AB Aurigae b: Protoplanet Candidate

• AB Aurigae b is a candidate embedded protoplanet identified at ~93 au via high-contrast imaging and polarimetric techniques.
• The source exhibits mixed near-infrared and UV/optical emissions, suggesting a blend of intrinsic radiation and scattered stellar light.
• Accretion tracers like Hα and Paβ, along with disk spiral features, imply weak or episodic accretion and a complex circumplanetary environment.

Searching arXiv for papers on AB Aurigae b to ground the article in the current literature. AB Aurigae b is a candidate embedded protoplanet in the protoplanetary disk of the Herbig Ae star AB Aurigae. In the current literature, the designation usually refers to the compact source detected at a projected separation of about $0.6''$, corresponding to roughly $93$–$100$ au, nearly due south of the star and within the millimeter dust cavity (Currie et al., 2022). Its interpretation remains disputed. One line of work treats it as a jovian protoplanet whose observed near-infrared emission is reprocessed radiation from an embedded source, possibly involving circumplanetary material; another finds that much of its UV and optical emission is consistent with scattered stellar light from a compact disk structure (Currie et al., 2022, Zhou et al., 2023). Subsequent H$\alpha$, Pa$\beta$, spectropolarimetric, and variability studies have narrowed the range of viable explanations without producing a universally accepted resolution (Zhou et al., 2022, Biddle et al., 2024, Dykes et al., 2024, Bowler et al., 20 Feb 2025, Currie et al., 25 Aug 2025).

1. Identification and observational history

The immediate antecedent to the modern AB Aurigae b discussion was not a direct point-source claim but a scattered-light anomaly. Radiative-transfer modeling of a darkened polarized-light region near $\sim 100$ au found that the observations were consistent with the absence of a planet and yielded an upper limit of $\leq 1\,M_J$, favoring a geometry/inclination/scattering explanation over a massive companion (Jang-Condell et al., 2010). That result established an important baseline: disk substructure at the relevant radius did not, by itself, require a planet.

The modern candidate emerged from direct imaging. Subaru/SCExAO-CHARIS and Hubble Space Telescope data revealed a compact source at $p \sim 0.59''$, corresponding to $\sim 93$ au, detected in nine datasets spanning four years and recovered in archival HST/NICMOS F110W total-intensity imaging from 2007 and in HST/STIS visible-light imaging from early 2021 (Currie et al., 2022). Over the 13-year HST baseline, its change in position angle was reported as consistent with counterclockwise orbital motion, and stationary-background interpretations were rejected at high significance (Currie et al., 2022).

Independent optical measurements placed the source at closely similar coordinates. HST/WFC3 H$\alpha$ imaging measured

$93$0

equivalent to a projected separation of $93$1 au (Zhou et al., 2022). UV/optical WFC3 imaging in F336W, F410M, and F645N gave an inverse-variance weighted mean of $93$2 and separation $93$3 mas (Zhou et al., 2023). Within the measurement uncertainties, these datasets refer to the same outer-disk source.

2. Morphology, spatial extent, and polarimetric behavior

In the near-infrared discovery data, AB Aur b was described as a bright concentrated source visible in $93$4, $93$5, and at least part of $93$6 band, with detection significance typically $93$7–12 and $93$8–12 in the highest-quality epochs (Currie et al., 2022). It was not merely point-like. In CHARIS, the apparent radius was $93$9, deconvolving to $100$0, or about $100$1 au (Currie et al., 2022). The HST/WFC3 H$100$2 study likewise found it somewhat broader than the instrumental PSF, with fitted FWHM values of $100$3 mas tangentially and $100$4 mas radially (Zhou et al., 2022). UV/optical WFC3 imaging inferred intrinsic sizes averaging $100$5 mas and $100$6 mas, corresponding to projected extents of about $100$7 au azimuthally and $100$8 au radially, with especially strong azimuthal elongation (Zhou et al., 2023).

Polarimetry has been central to the interpretation. The discovery paper reported that the source is detected in total intensity but not in polarized intensity, with a near-infrared polarized-intensity upper limit of about $100$9, lower than the surrounding disk’s $\alpha$0 (Currie et al., 2022). Later SCExAO/CHARIS integral-field spectropolarimetry reinforced this distinction: AB Aur b is detectable in complementary total-intensity data but is a non-detection in polarized light at $\alpha$1 (Dykes et al., 2024). In those data, any polarized signal near the source position is at most weak and featureless in the bluest $\alpha$2-band channels, while in $\alpha$3 and $\alpha$4 the emission is indistinguishable from the surrounding disk (Dykes et al., 2024).

