Low-Luminosity AGN: Accretion & Feedback

• Low-luminosity AGN are supermassive black holes with bolometric luminosities ≤10^42–10^43 erg s⁻¹ and low Eddington ratios, powered by radiatively inefficient accretion flows such as ADAF.
• They exhibit unique multiwavelength signatures including hard X-ray spectra, strong compact MIR/NIR emissions, and elevated CO(2–1)/CO(1–0) ratios, setting them apart from classical AGN.
• Their feedback is dominated by kinetic outflows and shocks that modify the circumnuclear ISM, influencing nuclear star formation and host galaxy evolution.

Low-luminosity active galactic nuclei (LLAGN) are accreting supermassive black holes (SMBHs) characterized by bolometric luminosities orders of magnitude below those found in luminous Seyferts, quasars, or classical AGN, typically with $L_\mathrm{bol} \lesssim 10^{42{-}43}$ erg s⁻¹ and Eddington ratios $$\lambda_\mathrm{Edd} \equiv L_\mathrm{bol}/L_\mathrm{Edd} \sim 10^{-6}$$–$10^{-2}$. LLAGN constitute the dominant population of AGN in the local universe and are found in diverse host environments, from late-type disks to massive ellipticals, spanning SMBH masses from $10^4$ to $10^{9}\,M_{\odot}$. Their low luminosities arise from low accretion rates and/or radiatively inefficient accretion modes, with spectral energy distributions, emission line diagnostics, and feedback mechanisms distinct from those in high-luminosity AGN.

1. Defining Characteristics and Accretion Physics

LLAGN are most robustly defined in terms of their Eddington-scaled accretion power and (rest-frame) X-ray and bolometric luminosities. Typical criteria include:

• Bolometric luminosity $L_\mathrm{bol} \lesssim 10^{42{-}43}$ erg s⁻¹ and $L_\mathrm{2-10\,keV} \lesssim 10^{40{-}41}$ erg s⁻¹ (Dumont et al., 2019, Annuar et al., 2020, Asmus et al., 2011).
• Eddington ratio $\lambda_\mathrm{Edd} \sim 10^{-6}$–$10^{-2}$ (Lusso et al., 24 Feb 2025, Zhang et al., 2022).

The physics of accretion in LLAGN is dominated by radiatively inefficient accretion flows (RIAFs), such as advection-dominated accretion flows (ADAF) (Nemmen et al., 2011). In these systems, low-density, hot, optically thin gas flows onto the SMBH, often exhibiting mass-loss via winds, and a significant portion of gravitational energy is advected or converted into kinetic power rather than radiated. The accretion structure is typically a truncated thin disk at radii $R_\mathrm{tr} \sim 30{-}300\,R_S$, with an inner ADAF and an outflowing relativistic jet (Nemmen et al., 2011, Yan et al., 12 Mar 2025).

Radiative efficiency drops to $\eta_\mathrm{rad} \sim 10^{-4}$–$10^{-3}$, and kinetic efficiency in jets $\eta_\mathrm{jet} \sim 10^{-4}$–$10^{-2}$, with jet kinetic power often exceeding bolometric luminosity ($P_\mathrm{jet} \gtrsim L_\mathrm{bol}$) (Nemmen et al., 2011). The corona, responsible for X-ray emission in luminous AGN, is generally weak or absent at the lowest $\lambda_\mathrm{Edd}$ (Yuk et al., 29 Jan 2025).

2. Multiwavelength Signatures: Spectral Energy Distribution and Emission Components

The SEDs of LLAGN deviate systematically from those of classical AGN (Dumont et al., 2019, Mason et al., 2012, Gilbert et al., 3 Aug 2025, Nemmen et al., 2011):

