AT 2022fpx: Dust-Reddened TDE in a Weak AGN

• AT 2022fpx is a nuclear transient characterized by a slow optical/UV decay, delayed X-ray brightening, and strong coronal line emission.
• It exhibits complex multi-wavelength behavior including a prominent mid-infrared dust echo, variable optical polarization, and evolving spectral signatures from Balmer, Bowen, and coronal lines.
• Astrophysical analysis indicates it may be a heavily dust-reddened tidal disruption event in a weak AGN, challenging conventional selection criteria for TDEs.

Searching arXiv for AT 2022fpx and related studies. AT 2022fpx, also known as ATLAS22kjn, ZTF22aadesap, and Gaia22cwy, is a nuclear transient in the galaxy SDSS J153103.70+532419.3 at $z=0.073$–0.0735 whose observed properties place it at the intersection of several rare classes of supermassive-black-hole transients. It has been described as a tidal disruption event (TDE)-like flare that simultaneously shows Bowen-fluorescence signatures, extreme coronal line emission, a long-lived slowly declining optical/UV outburst, delayed soft X-ray brightening, and variable low-level optical polarization (Koljonen et al., 2024). A later reanalysis emphasized its persistently red optical color, luminous mid-infrared echo, and evidence for weak pre-existing AGN activity, and argued that the source is most plausibly a heavily dust-reddened TDE in a weak AGN, while leaving a turn-on AGN interpretation as a remaining alternative (Lin et al., 7 Jul 2025).

1. Discovery and observational setting

AT 2022fpx was discovered by ATLAS on 2022-03-31 (MJD 59669) in the orange filter at magnitude 18.54. It is spatially coincident with the nucleus of SDSS J153103.70+532419.3, and Gaia pre-flare data showed the nucleus stable at $g\sim20.1$ mag. Follow-up spectroscopy classified it as a TDE candidate at redshift $z=0.073$; a later analysis adopted a median redshift $z=0.0735$ from six follow-up spectra (Koljonen et al., 2024).

The transient reached peak optical brightness in late July 2022. One study placed the ZTF $g,r$ peak at MJD 59784 with $r\approx17.6$ mag and $g\approx17.8$ mag, roughly 115 days after discovery, whereas a later phenomenological light-curve fit placed $t_{\rm peak}\sim59795$ (Koljonen et al., 2024). This difference reflects model-dependent peak definitions rather than a disagreement about the overall temporal sequence.

From the outset, AT 2022fpx stood out among nuclear flares because its optical/UV decay was unusually slow, its spectra combined high-ionization coronal lines with Balmer and helium emission, and its X-rays were absent near optical maximum but emerged hundreds of days later. These characteristics made it a candidate link between optical TDEs with delayed X-rays, Bowen-fluorescence flares, and extreme coronal line emitters.

2. Optical and ultraviolet evolution

The photometric evolution is complex. The combined ZTF and Swift/UVOT data show a precursor or flattening about 70 days before the main optical maximum, and the later study identified a sharp ATLAS $o$-band spike at MJD 59669.5 with $m_o=18.97\pm0.08$, followed by a return toward baseline for about a month before the main rise (Koljonen et al., 2024). Excluding the precursor, the 2025 analysis fit the $g\sim20.1$0-band rise and decline with a Gaussian-like rise joined to a $g\sim20.1$1 decay, obtaining $g\sim20.1$2 d, $g\sim20.1$3 d, $g\sim20.1$4 d, and $g\sim20.1$5 d in the rest frame (Lin et al., 7 Jul 2025).

Near peak, a UVOT+ZTF blackbody fit gave

$g\sim20.1$6

while power-law fits $g\sim20.1$7 generally described the UV/optical SED better, with $g\sim20.1$8 evolving from about 1 at early times to about 0 during the decay (Koljonen et al., 2024). The optical/UV decline was well fit by exponentials, $g\sim20.1$9, with $z=0.073$0 d in ZTF $z=0.073$1, $z=0.073$2 d in ZTF $z=0.073$3, $z=0.073$4 d in UVW1, $z=0.073$5 d in UVW2, and $z=0.073$6 d in UVM2 (Koljonen et al., 2024). These timescales are more than twice typical optical-TDE decay times and resemble Bowen-fluorescence flares such as AT 2021loi.

