--- title: 'Exomoons: Natural Satellites of Exoplanets' url: https://www.emergentmind.com/topics/exomoon type: topic --- # Exomoons: Natural Satellites of Exoplanets An **exomoon** is a natural satellite orbiting an exoplanet or another planetary-mass body outside the Solar System. Exomoons are inferred through their gravitational effects on host-planet transits, direct photometric or spectroastrometric signatures, radio modulation, phase-curve contributions, or Doppler motion of the host planet. No exomoon has been confirmed unambiguously; Kepler-1625 b-i and Kepler-1708 b-i remain candidates supported by different combinations of transit timing and photometric evidence [1810.02362; 2201.04643]. Their scientific importance extends from satellite formation and tidal evolution to planetary interiors, climate dynamics, and habitability. ## 1. Physical definition and dynamical architecture An exomoon is a gravitationally bound satellite of an exoplanet. In the standard hierarchical approximation, the planet and moon form a local two-body system whose barycentre follows an orbit around the star. The planet and moon separately orbit their common barycentre, so the planet executes a reflex motion that can be observed even when the moon itself does not measurably transit the star [1105.3189]. The principal parameters are the stellar, planetary, and satellite masses, $M_*$, $M_P$, and $M_S$; stellar and planetary radii, $R_*$ and $R_P$; the planet–star or planet–moon-barycentre semimajor axis, $a_P$; the satellite semimajor axis about the barycentre, $a_S$; the corresponding periods, $P_P$ and $P_S$; eccentricities, $e_P$ and $e_S$; inclinations, $i_P$ and $i_S$; arguments of periapsis, $\omega_P$ and $\omega_S$; and the satellite longitude of ascending node, $\Omega_S$. A useful orientation combination is $$ \varpi_S\equiv\omega_P+\Omega_S. $$ For a satellite within the Hill sphere, the nested two-body approximation gives $$ a_S=\mathfrak D\,a_P\left(\frac{M_P}{3M_*}\right)^{1/3}, $$ where $\mathfrak D$ is the satellite distance in Hill-radius units. Under $M_*\gg M_P\gg M_S$, Kepler’s third law gives $$ P_S=P_P\,\mathfrak D^{3/2}3^{-1/2}. $$ Thus, for a bound moon with $\mathfrak D<1$, $P_S<0.57735P_P$. Transit observations sample the satellite only once per planetary orbit, so the lunar signal is undersampled and generally aliased. A satellite orbit must lie between an inner disruption boundary and an outer stellar-perturbation boundary. The Roche limit is approximately $$ R_R\simeq2.44R_P\left(\frac{\rho_P}{\rho_M}\right)^{1/3}, $$ where $\rho_P$ and $\rho_M$ are planetary and lunar mean densities. The outer region is commonly represented as a fraction of the Hill radius, $$ R_H=a_P\left(\frac{M_P}{3M_*}\right)^{1/3}, $$ with stable prograde orbits often approximated by $a_M\lesssim\chi R_H$, where $\chi$ is commonly taken near $1/3$, or more conservatively near $1/4 [1704.01688]. A more refined prograde limit used in tidal-evolution calculations is approximately $$ a_{M,\max}=0.4895R_H \left(1.0000-1.0305e_P-0.2738e_M\right) $$ [1707.07040]. The permissible region may be narrow or absent in compact systems. ## 2. Transit timing, duration, and radius signatures The principal transit method exploits the planet’s motion around the planet–moon barycentre. The planet’s barycentric displacement is $$ a_{PB}=\frac{M_S}{M_P}a_S $$ for $M_S\ll M_P$. The resulting displacement along the transit direction changes the measured planetary transit time, producing a transit-timing variation (TTV). Its characteristic scaling is $$ \delta_{\rm TTV}\propto M_Sa_S. $$ Consequently, a more massive moon farther from the planet can produce the same TTV as a less massive moon on a closer orbit. TTV alone therefore has a mass–distance degeneracy and can be mimicked by additional planets, resonances, Trojan bodies, apsidal or nodal precession, starspots, stellar binarity, orbital decay, and other effects [1105.3189]. A moon also changes the planet’s projected velocity across the stellar disk. The associated velocity-induced transit-duration variation, TDV-V, scales approximately as $$ \delta_{\rm TDV-V}\propto M_Sa_S^{-1/2}, $$ because $P_S\propto a_S^{3/2}$. TTV is therefore principally a position signal, whereas TDV-V is a velocity signal. Their amplitude ratio approximately determines the lunar period: $$ \eta\equiv \frac{\delta_{\rm TDV-V}}{\delta_{\rm TTV}} \simeq \frac{\tilde T_B}{2\pi P_S} $$ for a nearly circular moon, where $\tilde T_B$ is the barycentric transit-duration parameter. The period yields $a_S$ through Kepler’s third law, after which either amplitude constrains $M_S$. In the circular, coplanar limit, $$ \Lambda_{\rm TTV}=\cos f_S,\qquad \Lambda_{\rm TDV-V}=\sin f_S, $$ so the two signals are displaced by $$ \Delta\phi=\frac{\pi}{2}. $$ This position–velocity phase relation is a distinctive dynamical diagnostic, although eccentricity, inclination, photometric contamination, and other perturbations can distort it. The inclination-induced duration component, TDV-TIP, arises from changes in the apparent impact parameter. It scales as $$ {\rm TDV-TIP}\propto M_Sa_S, $$ is usually smaller than TDV-V, and becomes important for inclined geometries and near-grazing transits. In the proposed framework, TDV-TIP can distinguish prograde from retrograde motion because it is in phase with TDV-V for prograde orbits and in antiphase for retrograde orbits, although the required signal-to-noise is substantially higher [1105.3189]. When a planet-only model is fitted to a planet–moon transit, three apparent indicators emerge: TTV, TDV, and transit-radius variation (TRV). A moon’s photometric contribution can alter the fitted planetary radius, especially when the planet and moon overlap or transit at widely separated positions. TRVs are primarily sensitive to lunar radius and projected geometry rather than lunar mass. Their apparent period can be half the moon’s orbital period because the projected planet–moon separation repeats after half an orbit [2004.02259]. The traditional TTV–TDV ellipse arises only when barycentric effects dominate and both signals remain approximately sinusoidal: $$ \left(\frac{\rm TTV}{A_{\rm TTV}}\right)^2+ \left(\frac{\rm TDV}{A_{\rm TDV}}\right)^2=1. $$ Low-density moons can produce substantial photometric distortions while generating relatively weak barycentric motion, destroying the ellipse. Failure to observe an ellipse therefore does not exclude a moon. TRVs, autocorrelation functions, and alias-aware periodograms may provide useful first-pass candidate-selection tools. ## 3. Photometric and spectroscopic detection methods A direct lunar transit may be too shallow to detect, particularly for a small moon orbiting a giant planet. Full photodynamical modeling instead combines the planet’s and moon’s orbital motion, transits, mutual occultations, limb darkening, timing shifts, duration changes, and transit-shape distortions. The open-source Python code **Pandora** implements an analytical limb-darkened transit model, nested Keplerian dynamics, mutual planet–moon eclipses, finite-exposure supersampling, and Bayesian inference compatibility [2205.09410]. Pandora models the planet–moon barycentre on a circumstellar orbit and the individual bodies on a local Keplerian orbit. It uses exact Mandel–Agol transit calculations where necessary, small-body approximations for $R_{\rm occ}/R_*<0.01$, hybrid interpolation at intermediate size ratios, and lookup tables for fixed limb-darkening parameters. The reported approximations maintain errors below approximately 1 ppm in the tested regimes. Demonstrations used four transits of a Jupiter–Neptune system on a one-year orbit, 10-minute cadence, 100-ppm white noise, and a photometrically quiet Sun-like star. The parameters were recovered with the UltraNest Bayesian sampler. The **exomoon corridor** is a population-level TTV diagnostic caused by aliasing. Because the moon’s orbit is sampled once per planetary transit, many physical lunar periods map to apparent TTV periods of approximately 2–4 planetary epochs: $$ P_{\rm TTV}= \frac{1}{ \left| \frac{1}{P_S}- \frac{{\rm round}(P_P/P_S)}{P_P} \right|}. $$ N-body simulations show that this corridor persists for systems containing up to five moons, including resonant and non-resonant chains [2106.13421]. The relationship is not a unique inversion of $P_S$ and does not prove a moon, but it can prioritize systems for follow-up. Increasing the number of moons reduces stability and generally requires lower total satellite mass ratios or finely tuned architectures. The simulations found stable fractions of approximately 99.8%, 92.1%, 23.2%, 7.4%, and 1.9% for fixed-host systems containing one through five moons, respectively. These are survival fractions for the simulated architectures, not occurrence rates. Other indirect methods exploit radiation or spectroscopy. **Phase-curve spectral contrast** compares multiwavelength phase curves in a planet-dominated band and a moon-enhanced band. Ordinary reflected-light moons are generally too faint, whereas strongly tidally heated or self-luminous moons may dominate at wavelengths beyond a few microns. Many plausible detections require photometric precision of approximately 10 ppm or better [1705.05203]. **Spectroastrometry** measures the wavelength-dependent photocentre of an unresolved planet–moon pair. If the planet dominates at one wavelength and the moon dominates at another, the centre of light shifts toward the corresponding body: $$ \mathbf c(\lambda)=\frac{\mathbf