AGN Unified Model

The AGN zoo and why it cried out for unification

By the late 1970s the catalogue of "active" extragalactic objects had grown into a small zoo. Carl Seyfert had isolated his class of spiral galaxies with bright nuclear emission lines (1943); Maarten Schmidt had identified 3C 273 as a quasi-stellar radio source at z = 0.158 (1963); BL Lacertae had been recognised as a featureless, polarised, violently variable continuum source; Fanaroff and Riley had divided radio galaxies into FR-I and FR-II by their morphology (1974); and intermediate "Seyfert 1.5/1.8/1.9" classifications were proliferating to handle objects whose broad-line strength was somewhere between Type 1 and Type 2.

The taxonomy was descriptive but incoherent. Seyfert 1s had broad permitted lines (full-width 1000–10 000 km/s) and narrower forbidden lines on top of a strong UV continuum. Seyfert 2s had only narrow lines, weaker UV continuum, more X-ray absorption, and (in the radio-loud version) extended radio emission. Quasars looked like very luminous Seyfert 1s. Radio galaxies came in narrow-line (FR-I, NLRG) and broad-line (BLRG) flavours. Blazars varied chaotically and had almost no spectral features. The question that had to be answered: are these intrinsically different things, or the same thing seen differently?

Antonucci 1993 — the unified picture

The synthesis is Robert Antonucci's 1993 Annual Review of Astronomy and Astrophysics article "Unified models for active galactic nuclei and quasars." Its central claim is simple. At the core of every AGN sits a supermassive black hole, typically 10⁶ to 10⁹⁻¹⁰ M☉, accreting through a thin disk that liberates 10⁴⁴–10⁴⁸ erg/s of UV-optical-soft-X-ray luminosity. The disk is encircled by structured gas at several radial scales:

• The broad-line region (BLR). Dense (n_e ~ 10⁹–10¹¹ cm⁻³) photoionised clouds at roughly 10⁻³ pc, orbiting at thousands of km/s. Permitted lines (Hα, Hβ, Mg II, C IV) get broadened by Doppler motion.
• The dusty obscuring torus. A geometrically and optically thick structure of molecular gas and dust at ~1 pc, extending well above and below the disk plane. Its inner edge is set by the dust sublimation radius (T < 1500 K).
• The narrow-line region (NLR). Lower-density (n_e ~ 10³–10⁶ cm⁻³) gas extending out to ~100 pc, photoionised by an unobscured fraction of the central source via the torus opening. Forbidden lines (e.g. [O III] 5007, [N II], [S II]) and narrow permitted components dominate. Velocities are tens to hundreds of km/s.
• (Sometimes) a relativistic jet. Launched perpendicular to the disk in about 10 % of AGN, extending up to megaparsecs in the most powerful sources.

Looking down the polar axis (face-on), the observer sees right through the torus opening into the BLR. Broad and narrow lines are both visible, the UV continuum is direct, and the X-rays are unabsorbed. That is a Type 1 (Seyfert 1 / quasar / broad-line radio galaxy). Looking edge-on, the torus blocks the disk and BLR but the larger NLR pokes out above the torus. Only narrow lines are seen, the UV is suppressed, and the X-rays show heavy obscuration. That is a Type 2 (Seyfert 2 / narrow-line radio galaxy). Looking along the jet, beamed jet emission dominates the spectrum — blazar.

Anatomy at a glance

Component	Scale	Density / state	Signature
Supermassive black hole	r_g ~ 10⁻⁷ pc (10⁸ M☉)	Spin a* ∈ [0, 1]	Sets ISCO, η, jet base
Accretion disk	10⁻⁶ – 10⁻³ pc	Ionised, 10⁴–10⁵ K	Big blue bump, UV continuum
Hot corona	~ 10 r_g	kT_e ~ 10⁹ K, optically thin	2–100 keV X-ray power law
Broad-line region (BLR)	~ 10⁻³ pc	n_e ~ 10⁹–10¹¹ cm⁻³	Broad permitted lines, 10³–10⁴ km/s
Dusty torus	~ 1 pc	Molecular + dust, T < 1500 K	Mid-IR thermal bump, X-ray absorption
Narrow-line region (NLR)	~ 10² pc	n_e ~ 10³–10⁶ cm⁻³	Forbidden lines, 10²–10³ km/s
Relativistic jet	up to Mpc	Γ ~ 5–30, B-field dominated	Radio, beamed γ-rays in blazars

The Type 1 / Type 2 dichotomy in detail

The dichotomy is operationally an emission-line one. Type 1 spectra contain Doppler-broadened permitted lines (Hβ FWHM > ~ 1200 km/s) superposed on narrow forbidden lines. Type 2 spectra contain only narrow components: Hβ FWHM and [O III] FWHM are comparable and modest. Sub-types fill in the gap — Seyfert 1.5 has both broad and narrow components clearly, 1.8 has very weak broad Hα only, 1.9 has broad Hα but no broad Hβ. The progression maps to increasing line-of-sight obscuration.

