High-Frequency Peaker (HFP): Early Radio AGN

• High-Frequency Peaker (HFP) are compact radio sources with convex spectra peaking above 5 GHz, often representing early evolutionary stages or Doppler-boosted blazar activity.
• Their high-frequency turnover is modeled using log-parabolic fits to capture both synchrotron self-absorption and free-free absorption, aiding in magnetic field and age estimates.
• HFP samples are constructed from high-frequency surveys and reveal ultracompact sizes (<1 kpc) alongside diverse host types, which inform studies on AGN duty cycles and evolution.

High-Frequency Peaker (HFP) denotes a class of compact radio sources with convex radio spectra that peak at unusually high radio frequencies, typically above a few gigahertz and often operationally above $5$ GHz. In the radio-AGN literature, HFPs are discussed together with Gigahertz-Peaked Spectrum (GPS) and Compact Steep Spectrum (CSS) sources as compact, peaked-spectrum systems whose small sizes and high turnover frequencies are commonly associated with an early evolutionary stage, although substantial contamination by blazar-like objects is also well established (Majorova et al., 5 Aug 2025, Orienti et al., 2010).

1. Spectral definition and taxonomic position

An HFP is defined observationally by a convex spectrum: flux density rises with frequency in the optically thick regime, reaches a maximum, and declines in the optically thin regime. One common convention writes the spectrum as $S_\nu \propto \nu^\alpha$, so $\alpha>0$ below the peak and $\alpha<0$ above it; other HFP papers use $S(\nu)\propto \nu^{-\alpha}$, which reverses the sign convention for “inverted” and “steep” spectra. This suggests that comparisons across HFP papers require checking the adopted sign convention explicitly (Majorova et al., 5 Aug 2025, Orienti et al., 2012).

The standard compact-source sequence used in HFP work places CSS, GPS, and HFP sources in adjacent spectral and size regimes. In one widely used formulation, CSS sources peak at $\nu_{\max}<0.5$ GHz and have linear sizes of $\sim 1$–20 kpc, GPS sources have $0.5<\nu_{\max}<5$ GHz and sizes $<1$ kpc, and HFP sources have $\nu_{\max}>5$ GHz and sizes $S_\nu \propto \nu^\alpha$0 kpc (Majorova et al., 5 Aug 2025).

Class	Peak frequency	Typical size
CSS	$S_\nu \propto \nu^\alpha$1 GHz	$S_\nu \propto \nu^\alpha$2–20 kpc
GPS	$S_\nu \propto \nu^\alpha$3 GHz	$S_\nu \propto \nu^\alpha$4 kpc
HFP	$S_\nu \propto \nu^\alpha$5 GHz	$S_\nu \propto \nu^\alpha$6 kpc

Some papers define HFPs through the observed peak frequency, $S_\nu \propto \nu^\alpha$7 GHz, while noting that other authors instead require the rest-frame condition $S_\nu \propto \nu^\alpha$8 GHz, with $S_\nu \propto \nu^\alpha$9 (Majorova et al., 5 Aug 2025). Other surveys begin from a strong low-frequency inversion and then confirm a convex spectrum with simultaneous multifrequency data; for example, one faint HFP sample retained sources satisfying $\alpha>0$0 or $\alpha>0$1 under the $\alpha>0$2 convention (0901.3068).

2. Spectral turnover and its physical interpretation

The defining observable of an HFP is the high-frequency turnover itself. In the classical compact-source framework, the low-frequency side is optically thick and the high-frequency side optically thin. A canonical synchrotron self-absorbed spectrum has approximately $\alpha>0$3, $\alpha>0$4, and a relatively narrow peak with $\alpha>0$5 decades in frequency under the $\alpha>0$6 convention (Majorova et al., 5 Aug 2025). In practice, HFP spectra are commonly fit with log-parabolic or related analytic forms in $\alpha>0$7-$\alpha>0$8 space to estimate the peak frequency, peak flux density, and spectral width (Majorova et al., 5 Aug 2025, 0901.3068).

The physical origin of the turnover is not unique. Synchrotron self-absorption (SSA) is the traditional explanation for very compact peaked-spectrum sources, but free-free absorption (FFA) can also shape or dominate the low-frequency cutoff. In a $\alpha>0$9 sample of radio-loud quasars with rest-frame turnovers of $\alpha<0$0–50 GHz, external inhomogeneous FFA accurately described the observed spectra for all nine well-constrained targets, and SSA-based magnetic fields for two sources were inconsistent with equipartition, arguing against a purely SSA origin in those cases (Shao et al., 2021). By contrast, the high-redshift blazar PKS 1614+051, an HFP with a stable spectral peak around $\alpha<0$1 GHz in the observer’s frame, was modeled as requiring both SSA and FFA, and SSA-based estimates gave magnetic field strengths peaking at $\alpha<0$2 mG (Sotnikova et al., 22 Jan 2025).