• The X-ray power-law continuum is often hard (photon index $\Gamma_X \sim 1.3$–2.3), lacking the “big blue bump” of the standard thin disk (Lusso et al., 24 Feb 2025).
• The radio-to-mm regime is dominated by jet synchrotron emission, with parabolic-to-conical jet morphologies seen in VLBI imaging (Yan et al., 12 Mar 2025).
• Strong, compact nuclear mid-infrared (MIR, $10{-}13\,\mu$m) and near-infrared (NIR, $2{-}3\,\mu$m) emission is detected in most LLAGN, correlating tightly with X-ray luminosity ($\log\,L_\mathrm{MIR} = \alpha + \beta\,\log\,L_X$ with $\beta\sim1.1$) (Asmus et al., 2011, Mason et al., 2012, Dumont et al., 2019). The MIR–X-ray relation appears independent of $\lambda_\mathrm{Edd}$ down to $\sim10^{-5}$ (Asmus et al., 2011).
• At the lowest Eddington ratios ($\lambda_\mathrm{Edd} \lesssim 10^{-4}$), NIR emission can exceed X-ray output, with $L_K/L_X$ rising toward low $\lambda_\mathrm{Edd}$ (Dumont et al., 2019).
• Strong silicate emission is present even when the canonical torus is weak or absent, indicative of optically thin, possibly clumpy dust geometries or low dust-to-gas ratios (Mason et al., 2012).
• Star-formation contributions to nuclear NIR/MIR flux are generally minor within tens of parsecs of the SMBH (Asmus et al., 2011, Mason et al., 2012), though some high-resolution SEDs show NIR excesses from young nuclear star clusters.

3. Emission Line Diagnostics and Molecular Gas Content

LLAGN are challenging to identify via standard optical emission-line diagnostics, especially in galaxies with strong star formation. Conventional BPT diagrams ([O III]/Hβ vs. [N II]/Hα) tend to miss LLAGN embedded in rapidly star-forming disks, as H II regions dominate the spectrum (Bär et al., 2016). More sensitive indices such as He II $\lambda4685$/Hβ are required to uncover AGN photoionization in high-SFR hosts, yielding a $\sim22\%$ increase in faint AGN identifications (Bär et al., 2016). X-ray confirmation (power-law spectra with $\Gamma_X \sim 1.6$–2.1, $L_{2-10}\sim 10^{40-41}$ erg s⁻¹) substantiates the AGN nature in weak-lined cases, ruling out pure starburst origins.

In molecular gas studies, LLAGN hosts show:

• CO(1–0) and CO(2–1) emission detected in nuclei, though integrated intensities and total gas masses ($$M_\mathrm{H_2}\sim 2\times10^5{-}5\times10^7\,M_\odot$$) are modest (Boeker et al., 2011).
• Elevated CO(2–1)/CO(1–0) line ratios ($R_{21}\gtrsim1.8$) in face-on LLAGN, compared to $R_{21}\sim0.9$ in quiescent spirals—a signature of AGN-induced molecular gas excitation, not explained by beam filling (Boeker et al., 2011).
• A scarcity of dense molecular gas (n(H$_2$) $\lesssim 2\times10^3$ cm⁻³), as evidenced by upper limits on HCN and HCO$^+$ emission, distinguishing LLAGN from powerful AGN with dense star-forming cores (Boeker et al., 2011).

4. Variability, Duty Cycle, and Identification Techniques

LLAGN identification via amplitude statistics of nuclear variability is highly effective. Optical and X-ray variability selection has uncovered populations missed by X-ray, MIR, and line-ratio surveys (Yuk et al., 29 Jan 2025, Ding et al., 2018, Young et al., 2012, Yuk et al., 2022). In high-cadence time-domain optical surveys, variability-selected LLAGN compose $\sim$2% of local bright galaxies for $\lambda_\mathrm{Edd}\sim10^{-3}$ (Yuk et al., 29 Jan 2025, Yuk et al., 2022), with a spectroscopic AGN confirmation rate of $\sim60$%. These sources often exhibit soft X-ray spectra (mean $\Gamma\sim2.4$) and are rarely MIR- or X-ray-selected (Yuk et al., 29 Jan 2025).

X-ray variability metrics (normalized excess variance, power spectral density break scaling as $f_b\propto M_\mathrm{BH}^{-1}\lambda_\mathrm{Edd}$) show that LLAGN variability amplitudes plateau at low luminosity, reproducing observed anti-correlation breaks in $\sigma^2_X$–$L_X$ relations (Ding et al., 2018, Young et al., 2012).

