Main

We know that the vast majority of stars host planets1—the focus of exoplanet science has now shifted to understanding their detailed physics. Determining the composition of exoplanetary atmospheres, geospheres and associated space weather conditions is considered paramount to assess the potential habitability of a planet2. While transmission spectroscopy is beginning to reveal the compositions of exoplanets3,4,5,6, the space weather and radiation environments that shape their atmospheric evolution and habitability remain relatively unknown.

The largest contributors to space weather in the Solar System are coronal mass ejections (CMEs)7,8, which launch a substantial amount of hot, dense plasma from the Sun into the Solar System. Persistent impacts of CMEs on a terrestrial planet have the potential to erode its atmosphere9,10 (as may the quiescent solar wind), particularly for planets without intrinsic magnetic fields11,12. Despite the importance of understanding the plasma physics of CMEs, there has yet to be an unambiguous detection of a CME from a star other than our Sun. The winds of low-mass main sequence stars are also generally too tenuous to detect via current technology13,14 for all but a handful of stars15. Furthermore, direct measurements of an exoplanet’s magnetic field have not yet been made. This information is crucial to understanding the atmospheric evolution of such a planet, as its magnetic field could protect the atmosphere from the impact of stellar plasma16,17.

Radio observations, particularly at low frequencies (≲300 MHz), are a unique probe of stellar and planetary plasma environments18,19,20. As observed on our Sun, CMEs are often accompanied by bursty, low-frequency radio emission that encodes the kinematics of the plasma as it is ejected into interplanetary space18,21. The incident solar wind on the magnetized planets also drives auroral emission22, particularly at radio wavelengths23. While optical emission from exoplanet aurorae will probably be difficult to observe even with extremely large telescopes24, the direct detection of auroral radio emission from an exoplanet has been a long-standing goal of the radio astronomy community. This would allow us to infer the presence, topology and strength of exoplanetary magnetic fields for the first time23,25,26. At other wavelength regimes, such information about the magnetic field of an exoplanet is either model dependent27 or untraceable20.

While the goals of detecting CMEs, stellar winds and planetary magnetic fields in exoplanetary systems have been pursued for decades, recent observational progress has been made due to both the maturation of low-frequency radio arrays and the increased sensitivity at gigahertz frequencies. For example, observations from the LOw-Frequency ARray (LOFAR)28, the Giant Metrewave Radio Telescope, the Karl G. Jansky Very Large Array, the Five-hundred-meter Aperture Spherical Telescope, and the Australian Square Kilometre Array Pathfinder29 have begun to uncover new and diverse signals from radio stellar systems30, some of which may be consistent with interactions between stars and planets analogous to phenomena observed in our Solar System31,32,33,34,35,36,37. This advancement in observational radio astronomy has also been paired with a revolution in optical and near-infrared capabilities, as exemplified by the Transiting Exoplanet Survey Satellite (TESS)38 and new near-infrared radial velocity39,40 and Zeeman–Doppler imaging41,42 surveys. The combination of these radio and optical/near-infrared facilities provides an unparalleled opportunity to trace key stellar activity indicators across the electromagnetic spectrum, informing us about whether emission is driven by coronal or magnetospheric processes and providing probes of extrasolar space weather environments.

This Perspective focuses on communicating recent observational advances that have been made in detecting the coherent radio signatures of star–planet interactions (SPIs), exoplanets, ultracool dwarfs (UCDs) and space weather, and the associated basic foundational theories—with the aim of being understandable and a primer for doctoral students and scholars new to the field.

Radio emission in the Solar System

The properties of radio emission observed from the Sun and Solar System planets are commonly used as templates for interpreting radio emission from stellar systems. Solar radio emission is often incoherent, implying that the electrons producing the radio emission do not act in phase, and is observed over a broad range of frequencies up to a few gigahertz18,43. Such incoherent radio emission is frequently associated with magnetic reconnection events in the solar corona, or active regions on the Sun, where mildly relativistic electrons produce gyrosynchrotron emission by their motion along magnetic field lines.

Coherent solar radio emission is predominantly generated by plasma emission via plasma waves. These plasma waves are excited by beams of energetic electrons that have been accelerated by magnetic reconnection events or shocks18,43. Phenomenologically, the two most important types of solar burst for tracing space weather are classed as type II and type III bursts. Type II bursts, lasting from several minutes to hours, are usually produced by the acceleration of electrons by shocks at the leading edges of outward-moving CMEs. Type II bursts tend to slowly decrease in frequency over several minutes, providing a measure of how the plasma density decreases as the wave propagates out of the solar corona. Type III bursts are short (few-second) events associated with electron beams accelerated by reconnection events, and often accompany type II bursts. A schematic of the structure of a CME and a dynamic spectrum showing type II and III bursts are provided in Fig. 1. In Fig. 2 we present a practical, heuristic guide to differentiating between the different radio emission mechanisms operating in stellar systems.

Fig. 1: Schematic representation of a CME and a dynamic spectrum of the event that shows type II and III bursts.
figure 1

Left: schematic of a CME. Right: radio dynamic spectrum of the event. The type II burst is produced in the coronal shock front, as represented by the region emanating from the red star in the left panel. The magnetic reconnection event that is allowing mass to escape the magnetosphere of the star is shown as the yellow region. Type III bursts are produced on open field lines surrounding the magnetic reconnection event. Some structure and higher-frequency harmonic emission is evident in and around the type II burst in the right panel. The dynamic spectrum is in total intensity.

