HVFs in AGN and Supernovae

• HVFs are defined as detached spectral features occurring at velocities much higher than typical photospheric lines, indicating extreme kinematics in both AGN disks and supernova ejecta.
• In AGN masers, HVFs arise from hyperfine-induced gain asymmetries in nearly edge-on, Keplerian accretion disks, enabling precise tracing of orbital motion around supermassive black holes.
• In Type Ia supernovae, HVFs manifest as blue-shifted absorption in Ca II and Si II lines, with their strength and evolution offering insights into explosion dynamics and progenitor environments.

High-velocity features (HVFs) are spectral components formed at velocities substantially above a reference systemic or photospheric component. In the literature considered here, the term is used in two distinct astrophysical settings: high-velocity water-vapor maser emission from the right and left wings of thin, edge-on Keplerian accretion disks in active galactic nuclei, and detached, blue-shifted absorption components in Type Ia supernova spectra, especially in Ca II and Si II lines (Nesterenok et al., 2010, Silverman et al., 2015). The shared terminology is therefore phenomenological rather than mechanistic. In both settings, HVFs probe outer or kinematically extreme material, but the governing physics differs: asymmetric maser gain profiles in AGN disks, and ejecta stratification, ionization, density enhancement, or circumstellar interaction in supernovae.

1. Terminology and observational definition

In Type Ia supernova work, HVFs are spectral features with minima indicating significantly higher, by greater than about $6000~\mathrm{km\,s^{-1}}$, velocities than typical photospheric-velocity features (PVFs). Near $B$-band maximum, PVFs usually lie at about $9000$–$15{,}000~\mathrm{km\,s^{-1}}$, whereas the HVFs are most prominently observed in Ca II H&K, the Ca II near-infrared triplet, and Si II $\lambda 6355$ (Silverman et al., 2015). The empirical distinction is often expressed through detached absorption components, multiple-Gaussian decompositions, or a pseudo-equivalent-width ratio between HVF and PVF components.

In AGN maser studies, high-velocity features are distinct emission components offset by large Doppler shifts from the systemic velocity. Within the accretion-disk model, the blueshifted features arise from disk regions where gas is moving toward the observer, and the redshifted features arise from regions where gas is receding; the systemic features originate from the sector where the velocity is perpendicular to the line of sight (Nesterenok et al., 2010).

This usage suggests that “HVF” is best treated as a kinematic descriptor. It does not, by itself, specify whether the line-forming region is produced by a detached shell, a density enhancement, a composition enhancement, a gain-profile asymmetry, or circumstellar interaction.

2. Maser HVFs in active galactic nuclei

For water-vapor masers in AGN, the emitting medium is modeled as a thin, edge-on accretion disk around a supermassive black hole. Gas in the disk orbits with high velocities, from hundreds to over a thousand $\mathrm{km\,s^{-1}}$, at radii of fractions of a parsec, and the velocity field follows the Keplerian law

$V = \sqrt{\frac{GM}{R}} .$

The pattern of maser spike velocities as a function of position traces this Keplerian curve, which fixes the disk origin of the high-velocity features (Nesterenok et al., 2010).

The specific mechanism proposed for the asymmetry of these HVFs is the hyperfine splitting of the ortho-$\mathrm{H_2O}$ $22.235~\mathrm{GHz}$ transition. The maser gain profile is a sum of several Gaussian components,

$$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$

with $B$0, $B$1, $B$2, $B$3, $B$4, and $B$5. At lower temperatures, such as $B$6–$B$7, the overlap and differing strengths of the hyperfine components make the gain profile asymmetric; thermal broadening makes it more symmetric at higher temperature (Nesterenok et al., 2010).

