Lunar Gravitational Wave Antenna (LGWA)

• LGWA is a mission concept that uses the Moon as an elastic antenna to detect gravitational waves through precise lunar seismometers and inertial sensors.
• The design employs cryogenic superconducting sensors achieving fm/√Hz sensitivity in the decihertz band, enabling long-duration and multiband observations.
• Advanced methodologies like layered lunar response modeling and seismic noise mitigation are key to improving detection accuracy and cosmological insights.

The Lunar Gravitational-wave Antenna (LGWA) is a proposed array of next-generation inertial sensors designed to monitor the response of the Moon to gravitational waves. In the mission studies, the Moon is treated as an elastic antenna and the readout is provided by seismic stations placed in permanently shadowed polar regions; in the payload description, the mission aims to detect the differential between the elastic response of the Moon and the motion of a suspended inertial-sensor proof mass induced by gravitational waves. Given the size of the Moon and the expected lunar seismic background, LGWA is intended to operate from about $1\ \mathrm{mHz}$ to $1\ \mathrm{Hz}$, thereby occupying the band between space-borne detectors such as LISA and future terrestrial detectors such as Einstein Telescope and Cosmic Explorer (Ajith et al., 2024, Heijningen et al., 2023).

1. Concept and measurement principle

LGWA differs from conventional interferometric observatories in that it does not measure an optical arm-length change between free masses. Its operating principle is to use the Moon itself as the test body: a passing gravitational wave excites elastic deformations of the lunar interior and surface, and precision inertial sensors measure the resulting ground motion. In the long-wavelength regime, the displacement field can be written as

$$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$

with $T_r(f)$ and $T_h(f)$ the radial and horizontal response functions per unit strain. The detector output is then a projection of this displacement field onto the local sensing axis (Bi et al., 2024).

In the seismometer description adopted for LGWA source studies, the observed strain is modeled as

$$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$

and, for a single seismometer in the long-wavelength approximation,

$$\stackrel{\leftrightarrow}{d} = \vec{e}_n \otimes \vec{e}_1,$$

where $\vec{e}_n$ is the local surface normal and $\vec{e}_1$ is the measured displacement direction. This formulation makes explicit that LGWA is fundamentally a lunar-response detector: the astrophysical strain is filtered by the elastic response of the Moon before it is recorded by the sensor (Song et al., 5 Feb 2025).

The response changes across frequency. At low frequencies, the description is naturally expressed in terms of global normal modes. Above about $0.1\ \mathrm{Hz}$, horizontally layered half-space models become computationally efficient and capture the role of near-surface geology, scattering, and shear-wave reflection. This division between normal-mode and layered-response regimes is central to present LGWA modeling (Bi et al., 2024).

2. Payload, stations, and inertial sensing technology

The baseline mission concept uses four stations in a permanently shadowed region at one lunar pole, and one commonly used implementation assumes that each station carries two horizontal lunar inertial GW sensors measuring two orthogonal horizontal displacements. The observation strategy is therefore an array measurement rather than a single-point experiment (Sharma et al., 2 Jun 2026, Song et al., 5 Feb 2025).

The payload development is centered on a cryogenic superconducting inertial sensor. The payload paper states that such a sensor “aims for fm/rtHz sensitivity or better down to 1 Hz” and is planned to be deployed in seismic stations (Heijningen et al., 2023). In the dedicated sensor study, the Cryogenic Superconducting Inertial Sensor is described as the key enabling instrument for LGWA, with a modeled displacement sensitivity at $1\ \mathrm{Hz}$0 “of 3 orders of magnitude better than the current state-of-the-art” and a modeled sensitivity of $1\ \mathrm{Hz}$1 at $1\ \mathrm{Hz}$2 (Badaracco et al., 2022).

The sensor architecture is a monolithic niobium Watt’s linkage carrying a $1\ \mathrm{Hz}$3 proof mass, operated at about $1\ \mathrm{Hz}$4, with superconducting magnetic actuators based on the Meissner effect and two readout systems: a cryogenic interferometric readout for ultra-high sensitivity and a Rasnik optical system for large dynamic range. The cryogenic niobium design is intended to achieve a mechanical quality factor of order $1\ \mathrm{Hz}$5, while the use of superconducting materials suppresses eddy-current losses that limited earlier room-temperature versions (Badaracco et al., 2022).

Mission-lifetime assumptions vary across analyses. One IMBH detectability study adopts a 10-year mission lifetime and truncates inspiral signals that would otherwise remain in band longer, using

$1\ \mathrm{Hz}$6

This long observation window is one of the defining instrumental advantages of LGWA for slowly evolving sources (Song et al., 5 Feb 2025).