Full size image
Fig. 2: Schematic for distinguishing the emission mechanism operating when a radio stellar system is detected.
figure 2

The emission characteristics listed to differentiate between the different mechanisms should be treated as approximate guidelines (see refs. 18,43 for a more physical differentiation based on the conditions of the plasma). δν/Δν (where ν is frequency) represents the amount of bandwidth δν the emission occupies relative to the available bandwidth Δν assuming a relatively large fractional bandwidth. Differentiating between plasma emission and emission from the ECM instability can be difficult if the time–frequency structure of the radio emission cannot be resolved. In that case, arguments can be made in favour of one emission mechanism based on the coronal scale height of the radio star, as derived from the stellar X-ray luminosity248,249. Fundamental and harmonic plasma emission have different circular polarization fractions (CPFs) and maximum brightness temperatures, with harmonic plasma emission able to reach the highest brightness temperatures but limited to ≲60% of the circular polarization fraction249. Note that the polarization fractions reported do not take propagation effects into account, which often suppress the fractional amount178.

Full size image

In contrast, radio emission from the magnetized Solar System planets is dominated by auroral processes. What this loosely refers to is emission that is generated close to the magnetic poles of the planet. A key ingredient that drives auroral emission is electron acceleration: the injection of a high-velocity electron population into a quiescent magnetosphere powers bright, coherent radio emission via the electron cyclotron maser (ECM) instability44,45. ECM emission is highly circularly polarized and occurs at the local cyclotron frequency (νc). The maximum frequency of ECM emission is directly proportional to the ambient magnetic field strength at the emitting point. The ECM emission is therefore a direct probe of the magnetic field strength of the emitting body18. Another characteristic of ECM emission is that it is beamed, in that the radio waves propagate outwards on the surface of a cone that is near perpendicular to the local magnetic field. One result of this beaming is that the emission is only generally visible for brief windows in time and can exhibit complicated periodicity46.

Several distinct mechanisms can produce the population of accelerated electrons in planetary magnetospheres. The first is due to the interaction of the solar wind directly with a planet’s magnetosphere. In this case, the incident solar wind is magnetized and compresses the planet’s intrinsic magnetosphere, resulting in magnetic reconnection on the nightside of the planet. This process is often referred to as the Dungey cycle22, and the interaction of the high-energy electrons with the atmosphere produces the aurora australis and aurora borealis on Earth. Interestingly, the observed auroral radio power from the magnetized Solar System planets is seen to directly scale with the power of the incident solar wind, both kinetically and magnetically23,47,48,49 — the kinetic relation is often referred to as the radiometric Bode’s law and the magnetic relation is referred to as the radio-magnetic scaling law.

Despite being the brightest low-frequency radio emitter in the Solar System, Jupiter’s auroral emission deviates from this narrative as two different processes dominate the auroral radio emission: the breakdown of co-rotation and Jupiter–satellite interactions50,51. In the breakdown of co-rotation model, Jupiter’s inner magnetosphere is continuously supplied with plasma from Io’s volcanic outgassing. The inner plasma is forced to co-rotate with the magnetic field of Jupiter as it is centrifugally expelled outwards. At a certain distance in Jupiter’s magnetosphere, the magnetic field is not strong enough to force co-rotation of the plasma, producing a shear against the outer plasma environment. This shearing generates a current system in which electrons are accelerated from the equator to the poles, producing the auroral ring52,53,54.

The brighter and more localized radio emission from Jupiter is associated with the magnetic field lines of Jupiter connecting it to its Galilean moons, particularly to Io51,55,56,57. The driving mechanism of this emission is thought to either be due to Alfvén waves generated by the perturbation of Jupiter’s magnetic field by the inner moons56 or magnetic reconnection occurring between the magnetic fields of Jupiter and the moons58,59,60. In either case, electrons are accelerated, subsequently producing bright circularly polarized radio emission via the ECM mechanism. One constraint on this form of emission is that it can only be produced if the perturbing body orbits inside the Alfvén surface of the host body for SPIs, a region in which the host body’s magnetic field dominates.

For new scholars to the field, Alfvén waves are transverse plasma waves that travel along magnetic field lines. They are produced when ions oscillate in response to the restoring force provided by the tension of magnetic field lines. The Alfvén surface is defined as the three-dimensional boundary that separates a star’s corona from its stellar wind—the boundary at which information in the stellar wind cannot propagate back to the surface of the star. This is the locus where the Alfvénic Mach number is unity.

The observed processes of radio emission from Jupiter serve as archetypes for what is expected from extrasolar planetary systems, where it is possible to produce emission that is orders of magnitude brighter than that observed in Solar System by scaling the interaction by the mass, obstacle size, magnetic field strengths of the stellar wind and the planetary obstacle and orbital distance (see the ‘Theory of magnetic SPIs’ section for quantitative expressions of this statement)23,49,61. These interactions are often referred to as magnetic SPIs in the literature; we will refer to them generally as SPIs here.