The spectral contrast between the source and the disk is comparably important. The CHARIS study found that AB Aur b’s total-intensity spectrum is blue relative to the surrounding disk regions, while the disk’s polarization spectrum at the same location is redder than AB Aur b’s total-intensity spectrum (Dykes et al., 2024). That result was advanced as evidence that the source is not simply a bright knot of scattered light with the same spectral behavior as the local disk. A plausible implication is that the system contains multiple emission components, even if the precise partition between intrinsic and scattered contributions remains unsettled.

3. UV, optical continuum, and H$\alpha$5 photometry

HST/WFC3 direct imaging in the narrow-band F656N filter detected a point-like H$\alpha$6 source at the AB Aur b position in two epochs separated by about $\alpha$7 days (Zhou et al., 2022). The combined photometry yielded

$\alpha$8

equivalently

$\alpha$9

with consistent fluxes in the two individual epochs (Zhou et al., 2022). Using published HST/STIS continuum photometry, the inferred line-to-continuum ratio was

$\beta$0

whereas adopting the Subaru/VAMPIRES continuum estimate reduced it to $\beta$1 (Zhou et al., 2022). The central star itself has an F656N band-average of $\beta$2 times the continuum, showing that the companion’s line-to-continuum behavior is not strongly distinct from the accreting host star’s (Zhou et al., 2022). The authors therefore concluded that the H$\beta$3 detection alone does not validate AB Aur b as an accreting protoplanet (Zhou et al., 2022).

UV and optical continuum imaging sharpened that caution. In HST/WFC3-UVIS2 observations obtained in F336W, F410M, and F645N, the source was recovered in all three bands with signal-to-noise ratios of $\beta$4, $\beta$5, and $\beta$6, respectively (Zhou et al., 2023). Its spectral energy distribution peaks around $\beta$7, drops toward the UV, and shows Balmer-jump absorption-like behavior “mimicking those of early-type stars” (Zhou et al., 2023). The measured colors, $\beta$8 mag and $\beta$9 mag from forward modeling, were found to be consistent with the colors of the disk spirals and inconsistent with planetary photospheric or accretion-shock models (Zhou et al., 2023). The paper therefore concluded that the UV and visible emission from AB Aur b does not necessitate the presence of a protoplanet and is consistent with scattered stellar light (Zhou et al., 2023).

This created a specific interpretive tension. The near-infrared and polarimetric results disfavor a purely ordinary disk knot, whereas the UV/optical spectral energy distribution is consistent with scattered stellar light. That tension has remained a defining feature of the subject.

4. Accretion tracers: Pa$\sim 100$0, H$\sim 100$1 variability, and MUSE spectroscopy

Because hydrogen recombination lines can trace infalling gas, Pa$\sim 100$2 and H$\sim 100$3 observations have been used to test whether AB Aur b is actively accreting. Deep Keck/NIRC2 narrow-band Pa$\sim 100$4 imaging obtained on UT 2022 September 17 used angular differential imaging with $\sim 100$5 of field rotation and a total deep exposure of $\sim 100$6 minutes (Biddle et al., 2024). The final PSF-subtracted image and signal-to-noise map showed no significant source anywhere in the $\sim 100$7 field of view. At the expected location of AB Aur b, the signal-to-noise was $\sim 100$8 for the position reported by Currie et al. and $\sim 100$9 for the position reported by Zhou et al. (Biddle et al., 2024). The reported $\leq 1\,M_J$0 contrast was $\leq 1\,M_J$1 mag at $\leq 1\,M_J$2 for a spatially resolved source and $\leq 1\,M_J$3 mag if unresolved, implying a $\leq 1\,M_J$4 lower limit of $\leq 1\,M_J$5 for a spatially resolved AB Aur b, given the contemporaneous estimate $\leq 1\,M_J$6 for the star (Biddle et al., 2024). The inferred equivalent-width limits for the companion agreed with the stellar line strength at only $\leq 1\,M_J$7 in the resolved case and $\leq 1\,M_J$8 in the unresolved case, so the data did not establish a distinct accretion-line signature (Biddle et al., 2024). The authors concluded that if AB Aur b is a protoplanet, it is either not heavily accreting or was accreting only weakly at the time of the observations (Biddle et al., 2024).