A mild optical/UV bump occurred around $z=0.073$7–300 days after peak, contemporaneous with the onset of X-ray brightening (Koljonen et al., 2024). The later paper further emphasized that after an initial decline, the optical and UV emission flattened into a slower decline or plateau around MJD $z=0.073$8 (Lin et al., 7 Jul 2025).

A major observational peculiarity is color. The later analysis found a persistently red optical color, $z=0.073$9, across most of the event, unlike the nearly constant blue colors typically used in optical TDE selection (Lin et al., 7 Jul 2025). This red color became central to the subsequent debate over dust reddening, line contamination, and sample-selection bias.

3. Spectroscopic diagnostics: Balmer, helium, Bowen, and coronal lines

Optical spectroscopy established AT 2022fpx as an emission-line-dominated nuclear flare. A Nordic Optical Telescope spectrum obtained 126.5 days after discovery, about 41 days after optical peak, showed a blue continuum together with strong H$z=0.0735$0, H$z=0.0735$1, He I, and He II emission (Koljonen et al., 2024). Gaussian measurements gave H$z=0.0735$2 FWHM $z=0.0735$3, H$z=0.0735$4 FWHM $z=0.0735$5, He II in the 4635–4685 Å blend with FWHM $z=0.0735$6, and N III in the same blend with FWHM $z=0.0735$7 (Koljonen et al., 2024). A later seven-spectrum campaign found broad Balmer components with FWHM $z=0.0735$8–$z=0.0735$9, narrow components $g,r$0–$g,r$1, He II $g,r$2 and He I $g,r$3 at luminosities of a few $g,r$4, persistent Fe II pseudo-continuum around 4400–4800 Å, and a substantial weakening of the broad Balmer lines by 2025 (Lin et al., 7 Jul 2025).

The 2024 study identified the 4635–4685 Å complex as clear Bowen fluorescence, including N III $g,r$5 and He II $g,r$6, and therefore placed AT 2022fpx in the growing class of Bowen-fluorescence flares (Koljonen et al., 2024). The 2025 reanalysis took a more cautious view, arguing that key N III and O III features are blended with H$g,r$7, He II, or [Fe VII], so the Bowen classification could not be confirmed unambiguously (Lin et al., 7 Jul 2025). This disagreement is one of the main spectroscopic subtleties of the source.

The most distinctive line diagnostics are the coronal features. AT 2022fpx shows [Fe XIV] $g,r$8, [Fe X] $g,r$9 or $r\approx17.6$0, [Fe XI] $r\approx17.6$1, [Fe VII] $r\approx17.6$2, and [S XII] $r\approx17.6$3 (Koljonen et al., 2024). In the 2024 spectrum, the [Fe XIV] $r\approx17.6$4 flux exceeded [O III] $r\approx17.6$5 by a factor of 1.7, comfortably satisfying the criterion that an extreme coronal line emitter has at least one coronal line more than 20% stronger than [O III] $r\approx17.6$6 (Koljonen et al., 2024). The later study measured coronal-line luminosities of $r\approx17.6$7 to $r\approx17.6$8 with FWHM $r\approx17.6$9–$g\approx17.8$0, present from early times and fading only weakly until late epochs (Lin et al., 7 Jul 2025).

Taken together, the spectra resemble a hybrid of ECLEs, some TDE-in-AGN candidates, and narrow-line Seyfert 1 outbursts more than a canonical blue optical TDE or an ordinary persistent AGN spectrum.

4. Polarization and geometric constraints

AT 2022fpx is one of the few TDE-like transients with multi-epoch optical polarimetry. Using NOT/ALFOSC and Liverpool Telescope/MOPTOP observations, the 2024 study found low but clearly variable intrinsic continuum polarization after correction for host dilution (Koljonen et al., 2024). In the $g\approx17.8$1 band, $g\approx17.8$2 was about $g\approx17.8$3 just before peak, dropped to about $g\approx17.8$4 at peak, and then ranged between $g\approx17.8$5 and $g\approx17.8$6 post-peak. In the $g\approx17.8$7 band, one detection near peak gave $g\approx17.8$8. In the $g\approx17.8$9 band there were no $t_{\rm peak}\sim59795$0 detections, with late-time $t_{\rm peak}\sim59795$1 upper limits $t_{\rm peak}\sim59795$2; at 99% confidence, two marginal detections at $t_{\rm peak}\sim59795$3–$t_{\rm peak}\sim59795$4 were reported around the onset of the X-ray flare (Koljonen et al., 2024).