As illustrated in Fig. 3, there are two sub-types of SPI that are relevant to producing ECM radio emission from radio stellar systems that contain a low-mass host star:

  1. (1)

    Sub-Alfvénic interactions. Analogous to the Jupiter–Io interaction. Radio emission occurs along the stellar magnetic field line connecting the star and planet, driven by interactions between the field line and the planet. It requires the planet to be orbiting inside the Alfvén surface, with either an intrinsic or induced magnetic field. These are sometimes referred to as magnetic SPIs. The power produced by these interactions seems to follow only the radio-magnetic scaling law62.

  2. (2)

    Wind–magnetosphere interactions. Auroral radio emission is produced in a ring-like configuration near the planet’s magnetic poles, driven by the interaction of the incident stellar wind and planetary magnetosphere. The power produced by this interaction seems to follow both the radiometric Bode’s law and the radio-magnetic scaling law for most auroral sources in the Solar System.

Fig. 3: Sketch illustrating the two putative sources of ECM emission in exoplanetary systems.
figure 3

Left: emission induced on a star by a close-in planet. If the planet orbits inside the Alfvén surface of the star, it can perturb the star’s magnetic field, producing Alfvén waves that propagate back towards the star. These waves interact with electrons, accelerating them towards the stellar surface. Electrons with sufficiently large pitch angles undergo a magnetic mirroring effect and are reflected, producing ECM emission that propagates in a hollow cone. Right: auroral emission induced on a planet by the interaction of its magnetosphere with the incident wind of its host star. The magnetic field carried by the stellar wind causes the planet’s magnetosphere to open up on the dayside (left). These field lines are pushed towards the nightside (right), where they subsequently reconnect. The energy released in the magnetic reconnection accelerates electrons back towards the poles along the field line highlighted in red, where they reflect and power ECM emission in a similar manner to that described for the left panel. For clarity, the emission cone is only shown for the northern hemisphere in both cases.

Full size image

Naturally, SPI terminology is no longer accurate once either the star or planet are replaced with a brown dwarf or moon, but the underlying physical processes are still relevant in those scenarios. For instance, an exomoon could produce detectable radio emission on an exoplanet via a sub-Alfvénic interaction. Stochastic flaring and breakdown of co-rotation63 are also pathways to generate coherent radio emission in these systems, but as these processes do not explicitly rely on the presence of a planet, they do not fall under the umbrella of SPIs. We also note that the term SPI is often used more broadly to refer to the irradiation of the planetary atmosphere by the host star’s light and its subsequent interactions with the wind of the host star64, as well as tidal interactions between the two bodies65. However, given that neither of these processes are expected to directly produce bright signatures in the radio region, we do not discuss them here.

Stellar flares and CMEs

Radio observations of the Sun and other solar-like and low-mass stars have shown that stellar flare production involves all layers of the stellar atmosphere, from the photosphere to the corona. The wide range of physical scales involved in producing flares implies that a variety of processes are responsible, such as coronal heating and particle acceleration66. There is growing consensus around a picture that connects the radio, optical, ultraviolet and X-ray flare luminosities of stars to coronal heating. However, tracing CMEs using time-resolved, simultaneous multiwavelength observations67 — both to understand stellar mass loss and to directly probe the space weather environments faced by exoplanets — remains a challenge.

There is a well-established model for the non-thermal incoherent radio emission from solar and stellar flares. The model is motivated by a remarkable correlation between the quasi-steady gigahertz-frequency radio Lν,rad and soft X-ray LX luminosities: ({L}_{{rm{X}}}propto {L}_{nu ,{rm{rad}}}^{0.73}) (refs. 68,69). This correlation is often referred to as the Güdel–Benz relation. Such a relation is canonically explained via the fact magnetic reconnection events produce streams of non-thermal accelerated coronal electrons. These streams impact the chromosphere and cause the evaporation of million-degree plasma, depositing their kinetic energy as thermal energy in the corona70. The free electrons produced by the magnetic reconnection event can be observationally traced by the gigahertz-frequency gyrosynchrotron emission as they radiate in the stellar magnetic field, while the heated plasma is identified by its thermal soft (0.2–2 keV) X-ray emission. It is important to note that while the explanation presented here is qualitatively plausible, it cannot fully describe the Güdel–Benz relation quantitatively71. The Güdel–Benz relation equally applies over many orders of magnitude to most solar and stellar flares, suggesting that very hot, active stellar coronae are heated by the integrated energy of flares — including those flares that may be too low in energy to detect individually72,73.

Small-scale stochastic flares could indeed be responsible for the quasi-steady soft X-ray and steady low circularly polarized, gigahertz-frequency radio emission from stellar systems. If the energy distribution E of stellar radio flares follows a power law, low-energy flares will be largely responsible for heating the corona if the power-law index α is >2 in dN/dE ∝ E−α. However, this has been challenging to test in the radio: observational flare distributions are limited at low E by the instrumental detection limit, depend on stellar distance and flare peak luminosity, and require long time series.

In the short UV, optical and non-thermal hard X-ray ranges, solar and stellar flare statistics do reveal power-law distributions down to the light curve noise, mostly with α ≳ 2 for active stars74,75,76,77,78,79,80,81,82,83. Optical stellar flare statistical studies have been revolutionized since the Kepler84 and TESS38 space telescopes produced hundreds of thousands of high-cadence, high-precision broadband optical light curves. Homogeneous statistical catalogues of flares83,85,86,87,88,89,90,91 have been used to investigate the population-level statistics92,93 and flare characteristics of planet hosts94,95,96, and have revealed Carrington-analogue superflares on solar-like stars97,98,99,100,101,102.

