Jupiter masses at orbital distances comparable to those beyond Saturn (at least 12 gas giants had been expected based on earlier models). The only previously unknown planet was 51 Eridani b, which was discovered via GPI as far back as 2014, a gas giant of two-and-a-half Jupiter masses in a Saturn-like orbit around a young star some 97 light years away, and one that had been previously unknown despite attempts to observe the star because no other instrument was able to sufficiently suppress the starlight to make the planet visible.
Image: Results of the survey of 531 stars and their exoplanets in the southern sky are plotted to indicate their distance from Earth. Gray dots are stars without exoplanets or a dust disk; red are stars with a dust disk but no planets; blue stars have planets. Dots with rings indicated stars imaged multiple times. Credit: Paul Kalas, UC Berkeley; Dmitry Savransky, Cornell; Robert De Rosa, Stanford.
Another useful find: A brown dwarf labeled HR 2562 B, 30 times more massive than Jupiter in a Uranus-like orbit. This brown dwarf and the other two imaged in the study shed light on planet vs. brown dwarf formation at wide separations from the host star. The question of brown dwarf vs. planet formation is long-standing. Whereas stars have been considered to form through the gravitational collapse of large clouds of gas and dust, planets are thought to have formed largely through core accretion, as small rocky bodies undergo collision and accumulation of mass.
Eugene Chiang (UC-Berkeley) is a co-author of the paper:
“What the GPIES Team’s analysis shows is that the properties of brown dwarfs and giant planets run completely counter to each other. Whereas more massive brown dwarfs outnumber less massive brown dwarfs, for giant planets, the trend is reversed: less massive planets outnumber more massive ones. Moreover, brown dwarfs tend to be found far from their host stars, while giant planets concentrate closer in. These opposing trends point to brown dwarfs forming top-down, and giant planets forming bottom-up.”
So our Solar System evidently doesn’t resemble many other systems that we’ve observed. Gas giants in outer orbits seem to be more common around significantly larger stars. Putting together a catalog of gas giants in the outer systems of other neighboring stars is going to take time — it’s telling that even the GPI can’t detect Jupiter-mass planets in these orbits — but the GPIES is the beginning of that process, and one that will soon publish additional results. Observations in the survey wrapped up in January with an examination of its 531st star. Moving toward their report on the complete dataset, the team is now following up candidate planets at the same time that it begins an upgrade on the Gemini Planet Imager itself.
The paper is Nielsen et al., ”The Gemini Planet Imager Exoplanet Survey: Giant Planet and Brown Dwarf Demographics from 10 to 100 au,” Astronomical Journal Vol. 158, No. 1 (12 June 2019). Abstract.
I wonder about this quote:
“Worlds several times more massive than Jupiter and above (the GPIES is not sensitive enough to pick up planets of as low a mass as Jupiter itself) tend to be hosted by stars more massive than the Sun. Our own wide-orbit Jupiter, then, may be a statistical outlier.”
Does the fact that planets *several times more massive* than Jupiter tend not to occur around stars like our sun really imply that our own, merely-Jupiter-sized Jupiter, is also a rarity around a sun-like star?
It’s tentative, to say the least, but I think it’s also suggestive. What scientists will be doing is gradually getting down to Jupiter-mass imaging in these outer orbits, which will either firm up the hypothesis or contradict it.
We’ll have to wait and see – it could be like the Hot Jupiters, where once we got a broader sampling of planets they turned out to be rare. Maybe Super-Jupiters are just rare in general, compared to Jupiters and Saturn-sized planets.
This one doesn’t do much for me I’m afraid. They can’t detect Jupiter sized planets. So is the detection limit double Jupiter’s mass or more (it does say several times the mass of Jupiter)? And the numbers are statistically quite small. Six versus 12 expected. When we can compare large data sets from all three major detection methods (transit, RV, and direct detection) we will have a much better idea of what is going on. Many of these systems examined could have Jupiter sized planets and larger orbiting around them at various distances and not be detected. This in fact could lead to quite misleading conclusions couldn’t it?
I agree, and with jonW, that all this paper concludes about planets, is that super-giant planets (> 5 Mj) in very wide (>10 AU) orbits are more likely around high mass(> 1.5 Msol) stars. That’s about it, no conclusions about Jupiters in Jupiter-like orbits.
And of course the statement about the two formation types, collapse versus accretion, is really interesting.
We may live in an oddball system. But, the jury is still out on that. These results on giant planets in wide orbits are very preliminary. The GPI is much more sensitive to hot start planets(disk instability) than cold start planets(core accretion) like Jupiter and Saturn. Even the proposed GPI 2.0 upgrade would only partially rectify this discrepancy.
