betaFeedback

WorksPhysics and Astronomy

Where Did the Angular Momentum Go? WASP-12 b May Have Circularized on a Retrograde Orbit

Abstract

WASP-12 b is the only exoplanet with confirmed orbital decay. Angular momentum transferred from its decaying orbit should have realigned the star's thin convective layer, yet an earlier Rossiter–McLaughlin (RM) measurement suggested a ∼\sim60∘^\circ spin–orbit misalignment. We present two new Keck Planet Finder RM measurements. We measure a sky-projected obliquity of λ=−1.3±8.7∘λ=-1.3\pm8.7^\circ; and the star's projected rotation velocity is vsin⁡i⋆=1.3±0.1 km s−1v\sin i_\star=1.3\pm0.1~\mathrm{km\,s^{-1}}, an order of magnitude lower than expected for a star of its mass and age. With WASP-12's low chromospheric activity, log⁡RHK′=−5.5\log R^\prime_{\rm HK}=-5.5, our analysis favors an aligned system in which vsin⁡i⋆v\sin i_\star reflects slow surface rotation with Prot∼60P_{\rm rot}\sim60 days, rather than a projection effect. Our evolution models favor a late-main-sequence star rather than a subgiant, with a central convective core found in 99% of posterior samples. This internal structure suppresses the central-focusing and nonlinear wave-breaking mechanisms for inward-propagating gravity waves. We propose that WASP-12 b was initially placed on a retrograde orbit after high-eccentricity migration, when its host rotated at vsin⁡i⋆≈14v\sin i_\star\approx14~km~s−1^{-1} (or Prot∼P_{\rm rot}\sim6 days). Tidal decay transferred orbital angular momentum to the convective envelope, aligning the surface. The retrograde interior acts as an angular-momentum reservoir, while interior–envelope coupling slows the surface rotation. A semi-empirical relation gives an envelope–interior angular-momentum exchange timescale of ∼4\sim4~Myr, similar to the observed orbital-decay timescale of ∼3.25\sim3.25~Myr. The angular-momentum budget also permits this initially retrograde evolution. In contrast, an initially prograde planet would have spun up the surface rotation to a rate substantially higher than observed.

Corresponding author: Elina Y. Zhang and Fei Dai

Email: [email protected], [email protected]

Keywords: Transits (1711), Hot Jupiters (753), Exoplanet dynamics (490), Planet-star interactions (2177), Stellar rotation (1629), Tidal interaction (1699), Stellar evolution (1599)

Introduction

WASP-12 b is a hot Jupiter with a 1.09 day orbital period and an inflated radius of ≃1.9 RJup\simeq1.9\,R_{\mathrm{Jup}} ([44, 31, 43, 2] (S.-L. Li et al. 2010)). Long-baseline transit and occultation timings reveal steady orbital-period decay, P˙≃−29 ms yr−1\dot{P} \simeq-29\ \mathrm{ms\,yr}^{-1}, implying a period-decay timescale of ∼3.25 Myr\sim3.25\ \mathrm{Myr} and an effective stellar tidal quality factor of Q⋆′≈1.8×105Q'_{\star} \approx1.8 \times10^{5} ([63, 73, 109, 106, 2, 12]). It remains the only exoplanet with confirmed orbital decay ([109, 17, 105]).

Although orbital decay is well established, its efficient tidal-dissipation mechanism remains uncertain and depends sensitively on the host star’s evolutionary state and internal structure ([104, 6, 109, 99]). N. N. Weinberg et al. [104] showed that if WASP-12 is a subgiant without a central convective core, tidally excited internal gravity waves can reach the stellar center, where geometric focusing drives nonlinear wave breaking and dissipation consistent with the observed decay. However, A. Bailey & J. Goodman [6] favor a main-sequence star retaining a convective core, preventing these waves from reaching the center. With only weak linear damping, the resulting gg-mode resonances are too narrow to reproduce the decay without fine-tuning. Recent nonadiabatic tidal calculations likewise find that linear radiative and convective damping in main-sequence models with a convective core cannot reproduce WASP-12 b’s observed decay ([97]). Alternative efficient-dissipation mechanisms include absorption at a corotation layer associated with a rapidly rotating interior ([6]) and magnetic conversion of inward-propagating internal gravity waves into outward-propagating magnetic waves that dissipate in the radiative interior ([24]). The latter can operate in main-sequence stars with convective cores and may reproduce the inferred strong dissipation. Planetary obliquity tides are another possible

source ([69]), but recent angular-momentum and radial-velocity constraints disfavor this scenario ([58 submitted]). Their radial-velocity analysis also hints at an observable stellar tidal bulge, suggesting equilibrium tides operate alongside other tidal effects and may transfer at least part of the planet’s lost orbital angular momentum to the stellar surface layer.