New radio facilities are finally capable of probing whether large optical flares are accompanied by CMEs. While this is true for the largest of the Sun’s flares103, commonly flaring M dwarfs have different magnetic geometries and stronger global fields42,104. Recent modelling efforts105 have shown that large overlying magnetic fields can prevent mass breakout, resulting in the occurrence of confined flares that do not have an accompanying eruption. The ongoing all-sky TESS mission uniquely complements modern radio telescopes for multiwavelength flare studies: simultaneous observations with the Australian Square Kilometre Array Pathfinder and TESS observations have traced a radio burst from Proxima Centauri32, as well as radio emission with no optical counterpart106. The TESS flare rates of LOFAR-detected stars correlate with their X-ray luminosities, consistent with the Güdel–Benz mechanism, but several show high ECM radio luminosities and few or no flares or X-ray emission. Such deviations have been interpreted as evidence for the radio emission being generated by star–planet magnetic interactions, since radio emission generated by plasma or gyrosynchrotron mechanisms is expected to correlate with magnetic events that often have an optical/X-ray flare counterpart107.

While type II bursts are the only unambiguous radio proof of CME material escaping from the stellar magnetosphere, there are no firm detections of type II bursts from other stars despite significant observational effort19,108,109,110. The non-detections of extrasolar type II bursts may be due to sensitivity limitations, the magnetic confinement of CMEs on radio-bright stars111 or an Alfvén speed that prevents shock formation112. Instead, radio observations of active M dwarfs have found other types of coherent radio burst, including hours-long events attributed to ECM emission19,110,113,114 that do not have a direct solar analogue. Other promising evidence of stellar CMEs includes blueshifts of chromospheric lines and extreme UV/X-ray coronal dimming115, but such measurements cannot determine whether the material was retained in the stellar magnetosphere — implying that the plasma never impacted a putative planet. Low-frequency wide-field surveys with <10 mJy sensitivity on timescales of minutes are the most likely avenue to detect an extrasolar type II burst and confirm that mass has been ejected, as such bursts are stochastic and the emission frequency is probably lower in weaker magnetic field strengths18,116. Such a detection would allow the plasma density to be measured as the radio emission occurs at the plasma frequency ({nu }_{mathrm{p}}propto {n}_{mathrm{e}}^{1/2}), where ne is the electron density, allowing us to trace the particle flux at the point of impact of an exoplanet117.

Finally, the Sun exhibits steady mass loss through the solar wind. Measuring stellar winds has been notoriously difficult: radio observations of thermal bremsstrahlung emission have so far only placed upper limits on stellar winds from young solar analogues13,14 and indirect UV wind measurements of M dwarfs suggest a fairly large spread of inferred mass loss rates with surface X-ray flux15. However, the Next Generation Very Large Array (ngVLA;118) may be able to detect mass loss from the steady stellar wind of nearby M dwarfs119.

Radio emission directly from exoplanets

Exoplanetary systems are observed in the radio primarily to search for auroral emission powered by the breakdown of co-rotation, wind–magnetospheric interactions or emission induced on the host star by an exoplanet via sub-Alfvénic interactions. However, the energetics of these different models imply that any radio detection of an exoplanet will probably be near the sensitivity limits of current radio facilities23,50,120,121. Therefore, direct detections of exoplanetary auroral radio emission have been lacking, despite decades of searching121,122,123. Following several foundational works23,48,49,124, a large collection of theoretical work predicting the intensity and frequency of the radio emission of exoplanets observed from Earth has been published (for example, refs. 26,50,125,126,127,128,129,130,131,132,133,134,135,136).

As stated in the ‘Radio emission in the Solar System’ section, radio emission directly from an exoplanet can be powered by the energy deposited on the planetary magnetosphere by the incident stellar wind23,48 or via a sub-Alfvénic interaction with a moon137,138,139. The radiometric Bode’s law48,26 and the radio-magnetic scaling law23,49,140 have been used to predict the radio intensity of exoplanets, directly relating the incident stellar wind power (kinetic or magnetic) to the emitted radio power. However, the uncertainty in the predicted flux density can be greater than an order of magnitude, and the uncertainty in the predicted maximum emission frequency can be off by a factor of 2–3 (ref. 140). These predictions should therefore be used with caution.

A large number of observational campaigns have been performed to search for radio emission from exoplanets20,141, resulting in clear non-detections25,120,133,142,143,144,145,146,147,148,149,150,151,152,153,154 and a few tentative, but contested, detections35,155,156,157,158,159. None of these claims of detection have yet been confirmed by follow-up observations. The causes for radio non-detections directly from exoplanets are degenerate121,122,123,135. Namely, it is possible that we have no clear detections because:

  1. (1)

    Observations have not been sufficiently sensitive

  2. (2)

    The planetary magnetic field is too weak, implying that ECM emission is not produced at accessible observing frequencies

  3. (3)

    Earth was outside the beaming pattern of the radio emission at the time of the observation127,160

  4. (4)

    The ECM conditions vary in response to changes in the stellar wind conditions and/or the electron velocity distribution, causing the radio emission to drop below the detection threshold of the observation161,162

  5. (5)

    The ECM process is inhibited by (for example) inflated ionospheres132,163,164,165,166

  6. (6)

    The wind of the host star absorbs or prevents the escape of the emission135

  7. (7)

    The presence of high plasma densities in the planetary magnetosphere or surrounding stellar wind suppresses wave propagation18,164

  8. (8)

    There are difficulties in disentangling whether the detected coherent emission is stellar or exoplanetary in origin as stars can also intrinsically produce coherent emission (for example, refs. 110,154)

It is possible that all of these effects could come into play for each individual system, although their relative contributions have not yet been quantified. This means that it is difficult to infer physically meaningful information about a single system from a non-detection of ECM emission.