The radiative-transfer description is the standard maser form,

$B$8

with exponential amplification in the maser regime. Because the gain profile is asymmetric, the blueshifted high-velocity features sample a stronger part of the profile than the redshifted ones, so the blueshifted features are predicted to be brighter. The paper argues that relativistic effects, including kinematic Doppler, gravitational redshift, and transverse Doppler terms, are much smaller than this hyperfine-induced asymmetry. The effect is maximized when the disk inclination to the line of sight is less than a critical small value, about $B$9; at larger inclinations, line-of-sight integration through several disk layers smears out the asymmetry (Nesterenok et al., 2010).

UGC 3789 is presented as the observational example in which the blueshifted high-velocity features are brighter than the redshifted ones. In that source, the asymmetry is explained as a direct consequence of the gain-profile asymmetry rather than foreground absorption (Nesterenok et al., 2010).

3. Identification and measurement in Type Ia supernova spectra

The supernova literature treats HVFs primarily as detached blue-shifted absorption components. Their decomposition is line-dependent. Ca II H&K is often fitted with up to three components because of blending with Si II $9000$0; the Ca II near-infrared triplet is modeled with one or two triplet components; Si II $9000$1 is fitted with one or two Gaussians (Silverman et al., 2015). A standard strength measure is

$9000$2

and a complementary kinematic diagnostic is the absorption-weighted velocity

$9000$3

which is used, for example, in comparisons between Ca II IR3 and Si II $9000$4 (Meng, 2019).

Large surveys have emphasized that inferred HVF frequencies depend on phase, line, and fitting protocol. In the 445-spectrum study of 210 low-redshift SNe Ia, Ca II HVFs were present in $9000$5 of SNe Ia overall and in about $9000$6 of spectra earlier than $9000$7 days, while Si II HVFs were detected in only $9000$8 overall and about $9000$9 at epochs earlier than $15{,}000~\mathrm{km\,s^{-1}}$0 days; Ca II HVFs were never found in underluminous/Ia-91bg SNe (Silverman et al., 2015). In the ZTF SN Ia DR2 analysis, each Si II $15{,}000~\mathrm{km\,s^{-1}}$1 feature was fit with both a single-component and a double-component model using MCMC followed by BIC testing, and reliable identification required $15{,}000~\mathrm{km\,s^{-1}}$2 per feature, dispersion $15{,}000~\mathrm{km\,s^{-1}}$3, and $15{,}000~\mathrm{km\,s^{-1}}$4. After correcting for detection efficiency and false positives, about $15{,}000~\mathrm{km\,s^{-1}}$5 of spectra before $15{,}000~\mathrm{km\,s^{-1}}$6 days showed Si II HVFs, dropping to $15{,}000~\mathrm{km\,s^{-1}}$7 between $15{,}000~\mathrm{km\,s^{-1}}$8 and $15{,}000~\mathrm{km\,s^{-1}}$9 days and $\lambda 6355$0 between $\lambda 6355$1 days and maximum light (Harvey et al., 6 Feb 2025).

Temporal evolution is a central observational property. Si II HVFs generally fade earlier than Ca II HVFs. In the ZTF sample, HVFs with larger velocity separations from the photosphere faded earlier, leaving only double components with smaller separations toward maximum light (Harvey et al., 6 Feb 2025). In the earlier large-sample study from early to maximum phases, Si II HVFs were weaker and disappeared earlier than Ca II near-infrared HVFs, while Ca II HVFs were often present even near maximum light (Zhao et al., 2015).

Methodological caution is recurrent. The compact-shell analysis of SN 2011fe argued that Gaussian decomposition of the observed Ca II near-infrared profile can falsely identify HVFs or other components because the physical line shape is non-Gaussian (Mulligan et al., 2015). This is consistent with broader concerns that some fitted “high-velocity components” may reflect decomposition choices rather than fully detached line-forming regions.