Class	Viewing angle	Broad lines?	Narrow lines?	X-ray N_H	Jet direction
Seyfert 1 / Type 1 quasar	Face-on (i ≲ 45°)	Yes	Yes	< 10²² cm⁻²	—
Seyfert 1.5	Intermediate	Weak	Yes	~ 10²²	—
Seyfert 1.8 / 1.9	Near torus edge	Very weak (Hα only)	Yes	10²²–10²³	—
Seyfert 2 / Type 2 quasar	Edge-on	Hidden by torus	Yes	10²³–10²⁵ (often Compton-thick)	—
Broad-line radio galaxy (BLRG)	Off-axis	Yes	Yes	Moderate	Side-on jet, double lobes
Narrow-line radio galaxy (NLRG)	Edge-on	Hidden	Yes	Heavy	Side-on jet, double lobes
FSRQ blazar	Jet aligned, lines present	Yes (rest frame)	Yes	Low	Toward us
BL Lac	Jet aligned, lines weak	Weak or absent	Weak	Low	Toward us

The Antonucci-Miller test — NGC 1068, 1985

The single most influential observation supporting the unified model is the spectropolarimetry of the Seyfert 2 NGC 1068 by Antonucci and Miller in 1985. Their key finding: although the total-flux spectrum showed no broad lines, the polarised flux spectrum revealed broad Balmer and Mg II lines at FWHM ≳ 3000 km/s — exactly what a Seyfert 1 looks like.

The geometric interpretation is direct. The torus blocks the direct view of the BLR, but UV photons and broad-line photons escape upward through the polar opening, scatter off electrons and dust above the torus, and arrive at Earth polarised perpendicular to the scattering plane. The polarisation tags those photons as having taken an indirect route, and dispersing the polarised flux reveals the hidden Seyfert 1 sitting behind the torus. Antonucci's 1993 review elevates this from a curious NGC 1068 result to a general organising principle: most Seyfert 2s harbour hidden BLRs, and the differences between Type 1 and Type 2 are mainly orientation, not intrinsic engine differences.

The polarisation argument has been reproduced in many Seyfert 2s and Type 2 quasars since (e.g. Tran 2003 surveys). It remains the textbook evidence for orientation-based unification.

Radio-loud versus radio-quiet — the second axis

About one in ten AGN are radio-loud — they launch a powerful, collimated, relativistic jet that dominates their radio luminosity and, when pointed near the line of sight, the rest of their spectrum too. The radio-loud / radio-quiet division is real and not just orientation; it correlates with black-hole spin, accretion mode, and host morphology. But within each branch, orientation still rules. A radio-loud AGN seen at three different angles produces three classes:

Jet angle to line of sight	Radio-loud → class	Radio-quiet → class
0° (jet at us)	Blazar (FSRQ if lines; BL Lac if not)	Type 1 quasar / Seyfert 1
30°–60° (off-axis)	Broad-line radio galaxy (FR-I core / FR-II)	Seyfert 1 / Seyfert 1.5
~ 90° (edge-on)	Narrow-line radio galaxy	Seyfert 2 / Type 2 quasar

FR-I (Fanaroff-Riley I) sources have lower jet power and edge-darkened morphology; FR-II have higher power and edge-brightened lobes with hot spots. The split correlates with environment and accretion rate.

Blazars, FSRQs, and BL Lacs

Blazars are the most extreme members of the unified zoo. Looking directly down the jet, you see emission Doppler-boosted by factors of D⁴ ~ 10⁴ in luminosity at Γ ~ 10 and i ~ 1/Γ. Variability is rapid (intra-day in many cases), polarisation is high, and the spectrum is a non-thermal two-bump structure: a synchrotron peak in IR-to-X-ray plus an inverse-Compton peak in MeV-to-TeV γ-rays.

Two flavours are distinguished by emission lines:

• FSRQs (flat-spectrum radio quasars). The jet sits in front of a luminous accretion disk and a BLR. Broad emission lines are visible in the rest frame, and the inverse-Compton bump is fed by external radiation (BLR clouds, dusty torus IR). High accretion rate.
• BL Lacs. The jet sits in front of a low-luminosity, low-accretion-rate engine without a strong BLR. Continuum is almost featureless. Inverse-Compton is dominated by synchrotron self-Compton (jet photons scattered by jet electrons themselves).

The "blazar sequence" — first proposed by Fossati et al. (1998) — connects FSRQs and BL Lacs through a continuum of decreasing accretion rate, decreasing line strength, and shifting synchrotron peak frequency, broadly consistent with a single jet phenomenon at different fuel levels.

Where the simple model breaks — and the modifications it needs

The 1993 picture survives, but a generation of follow-up work has refined it on three fronts.