Progress in detecting radio emission directly from an exoplanet will largely come from complete phase coverage of the orbit of the exoplanet (preferably several times) and increased sensitivity at low frequencies. The development of sensitive instruments that observe at ≲300 MHz is vital if we want to probe systems with Jupiter masses or below. This is because ECM emission scales linearly with the magnetic field strength of the body, and according to dynamo theory it is unlikely that planets with masses less than ten times that of Jupiter will possess magnetic field strengths significantly greater than ~100 G (ref. 167; Jovian radio emission cuts off above ~40 MHz as its magnetic field does not exceed ~14 G)168.

Auroral radio emission from UCDs

Owing to the difficulty in conclusively detecting radio emission directly from exoplanets, we can leverage the considerable effort that has gone into observing UCDs (spectral types > M7) to examine magnetic processes across the substellar regime, which are expected to have the same underlying physics as direct emission from exoplanets and SPIs. Some UCDs have exhibited bright radio bursts169 generated by ECM emission170, allowing us to trace auroral processes in magnetospheres that are more similar to Jupiter than the Sun171,172,173. The Jupiter-like analogy for UCDs has recently gained further support with the detection of synchrotron radiation belts around a UCD174,175, a key radio morphological characteristic possessed by Jupiter’s magnetosphere.

The serendipitous discovery of bursting radio emission from a brown dwarf169, and ensuing UCD radio detections, demonstrated that their radio activity can strongly depart from well-established stellar coronal/flaring relationships176,177. In light of the subsequent discovery that UCD radio bursts can exhibit periodic timing178, their strong circular polarization and high brightness temperatures imply that the ECM process is at least partly responsible for the radio emission from some UCDs170,179. Despite thorough gigahertz-frequency radio searches of UCDs, detections rates have remained persistently low at approximately 10% (refs. 180,181,182,183,184,185,186,187,188).

Periodically bursting radio emission traces extrasolar aurorae in UCDs. Simultaneous radio and optical observations of the M8.5 dwarf LSR J1835+3259 showed that the impacting electron current traced by its bursting radio emission creates a surface feature that is spectrally distinct from typical magnetic spots seen on stars171. UCDs with optical–infrared periodic variability or Hα emission can have an 80% radio detection rate172. Magnetospheric current systems modelled using Jupiter can explain the generally correlated radio and optical/atmospheric behaviours across the UCD population173,189. In this picture, brown dwarfs are magnetic analogues to gas giants and laboratories for studying planetary dynamos172,190,191. Furthermore, the discovery of previously unknown brown dwarfs via their radio emission189,191,192 demonstrates that large-field radio surveys can access a new discovery space for brown dwarfs or planets that other observational means may overlook. In particular, it is expected that the next generation of wide-field radio surveys will identify lower-mass or more-distant UCDs, as infrared surveys are reasonably complete within 25 pc for UCDs of spectral types earlier than T4193.

The broad success of the UCD auroral paradigm naturally leads to a substantive open question: what is the source of the magnetospheric plasma in such systems? This magnetospheric condition may determine the overall low radio detection rate (<10%)186,187. One potential source is exoplanet companions173,194: in the same way that Io seeds the Jovian magnetosphere with plasma, these satellites could provide the plasma for their UCD hosts. Although challenging, searches for planets around UCDs are ongoing195,196,197, with TRAPPIST-1198 providing a prototypical example. The radio emission itself is also enabling companion searches, and astrometric radio monitoring via very-long-baseline interferometry has already yielded evidence of a companion around an auroral UCD199.

A statistically significant coincidence between UCD planetary systems and bright radio emission remains unproven. Nevertheless, because a satellite population is likely to exist, the strong magnetic fields of the lowest-mass stars and brown dwarfs190 make them excellent candidates for searching for magnetic host–planet interactions200. Alternatively, the radio emission could be similar to the breakdown of co-rotation seen in the Jupiter system, with the plasma supplied by magnetic reconnection events in the UCD’s magnetosphere. Long-term radio monitoring of these systems and determination of the UCD rotation period are required to identify which model is correct, as the periodicity will either correlate with the rotation of the UCD or the orbit of a putative satellite.

Radio emission from SPIs

As well as radio emission directly from a planetary magnetosphere, it may be possible to drive radio emission via the magnetic connection between a close-in planet and its star similar to a scaled-up Jupiter–Io interaction23,49,140,141,201, with a star taking the place of Jupiter and a planet taking the place of Io.