The polarization angle evolved smoothly and significantly. An early $t_{\rm peak}\sim59795$5-band observation gave $t_{\rm peak}\sim59795$6; around peak, $t_{\rm peak}\sim59795$7 rotated to about $t_{\rm peak}\sim59795$8, a total change of roughly $t_{\rm peak}\sim59795$9 over a few tens of days. The inferred $o$0-band rotation rate was

$o$1

and the implied rate between the last $o$2-band detection and a later $o$3-band detection was $o$4, consistent within $o$5 with a constant rate (Koljonen et al., 2024). Several hundred days after peak, polarization became essentially undetectable.

The instrumental and astrophysical corrections were explicitly treated. Host-galaxy dilution was estimated from Pan-STARRS magnitudes converted to Johnson $o$6, and simulations with realistic flat-field residuals and sky backgrounds were used to verify polarization uncertainties. Strong Balmer lines contributed only 3% of the total flux in $o$7 and 8% in $o$8, so line depolarization was not expected to dominate (Koljonen et al., 2024).

These measurements disfavor a purely static electron-scattering photosphere. Two explanatory frameworks were proposed. In the outer-shock scenario, pericenter shocks dominate during the rise and outer shocks later add a second polarized component with a different position angle, so the vector sum produces depolarization and angle rotation. In the clumpy-torus or precessing-disk scenario, a compact scattering torus around a misaligned accretion flow yields smooth position-angle rotation through Lense–Thirring precession. The authors explicitly noted that the two pictures are not mutually exclusive.

5. X-ray, infrared, and continuum-SED behavior

Near optical maximum, AT 2022fpx was X-ray faint or undetected. Around MJD 59784, Swift/XRT gave a $o$9 upper limit of $m_o=18.97\pm0.08$0 in 0.3–1 keV, corresponding to $m_o=18.97\pm0.08$1 if the later blackbody spectrum is assumed (Koljonen et al., 2024). A soft X-ray flare emerged only around MJD $m_o=18.97\pm0.08$2, about 250 days after optical peak. Joint Swift/XRT and XMM-Newton spectral fitting with tbabs * zbb, fixing Galactic $m_o=18.97\pm0.08$3 and redshift $m_o=18.97\pm0.08$4, yielded

$m_o=18.97\pm0.08$5

at the XMM epoch, with Cash statistic $m_o=18.97\pm0.08$6 for 44 degrees of freedom (Koljonen et al., 2024). The temperature remained roughly constant while the flux changed, suggesting either a varying emitting area or variable obscuration.

The later X-ray analysis extended the monitoring to 75 Swift/XRT epochs plus an Einstein Probe stack. It resolved three phases: an early stacked low state with count rate $m_o=18.97\pm0.08$7, a high state at MJD 60071–60154 with $m_o=18.97\pm0.08$8 and $m_o=18.97\pm0.08$9, and a later low state with $g\sim20.1$00 and $g\sim20.1$01 (Lin et al., 7 Jul 2025). The Einstein Probe spectrum remained very soft, with $g\sim20.1$02 and $g\sim20.1$03 (Lin et al., 7 Jul 2025). The later paper therefore characterized the X-ray evolution not merely as a delayed flare but as a $g\sim20.1$04-day soft X-ray plateau with little spectral evolution.

Mid-infrared observations revealed a strong dust echo. Host-subtracted WISE/NEOWISE photometry was fit with a single-temperature blackbody peaking at $g\sim20.1$05, with dust temperature declining from $g\sim20.1$06 K to $g\sim20.1$07 K and effective radius expanding from $g\sim20.1$08 cm to $g\sim20.1$09 cm (Lin et al., 7 Jul 2025). This places hot dust at sub-parsec scales and implies powerful reprocessing of EUV/X-ray radiation by circumnuclear dust.

Continuum-SED modeling became a major interpretive issue. The 2025 study showed that optical-only blackbody fits for AT 2022fpx gave $g\sim20.1$10 K near peak, whereas optical+UV fits gave $g\sim20.1$11 K; after adopting a conservative internal extinction $g\sim20.1$12 mag from the Balmer decrement, the temperatures became $g\sim20.1$13 K and $g\sim20.1$14 K, respectively (Lin et al., 7 Jul 2025). Over $g\sim20.1$15–$g\sim20.1$16 Å, power laws $g\sim20.1$17 fit better than single blackbodies, with near-peak slopes around $g\sim20.1$18 in the observed SED and steeper values after extinction correction (Lin et al., 7 Jul 2025). A plausible implication is that much of the radiative output lies at shorter wavelengths than the optical bands alone would suggest.