Orbital decay also poses an angular momentum budget problem: the lost orbital angular momentum must go elsewhere in the system. Substantial initial deposition in the host’s shallow convective envelope should tend to produce surface spin-up and orbit–surface realignment, although internal coupling may redistribute it throughout the star. Yet a previous RM observation yielded λ=59−20+16\lambda= 59^{+16}_{-20}, suggesting substantial sky-projected orbit–surface misalignment ([3]). Moreover, we measure a low projected rotation velocity, vsin⁡i⋆=1.3±0.1 km s−1v\sin i_{\star} = 1.3 \pm0.1\ \mathrm{km\,s}^{-1} (§ 2), consistent with the spectroscopic 2.2±1.5 km s−12.2 \pm1.5\ \mathrm{km\,s}^{-1} reported by L. Hebb et al. [44] and compiled by A. S. Bonomo et al. [16]. WASP-12 also has exceptionally weak Ca II H&K and Mg II emission ([57, 43, 30]). H. A. Knutson et al. [57] reported log⁡RHK′=−5.50\log R'_{\mathrm{HK}} = -5.50 without uncertainty; we adopt a typical 0.05 dex. Its activity lies near the low end for similar stars (Figure 3). The slow rotation and previously reported misalignment could arise if tidal angular momentum is deposited below the surface convective layer rather than retained in the envelope. Accurate spin–orbit and surface-rotation measurements, with improved stellar internal-structure constraints, therefore help identify the ultimate sink of the planet’s lost orbital angular momentum.

Joint posterior distributions, orientation-marginalized likelihood, and chromospheric activity–Rossby diagram

Figure 3. Top: Joint posterior distributions of true obliquity, ψ\psi, and azimuth, ϕ\phi, for the fast-rotation (upper) and slow-rotation (lower) models. Colors show log⁡10(p/pmax⁡)\log_{10}(p/p_{\max}), normalized separately in each panel. Middle: Orientation-marginalized likelihood of the equatorial rotation velocity, normalized to its maximum. The pink dashed line marks the aligned solution; the purple dashed line and shaded band show the prediction of K. C. Schlaufman [88] and its 1σ1\sigma uncertainty. Bottom: Chromospheric activity versus solar-normalized Rossby number for the LAMOST–Kepler sample of H. Yang et al. [108], colored by effective temperature. Black outlines identify stars with ∣Teff−6155 K∣≤200 K|T_{\rm eff} - 6155\ {\rm K}| \leq200\ {\rm K} and ∣log⁡g−4.1428∣≤0.05|\log g - 4.1428| \leq0.05. The star and diamond mark WASP-12 assuming the predicted rotation period of 5.89 d and the vsin⁡i⋆v\sin i_{\star}-based maximum period of 64 d, respectively. The shaded region marks log⁡RHK′<−5.0\log R'_{\rm HK} < -5.0, where the LAMOST activity measurements become increasingly uncertain.

We present two new Rossiter–McLaughlin (RM, [85, 66]) observations of WASP-12 b with the Keck Planet Finder (KPF, [34, 35, 36]), jointly analyzed with TESSTESS ([84]) photometry. The sky-projected obliquity λ=−1.3±8.7∘\lambda= -1.3 \pm8.7^{\circ} and vsin⁡i⋆=1.3±0.1 km s−1v\sin i_{\star} = 1.3 \pm0.1\ \mathrm{km\,s}^{-1} favor alignment and anomalously slow surface rotation. Our stellar-evolution and magnetic-braking models favor a late-main-sequence host with a central convective core, a thin outer convective envelope, and much faster surface rotation than observed. We propose that orbital angular momentum was deposited partly in the surface convective layer, aligning it with the orbit. If WASP-12 b began on a retrograde orbit, a retrograde stellar interior could absorb this angular momentum by slowing its retrograde rotation rather than spinning up prograde. Interior–envelope coupling is likely fast enough for this redistribution, and the angular-momentum budget permits the required transfer. Section 2 presents the stellar-obliquity measurement, and Section 3 describes the stellar-evolution models. Section 4 combines these results to explore an initially retrograde WASP-12 b; Section 5 summarizes our findings.