Radio emission from SPIs via ECM emission could reach higher frequencies than planetary magnetospheric radio emission (up to gigahertz frequencies for some M dwarf, massive or young stars due to their relatively strong magnetic fields) as the auroral emission is occurring in the magnetosphere of the star. This higher frequency makes radio emission from SPIs easier to detect with ground-based radio observatories than the emission expected directly from exoplanetary magnetospheres. Note that the radio emission from SPIs only gives a direct measure of the magnetic field of the star, not the planet. However, it may be possible to model a planet’s magnetic field in such an interaction (see, for example, refs. 123,202,203) in a similar manner to models for Ganymede204. Emission from SPIs may also manifest as variation in X-rays205 and UV/optical/near-infrared activity indicators27,206,207,208—however, all claims of SPI detections at these wavelengths are disputed209,210,211,212.

Similar to radio emission from exoplanets, there has been no confirmed detection of radio emission from SPIs. However, some recent publications showed encouraging signals that merit follow-up. LOFAR observations have shown low-frequency radio emission from the quiescent M dwarf GJ 1151, and several others, at 144 MHz (refs. 31,34), which has been attributed to the interaction of the star with a close-in Earth-sized planet due to the emission and stellar properties. An intensive campaign of precise radial velocity observations with HARPS-N213, HPF214 and CARMENES215 detected a >10.6 M⊕ (where M⊕ is the mass of the Earth) companion to GJ 1151 in a 390 day orbit at a separation of 0.57 au, but only a 1.2 M⊕ upper limit on the mass and orbit of any companion close enough to be in the sub-Alfvénic region of the star216.

Several other recent unconfirmed SPI detections also merit follow-up with intensive observing campaigns, particularly in synergy with spectropolarimetric monitoring that aims to model the large-scale field of the host stars. At gigahertz frequencies there have been suggested detections of ECM emission that may be modulated by planetary orbits from Proxima Centauri33 and YZ Ceti158,217. Modulations of optical activity tracers close to the period of AU Microscopii b have been interpreted as SPIs208, and suggest AU Mic as a promising target for SPI detections. However, only evidence of stellar rotation modulation in the radio has been detected from AU Mic (ref. 154).

Theory of magnetic SPIs

Considering the likelihood of the detection of radio emission from SPIs with the current generation of instruments, here we outline the theory of SPIs and how it can be used to derive the physical characteristics of systems with SPIs.

SPI radio emission is produced by the ECM process near the local electron νc, which peaks at νc = 2.8 B* MHz, where B* is the magnetic field strength of the star in gauss. Such emission is produced in a rarefied (νc ≫ νp) plasma by an unstable electron population with characteristic energy of ~1–20 keV (refs. 45,218,219,220,221,222).

Furthermore, SPI radio emission is predominantly emitted via the extraordinary (x) magneto-ionic mode, implying that the polarization of the emission is highly circular or weakly elliptical. The emission is beamed at a large angle from the magnetic field, typically 60–90°, and in a thin conical sheet with a thickness of 1–2° (ref. 218). We schematically show such an emission geometry in Fig. 3. Modelling the ECM emission often involves assumptions about the stellar magnetic field topology, making it possible to produce a ‘visibility curve’ for radio emission for a system with SPIs135,223. In the limit of a dipolar stellar magnetic field, radio emission from SPIs can have a very low duty cycle (<10%) and a sensitive dependence on the obliquity between the axes of the stellar dipole, rotation and planetary orbit46.

The radio power produced by variable ECM emission is difficult to predict from first principles: ECM growth rates and the wave path along which they operate depend on the details of the distribution of the unstable ~1–20 keV electrons in the auroral regions and on the magnetic field topology and ambient plasma distribution in the radio sources. However, observations collected in the Solar System reveal that the emitted auroral radio power Pr averaged over time, frequency and solid angle is approximately proportional to both the incident kinetic Pkin and the Poynting PB fluxes23,141:

$${P}_{mathrm{r}}=alpha {P}_{mathrm{kin}}=alpha {rho }{v^3}uppi {{R}_{{rm{obs}}}}^{2}$$
(1)
$${P}_{mathrm{r}}=beta {P}_{mathrm{B}}=beta frac{{B}_{perp }^{2}}{mu }vuppi {{R}_{{rm{obs}}}}^{2},$$
(2 )

where ρ is the density of th e wind incident on the planet’s magnetosphere, Robs is the effective size of the obstacle, v is the incident flow velocity in the obstacle’s frame, B⊥ is the magnetic field in the flow perpendicular to v and μ is the permeability of space. The proportionality factors α and β are the overall efficiencies of the conversion of Pkin and PB into the emitted radio power, which are empirically estimated to be on the order of 10−5 and 10−3, respectively, from Solar System observations49,62. When auroral and satellite–Jupiter radio emissions are considered together, only equation (2) holds. Its high efficiency (β ≈ 10−3) explains why the Jupiter–Io interaction is the brightest object in the sky below 40 MHz, outshining even the quiet Sun.

The detailed magnetohydrodynamics of SPI models lead to a wide range of estimates of the Poynting flux. For sub-Alfvénic interactions we can have (1) force-free models, in which the electric current and magnetic field vectors are parallel224 or (2) Alfvén wing models61,225 that include pairs of standing waves in the magnetized wind, which may separately connect the planet either to the star or to interplanetary space.

