Anomalous Microwave Emission (AME): Insights

• AME is a dust-correlated microwave emission peaking near 20–30 GHz, primarily attributed to rapidly spinning ultrasmall grains.
• It is identified by a broad spectral bump in the 10–60 GHz range, discerned through advanced SED construction and component separation techniques.
• Observational studies and modeling techniques, including polarization limits, help distinguish AME from other ISM emissions like free–free and synchrotron.

Anomalous microwave emission (AME) is a dust-correlated excess of microwave radiation observed between roughly 10 and 60 GHz, or more broadly 10 and 100 GHz, that cannot be accounted for by the canonical radio-to-millimetre components of the interstellar medium: synchrotron emission, thermal free–free emission, or vibrational thermal dust emission. In total intensity it appears as a broad microwave “bump” in the spectral energy distribution (SED), typically peaking near 20–30 GHz, and it has been detected in diffuse Galactic clouds, molecular complexes, photodissociation regions, some H II-region environments, and a small number of external galaxies. The leading interpretation is electric dipole radiation from rapidly rotating ultrasmall grains (“spinning dust”), while magnetic dipole emission and other dust-physics alternatives remain under investigation (Vidal et al., 25 Jun 2026, Dickinson et al., 2018).

1. Spectral identity and observational phenomenology

AME was first detected in 1996 and recognized as a distinct dust-correlated component in 1997, after microwave data showed an excess that could not be reproduced by synchrotron, free–free, or thermal dust alone (Dickinson et al., 2018). Its defining observational property is a convex or broad bump-like spectrum: it rises through the centimetre band, usually peaks in flux density at about 20–40 GHz, and then declines toward higher frequencies, unlike synchrotron and optically thin free–free, which are power laws, or thermal dust, which rises steeply above about 100 GHz (Collaboration et al., 2011).

In standard foreground language, the neighbouring components have well-defined scalings. Optically thin free–free is nearly flat in flux density, $S_\nu \propto \nu^{-0.1}$, synchrotron is a falling power law, typically $S_\nu \propto \nu^{-\alpha}$ with $\alpha \approx 0.7$–1, and thermal dust is described by a modified blackbody, $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$ (Dickinson et al., 2018). AME departs from all three by showing spectral curvature across the 10–60 GHz window (Dickinson, 2013).

The most common Galactic peak frequencies lie near 20–30 GHz, although higher values are observed in some environments. In the Planck cloud sample, fitted spinning-dust peaks lay predominantly at 20–35 GHz, with one clear outlier, the California Nebula, at about 50 GHz (Collaboration et al., 2013). In the 2025 cloud compilation of 144 Galactic sources, the peak-frequency distribution had a median $\nu_{\rm AME} = 22.4$ GHz and a Gaussian mean of 21.9 GHz with standard deviation 3.7 GHz, while the Galactic-plane QUIJOTE analysis found a median $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$ GHz for significant pixels (Cepeda-Arroita et al., 6 Oct 2025, Fernández-Torreiro et al., 2023).

AME is not a minor curiosity in total intensity. On the Galactic plane it contributes about half of the total intensity around 30 GHz, and in robust Galactic cloud detections the AME fraction at 28–30 GHz is often large (Vidal et al., 25 Jun 2026, Collaboration et al., 2013). This makes it both an ISM diagnostic and a practical foreground for microwave cosmology.

2. Physical mechanisms and carrier models

The canonical spinning-dust picture treats AME as electric dipole radiation from ultrasmall grains with permanent dipole moments rotating at GHz frequencies. For a grain with dipole moment $\mu$ rotating at angular frequency $\omega$, the emitted power is

$$P(\omega) = \frac{2}{3}\frac{\mu^2 \omega^4}{c^3}, \qquad \nu = \frac{\omega}{2\pi}.$$

The ensemble emissivity is obtained by integrating over the grain-size distribution and rotational-state distribution (Bell et al., 2015, Dickinson et al., 2014):

$$j_\nu = \int n(a)\int f(\omega\,|\,a)\,\frac{2}{3}\frac{\mu(a)^2 \omega^4}{c^3}\, \delta\!\left(\nu - \frac{\omega}{2\pi}\right)\, d\omega\, da.$$

This framework makes the AME spectrum sensitive to grain size, dipole moments, charge state, gas density, gas temperature, ionization fraction, and radiation-field strength. In typical cold neutral medium conditions the peak is near 30 GHz, while changes in density, radiation field, and the small-grain population can shift both peak and width (Dickinson et al., 2014, Vidal et al., 25 Jun 2026).