Stellar Obliquity Measurements

TESS Light Curves

The Transiting Exoplanet Survey Satellite (TESS; G. R. Ricker et al. [84]) observed WASP-12 in Sectors 20, 43, 44, 45, 71, and 72. We analyzed 120-s-cadence light curves from the TESS Science Processing Operations Center (SPOC; J. M. Jenkins et al. [53]), publicly available through the Mikulski Archive for Space Telescopes (MAST)3 at DOI: 10.17909/0rhv-mm50.

We used the Presearch Data Conditioning Simple Aperture Photometry (PDCSAP) light curves provided by the SPOC pipeline (J. C. Smith et al. [90]; M. C. Stumpe et al. [96]). The light curves were downloaded using the lightkurve package (Lightkurve Collaboration et al. [62]). Any NaN values were removed and significant outliers were rejected using a 10σ10\sigma clipping threshold.

To isolate WASP-12 b’s transit signal, we masked transits within windows of 1.5 times the known duration. We detrended the masked light curves with lightkurve’s flatten routine, which uses a Savitzky–Golay filter (A. Savitzky & M. J. E. Golay [87]) to remove low-frequency stellar-activity variability and residual instrumental trends. The normalized light curves preserve transit profiles and photometric uncertainties for subsequent transit and obliquity analyses.

KPF Rossiter–McLaughlin Observations

We obtained high-resolution spectra of two WASP-12 b transits with the Keck Planet Finder (KPF; S. R. Gibson et al. [34, 35, 36]) on Keck I at the W. M. Keck Observatory. KPF covers 445–870 nm at resolving power R∼98,000R \sim98{,}000. Observations on UT 2024 March 23 and UT 2026 February 2 yielded 22 and 45 reduced spectra, respectively, with 600 s integrations. The first sequence covers the full transit and roughly 30 minutes each before ingress and after egress; the second extends approximately 140 minutes on either side. Extracted spectra typically have a signal-to-noise ratio of approximately 100 per pixel at 550 nm. Substantially poorer seeing during the latter half of the 2024 March 23 observations reduced this to ∼50\sim50 per pixel at 550 nm and increased RV uncertainties for several late exposures.

The spectra were reduced using the KPF Data Reduction Pipeline (DRP), version 2.12.0, applying bias subtraction, flat-field correction, order extraction, and wavelength calibration. Radial velocities were calculated using the cross-correlation function (CCF) method with numerical stellar masks adapted from those developed for ESPRESSO (F. Pepe et al. [79]). Table 1 lists the RV measurements used to model the RM signal.

TimeRVσRV\sigma_{\mathrm{RV}}TimeRVσRV\sigma_{\mathrm{RV}}
(BJD)(km/s)(m/s)(BJD)(km/s)(m/s)
3392.744319.106573.044073.703719.202512.74
3392.749519.105363.134073.711619.201152.66
3392.758119.095222.554073.719219.189612.67
3392.765319.093742.714073.726319.187733.41
3392.773319.086702.954073.734219.175132.93
3392.780519.078002.774073.741919.167752.77
3392.787919.068332.864073.749319.157882.67
3392.795919.056873.124073.756919.150442.71
3392.803219.048662.854073.764219.142862.73
3392.810619.033222.824073.771819.136732.72
3392.818219.019742.914073.779219.121972.98
3392.825719.007822.884073.786819.113502.95
3392.832918.999843.164073.794519.105442.93
3392.840718.985293.534073.802019.099682.89
3392.857318.965446.264073.809319.094412.75
3392.864918.952426.034073.816919.083192.77
3392.872818.966975.534073.824319.076402.80
3392.880218.962295.184073.831919.064552.94
3392.887518.940055.254073.839519.056152.80
3392.895218.929285.474073.846919.046402.61
3392.902618.926385.344073.854419.029162.60
3392.909818.916985.864073.861919.016822.60
–––4073.869419.000592.76
–––4073.877118.994522.74
–––4073.884318.978213.07
–––4073.891818.968103.07
–––4073.899418.954483.11
–––4073.907218.945102.85
–––4073.914418.941012.59
–––4073.921918.940262.66
–––4073.929418.928612.70
–––4073.937018.920942.79
–––4073.944618.912722.49
–––4073.959418.896062.44
–––4073.967018.887492.48
–––4073.974418.876722.61
–––4073.982018.870922.58
–––4073.989518.863482.61
–––4073.997018.859212.60
–––4074.004418.852552.79
–––4074.011918.850003.08
–––4074.019518.841613.49
–––4074.026918.842924.52
–––4074.034518.829655.67
–––4074.042018.815647.82