For the Alfvén wing model, the power transmitted in one Alfvén wing was first estimated as PAW = ϵPB, where (epsilon ={(1+{M}_{mathrm{A}}^{-2})}^{-1/2}). MA is the Alfvén Mach number, which is defined as the plasma flow speed divided by the Alfvén velocity. For small MA, ϵ ≈ MA (refs. 23,49). For reference, MA = 0.3 ± 0.1 for Io’s interaction226. Based on a nonlinear solution of Alfvén wings56, the total power can be expressed in closed form for small MA as61:

$$begin{array}{r}{P}_{mathrm{AW}}=2uppi {R}^{2}{bar{alpha }}^{2}{M}_{mathrm{A}}frac{{B}_{perp }^{2}}{{mu }_{0}}{v}_{0} .end{array}$$
(3)

Here (bar{alpha }) is the interaction strength ((0le bar{alpha }le 1)) and R is the radius of the central Alfvén tube carrying the energy away along the background field. It is often approximated by the effective radius of the obstacle61. (bar{alpha }) is expected to be close to one if the planet possesses a dense atmosphere/ionosphere. This Alfvén wing expression is widely used60,161,227, and is a more exact solution than the early estimate ~MAPB presented in ref. 23. The energy flux in equation (3) is a factor of two larger as than that presented in ref. 23 as it considers the energy fluxes inside and outside the central Alfvén tube defined by (uppi {R}_{mathrm{obs}}^{2}) (ref. 61). The dependence on MA also highlights that low-density stellar winds will have a lower amount of power emitted than high-density stellar winds.

In force-free models of SPI powered by reconnection, the energy fluxes have been estimated as PFF = γPB with an efficiency factor 0 ≤ γ ≤ 1 (ref. 224). These fluxes are of roughly the same order as those in the Alfvén wing models. In force-free models of the Jupiter–Io interaction, such as the unipolar inductor, the energy fluxes are controlled by the conductance of Jupiter’s ionosphere and not by the wave impedance of the Alfvén waves200,228.

One commonly used form of a force-free model is called the stretch-and-break model229,230, in which the planetary polar magnetic field BP is stretched and energy is released by reconnection. In this case the power produced is:

$$begin{array}{rc}{P}_{mathrm{SB}}&=2uppi {R}_{mathrm{P}}^{2}frac{{B}_{mathrm{P}}^{2}}{{mu }_{0}}{v}_{0}{f}_{mathrm{AP}},end{array}$$
(4)

where RP is the radius of the planet and fAP is the area fraction of the planet where magnetic field lines connect to the stellar wind. fAP might have a typical value on the order of 0.1 depending on the strength and configuration of the magnetic fields of the planet and wind. The direct occurrence of the planet’s internal magnetic field BP is the primary difference between the Alfvén and stretch-and-break model. It would mean that PSB/PAW can be 100–1,000 (ref. 230). However, the Poynting flux in equation (4) is primarily perpendicular and not directed along the flux tube towards the star, and the flow velocities v0 over the poles will also be reduced. The fluxes in equation (4) have not been seen in numerical simulations227 or at Jupiter’s magnetized moon Ganymede. We suggest that further work on the applicability of equation (4) is necessary if the stretch-and-break model is to be used for predictions.

A general prediction from all of these models is that the brightest radio emission from SPIs will be produced by the largest planet as close as possible to a star with the strongest magnetic field. This heuristic can inform future SPI searches but needs to be used with caution, as such a heuristic neglects potentially important second-order effects, such as ECM inhibition due to inflated ionospheres132,163,164,165,166.

Stellar wind environments around planet-hosting stars

Whether or not radio emission from SPIs occurs (as well its morphology and strength) depends on the plasma environment that the planet is embedded in. We require detailed knowledge of this plasma if we are to predict and interpret signatures of SPIs. For low-mass stars, the primary source of the interplanetary plasma is its stellar wind, which expands outwards from the stellar surface, probably via a combination of magnetic and thermal pressure forces231. Additional sources may further add to the interplanetary environment, such as atmospheric mass loss from the planet itself232 and CMEs233.

Unlike our Sun, we lack detailed knowledge of the stellar wind environments around other low-mass stars234. This is primarily because their winds are very tenuous, implying that they do not produce strong observational signatures. However, sophisticated magnetohydrodynamic (MHD) models can be deployed to obtain three-dimensional snapshots of the winds of low-mass stars. Measurements of the mass loss rates of the winds of low-mass stars are currently limited to a handful of cases, and rely on indirect methods (see refs. 13,235,236). Nevertheless, this has not stopped trends relating mass loss rates to more readily available information such as surface X-ray fluxes from being established117. These trends are in turn used to inform MHD models.

Another key constraint that can be implemented in MHD models of the winds of low-mass stars is reconstructed surface magnetic field maps. Since the inception of the Zeeman–Doppler imaging method41, such data have become readily available for a relatively large population of low-mass stars42,208. The magnetic field is embedded in the stellar wind and strongly influences the dynamics of the flow135. The magnetic field of the stellar wind is also essential in predicting and interpreting signatures of SPIs for at least two reasons. First, it is necessary to estimate the location of the Alfvén surface and thus predict whether the planet has a sub- or super-Alfvénic orbit237. Second, the local magnetic field is also required to estimate the Poynting flux reaching the planet, which dictates the strength of the interaction161. By including such maps as boundary conditions in MHD simulations of stellar winds, we can construct models of the plasma environment around the star. These can then be used to estimate quantities relevant to SPIs.