r_P}{d}+f_M(\lambda)\boldsymbol\beta. $$ Repeated centroid measurements can trace the projected lunar orbit, determine the combined planet–moon mass, and help separate their spectra. Simulations of an Earth–Moon analogue around $\alpha$ Cen A recovered a combined mass of $1.03\pm0.12\,M_\oplus$ under idealized assumptions. The technique requires broad spectral coverage, particularly to approximately $3\,\mu{\rm m}$, where strong water bands can produce moon-dominated channels [1509.01615]. For directly imaged planets, a sufficiently massive moon induces a Doppler reflex signal in the host planet. The planetary radial-velocity semiamplitude is $$ K_P= \left[ \frac{2\pi G}{P_M(1-e_M^2)^{3/2}} \frac{M_M^3\sin^3i_M}{(M_P+M_M)^2} \right]^{1/3}. $$ For $M_P\gg M_M$, $$ K_P\propto M_M\,M_P^{-2/3}\,P_M^{-1/3}\sin i_M. $$ A Neptune-mass moon around a $10M_J$ planet with a period near 1.8 days would induce approximately $200\,{\rm m\,s^{-1}}$, while similar moons around Jupiter-mass planets could produce signals near $900\,{\rm m\,s^{-1}}$ [1805.01903]. Planetary rotation, illumination-induced velocity shifts, atmospheric activity, peak-pulling from moonlight, and disk clumps are important contaminants. A radio method extrapolates from the Jupiter–Io interaction. A conducting moon moving through a planetary magnetosphere generates a motional electromotive force and Alfvén wings. Field-aligned currents accelerate electrons near the planetary poles, producing electron-cyclotron maser emission. The characteristic frequency is $$ f_C=\frac{eB_{\rm pole}}{2\pi m_e}. $$ The proposed signal is periodic or phase-dependent modulation of planetary radio emission. Under Jupiter–Io-like assumptions, large moons around nearby giant planets could produce detectable signals at tens of megahertz, but the method depends strongly on uncertain magnetic fields, plasma densities, beaming, efficiencies, and observing geometry [1308.4184]. ## 4. Candidate systems and observational status Kepler-1625 b-i was proposed as a candidate after a Hubble Space Telescope epoch showed a planetary transit approximately 77.8 minutes earlier than predicted and a post-transit flux decrement of approximately 500 ppm [1810.02362]. A self-consistent photodynamical model favored a planet–moon interpretation over planet-only, free-TTV, and zero-radius-moon models. The inferred satellite was generally Neptune-sized, with model-dependent radii of approximately $3.1$–$4.9\,R_\oplus$, periods near 22–24 days, and a separation of approximately $36$–$45R_P$. The candidate interpretation remains unresolved. The revised Kepler analysis weakened the earlier evidence, the photometric signal depended on HST visit-long detrending, and an external planet could reproduce the timing variation. The formal Bayes factor was conditional on model choices, priors, and instrumental-systematics treatment. Confirmation requires repeated, phase-consistent moon transits, planetary TTVs, TDVs, and a stable orbital solution. Kepler-1708 b-i was identified in a survey of 70 cool giant Kepler candidates [2201.04643]. Kepler-1708 b is a Jupiter-sized planet with $P_P\simeq737$ days, $a_P\simeq1.64$ AU, and an approximately Sun-like host. The candidate moon has $$ R_S/R_P=0.263^{+0.040}_{-0.042}, $$ or approximately $2.6R_\oplus$, with an inferred period of $4.6^{+3.1}_{-1.8}$ days and separation of approximately $11.7R_P$. The photometric evidence yielded $\Delta\chi^2=23.2$, approximately $4.8\sigma$ as a likelihood-based quantity, and a planet–moon to planet-only Bayes factor of 11.9. Eight detrending reductions produced consistent fit improvements, and injection recovery gave a source-specific false-positive estimate of $1.0^{+0.7}_{-1.0}\%$. This candidate also remains unconfirmed. Only two planetary transits were observed, the inferred satellite mass is constrained only by an upper limit, and the evidence depends on sparse long-cadence photometry and model assumptions. A second transiting planet, stellar activity, residual correlated noise, or instrumental effects cannot be excluded completely. Future observations must reproduce the predicted approximately 500-ppm moon transit and any associated 1.2–77-minute TTV signal. Isolated planetary-mass objects provide another observational context. Their lack of stellar glare permits precision infrared photometry without high-contrast suppression. Transit probabilities for close moons can be approximately 5%–25%, and transiting-moon systems are estimated to occur around 10%–15% of isolated planetary-mass objects under the stated assumptions [2108.08323]. JWST/NIRSpec forecasts indicate that at least 30–33 of 57 known or candidate isolated planetary-mass objects could permit single-transit detection of Io-like or Titan-like moons. A fading event in the