Table 1. WASP-12 KPF RM Data

Joint Light Curve and Rossiter–McLaughlin Modeling

To characterize WASP-12’s spin–orbit architecture, we jointly fit the TESS transit photometry and KPF radial velocities with a modified allesfitter4 package (M. N. Günther & T. Daylan [41, 40]; X.-Y. Wang et al. [103]; X.-Y. Wang [102]). The modification incorporates the analytic RM prescription of T. Hirano et al. [50], extending the formulation of Y. Ohta et al. [71]; T. Hirano et al. [49]. Simultaneously modeling the transit light curve and spectroscopic RM anomaly enables self-consistent measurement of the projected stellar obliquity λ\lambda and other transit parameters.

The TESS data contain multiple 2-min-cadence transits of WASP-12 b (§ 2.1), while the spectroscopic data comprise two KPF transit sequences from UT 2024 March 23 (KPF1) and UT 2026 February 2 (KPF2), treated as independent radial-velocity datasets. The transit model includes the planet-to-star radius ratio Rp/R⋆R_{\mathrm{p}}/R_{\star}, scaled sum of radii (R⋆+Rp)/a(R_{\star}+R_{\mathrm{p}})/a, orbital-inclination cosine cos⁡ip\cos i_{\mathrm{p}}, and period PP. Because WASP-12 b shows measurable transit-timing variations from orbital decay, we fitted independent transit midpoints (T0)(T_{0}) for KPF1 and KPF2. We assumed a circular orbit, justified by the planet’s short period and available RV data.

We modeled the radial velocities as the RM signal plus an independent 3rd-order polynomial baseline for each KPF dataset. Because WASP-12 b has a short period (P≃1.09P \simeq1.09 d), stellar radial velocity changes measurably over the ∼3.0\sim3.0 hr transit. The baselines flexibly describe low-frequency variations, including orbital motion, the tidal RV effect, and possible stellar or instrumental variability. Since the localized in-transit anomaly primarily Note—Times are given as BJD−2457000. Radial velocities are reported in km s−1^{-1}, while their uncertainties are reported in m s−1^{-1}.

constrains the RM signal, these low-order trends mainly absorb longer-timescale variations and are not expected to significantly affect λ\lambda. Fitting the out-of-transit RV variation with the planet’s Keplerian orbit gave nearly identical obliquity constraints. The RM model adds two free parameters: projected stellar rotation vsin⁡i⋆v\sin i_\star and projected spin–orbit angle λ\lambda.

We assumed quadratic limb darkening for the photometric and spectroscopic datasets, parameterized by the q1q_1 and q2q_2 parameters of D. M. Kipping [56], with independent coefficients for the TESS and KPF bandpasses. To account for white noise beyond formal uncertainties, we included a photometric error-inflation term, ln⁡σTESS\ln\sigma_{\mathrm{TESS}}, and independent radial-velocity jitter terms, ln⁡σjitter;KPF1\ln\sigma_{\mathrm{jitter;KPF1}} and ln⁡σjitter;KPF2\ln\sigma_{\mathrm{jitter;KPF2}}, added in quadrature to the two KPF RV uncertainties.