“Changing-look” LLAGN, such as NGC 4614, demonstrate transitions in optical type and X-ray flux coincident with large drops in $\lambda_\mathrm{Edd}$, providing evidence for state transitions (thin disk to RIAF) and the disappearance of broad line regions at critical low accretion rates (Lusso et al., 24 Feb 2025). Faded changing-look quasars at $L_\mathrm{bol}/L_\mathrm{Edd} \lesssim 10^{-3}$ show SED evolution consistent with the low/hard states of X-ray binaries, indicating physical continuity in accretion flow transformation across mass scales (Gilbert et al., 3 Aug 2025).

5. Circumnuclear Environment and Feedback

LLAGN exhibit distinct feedback and ISM-processing signatures compared to their higher-luminosity cousins. Multiwavelength observations resolve:

• Powerful parsec-scale mass outflows and jets, producing fast (v$_\mathrm{out}\sim 200$–$500$ km s⁻¹), shock-excited, coronal line—emitting gas. In NGC 1386, the outflow rate is $\dot{M}_\mathrm{out} \sim 11\,M_\odot$ yr⁻¹, with kinetic power $\dot{E}_\mathrm{kin} \sim 10^{41}$ erg s⁻¹, often $\gtrsim5$\% of $L_\mathrm{bol}$ (Rodríguez-Ardila et al., 2017).
• Nuclear feedback operates in the “kinetic mode”: LLAGN with $\lambda_\mathrm{Edd} \sim 10^{-6}$–$10^{-3}$ exhibit strong shocks, which dominate the excitation of both ionized and molecular phases on $\lesssim100$ pc scales, as established by JWST and Spitzer spectroscopy of diagnostic fine-structure lines ([Ne V], [O IV]) and H$_2$ bands (Zhang et al., 2022, Zhang et al., 12 Jan 2026).
• The kinetic feedback imprints on the ISM: small PAH molecules are destroyed—evidenced by suppressed PAH $6.2/7.7$ and enhanced $11.3/7.7$ band ratios, with most surviving PAH population consisting of large, neutral molecules ($N_C\gtrsim200$) (Zhang et al., 2022, Zhang et al., 12 Jan 2026).
• H$_2$ rotational transitions indicate slow, jet-driven molecular shocks at velocities $\leq 10$ km s⁻¹, incompatible with pure radiative (PDR or XDR) excitation. A few percent of the jet mechanical power suffices to power the observed H$_2$ emission (Zhang et al., 12 Jan 2026).
• Compared to Seyferts and quasars, which manifest “radiative mode” UV/X-ray feedback and strong suppression of all PAH bands, LLAGN display a kinetic–mechanical regime that directly modifies their immediate molecular environment (Zhang et al., 2022).

6. Demographics, Host Properties, and Environmental Dependence

LLAGN are prevalent in both bulge and disk galaxies across the local universe. Narrow-line, low-mass LLAGN (FWHM Hβ < 2200 km s⁻¹) show sub-Eddington accretion, weak Fe II, strong optical variability, but are physically distinct from classical NLSy1 galaxies, forming a “low-tail” in 4DE1 parameter space and representing a low-accretion analogue within the broader narrow-line AGN family (Sharma et al., 5 Dec 2025).

Variability- and emission-line-selected LLAGN are found in diverse environments, but modern statistical surveys report a $\sim3\times$ higher fraction of LLAGN in galaxy clusters relative to the field, contrasting with the tendency for luminous AGN to avoid clusters (Yuk et al., 29 Jan 2025). Proposed triggering mechanisms at low $\lambda_\mathrm{Edd}$ include tidal interactions with the cluster potential, cooling flows, and minor mergers rather than major mergers or violent gas inflows.