unresolved binary 2MASS J1119-1137 AB had a depth of approximately 0.53%–0.63%, duration near 36–38 minutes, and an equivalent radius of approximately $1.6$–$1.7R_\oplus$. It was compatible with a transiting habitable-zone moon but also with intrinsic atmospheric variability. The event occurred only once, appeared in one infrared band, and was therefore inconclusive. ## 5. Habitability and climate Exomoon habitability depends on stellar irradiation, planetary reflected light, planetary thermal emission, eclipses, tidal heating, atmospheric retention, mass, composition, and orbital evolution. A conservative energy-budget criterion compares the globally averaged absorbed radiative flux plus tidal heating with a runaway-greenhouse threshold: $$ F_{\rm RG}>\bar F_S^{\rm glob}. $$ For an Earth-like moon, the adopted threshold is approximately $295\,{\rm W\,m^{-2}}$; for a $0.25M_\oplus$ Super-Ganymede it is approximately $266\,{\rm W\,m^{-2}}$ [1209.5323]. These values identify configurations in which runaway greenhouse is not forced by the model; they do not establish liquid water, a stable atmosphere, or life. Planetary illumination has reflected and thermal components, both declining approximately as $a_{PS}^{-2}$. Eclipses can substantially reduce stellar illumination, particularly for low-inclination moons. Tidal heating is highly sensitive to orbital distance and eccentricity. At small eccentricity, $$ h_S\propto M_P^2R_S^3a_{PS}^{-9}e_{PS}^2. $$ Consequently, moving a moon inward or increasing its eccentricity can transform a potentially temperate satellite into an Io-like volcanic body or a runaway-greenhouse world. Resonances with additional moons can maintain eccentricity after isolated two-body tides would otherwise circularize the orbit. Mass and atmospheric retention impose additional constraints. Approximate lower mass requirements discussed for robust habitability include $0.1M_\oplus$ for a long-lived magnetic field, $0.12M_\oplus$ for a substantial atmosphere, and $0.23M_\oplus$ for tectonics and a carbon–silicate cycle. A practical lower limit near $0.25M_\oplus$ is suggested, while an upper limit near $2M_\oplus$ is motivated by increasing interior pressure and viscosity that could suppress convection, dynamos, and plate tectonics. These are model-dependent thresholds rather than universal boundaries. Climate models emphasize that synchronous rotation with the planet produces a fixed planetary illumination pattern but a moving stellar day. A general-circulation model of a $0.25M_\oplus$ moon around a $10M_J$ planet found that planetary infrared heating is absorbed primarily in the upper atmosphere, strengthening subplanetary convection, poleward energy transport, polar warming, and upper-atmospheric humidity [1806.06822]. Moderate planetary heating and geothermal flux can produce polar amplification without ice–albedo feedback. Strong planetary illumination of $500\,{\rm W\,m^{-2}}$ generated modeled stratospheric humidities of $3.3\times10^{-3}$ and $1.4\times10^{-2}$ and outgoing longwave radiation of $398.3$ and $482.2\,{\rm W\,m^{-2}}$, conditions associated with moist-greenhouse and runaway-like behavior. Low-mass stellar systems are particularly restrictive. Their habitable zones lie close to the star, forcing circumplanetary orbits to be compact. Coupled dynamical and tidal simulations found that stars with $M_*\lesssim0.2M_\odot$ are very unlikely to host habitable prograde exomoons in their habitable zones. Stellar perturbations may remain consequential up to approximately $0.5M_\odot$, depending on planetary mass and location [1707.07040]. Close moons can spiral inward rapidly when the planet is synchronized to the star; Saturn-like cases may reach the Roche limit in less than $10^7$ years, while Jupiter-like cases can have estimated lifetimes below approximately 200 Myr for Io-like initial orbits. Compact planetary systems can lack dynamically viable satellites altogether. In the TRAPPIST-1 system, planets b through e cannot host moons of reasonable density under conservative reduced-Hill criteria, while outer planets have more room but remain vulnerable to resonances, tidal evolution, and planetary perturbations [1704.01688]. A moonless planet is not necessarily uninhabitable, but the absence of a substantial moon may affect obliquity evolution, rotational history, oceanic tides, and tidal-pool environments. ## 6. Tidal evolution, formation, and scientific significance Tidal evolution can drive a moon outward beyond the stable planetary region or inward toward the Roche limit. In a constant-time-lag or constant-phase-lag framework, the sign of migration depends on the relative planetary spin and lunar mean motion. If $\Omega_P>n_S$, the moon generally migrates outward; if $\Omega_P