4. Statistical behavior, subclasses, and environments in SNe Ia

Near maximum light, strong Ca II IR3 HVFs are linked to slower light-curve decline and are absent in rapidly declining events. One 58-supernova study found that average HVF strength increases with decreasing $\lambda 6355$2, and that rapidly declining SNe Ia with $\lambda 6355$3 show no HVFs in their maximum-light spectra (Childress et al., 2013). The larger early-time sample that included very early spectra clarified that this tendency applies to both Si-HVFs and Ca-HVFs in the earlier phase: HVF-strong SNe Ia have smaller decline rates, and all early-phase Si II HVF-strong SNe have $\lambda 6355$4 (Zhao et al., 2015).

Host-environment correlations are strongest for Ca II IR3 HVFs around maximum brightness. Strong Ca II IR3 HVFs are associated with relatively younger stellar populations, and such events tend to occur in late-type galaxies or in early-type galaxies with significant recent or ongoing star formation. Pixel statistics based on $\lambda 6355$5 and near-UV emission show that SNe Ia with strong Ca II IR3 HVFs are closer to the distribution of star-formation tracers than those with weak HVFs. In the same work, all 1991T-like SNe Ia in the sample had very strong HVFs, while nearly all 1991bg-like SNe Ia had weak or absent HVFs (Meng, 2019).

A useful empirical relation links Ca II IR3 HVF strength to the difference between Ca II and Si II absorption-weighted velocities:

$\lambda 6355$6

This relation was proposed as a practical diagnostic of whether there is a high-velocity component in the Ca II IR3 absorption feature (Meng, 2019).

At the same time, the literature does not support a single population-level rule for all lines or all phases. The ZTF DR2 study found no significant difference between SNe Ia that do and do not form Si II $\lambda 6355$7 HVFs in SALT2 $\lambda 6355$8, peak magnitude, decline rate, host mass, or host colour, supporting the view that Si II HVFs are ubiquitous across the normal SN Ia population in the pre-peak regime (Harvey et al., 6 Feb 2025). A plausible implication is that Ca II HVF strength and Si II HVF presence are tracing overlapping but not identical physics.

The relation between HVFs and the Wang “high-velocity” subclass is also phase-sensitive. Maximum-light studies showed that SNe Ia with lower $\lambda 6355$9 have stronger Ca II HVFs, while the high-velocity subclass with $\mathrm{km\,s^{-1}}$0 exhibits no HVFs in maximum-light spectra (Childress et al., 2013). By contrast, very early Si II HVFs are more common in fast-expanding SNe Ia, and the 2015 early-time analysis argued that these HVFs in fast-expanding objects usually disappear more rapidly and become blended with the photospheric component as maximum light approaches (Zhao et al., 2015). This distinction corrects a common misconception: a “high-velocity” SN Ia in the Wang sense is not equivalent to an HVF-strong object at all phases.

A later 145-object study further separated PVF and HVF phenomenology. It found that the Ca II PVF velocity distribution is predominantly unimodal, unlike the well-known bimodal Si II PVF distribution, and that HVFs do not significantly bias PVF velocity distributions. It also found a significant negative correlation between Si II PVF velocity and $\mathrm{km\,s^{-1}}$1 only for HVF-weak SNe Ia; this trend vanishes in HVF-strong events, likely due to circumstellar interaction (Hakobyan et al., 28 Nov 2025).

5. Proposed physical origins and modeling frameworks

The supernova literature does not support a single origin for all HVFs. The principal scenarios are abundance enhancement, density enhancement, ionization effects, circumstellar interaction, and geometric asymmetry. The 2015 early-to-maximum analysis concluded that ionization and/or thermal processes alone cannot explain the observed relationships between Si II and Ca II absorption features, and that different mechanisms are required for the creation of the HVF-forming region in SNe Ia (Zhao et al., 2015).