3. Lunar response modeling, seismic background, and calibration limits

Accurate prediction of LGWA sensitivity requires accurate prediction of the lunar transfer function. A major development in this direction is the extension of Freeman Dyson’s half-space model to horizontally layered geologies, which allows “computationally efficient calculations of the lunar GW response above $1\ \mathrm{Hz}$7” and shows that “modifications of the geological model as required to explain Apollo seismic observations can boost the lunar GW response” (Bi et al., 2024). In the layered calculations, low near-surface shear-wave speeds and deep shear-reflecting structures can increase horizontal response markedly; a plausible implication is that shallow lunar structure is not a secondary correction but part of the detector calibration.

Seismic background is not merely an engineering nuisance but a potential fundamental limitation. In the analytic mitigation study, LGWA is modeled as an array of accelerometers in an isotropic, random, Gaussian seismic field, and the resulting optimization shows that mitigation depends critically on station spacing relative to the seismic-correlation length. For two stations, “optimal placement of the two stations can yield significant improvements in the equivalent seismic noise amplitude spectrum density (ASD), approximately a factor of 2.3 at $1\ \mathrm{Hz}$8, compared to the measurement with a single station” (Yan et al., 27 Apr 2026).

At the same time, the current literature contains an explicit sensitivity caution. An IMBH study that used the original LGWA power spectral density also examined an updated PSD based on a more realistic lunar-response calculation; in that update, sensitivity around $1\ \mathrm{Hz}$9 is worse by about two orders of magnitude, and the IMBH horizon redshift is reduced from $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$0 to $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$1 (Song et al., 5 Feb 2025). This is not a contradiction in the literature so much as a reflection of model dependence. This suggests that LGWA science forecasts remain coupled to unresolved questions in lunar interior structure, scattering, and near-surface response.

4. Sensitivity band and principal source classes

Published LGWA analyses use slightly different working bands. The mission study frames the detector as a $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$2 to $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$3 observatory (Ajith et al., 2024); one cosmology forecast uses $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$4–$$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$5 (Sharma et al., 2 Jun 2026); one dispersion forecast uses $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$6–$$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$7 (Praveen et al., 30 Apr 2025); and the IMBH study describes LGWA as operating “from 1 mHz to a few Hz, with optimal sensitivity in the decihertz band” (Song et al., 5 Feb 2025). A plausible implication is that the quoted frequency range is analysis-dependent and tied to the adopted sensitivity curve and waveform integration limits rather than to a single immutable design number.

Intermediate-mass black hole binaries are a recurring flagship science case. In the IMBH detectability study, LGWA is “more sensitive to nearby binaries ($$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$8) with the primary mass $$\vec{\xi}(\hat{r}) = 2 T_h(f)\,\mathbf{h}\cdot\hat{r} + \bigl(T_r(f) - 2T_h(f)\bigr)\,(\hat{r}\cdot \mathbf{h}\cdot \hat{r})\,\hat{r},$$9, while it prefers distant binaries ($T_r(f)$0) with $T_r(f)$1.” With an SNR threshold of 10, the original sensitivity implies detectability “up to $T_r(f)$2.” The same study finds relative primary-mass errors $T_r(f)$3 for $T_r(f)$4, redshift errors $T_r(f)$5 for $T_r(f)$6, and sky localization within $T_r(f)$7 for binaries with $T_r(f)$8 at $T_r(f)$9 (Song et al., 5 Feb 2025).

Stellar-mass and massive binary black holes are also central. In the population study of BBHs up to GWTC-4.0, LGWA “alone would have been able to observe more than one third of the events detected so far,” and in simulated populations it “could detect $T_h(f)$0 events merging in the ground-based band per year out to redshifts $T_h(f)$1” (Iacovelli et al., 10 Dec 2025). The characteristic strain formalism used throughout these studies is standard: $T_h(f)$2 and matched-filter detection is based on

$T_h(f)$3

These definitions are important because the LGWA science case is fundamentally a long-duration inspiral case rather than a short merger case (Song et al., 5 Feb 2025).

5. Multiband astronomy, localization, and parameter inference

LGWA’s most distinctive astrophysical role is as a multiband detector. The BBH parameter-estimation study finds that “the short time to merger from the deci-Hz band to the Hz-kHz band (typically months to a year) allows for early warning, targeted follow-up, and archival searches,” and that current detectors at design sensitivity “could detect thousands of BBHs per year, with one to a few hundred multiband counterparts in LGWA.” Third-generation detectors can observe most of the BBHs detected by LGWA in the simulated mass range $T_h(f)$4, enabling systematic joint analyses (Iacovelli et al., 10 Dec 2025).