Changing-look AGN

A growing population of AGN flip optical type — Type 1 to Type 2, or back — on timescales of months to years. NGC 2617, Mrk 1018, NGC 1566, 1ES 1927+654, and dozens more catalogued in the past decade. Months is far too short for the torus to reorient (which would take ~10⁵ yr), so the changing look cannot be a viewing-angle change. The current consensus interpretation is an accretion-rate change: when ṁ = Ṁ/Ṁ_Edd drops below a threshold (~10⁻²), the BLR fades because the inner disk loses the radiation needed to maintain its photoionised clouds. The unified model accommodates this by adding accretion rate as an independent axis on top of orientation.

True Type 2 AGN

A small fraction of Seyfert 2s show no broad lines in either total flux or polarised flux — implying there is no hidden BLR, not just an obscured one. NGC 3147 is the canonical example. These objects again sit at low accretion rate and support the interpretation that the BLR is a property of the accretion flow, not a universal AGN feature.

The torus is clumpy, not smooth

Mid-infrared interferometry (VLTI/MIDI, then MATISSE) and X-ray variability resolve the obscuring structure as clumpy molecular clouds rather than a smooth doughnut. A cloud crossing the line of sight on month-to-year timescales is now invoked to explain X-ray N_H variability and even some short-timescale optical type changes. The geometry-and-orientation logic survives; the implementation is messier.

The receding torus

Higher-luminosity AGN have larger dust sublimation radii — the torus inner edge recedes — which makes high-luminosity Type 2s rarer in luminosity-limited samples (the so-called Lawrence 1991 receding torus). This is a refinement to the simple geometry: the torus opening angle increases with luminosity.

Some canonical real systems

• NGC 1068. The first clear hidden BLR. Polarised broad Balmer and Mg II lines; mid-IR resolved torus; H₂O megamaser disk pinning the SMBH mass at 1.7 × 10⁷ M☉.
• NGC 4151. Seyfert 1.5, archetypal reverberation-mapped BLR (rest-frame Hβ time lag ~ 6 days → ~ 1.6 × 10⁻³ pc).
• 3C 273. The first quasar (z = 0.158), bright Type 1 with a one-sided 50 kpc jet (so radio-loud) and broad Balmer/Mg II lines.
• Mrk 421 / Mrk 501. Nearby BL Lacs, prototypes of TeV-detected high-frequency-peaked blazars.
• M87. Low-luminosity AGN with a relativistic jet imaged by VLBI and an EHT-imaged ring at the horizon. 6.5 × 10⁹ M☉, accreting at ~ 10⁻⁵ L_Edd — a radiatively inefficient flow that fuels a jet but no BLR.
• 1ES 1927+654. A spectacular changing-look event in 2018 — Type 2 to Type 1 in months, followed by complete X-ray drop-out and rebuild. Used to time-resolve corona formation.
• NGC 3147. A low-luminosity AGN with no polarised hidden BLR — a true Type 2 candidate.

How observers tell the classes apart in practice

A short decision tree from a single optical/X-ray spectrum:

1. Broad permitted lines (FWHM > 1000 km/s)? Yes → Type 1; No → step 2.
2. Narrow forbidden lines with [O III]/Hβ > 3, [N II]/Hα diagnostics in the AGN region of a BPT diagram? Yes → Type 2; No → not an AGN.
3. For Type 2: column density from X-ray fitting. N_H > 10²⁴ cm⁻² → Compton-thick; we are looking through tens of optical depths of gas.
4. Radio loudness R = F_5GHz / F_B-band. R > 10 → radio-loud; check for compact, flat-spectrum core for blazar candidacy; check for extended lobes for FR-I/FR-II classification.
5. For Type 2: deep spectropolarimetry. Polarised broad lines → hidden BLR confirmed, classic unified-model Type 2; no polarised broad lines → true Type 2 candidate.

Common pitfalls

• Equating "Type 2" with "the engine is different." The unified model says the engine is the same; the obscurer is what differs. Treating Seyfert 2s as a different physical class biases population studies and luminosity functions.
• Forgetting that orientation is not the only axis. Accretion rate, jet activity, and SMBH spin matter too. The 1993 picture is "orientation first," not "orientation only." A 2026 statement of the unified model has four axes: orientation, obscuration, accretion rate, jet on/off.
• Confusing radio-loud edge-on (NLRG) with radio-quiet edge-on (Seyfert 2). Both look obscured in the optical. The radio map disambiguates.
• Reading every changing-look AGN as a viewing-angle event. Months-to-years type changes are too fast for the torus to spin around — they are accretion-state changes by default.
• Counting a single Compton-thick X-ray N_H as an unambiguous Type 2. Some Type 1s can show transient absorption from clumpy obscurers; the optical broad lines remain the cleanest tag.
• Assuming every AGN should have a hidden BLR. True Type 2 candidates exist; at low Ṁ/Ṁ_Edd the BLR itself may not form. The model needs to admit "no BLR" as a state, not enforce one universally.