Compact-shell interaction models are a prominent density-enhancement framework. For SN 2011fe, interaction with compact circumstellar shells of $\mathrm{km\,s^{-1}}$2–$\mathrm{km\,s^{-1}}$3 was used to reproduce the evolution of the Ca II near-infrared HVF. Models with a shell fit the data better than models without a shell, and a shell of $\mathrm{km\,s^{-1}}$4 tended to be better fitting than the other shells. The analysis also found that the optical-depth evolution of Ca II implies that the ionization state of calcium within the ejecta and shell is not constant (Mulligan et al., 2015). TARDIS calculations based on hydrodynamic shell models reached a related conclusion: a compact shell of $\mathrm{km\,s^{-1}}$5 with super-solar calcium abundance can approximately reproduce the blended HVF and PVF at 2 and 5 days in SN 2011fe, but the distinct 9-day HVF is not reproduced, and helium-shell double-detonation models tend to give either no HVF or excessively strong features (Mulligan et al., 2019).

Surface-composition models connect HVFs to pre-explosion nucleosynthesis. Successive helium shell flashes on mass-increasing carbon-oxygen white dwarfs can build silicon-rich outer layers before explosion. For a $\mathrm{km\,s^{-1}}$6 white dwarf accreting at $\mathrm{km\,s^{-1}}$7, the surface layer was calculated to consist of $\mathrm{km\,s^{-1}}$8 $\mathrm{km\,s^{-1}}$9, $V = \sqrt{\frac{GM}{R}} .$0 $V = \sqrt{\frac{GM}{R}} .$1, $V = \sqrt{\frac{GM}{R}} .$2 $V = \sqrt{\frac{GM}{R}} .$3, and a few percent $V = \sqrt{\frac{GM}{R}} .$4 by weight; the paper proposed that such a silicon-rich layer could be the origin of Si II HVFs, while possible Ca production was discussed for very massive white dwarfs $V = \sqrt{\frac{GM}{R}} .$5 (Kato et al., 2018). A distinct pre-explosion route invokes g-mode excitation during the simmering phase. In that picture, mode breaking deposits energy into a shell of mass $V = \sqrt{\frac{GM}{R}} .$6, raises the surface temperature by $V = \sqrt{\frac{GM}{R}} .$7, ignites helium, synthesizes $V = \sqrt{\frac{GM}{R}} .$8, $V = \sqrt{\frac{GM}{R}} .$9, $\mathrm{H_2O}$0, and some $\mathrm{H_2O}$1, and ejects these ashes at velocities of $\mathrm{H_2O}$2, which may explain the HVFs seen in many SNe Ia (Piro, 2011).

Explosion-model interpretations remain contested. Early-color and early-spectroscopy work argued that the He-detonation model, combined with geometric or projected effects, may account for the very high Ca II NIR HVF velocities in high-velocity SNe Ia, the absence or weakness of C II $\mathrm{H_2O}$3, and the correlations with redshifted nebular [Fe II] velocities and massive hosts. In that sample, CaIR3 HVF velocities in high-velocity objects typically exceeded $\mathrm{H_2O}$4 at 2–3 days, about $\mathrm{H_2O}$5 higher than in normal-velocity objects at the same epochs (Li et al., 2020). By contrast, the recent six-object TARDIS study found that a single Gaussian density enhancement at high velocity can reproduce the observed Si II HVF evolution for Si II only, but cannot simultaneously produce the observed silicon and calcium HVF evolution. It further concluded that neither delayed-detonation nor double-detonation models, as currently modeled, can produce these HVFs, suggesting that something may be missing from the models (Harvey et al., 15 Dec 2025).

The balance of evidence therefore favors a plural framework. Some events are consistent with compact shells or circumstellar interaction; some support compositionally stratified outer layers; some can be described by ionization suppression; and current 1D calculations do not yet yield a unified explanation for detached Si II and Ca II HVFs across the observed population.