Massive binaries illustrate the measurement regime particularly clearly. For a GW231123-like system, the same study states that the source accumulates “$T_h(f)$5 inspiral cycles in LGWA,” and that deci-Hz observations “can measure the chirp mass even better than 3G instruments and yield good sky localization and inclination measurement, even with a single observatory” (Iacovelli et al., 10 Dec 2025). A separate geometry study, using GW250114 as a case study, sharpens that statement: “Two minutes before its merger, the LGWA would have measured its chirp mass to a precision of 0.0002 solar masses (90% symmetric) and constrained its sky position to within 65 square degrees (90% HPD area); these constraints are tighter than those obtained by the LIGO-Virgo-KAGRA detectors, despite a lower signal-to-noise ratio” (Tissino et al., 3 Jun 2026).

The same geometry work argues that long-duration LGWA inference “must be treated as a geometrical problem,” because detector motion, reference-frame choice, and signal evolution jointly determine both parameter constraints and computational efficiency. In that study, adopting a frame comoving with the Solar System barycenter but choosing an origin that minimizes timing uncertainty reduces the sampling time “by an order of magnitude” (Tissino et al., 3 Jun 2026). This is technically important: for LGWA, orbital geometry is not a small correction to a static detector response but part of the inference problem itself.

At the population level, LGWA+ET forecasts for IMBH binaries show that LGWA is strongest for high-mass IMBH mergers while ET is strongest for low-mass IMBH binaries, and that the joint network “possesses strong detection capabilities across the full IMBH mass spectrum” and can “significantly and effectively recover the IMBH population distributions” (Dong et al., 14 Jul 2025). Here the role of LGWA is not merely additional SNR but access to the part of inspiral that neither LISA-like mHz detectors nor hecto-Hz ground detectors observe optimally.

6. Cosmology, stochastic backgrounds, and unresolved issues

LGWA has been proposed as a cosmology instrument in several distinct senses. For stochastic backgrounds, a dedicated lunar-seismometer study finds that a future LGWA network could constrain a flat stochastic background to

$T_h(f)$6

at SNR $T_h(f)$7, for a 1-year integration over $T_h(f)$8–$T_h(f)$9 (Yan et al., 2024). That result is specifically framed as better than previous constraints in the mid-frequency band.

For dark-siren cosmology, a forecast with Taiji, LGWA, and Einstein Telescope concludes that the three-band network outperforms all two-detector configurations considered there. In flat $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$0CDM, the TJ-LGWA-ET network constrains the Hubble constant and matter density to $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$1 and $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$2, respectively; in $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$3CDM, a 4-year dark-siren sample alone constrains $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$4 to $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$5, improving to $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$6 when BAO and Type Ia supernova data are added (Song et al., 13 Mar 2026). In that framework, LGWA is the decihertz bridge that improves both luminosity-distance precision and sky localization.

A more restrictive bright-siren curvature forecast reaches a subtler conclusion. In a network with two Cosmic Explorers and Einstein Telescope, adding LGWA does not significantly improve $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$7, because “the additional signal-to-noise ratios accumulated in their bands are modest.” Yet the same study finds that LGWA improves sky localization substantially: the median 90% credible sky area changes from $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$8 for 2CE+ET to $$h = h_{+} \,\stackrel{\leftrightarrow}{e}_{+} : \stackrel{\leftrightarrow}{d} + h_{\times} \,\stackrel{\leftrightarrow}{e}_{\times} : \stackrel{\leftrightarrow}{d},$$9 for 2CE+ET+LGWA (Sharma et al., 2 Jun 2026). This distinction matters. Direct per-event cosmological improvement can be small in one Fisher setup even when the practical ability to identify hosts and realize bright sirens improves sharply.

Fundamental-physics studies make a related point about the decihertz band itself. In a multiband dispersion analysis, CE-ET+LGWA+LISA can probe the effective-theory energy scale of modified-gravity scenarios with a precision of approximately $$\stackrel{\leftrightarrow}{d} = \vec{e}_n \otimes \vec{e}_1,$$0 using only $$\stackrel{\leftrightarrow}{d} = \vec{e}_n \otimes \vec{e}_1,$$1 high-SNR events, and the authors explicitly identify “the operation of the gravitational wave detector in the deci-Hertz frequency band” as the key ingredient (Praveen et al., 30 Apr 2025).

The unresolved issues are correspondingly clear. LGWA forecasts remain sensitive to the adopted lunar response model, the final realized power spectral density, the degree to which seismic background can be mitigated, the realism of multiyear lunar deployment, and, in cosmological applications, the depth and completeness of host-galaxy catalogs (Song et al., 5 Feb 2025, Yan et al., 27 Apr 2026, Song et al., 13 Mar 2026). This suggests that LGWA should be understood not as a single settled instrument specification but as a mission concept whose final astrophysical reach is inseparable from lunar geophysics, cryogenic inertial sensing, and array-level data analysis.