Outlook

We expect that the emerging trend of claimed radio emission from stellar CMEs, SPIs and planetary aurora from LOFAR, the Five-hundred-meter Aperture Spherical Telescope, the Giant Metrewave Radio Telescope, the Karl G. Jansky Very Large Array and the Australian Square Kilometre Array Pathfinder will become a flood in the next decade, as the expected construction and commissioning of the much more sensitive Square Kilometre Array (SKA)238 and ngVLA are completed. Scaling from the sample of LOFAR-detected stars would suggest that SKA1-Low (~50–350 MHz) could detect ~103 M dwarf systems34,239 and SKA1-Mid (~350 MHz–15.4 GHz) and ngVLA (1.2–116 GHz) will be sensitive even to quiescent radio emission from the nearest stars240. The relatively higher frequencies of SKA1-Mid/ngVLA (~1–2 GHz) will probably only probe stellar magnetic fields via SPIs, whereas the lower frequencies of LOFAR/SKA-Low could directly detect auroral emission from Jupiter-mass and larger planets. To trace aurora from lower-mass planets (such as an Earth twin), observations will need to be conducted from space as the Earth’s ionosphere reflects emission ≲10 MHz back into space. In the long term, low-frequency exoplanet science will require radio interferometers on the far side of the Moon241,242, the best location in the Solar System for avoiding terrestrial radio frequency interference—provided that lunar satellites and missions do not pollute that pristine location by the time such telescopes can be established.

The new ground-based radio facilities will need to be matched by both improved theoretical models of stellar plasma environments and multiwavelength observations that independently constrain stellar activity and exoplanet companions. One of the most important questions that MHD modelling could begin to answer is the extent to which sub-Alfvénic SPIs affect planetary atmospheric mass loss and habitability.

Better statistical robustness is also needed in the field before the detection of radio periodicity from SPIs can be claimed, given that the emission mechanism is inherently variable and often dependent on unknown stellar magnetic field properties that can vary in time243. The expected periodicity can also have a complicated relationship with the orbital and rotational periods46. We suggest that the following rule of thumb should be adopted in the field: at least three detections of a stellar system at the same phase, preferably distinct from the rotational phase of the star, should be required before a reliable claim of the detection of SPIs can be made. Such a standard would prevent the premature declaration of a detection of radio emission from an SPI, with off-phase observations also vital to ensure that the alignment of detections at a set phase is not due to a biased sampling window. No claim of SPIs in the literature has met this standard.

The detection of these variable and periodic signals may be facilitated by methods with greater statistical sophistication. For example, using Gaussian processes244 or generalized periodograms245 could potentially aid in a reliable determination of periodicity from heavily undersampled datasets. A firm attribution of observed radio emission to SPIs also depends critically on knowledge of the stellar magnetic field, implying that it will be essential to complement SKA facilities with dedicated facilities for contemporaneous spectropolarimetric monitoring of interesting targets.

As radio emission from SPIs may even be biased against transiting or edge-on systems46, it will be necessary to continue precise radial velocity surveys to determine the existence and orbits of exoplanets inferred from the radio. Red-sensitive and near-infrared instruments (such as SPIRou246, the Habitable Planet Finder40 and CARMENES39) will be especially important to probe the population around M dwarfs where the Alfvén surface encompasses a large fraction of planetary systems and reaches into habitable-zone orbits. As well as stabilized spectrographs for Zeeman–Doppler imaging and precise radial velocities, short-cadence time series photometry is essential to detect stellar flares. Photometry by itself can determine a stellar flare rate and can help distinguish between SPIs and coronal emission107, but simultaneous photometry is ideal as correlations between some radio bursts and optical flares have been observed32.

Considering the significant commitment of observatory resources required to determine radio detections of SPIs, it is important that the field invests wisely in the most likely candidates. To first order, it is possible to identify the systems that are most likely to produce SPI radio emission based on the size of the exoplanet, its proximity to its host star and the distance of the stellar system from Earth121,247. In this case, the top five candidate systems are 51 Pegasi, HIP 65 A, Tau Boötis A, 55 Cancri A and WASP-18—many of which have been searched for radio signatures already.

However, many of these stars probably do not have strong enough magnetic fields to produce ECM emission at ≳150 MHz, where our most sensitive telescopes currently operate. Such stars are also likely to have complicated magnetic field topologies, implying that radio emission may not be consistently beamed to Earth. Therefore, planets close to M dwarfs may be the best systems to follow up first, considering that M dwarfs can possess kilogauss-strong dipolar magnetic fields. GJ 367, GJ 436, GJ 1252, GJ 3253, GJ 625, YZ Ceti, AU Mic and Proxima Centauri could be ideal candidates in such a case, many of which have been searched33,154. The main issue with focusing on M dwarfs for SPI radio signatures is that the star itself is also known to produce radio emission. As discussed above, this implies that a robust detection of SPIs will probably involve ≳200 h on radio telescopes, given that the planets will have an ~1–10 d orbital period.

Regardless of the difficulties faced in determining radio emission from SPIs, exoplanets and stars, the potential scientific return is invaluable. Radio observations of stellar systems can provide information about the planet, star and space weather that are not directly attainable at any other wavelength. Determining the potential habitability of exoplanets will be a focus of astronomy in the coming decades—and radio astronomy is poised to provide key pieces of this puzzle.