This mix of SSA and FFA results implies that the HFP label is observational rather than uniquely mechanistic. A high turnover frequency can reflect a very small synchrotron-emitting region, a dense absorbing environment, or both. That ambiguity is central to current HFP interpretation.

3. Survey construction, sample properties, and host populations

Large HFP samples have been built through high-frequency or inverted-spectrum selection. A recent AT20G-based study formed a homogeneous sample of $\alpha<0$3 HFP candidates by requiring the optically thick spectral index $\alpha<0$4, then constructing broadband spectra and measuring $\alpha<0$5, $\alpha<0$6, $\alpha<0$7, $\alpha<0$8, and peak width. All but six sources had $\alpha<0$9 GHz; $S(\nu)\propto \nu^{-\alpha}$0 had ultra-inverted spectra with $S(\nu)\propto \nu^{-\alpha}$1, corresponding to $S(\nu)\propto \nu^{-\alpha}$2 of the full AT20G catalogue and about $S(\nu)\propto \nu^{-\alpha}$3 of the HFP sample. Optical identification showed that $S(\nu)\propto \nu^{-\alpha}$4 of the hosts are quasars, radio luminosities at $S(\nu)\propto \nu^{-\alpha}$5 GHz span $S(\nu)\propto \nu^{-\alpha}$6–$S(\nu)\propto \nu^{-\alpha}$7, angular sizes inferred for the emitting regions are $S(\nu)\propto \nu^{-\alpha}$8–$S(\nu)\propto \nu^{-\alpha}$9 mas, and projected linear sizes are $\nu_{\max}<0.5$0–$\nu_{\max}<0.5$1 pc (Majorova et al., 5 Aug 2025).

Lower-flux-density selection yields related but not identical populations. The “faint” HFP sample restricted to the North Galactic Cap used compactness checks with FIRST and simultaneous VLA spectra from $\nu_{\max}<0.5$2 to $\nu_{\max}<0.5$3 GHz, producing a final list of $\nu_{\max}<0.5$4 confirmed faint HFPs with $\nu_{\max}<0.5$5 (0901.3068). In a separate multi-frequency catalogue behind the Large Magellanic Cloud, HFP classification was based on convex spectra peaking above $\nu_{\max}<0.5$6 GHz; that paper’s abstract states six HFP candidates, whereas the body and Table 12 list $\nu_{\max}<0.5$7 HFP sources, and the final classification table uses $\nu_{\max}<0.5$8 (Filipović et al., 2021). This discrepancy illustrates the dependence of HFP counts on catalogue conventions and peak constraints.

Sample incompleteness is an important recurring issue. In the AT20G HFP sample, $\nu_{\max}<0.5$9 of sources lack data below $\sim 1$0 GHz, limiting constraints on the optically thick slope and on possible older, low-frequency emission (Majorova et al., 5 Aug 2025). High-frequency coverage above $\sim 1$1 GHz is also sparse in many surveys, so some “HFP” spectra remain lower limits on the true peak frequency.

4. Morphology, variability, and the young-source–blazar dichotomy

Parsec-scale imaging shows that the HFP class is heterogeneous. In VLBA observations of $\sim 1$2 faint HFPs at $\sim 1$3 and $\sim 1$4 GHz, $\sim 1$5 ($\sim 1$6) were at least marginally resolved and $\sim 1$7 ($\sim 1$8) remained unresolved even at the higher frequency. The final classification was $\sim 1$9 confirmed CSO, $0.5<\nu_{\max}<5$0 CSO candidates, $0.5<\nu_{\max}<5$1 blazar-like, and $0.5<\nu_{\max}<5$2 uncertain. Equipartition magnetic fields in individual components were typically $0.5<\nu_{\max}<5$3–$0.5<\nu_{\max}<5$4 mG, with a blazar core reaching $0.5<\nu_{\max}<5$5 mG, and the magnetic-field differences among components were taken as evidence against simple self-similar early evolution (Orienti et al., 2012).

Multi-epoch spectroscopy reinforces the same division. In the faint HFP sample followed over several VLA epochs, $0.5<\nu_{\max}<5$6 of $0.5<\nu_{\max}<5$7 sources showed no significant variability and retained convex spectra, $0.5<\nu_{\max}<5$8 were variable but remained peaked, and $0.5<\nu_{\max}<5$9 became flat-spectrum in at least one epoch. The same study concluded that $<1$0 sources were consistent with young radio source candidates, while $<1$1 behaved like flat-spectrum blazar objects; $<1$2 variable but peaked sources had spectral changes consistent with adiabatic expansion (Orienti et al., 2010).

The blazar PKS 1614+051 illustrates a stable HFP-blazar hybrid. Over $<1$3–$<1$4, it showed low overall variability indices of $<1$5–$<1$6, a spectral peak around $<1$7 GHz that remained stable during long-term monitoring, radio time delays of $<1$8 to $<1$9 years between frequencies, and rest-frame variability timescales of $\nu_{\max}>5$0 to $\nu_{\max}>5$1 years (Sotnikova et al., 22 Jan 2025). This suggests that an HFP spectrum alone does not distinguish a nascent symmetric source from a Doppler-boosted jet source.