6. Host galaxy, black-hole estimates, classification, and significance

Host-galaxy inference is model-dependent. One CIGALE analysis, using SDSS, Pan-STARRS, 2MASS, and AllWISE photometry, found $g\sim20.1$19, an old stellar population with average age $g\sim20.1$20 Gyr, a more recent starburst of age $g\sim20.1$21 Gyr, negligible current star formation with $g\sim20.1$22, $g\sim20.1$23 mag, and $g\sim20.1$24 (Koljonen et al., 2024). A later CIGALE fit, built from SDSS, Pan-STARRS, DESI Legacy Survey, 2MASS, unWISE, and AllWISE photometry, instead gave $g\sim20.1$25 and $g\sim20.1$26, explicitly indicating a weak AGN component (Lin et al., 7 Jul 2025). This suggests that the host characterization is sensitive to the adopted photometric set and SED assumptions.

Black-hole mass estimates are similarly broad. TDE-oriented modeling in the 2024 study yielded $g\sim20.1$27 from MOSFiT and $g\sim20.1$28 from the TDEmass outer-shock prescription, while empirical scalings gave $g\sim20.1$29 from peak luminosity, $g\sim20.1$30 from radiated energy, and $g\sim20.1$31 from host mass, with a global mean $g\sim20.1$32 (Koljonen et al., 2024). The 2025 study, using the $g\sim20.1$33 relation for low-mass AGN, derived $g\sim20.1$34 (Lin et al., 7 Jul 2025). Both analyses place the system in the broad regime usually associated with TDE hosts.

The classification debate has three components. First, a supernova origin is strongly disfavored: the later paper argued that the very long rise, ultra-soft and long-lived X-rays, persistent luminous coronal lines, exceptionally bright MIR echo, and absence of late nebular SN features are all atypical of SNe IIn (Lin et al., 7 Jul 2025). Second, a purely ordinary persistent AGN interpretation is also problematic, because the event shows a well-defined rise, peak, decline, delayed ultra-soft X-rays, and broad-line fading rather than stochastic AGN variability (Koljonen et al., 2024). Third, a weak pre-existing AGN is plausible: the 2025 study found two low-level pre-outburst MIR flares in NEOWISE and a non-zero AGN dust component in the host SED, while the 2024 study found no strong pre-flare variability in CRTS, NEOWISE, or Gaia and no radiatively dominant AGN in its SED fit (Lin et al., 7 Jul 2025).

Accordingly, the 2024 study classified AT 2022fpx as a TDE-like flare that bridges long-lived Bowen-fluorescence TDEs, ECLEs, and optical TDEs with delayed X-rays (Koljonen et al., 2024). The 2025 study favored a heavily dust-reddened TDE of a giant star in a weak AGN, but retained a turn-on AGN interpretation as the principal alternative (Lin et al., 7 Jul 2025). The source therefore occupies an important boundary case in the TDE–AGN interface.

Its broader significance extends beyond source taxonomy. Because the observed color remained red, $g\sim20.1$35, AT 2022fpx would have been rejected by many optical TDE selection pipelines using a blue-color cut. The later paper showed that line contamination can redden $g\sim20.1$36 by up to $g\sim20.1$37 mag, but that the continuum is still intrinsically red in the observed frame; with a conservative extinction correction of $g\sim20.1$38 mag, the intrinsic color becomes blue and the SED becomes consistent with TDE-like continua (Lin et al., 7 Jul 2025). This suggests that color-based optical TDE samples may exclude dust-reddened or line-dominated events, especially in gas-rich or weakly active nuclei.

AT 2022fpx is therefore significant both as an individual transient and as a selection-function test case. It demonstrates that ECLE-like spectra, delayed soft X-rays, variable optical polarization, strong dust echoes, and red observed colors can coexist within a single nuclear flare, and that the boundaries between quiescent-host TDEs, TDE-in-AGN events, Bowen-fluorescence flares, and ECLEs are at least partly observational rather than categorical.