Multiwavelength studies indicate that nuclear star formation is suppressed in LLAGN hosts, linked to low dense-gas fractions and mechanical heating by weak jets or AGN-driven feedback (Boeker et al., 2011). The overall LLAGN fraction in volume-limited surveys is $\sim2\%$ at $\lambda_\mathrm{Edd}\gtrsim 2\times10^{-3}$, constraining the local AGN duty cycle and black-hole growth scenarios (Yuk et al., 2022).

7. Future Prospects and Observational Diagnostics

The unexpected strength of nuclear K-band and MIR emission in LLAGN, alongside tight correlations with X-ray luminosity, highlight the efficacy of IR selection, particularly with JWST/NIRSpec and MIRI (Dumont et al., 2019). JWST enables spatially resolved identification of hot-dust AGN emission and separation from stellar backgrounds down to $L_K \sim 10^{37}$–$10^{40}$ erg s⁻¹ at distances up to tens of Mpc.

PAH band ratios and H$_2$ excitation provide a robust mid-IR diagnostic to distinguish kinetic-mode feedback in LLAGN from radiative-mode processes in more powerful AGN (Zhang et al., 2022, Zhang et al., 12 Jan 2026). The employment of variability-based, emission-line, and multiwavelength selection, coupled with high-resolution IFU spectroscopy, is essential to correct for misclassification of Compton-thick, obscured, or host-dominated LLAGN and to avoid significant underestimates in low-$L_X$ AGN space density (Lambrides et al., 2020, Ding et al., 2018).

The transition between radiative- and kinetic-dominated feedback as a function of $\lambda_\mathrm{Edd}$, the survival and dissipation timescale of the dusty torus, the structure of accretion flows, and the nature of accretion state changes remain central topics for future theoretical and observational work—areas in which deep, high-angular-resolution, multi-epoch IR and X-ray spectroscopy will be decisive.

---

Selected Key Quantitative Results and Correlations in LLAGN:

Observable or Ratio	LLAGN Regime	Typical Value/Trend	Reference
$L_\mathrm{bol}$	LLAGN	$\lesssim 10^{42{-}43}$ erg s⁻¹	(Dumont et al., 2019, Annuar et al., 2020)
$\lambda_\mathrm{Edd}$	LLAGN	$10^{-6}$–$10^{-2}$	(Lusso et al., 24 Feb 2025, Zhang et al., 2022)
$L_K/L_X$	NIR-to-X ratio ($<10^{-4}$)	$L_K$ exceeds $L_X$ as $\lambda_\mathrm{Edd}$ drops	(Dumont et al., 2019)
MIR–X-ray correlation	log–log slope across $10^{40{-}45}$ erg/s	$B\approx1.12$; no dependence on $\lambda_\mathrm{Edd}$	(Asmus et al., 2011)
Silicate ($S_{9.7}/N_H$)	High in high $S_{9.7}$ LLAGN	$0.5$–$20\times10^{-23}$ cm² (order mag above Seyferts)	(Mason et al., 2012)
$$R_{21}=\mathrm{CO}(2\text{–}1)/\mathrm{CO}(1\text{–}0)$$	LLAGN (face-on disks)	$R_{21}\gtrsim1.8$ ($>99.9\%$ confidence, elevated)	(Boeker et al., 2011)
Dense H$_2$ (HCN/CO ratio)	LLAGN nuclei	$r_{\mathrm{HCN}}<0.06$ (low density)	(Boeker et al., 2011)
Nuclear X-ray photon index	LLAGN (variable)	Mean $\Gamma=1.93\pm0.13$ (soft, unobscured)	(Young et al., 2012, Ding et al., 2018)
BLR radius–luminosity scaling	Down to low $L$	$R_\mathrm{BLR} \propto L^{0.533}$, $\sigma\sim0.13$–0.19 dex	(Bentz et al., 2013)
PAH band ratios	LLAGN (shock-dominated)	$6.2/7.7\downarrow$, $11.3/7.7\uparrow$, $N_C\gtrsim200$	(Zhang et al., 2022, Zhang et al., 12 Jan 2026)
Optical variability, log σ	Low-mass LLAGN	median log σ ≈ –0.68	(Sharma et al., 5 Dec 2025)

---