A common misconception is therefore that all HFPs are simply the youngest radio galaxies. The observational record does not support that simplification. HFP samples contain genuinely young compact sources, beamed blazars with temporarily or persistently peaked spectra, and objects whose classification remains ambiguous until variability and milliarcsecond morphology are known.

5. Sizes, ages, and duty-cycle interpretation

The main evolutionary argument for HFPs is the anti-correlation between turnover frequency and size. In the AT20G HFP galaxies, the relation between intrinsic peak frequency and angular size was fitted as

$\nu_{\max}>5$2

or $\nu_{\max}>5$3, in good agreement with earlier CSS/GPS relations. The same study used previously established CSS/GPS size relations, $\nu_{\max}>5$4 and $\nu_{\max}>5$5, to show that HFP galaxies occupy the smallest-size, highest-peak-frequency extension of the compact-source sequence (Majorova et al., 5 Aug 2025). This supports the “youth scenario,” in which HFPs precede GPS, then CSS, and ultimately large Fanaroff-Riley radio galaxies.

Age estimates in the HFP literature range from $\nu_{\max}>5$6–$\nu_{\max}>5$7 years for the youngest HFPs to $\nu_{\max}>5$8–$\nu_{\max}>5$9 years for GPS sources and $S_\nu \propto \nu^\alpha$00–$S_\nu \propto \nu^\alpha$01 years for CSS sources under the same broad framework (Orienti et al., 2010, Orienti et al., 2012). At the same time, LOFAR-based work on compact sources emphasizes that some CSS, GPS, and HFP systems may not simply evolve monotonically outward. They may die prematurely, remain confined within the host galaxy, or represent renewed activity inside much older low-surface-brightness lobes. Low-frequency searches for diffuse outer emission are therefore used to constrain the duty cycle of compact radio sources, including HFPs (Brienza et al., 2015).

A detailed illustration comes from the CSS source B2 0258+35, whose $S_\nu \propto \nu^\alpha$02 kpc compact source has a spectral age of $S_\nu \propto \nu^\alpha$03 yr, while diffuse outer lobes extend over $S_\nu \propto \nu^\alpha$04 kpc and were interpreted as $S_\nu \propto \nu^\alpha$05 Myr-old plasma from a previous cycle (Brienza et al., 2015). Although that source is not itself an HFP, it provides a concrete template for how an HFP plus relic lobes system would constrain quiescent intervals between activity episodes. This suggests that HFPs are best understood not only as “young” objects, but also as possible markers of recurrent radio-AGN activity.

6. Special cases, high-redshift systems, and terminological ambiguity

Several individual objects illustrate the breadth of the HFP category. In a $S_\nu \propto \nu^\alpha$06 quasar sample observed with the VLA and uGMRT, all ten targets showed evidence for spectral turnover, with rest-frame turnover frequencies of $S_\nu \propto \nu^\alpha$07–50 GHz, making them GPS or HFP candidates; for nine well-constrained sources, external inhomogeneous FFA fit the full radio spectra, and the derived radio loudness values $S_\nu \propto \nu^\alpha$08 ranged from $S_\nu \propto \nu^\alpha$09 to $S_\nu \propto \nu^\alpha$10 (Shao et al., 2021). At the opposite extreme of observational circumstance, the source M17 JVLA 35 has a spectrum consistent with an extragalactic HFP, with a turnover at $S_\nu \propto \nu^\alpha$11 GHz and $S_\nu \propto \nu^\alpha$12 at higher frequencies, but an apparent size of $S_\nu \propto \nu^\alpha$13 at $S_\nu \propto \nu^\alpha$14 GHz; the preferred explanation is that it is an HFP whose image is broadened by plasma scattering in the foreground H II region M17, which predicts an angular-size dependence $S_\nu \propto \nu^\alpha$15 (Rodriguez et al., 2014).

HFP status can also apply to a component rather than an entire source. In the radio galaxy 0932+075, one VLBA component had an inverted spectrum throughout $S_\nu \propto \nu^\alpha$16–$S_\nu \propto \nu^\alpha$17 GHz and was labelled an HFP, but two-epoch proper motions of surrounding components could not be reconciled with the compact symmetric object paradigm. The western part of the source was therefore rejected as a CSO, and the HFP component could not be securely identified as the core (Marecki et al., 2014). This is a direct reminder that an HFP spectrum does not by itself define dynamical role.

Finally, the acronym itself is not unique outside radio astronomy. In unrelated computer-vision literature, “HFP” may denote “High-Frequency Perception,” and one super-resolution paper states explicitly that it does not use the phrase “High-Frequency Peaker” at all (Wu et al., 3 Jun 2025). Within radio astronomy, however, HFP has a stable technical meaning: a compact peaked-spectrum radio source with turnover at high radio frequency.