That interpretation was contested almost immediately. A reanalysis argued that AB Aur b is decisively imaged with SCExAO/CHARIS at wavelengths covering Pa$\leq 1\,M_J$9, with contrast around $p \sim 0.59''$0, and that the Keck/NIRC2 non-detection resulted from much poorer image quality and an inaccurate source model (Currie, 2024). In that analysis, the NIRC2 stellar PSF had $p \sim 0.59''$1 and Strehl ratio $p \sim 0.59''$2, compared with near-infrared Strehl ratios of $p \sim 0.59''$3–$p \sim 0.59''$4 in the CHARIS data (Currie, 2024). Using an extended Gaussian source model with HWHM $p \sim 0.59''$5, yielding a convolved FWHM of $p \sim 0.59''$6, the paper revised the $p \sim 0.59''$7 upper limit on the Pa$p \sim 0.59''$8 line flux density upward to about $p \sim 0.59''$9, roughly three times higher than previously reported (Currie, 2024). It further argued that single-band Pa$\sim 93$0 imaging is intrinsically ill suited to conclusively identifying accretion onto AB Aur b (Currie, 2024).

H$\sim 93$1 monitoring then introduced a different diagnostic. A five-epoch HST/WFC3 program spanning $\sim 93$2 months used “accretion light echoes” to test whether the companion’s H$\sim 93$3 variability was simply delayed stellar variability (Bowler et al., 20 Feb 2025). For the inferred geometry, the true separation was $\sim 93$4 au, corresponding to $\sim 93$5 hr and a predicted observed echo delay of $\sim 93$6 hr (Bowler et al., 20 Feb 2025). Across the campaign, AB Aur’s H$\sim 93$7 flux changed by about $\sim 93$8, whereas AB Aur b varied by about $\sim 93$9, more than $\alpha$0 times the host-star variability (Bowler et al., 20 Feb 2025). The brightness changes were not correlated, which ruled out unobstructed scattered starlight from the host star as the only source of AB Aur b’s H$\alpha$1 emission, while remaining consistent with an independently accreting protoplanet, inner-disk shadowing effects, or a physically evolving compact disk structure (Bowler et al., 20 Feb 2025).

The strongest line-profile evidence to date came from VLT/MUSE. Medium-spectral-resolution $\alpha$2 observations in four 2022 observing blocks detected AB Aur b in emission at $\alpha$3–$\alpha$4 \AA, corresponding to roughly $\alpha$5, and in absorption at $\alpha$6–$\alpha$7 \AA, corresponding to about $\alpha$8 (Currie et al., 25 Aug 2025). The detection reached $\alpha$9 in the best reduction, and the resulting spectrum was found to be inconsistent with that of the host star or the average residual disk spectrum (Currie et al., 25 Aug 2025). The authors judged the shape to resemble an inverse P Cygni-like profile, which in classical stellar spectroscopy is interpreted as evidence of infalling material, while explicitly noting that nonaccretion origins could not be formally ruled out (Currie et al., 25 Aug 2025). This moved the debate from simple line excess to line morphology.

5. Relation to disk structure and neighboring features

AB Aurigae b is embedded in a disk with pronounced spirals, ring asymmetries, and compact inner structures. Near-infrared and thermal-infrared imaging with LBT/LMIRCam recovered the inner spiral structures inside the cavity at separations $93$00 and a broader ring-like structure centered near $93$01, roughly $93$02 au, with a strong southeast/northwest brightness asymmetry (Jorquera et al., 2022). Those data did not recover a planetary point source at the outer candidate location $93$03, arguing that the previously claimed source there is probably not a real companion, while leaving the inner $93$04 candidate at $93$05 unconstrained because of the $93$06 inner working angle (Jorquera et al., 2022). The ring’s location interior to the ALMA dust cavity was interpreted as evidence for dust trapping, though time-variable illumination was also discussed as an alternative explanation for the asymmetry (Jorquera et al., 2022).

Integral-field spectropolarimetry later strengthened the connection between AB Aur b and the surrounding spiral system. In SCExAO/CHARIS data covering $93$07–$93$08, the known inner spirals at $93$09 were recovered in every channel, and at the longest wavelengths the western spiral appeared to extend to $93$10–$93$11, just interior to and then clockwise from AB Aur b, coincident with the ALMA-detected CO gas spiral (Dykes et al., 2024). That spatial coincidence was presented as evidence for a physical connection between the near-infrared scattered-light spiral and the molecular gas spiral (Dykes et al., 2024).