Spinning-dust calculations are implemented in codes such as SPDUST, SPDUST2, and SpyDust. These map environmental parameters to predicted emissivity curves and show that the observed spectrum can broaden or shift when multiple phases coexist along the line of sight (Vidal et al., 25 Jun 2026). This is consistent with recent large samples in which many observed AME spectra are broader than the standard single-phase predictions (Cepeda-Arroita et al., 6 Oct 2025).

Magnetic dipole emission remains the principal alternative mechanism. In this class of models, thermal fluctuations in the magnetization of ferromagnetic or ferrimagnetic nanoparticles—metallic Fe, magnetite, maghemite, or Fe inclusions—produce microwave radiation. The emissivity depends on the dissipative part of the magnetic susceptibility, and the polarization properties can differ sharply from spinning dust; free-flying aligned magnetic grains can in principle reach much higher polarization fractions than spinning dust (Dickinson et al., 2014, Vidal et al., 25 Jun 2026).

A more specialized alternative is the cosmic amorphous dust model, in which AME is attributed to resonance transitions of two-level systems in amorphous carbon grains. In that formulation, the AME band is produced by the microwave dielectric response of amorphous carbon, while submillimetre polarization is carried primarily by amorphous silicates (Nashimoto et al., 2020). This model remains a minority interpretation, but it illustrates that AME is also a problem in the solid-state physics of interstellar grains, not only in rotational dynamics.

3. SED construction and component separation

Observationally, AME is identified by building broadband radio-to-infrared SEDs and subtracting the standard components. A common decomposition is

$S_\nu \propto \nu^{-\alpha}$0

with terms for free–free, synchrotron, thermal dust, CMB anisotropy, and spinning dust (Collaboration et al., 2013). In H II-region work the same logic is often written without the explicit CMB term, as a sum of free–free, synchrotron, thermal dust, and AME (Dickinson, 2013).

For free–free, the relevant complication is the transition between optically thin and optically thick regimes. A widely used expression is

$S_\nu \propto \nu^{-\alpha}$1

with optical depth approximated by

$S_\nu \propto \nu^{-\alpha}$2

so that the spectrum rises as $S_\nu \propto \nu^{-\alpha}$3 in the optically thick limit and approaches $S_\nu \propto \nu^{-\alpha}$4 in the optically thin limit (Dickinson, 2013). Thermal dust is modelled as a modified blackbody,

$S_\nu \propto \nu^{-\alpha}$5

or equivalently $S_\nu \propto \nu^{-\alpha}$6 in Rayleigh–Jeans notation (Dickinson, 2013).

Large statistical studies generally smooth all maps to a common beam, perform aperture photometry, and fit the resulting SEDs. The Planck Galactic-cloud analysis used 1°-smoothed maps from 0.408 to 3000 GHz and found that 42 of 98 candidate sources had significant $S_\nu \propto \nu^{-\alpha}$7 excess emission between 20 and 60 GHz; after filtering likely ultracompact-H II contamination, 27 remained as a robust AME sample (Collaboration et al., 2013). More recent work has tightened spectral constraints by adding low-frequency anchors from S-PASS, C-BASS, and QUIJOTE, together with parametric MCMC fitting (Cepeda-Arroita et al., 6 Oct 2025).

A practical empirical model now often used for AME is a log-Gaussian or log-normal profile,

$S_\nu \propto \nu^{-\alpha}$8

which captures the peak frequency and spectral width without committing to a unique physical environment (Cepeda-Arroita et al., 6 Oct 2025, Fernández-Torreiro et al., 2023). This has proved especially useful in large samples, where line-of-sight averaging broadens real spectra relative to single-phase templates.

Coverage below about 20 GHz is critical. In the Galactic plane, not including QUIJOTE-MFI data leads to underestimation of the AME signal by up to 50% in favour of free–free emission (Fernández-Torreiro et al., 2023). In specific regions such as S140, the addition of Green Bank Telescope measurements between 4 and 8 GHz sharply reduced the degeneracy between spinning dust and optically thick free–free, yielding a spinning-dust peak at $S_\nu \propto \nu^{-\alpha}$9 GHz if the AME interpretation is adopted (You et al., 2018).

4. Galactic environments and empirical trends

The best-established AME detections are in dusty, extended Galactic environments rather than in compact classical H II cores. Planck’s early detailed spectra of the Perseus and $\alpha \approx 0.7$0 Ophiuchi clouds showed AME detections at 17.1$\alpha \approx 0.7$1 and 8.4$\alpha \approx 0.7$2, respectively, with peaks in the 20–40 GHz range and SEDs well fitted by spinning-dust models (Collaboration et al., 2011). The Pleiades reflection nebula yielded a 17.7$\alpha \approx 0.7$3 residual AME detection at 22.8 GHz, with COSMOSOMAS upper limits at 10.9 and 14.7 GHz that pinned down the low-frequency side of the spectrum (Genova-Santos et al., 2011).