Needed are larger instruments/greater sensitivity for the fainter cold start objects and smaller inner working angle.
If this uses direct imaging, this depend also on the planet being in favorable position to see it. If the planet is “behind” its sun from our point of view, it will be illuminated (on at least half of its surface) and we will see it. If it is closer, we will “see” only the dark side or perhaps a small crescent … and we will not detect it.
Similarly as we can easily see full moon, but we can’t see anything if the moon is at the opposite phase, with sun illuminating its opposite side.
So statistically, if half-illuminated planet is enough for a detection here, there would be twice as many planets as actually detected by this survey.
Martin, that’s a very good point, plus many of these planets would be observed from above the plane of their orbits. This would give half crescent illumination or less in many cases depending on their orbit position in the four years of the survey.
Since this device is described as a planet imager, how does it determine whether a jovian planet is more massive than Jupiter or not?
My understanding ( sic) of gas giant structure is that the mass-radius is rather flat and might even be slightly negative minus a sources of interior
heat only available to brown dwarfs ( deuterium fusion) and stars ( hydrogen fusion). At first blush, wouldn’t gas giants look much alike?
The moons of those super gas-giants (at least 10-20 mass of Jupiter) should be very interesting, imaging quasi-Europa is as big as Mars or even bigger.
I have to concur that this looks like a case of “Absence of evidence is not evidence of absence”. The instrument must be able to detect Jupiter sized planets sufficiently and with enough frequency to ensure the hypothesis is confirmed. At this point they have detected 6 planets in the orbital distances (10 – 100 AU), all of which are at least 2,6 Jupiter masses.
From this small data set, they build an occurrence model of abundances from very little data.
This is important, because I suspect the modelers will build planetary formation models to that support this outcome, but they may be biasing their parameters to achieve this, going off-track if the observations do not reflect reality.
Bottom line: It isn’t clear to me that the lack of detection of planets of around 1x Jupiter mass is due to true absence or inability to detect the planet. Our own sun would show no data as Jupiter is just 5 AU from the Sun and therefore outside their range for detection (10 – 100 AU).
Strangely if we think about the results,
it actually expands slightly the possibility of stable worlds
with high habitability index in sun like stars .
This result means that highly massive disruptive (mass > Jupiter) gas giants are not out there to create chaos in the early formations of solar systems. Our own Possible planet X gives a small sampling of what
maybe out there. Objects greater than Jupiter out there would complicate a solar systems’ final structure/arrangement.
Did we know this for sure before the findings presented; there was
a hint of it in Kepler data, but its nice to rule out this “fly in the ointment” in matters of solar system formation.
So the popular viewpoint among the comments re these results (which I too agree with) is that the team has extrapolated too far in concluding that true Jupiter analogs (close in both mass and orbit) are rare. To add to the good points others have made there is the natural trend for less massive to normally outnumber the more massive. That’s how I would defend my not wanting to believe their ‘Jupiters are rare’ prediction, but I wonder how much our thinking is being colored by hopes rather than cold logic. The prediction is based on hard (albeit somewhat scanty) data, whereas our arguments against the conclusion may be based wishful thinking.
One of the tenants of the Rare Earth theory is the need for a Jupiter like planet. But we don’t WANT Earths to be rare, therefore we don’t want Jupiters too be rare either.
The fact that cold jupiters don’t seem to be that common is something that has been discussed before. I have certainly noticed a few years ago.
Warm jupiters (sort of the HZ) and hot jupiters seem to be a lot more common.
You can see it in mass/period diagrams like this one :
There seem to be two main clusters of hot and warm jupiters.
I remember reading that Gaia should find all jupiters < 150 ly. That would help.
I think we can’t jump to conclusions the Jupiter mass planets are rare around Sun-like stars because first the planets than instruments can detect right now are only young self illuminate planets and second I never notice this telescopes instruments detect a mature Jupiter by reflect light of it parent star. The question is when the instruments will be capable to detect reflect light mature around it’s stars? I still don’t direct image of Epsilon Indi b or Gliese 832 b for example that are mature planets not so far and mature.
300 star systems out of 400 billion in the Milky Way galaxy? When they have scanned many thousands of star systems and come to similar data finds, then come talk to us.
It is hard not to wonder if there is an underlying nonscientific agenda when I see things like this. Yes, they can make assumptions based on the data, but it is an awfully small sample. However, it seems someone or a group of someones wants to make Earth and our Sol system unique with all that it implies.
Those past downgrades of our literal place in the Universe can be tough after millennia of our species thinking we are the only ones and that everything was made just for us, I get that.