Multi-epoch SPHERE/IRDIS polarimetry over $93$12 years found that the disk globally follows Keplerian rotation but departs from that behavior at radii smaller than $93$13 au, with a deviation as large as $93$14 over $93$15 years at $93$16 au, indicating sub-Keplerian rotation (Boccaletti et al., 26 May 2026). In that analysis, the two bright spirals inside the cavity had different dynamic trends. The western spiral, labeled $93$17, with mean angular speed about $93$18, was the only one whose motion could plausibly be dynamically linked to AB Aur b; the eastern spiral $93$19 rotated too slowly to be simply explained by a perturber at $93$20 (Boccaletti et al., 26 May 2026). In the same paper, H$93$21 imaging with SPHERE/ZIMPOL did not robustly detect AB Aur b: the continuum-subtracted flux at the expected position was $93$22–$93$23, while the extended inner feature $93$24 had $93$25, about $93$26 times brighter (Boccaletti et al., 26 May 2026). This reinforced the view that the inner disk hosts multiple compact structures whose morphologies and line fluxes should not be conflated with AB Aur b itself.

The system may contain additional localized phenomena unrelated to AB Aur b. M-band IRTF/iSHELL spectroastrometry detected an off-centered, low-temperature, compact CO-emitting source localized to about $93$27 au and $93$28, explicitly distinct from AB Aur b, which was not detected in those CO ro-vibrational lines (Kozdon et al., 27 Feb 2026). The existence of a separate compact source emphasizes that AB Aurigae is not a two-component system of star plus single candidate planet, but a structurally crowded environment in which different tracers can isolate different embedded features.

6. Physical interpretations, circumplanetary environment, and current status

The original protoplanet interpretation was physically specific. The discovery paper argued that a purely thermally emitting object of the observed size and temperature would be far too luminous, while an object with the measured luminosity and size would be implausibly cold to explain the near-infrared flux (Currie et al., 2022). It therefore modeled the source as reprocessed radiation from an embedded object, with the observed near-infrared signal arising from a $93$29 K source embedded in the disk and its light being scattered or processed by nearby dust (Currie et al., 2022). In that framework, the observed emission could be dominated by a circumplanetary envelope or disk rather than a bare planetary photosphere (Currie et al., 2022).

Predictions for a circumplanetary disk have been explored in the millimeter. A dedicated CPD study used $93$30 and $93$31, following the parameters inferred from the early imaging work, and found that the expected $93$32 mm dust continuum flux is below ALMA’s $93$33 non-detection limit of $93$34, consistent with the existing non-detection (Shibaike et al., 2024). However, once extinction by small grains is included, the inferred parameters shift to $93$35 and $93$36 or $93$37, and the predicted Band 6 continuum becomes stronger than the $93$38 limit for a typical inflow dust-to-gas ratio $93$39 over parts of parameter space (Shibaike et al., 2024). The paper therefore concluded that if AB Aur b is a planet with a CPD, the dust supply to its vicinity must be very small; alternatively, the non-detection is consistent with the possibility that the source is not a planet at all but a scattered-light feature (Shibaike et al., 2024). It recommended shorter-wavelength continuum observations, particularly ALMA Band 7, as a more decisive test (Shibaike et al., 2024).

At present, the literature supports several propositions simultaneously. AB Aur b is not a transient reduction artifact: it has been recovered in multiple instruments, in total intensity, over a long temporal baseline, with astrometry consistent across independent datasets (Currie et al., 2022, Zhou et al., 2022, Zhou et al., 2023). The simplest “static, unobstructed scattered-light echo of the star” model is disfavored by the H$93$40 light-echo experiment and by the MUSE line profile (Bowler et al., 20 Feb 2025, Currie et al., 25 Aug 2025). At the same time, the UV/optical spectral energy distribution, the modest line-to-continuum ratios in some datasets, and the Pa$93$41 non-detection or non-confirmation prevent a clean identification with a canonical accretion-shock source of the PDS 70 type (Zhou et al., 2022, Zhou et al., 2023, Biddle et al., 2024).

A plausible synthesis is that AB Aur b is embedded in, and observationally entangled with, a structured dusty environment whose scattered-light contribution is substantial, while some datasets may also contain locally generated emission related to accretion or infall. That synthesis remains an inference rather than a settled conclusion. Planned or proposed follow-up with higher-fidelity optical coronagraphy and polarimetry, including Roman CGI observations targeting $93$42–$93$43 contrast, has been motivated precisely by the need for less biased astrometry, photometry, and discrimination among emission sources (Currie et al., 2 Sep 2025). As of the current literature, AB Aurigae b remains one of the most intensively scrutinized and methodologically instructive protoplanet candidates in a natal disk.