At cloud scale, AME regions are typically more extended than non-AME regions. In the 98-source Planck sample, deconvolved 28.4 GHz sizes were larger for AME sources than for non-AME sources, and the AME regions tended to be associated with cooler dust, $\alpha \approx 0.7$4–20 K, compared with 20–27 K for non-AME regions; their mean dust emissivity index was also slightly higher, $\alpha \approx 0.7$5 versus 1.72 (Collaboration et al., 2013). The same work found a clear anti-correlation between AME emissivity and column density, $\alpha \approx 0.7$6, consistent with depletion of the smallest grains in denser material (Collaboration et al., 2013).

Recent larger samples have reinforced and refined these trends. In 144 Galactic clouds, the AME amplitude correlated most strongly with the thermal dust peak flux and dust radiance, with about 30% scatter and sublinear scaling. The same sample found that the AME peak frequency increases with thermal dust temperature, with a linear slope of $\alpha \approx 0.7$7 GHz K$\alpha \approx 0.7$8, and that the median spectral width is $\alpha \approx 0.7$9, broader than single-phase spinning-dust predictions (Cepeda-Arroita et al., 6 Oct 2025). On degree scales along the Galactic plane, QUIJOTE found $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$0 correlated with $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$1 and with the radiation-field proxy $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$2, while the AME emissivity normalized by $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$3 also rose with $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$4 (Fernández-Torreiro et al., 2023).

The $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$5 Orionis ring provides a spatially resolved example of environmental control. Combining QUIJOTE and C-BASS with ancillary maps, AME was detected at $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$6 around the ring, with a radial decrease in peak frequency from about 35 GHz near the free–free region to about 21 GHz in the outer ring. The peak frequency correlated strongly with emission measure and significantly with dust temperature, while the AME amplitude normalized by optical depth correlated with the radiation field (Cepeda-Arroita et al., 2020). This is the first detection of AME spectral variations across a single region at degree scales (Cepeda-Arroita et al., 2020).

H II regions remain a special case because free–free confusion is severe. Classical H II regions are dominated by warm-ionized-gas free–free below about 100 GHz and warm dust above that, and compact ultracompact or hypercompact H II regions can remain optically thick into the 15–40 GHz range if their emission measures reach $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$7 (Dickinson, 2013). Consequently, several H II-associated AME claims are modest in significance and vulnerable to reinterpretation. W40 and LPH96, for example, were revised downward after improved analyses, and robust high-significance AME within classical compact H II regions remains uncommon (Dickinson, 2013).

5. Carriers, polarization, and the emissivity problem

The identity of the carrier is not settled. Several regional studies support a link between AME and the smallest carbonaceous grains, but they do not point to a single universally dominant tracer. In $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$8 Orionis, hierarchical Bayesian dust SED modelling showed that AME correlates more strongly with PAH mass than with total dust mass, supporting a spinning-PAH interpretation in that region (Bell et al., 2019). At arcminute scales in the Perseus cloud, however, the microwave–infrared correlation peaked at 24 $$I_{\nu,\mathrm{dust}} \propto \nu^\beta B_\nu(T_{\rm d})$$9m rather than 8 $\nu_{\rm AME} = 22.4$0m, suggesting stochastically heated very small grains rather than PAHs as the dominant emitters there (Tibbs et al., 2013).

All-sky analyses complicate the picture further. Using the Planck Commander AME map, Hensley and collaborators found that the best predictor of AME strength is the dust radiance, not the PAH-abundance proxy, and that fluctuations in AME intensity per dust radiance are uncorrelated with PAH abundance. They argued that this result casts doubt on spinning PAHs as the dominant source of Galactic AME and warrants renewed consideration of ultrasmall silicates or other carriers (Hensley et al., 2015). AKARI analyses of Perseus and $\nu_{\rm AME} = 22.4$1 Ophiuchi likewise concluded that, once an approximate interstellar-radiation-field correction is applied, the 30 GHz morphology follows the 100 $\nu_{\rm AME} = 22.4$2m structure at least as well as the 9 $\nu_{\rm AME} = 22.4$3m PAH tracer, especially in complex star-forming environments (Bell et al., 2015).