6. Benchmark events, comparative phenomenology, and unresolved issues

Individual SNe Ia have supplied the best-resolved empirical benchmarks. SN 2009ig provided a complete observational sequence from HVF-dominated to PVF-dominated spectra. High-velocity features were identified in Si II, Si III, S II, Ca II, and Fe II. At $\mathrm{H_2O}$6 and $\mathrm{H_2O}$7 days, the spectra clearly resolved Si II $\mathrm{H_2O}$8 HVF as separate absorptions from a detached line-forming region; from $\mathrm{H_2O}$9 to $22.235~\mathrm{GHz}$0 days, HVFs and PVFs were detected simultaneously with a roughly constant separation of about $22.235~\mathrm{GHz}$1; after $22.235~\mathrm{GHz}$2 days, all absorption features were PVFs. Using mean velocities, the HVF region was inferred to lie at about $22.235~\mathrm{GHz}$3 in homologous expansion (Marion et al., 2013).

SN 2012fr is a canonical case of exceptionally clean HVF/PVF decoupling. In Si II $22.235~\mathrm{GHz}$4, the HVF dominated at $22.235~\mathrm{GHz}$5 days with $22.235~\mathrm{GHz}$6, faded by phase $22.235~\mathrm{GHz}$7, and gave way to a photospheric component with a very narrow velocity width and a long plateau at about $22.235~\mathrm{GHz}$8. The Ca II infrared triplet showed an even higher HVF, beginning at $22.235~\mathrm{GHz}$9 at $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$0 days and remaining detectable to about $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$1 days (Childress et al., 2013).

SN 2000cx and SN 2013bh represent an extreme, rare class. Both showed separate PVFs at about $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$2 and Ca II HVFs at about $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$3, with additional HVFs of Ti II and Fe II. Their environments lacked narrow Na I D absorption, and the favored interpretation was clumpy, high-velocity ejecta produced by an energetic delayed-detonation-like explosion rather than circumstellar interaction (Silverman et al., 2013).

Recent events have sharpened the connection between HVFs and other early observables. SN 2021hpr showed $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$4 excess early emission above a homogeneously expanding fireball model and prominent detached HVFs of Si II and Ca II. At $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$5 days, the Si II HVF was measured at about $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$6 and the Ca II near-infrared HVF at about $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$7. A Kolmogorov-Smirnov test on early-phase velocities in a comparison sample gave $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$8, supporting a correlation between early excess and prominent HVFs; the favored interpretation was interaction with a companion or immediate circumstellar matter, though more complete burning in the outer layers was also discussed (Iskandar et al., 4 Mar 2025). SN 2024gy showed Ca II near-infrared HVFs exceeding $$\varphi_{\mathrm{HF}}(u) = a e^{-\left(\frac{u-u_a}{v_T}\right)^2} + b e^{-\left(\frac{u-u_b}{v_T}\right)^2} + c e^{-\left(\frac{u-u_c}{v_T}\right)^2},$$9 while Si II $B$00 was at about $B$01 at the same epoch, and a nebular $B$02; the paper argued that these data support a delayed-detonation origin, with prominent Ca II HVFs possibly arising from ionization suppression within the highest-velocity ejecta (Li et al., 2 Aug 2025).

Element-dependent behavior remains one of the main open problems. A recent study of Ca II NIR and Ca II H&K found an anti-correlation between the HVF and photospheric component strengths, with $B$03, and a positive correlation between $B$04 and $B$05. By contrast, Si II $B$06 and O I $B$07 showed positive HVF–PHO strength correlations and no clear $B$08–$B$09 correlation. The proposed interpretation was that calcium, synthesized in deeper layers, experiences much more serious blocking by substances in outer layers than silicon and oxygen (Zhao, 2024).

The present literature therefore supports several secure statements. HVFs are common in early SN Ia spectra, especially in Ca II; their prevalence and strength depend strongly on phase; Ca II and Si II HVFs are not interchangeable diagnostics; and current models do not yet provide a single, self-consistent account of the detached, line-dependent high-velocity structure. In AGN masers, by contrast, the evidence summarized here is narrowly focused: HVF asymmetry is explained through hyperfine-split gain asymmetry in nearly edge-on Keplerian disks, with UGC 3789 as the specific motivating example (Nesterenok et al., 2010).