# Shortest and longest wavelengths at which exoplanets have been resolved from their primaries?

Phys.org's Radio signals from distant stars suggest hidden planets summarizes the Nature Astronomy paper The population of M dwarfs observed at low radio frequencies (also in arXiv) but this is an indirect measurement.

Here I'd like to ask about techniques that provide resolved imaging of the system, where the exoplanet can be separated spatially from its primary. In other words some kind of image where one can say "here's the exoplanet, north-east of the primary at X micro-arc seconds."

Question: What are the shortest and longest wavelengths at which exoplanets have been resolved from their primaries?

I think these are usually done using adaptive optics in red or near infrared light and silicon CCDs, so perhaps something like 700 to 1000 nm. But has any exoplanet been imaged in visible light? Or UV, perhaps from the HST?

On the other side, have large radio telescope arrays ever resolved exoplanets? I'd guess not as their Suns would not illuminate them with sufficient brightness (How far have individual stars been seen by radio telescopes?)

This is a particularly nice example source:

Motion interpolation of seven images of the HR 8799 system taken from the W. M. Keck Observatory over seven years, featuring four exoplanets

• I'd think this one could be resolved with an exhaustive look at all the planet's we've found by direct image, which wouldn't be all that exhaustive since we've only imaged a small fraction of the ones we've found. Oct 12, 2021 at 14:21
• @zephyr I'd thought it would be the other way around; that nobody would dedicate the observing time necessary to interferometrically image random stars in hopes there would be a resolvable planet. Wouldn't they comb the list of known systems discovered by parallax or spectroscopically for the best candidates for imaging, then use that observing time on a target that's very likely to be successfully resolved?
– uhoh
Oct 12, 2021 at 14:46
• Yeah, that's how they'd do it. I just mean that, the number of exoplanets that have been directly imaged is very small. Since you seem to be interested in the range of wavelengths used in imaging exoplanets, and since the number of imaged exoplanets is so small, it shouldn't be too much work to just find all imaged exoplanets and look at the range of the very small set. Oct 12, 2021 at 20:53
• @uhoh In fact, the Gemini Planet Imager Exoplanet Survey targeted $\sim$600 young stars (without pre-existing planet detections by other means) specifically in hopes of finding planets. (arxiv.org/abs/1904.05358) Oct 13, 2021 at 10:42
• Planets detected by spectroscopy (radial velocity measurements) or occultations would not be good candidates for direct imaging, because they would be too close to their host stars. Oct 13, 2021 at 10:44

I'm pretty sure it's restricted to the near-IR, with the shortest wavelengths being $$Y$$-band or $$J$$-band (i.e., 1 or 1.2 microns) and the longest being $$L$$-band (i.e., 3.8 microns). This is probably because 1) classical adaptive optics systems use the optical for corrections that are applied in the near-IR (for reasons I discussed in this answer); and 2) young, hot planets may radiate enough in the near-IR to be brighter relative to the star than they would be in the optical.
"Fomalhaut b" used to be an example of an exoplanet imaged (with the Hubble Space Telescope) in the optical, with wavelengths as short as $$\sim 0.4$$ microns, but the consensus now seems to be that's it's not a planet at all, and is instead something like an expanding cloud of dust from the collision of two protoplanetary objects.