Polarization provides one of the strongest constraints on mechanism. Spinning-dust emission is expected to be weakly polarized because ultrasmall grains align poorly, whereas free-flying magnetic nanoparticles can reach much larger polarization fractions (Dickinson et al., 2014, Vidal et al., 25 Jun 2026). Observationally, the upper limits are stringent: QUIJOTE reported $\nu_{\rm AME} = 22.4$4 at 28.4 GHz for $\nu_{\rm AME} = 22.4$5 Oph, $\nu_{\rm AME} = 22.4$6 at 28.4 GHz for Perseus, and $\nu_{\rm AME} = 22.4$7 at 33 GHz for W43, while diffuse-sky analyses gave $\nu_{\rm AME} = 22.4$8 at 95% confidence under synchrotron priors (Vidal et al., 25 Jun 2026). These limits are compatible with weakly polarized spinning dust and strongly disfavor highly polarized free-flying magnetic nanoparticles as the dominant AME source.

Another long-standing issue is how to define “AME emissivity.” A widely used quantity is the ratio of 1 cm AME intensity to 100 $\nu_{\rm AME} = 22.4$9m dust intensity, often expressed in $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$0. This normalization is strongly temperature biased because 100 $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$1m lies near the Wien side of the dust SED. For fixed AME intensity, the 100 $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$2m intensity increases by a factor of about 11.1 between dust temperatures of 20 K and 30 K, so the nominal AME emissivity decreases by the same factor (Tibbs et al., 2012). This bias is especially acute near H II regions, where dust is commonly warm. The recommended alternatives are normalization by dust optical depth, dust column density, or $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$3, all of which are closer to the physics of a grain-related process (Tibbs et al., 2012, Dickinson, 2013).

6. Extragalactic AME and future observational regimes

Extragalactic AME is much rarer than Galactic AME, but it is not absent. The strongest integrated detection is in M31, where Sardinia Radio Telescope C-band observations combined with WMAP, Planck, and infrared data yielded an AME component of $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$4 Jy peaking near 25 GHz. In the integrated SED of M31, synchrotron dominates below 10 GHz, AME overtakes synchrotron and free–free between about 20 and 50 GHz, and thermal dust dominates above 60 GHz (Battistelli et al., 2019). Localized detections are also reported in star-forming regions of NGC 6946 and in NGC 4725 (Vidal et al., 25 Jun 2026).

Non-detections are also informative. K-band Sardinia observations of four nearby spirals—NGC 3627, NGC 4254, NGC 4736, and NGC 5055—yielded only upper limits, but these limits remained consistent with an average extragalactic AME emissivity of $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$5, or equivalently $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$6 (Bianchi et al., 2022). The implication is that AME may be common but strongly diluted in global galaxy SEDs unless radio emission is relatively weak compared with dust.

The next major step is high-fidelity low-frequency mapping. SKA1-MID Band 5 covers 4.6–13.8 GHz, with Bands 3 and 4 supplying lower-frequency leverage, and the full SKA is expected to extend toward about 24 GHz (Dickinson et al., 2014). Updated SKAO forecasts give Band 5a at 4.6–8.5 GHz and Band 5b at 8.3–15.4 GHz, with representative continuum sensitivities of $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$7 in 1 hour at arcsecond resolution and $\nu_{\rm AME} = 20.7^{+2.0}_{-1.9}$8 in a 100-hour Band 5b polarization observation (Vidal et al., 25 Jun 2026). These data are expected to constrain the low-frequency rise, spectral width, and polarization of AME in PDRs, disks, and galaxies at angular resolutions where beam mixing becomes much less severe.

For galaxies near and far, the SKA regime is not uniform. Modelling of SKA observations indicates that thermal free–free dominates the radio continuum of distant galaxies near 10 GHz, with negligible AME contribution, whereas high-angular-resolution observations of nearby galaxies that resolve individual star-forming regions may require multi-frequency measurements to avoid bias from localized AME when inferring star-formation rates from a single band (Yoon et al., 24 Jun 2026). This suggests a bifurcated future for AME studies: large-area surveys will constrain its role as a foreground and statistical dust diagnostic, while targeted high-resolution observations will test carrier physics directly through morphology, spectral curvature, and polarization.

Taken together, current evidence supports AME as a real and widespread microwave component whose dominant manifestation is dust-related, weakly polarized, and spectrally peaked. The empirical case for spinning dust is strong, but the identity of the dominant carrier population, the degree of contribution from alternative mechanisms, and the environmental origin of the observed peak-frequency and width variations remain active research problems (Cepeda-Arroita et al., 6 Oct 2025, Vidal et al., 25 Jun 2026).