Uniform priors, U(a,b)\mathcal{U}(a,b), were assigned to all fitted parameters unless noted otherwise. We imposed a Gaussian prior on stellar mean density from the mass and radius in Table 3 (§ 3). Following T. Hirano et al. [50], we also modeled the micro- and macroturbulent velocities, VξV_\xi and VζV_\zeta, with 1 km s−1^{-1} Gaussian priors centered on empirical calibrations. We used H. Bruntt et al. [18] and A. P. Doyle et al. [23] where applicable and otherwise the Gaia–ESO Survey calibrations of S. Blanco-Cuaresma et al. [15]; S. Blanco-Cuaresma [13].

ParameterValueUnit
Modeling inputs
Effective temperature TeffT_{\mathrm{eff}}6155±1006155 \pm100K
Metallicity [M/H]0.15±0.060.15 \pm0.06dex
Mean stellar density ρˉ\bar{\rho}0.4399±0.00970.4399 \pm0.0097g cm−3^{-3}
Luminosity L⋆L_{\star}3.75±0.133.75 \pm0.13L⊙L_{\odot}
Estimated parameters
Initial helium fraction YinitY_{\mathrm{init}}0.252±0.0300.252 \pm0.030
Mixing-length parameter αMLT\alpha_{\mathrm{MLT}}2.10±0.382.10 \pm0.38
Mass M⋆M_{\star}1.353±0.0591.353 \pm0.059M⊙M_{\odot}
Radius R⋆R_{\star}1.632±0.0251.632 \pm0.025R⊙R_{\odot}
Age tt3.44±0.563.44 \pm0.56Gyr
Surface gravity log⁡g\log g4.1428±0.00944.1428 \pm0.0094cm s−2^{-2}
Convective core prob.0.990±0.0020.990 \pm0.002
Convective envelope prob.1.000±0.0001.000 \pm0.000
BCZ radius1^{1} rbczr_{\mathrm{bcz}}1.281±0.0971.281 \pm0.097R⊙R_{\odot}
Norm. BCZ radius rbcz/R⋆r_{\mathrm{bcz}}/R_{\star}0.783±0.0560.783 \pm0.056
BCZ mass coordinate mbczm_{\mathrm{bcz}}1.345±0.0581.345 \pm0.058M⊙M_{\odot}
Norm. BCZ mass coordinate mbcz/M⋆m_{\mathrm{bcz}}/M_{\star}0.9951±0.00610.9951 \pm0.0061
ICZ radius1^{1} riczr_{\mathrm{icz}}0.1072±0.00590.1072 \pm0.0059R⊙R_{\odot}
Norm. ICZ radius1^{1} ricz/R⋆r_{\mathrm{icz}}/R_{\star}0.0658±0.00330.0658 \pm0.0033
ICZ mass coordinate1^{1} miczm_{\mathrm{icz}}0.0973±0.00650.0973 \pm0.0065M⊙M_{\odot}
Norm. ICZ mass coordinate1^{1} micz/M⋆m_{\mathrm{icz}}/M_{\star}0.0720±0.00400.0720 \pm0.0040
Envelope moment of inertia log⁡Ienv\log I_{\mathrm{env}}52.92±0.5652.92 \pm0.56g cm2^{2}
Total moment of inertia log⁡I⋆\log I_{\star}54.224±0.06954.224 \pm0.069g cm2^{2}

Table 3. WASP-12 Stellar Modeling Results. Note—1. Estimated based on models that possess an inner convection zone (core).

Posterior distributions were sampled using the nested sampling algorithm implemented in the dynesty package (E. Higson et al. [48]; J. S. Speagle 2020), as integrated within allesfitter. The sampling was initialized with 500 live points and employed a random-walk (rwalk) proposal strategy. The final posteriors contain 45,918 samples. We assessed convergence using the default criteria, posterior inspection, and the Anderson–Darling, Augmented Dickey–Fuller, Durbin–Watson, and Ljung–Box residual diagnostics provided by allesfitter. Figure 1 shows the best-fit RM model and detrended TESS transits; Table 2 summarizes median posteriors, 68% credible intervals, and derived physical properties.

KPF Rossiter–McLaughlin observations, TESS transit light curve, and HIRES–KPF RM comparison

Figure 1. Left: KPF Rossiter–McLaughlin observations obtained on UT 2024 March 23 (KPF1) and UT 2026 February 2 (KPF2). For each observing sequence, the left panels show the full radial-velocity time series including the fitted baseline and RM model, while the right panels show the isolated RM anomaly after baseline removal. The best-fit model yields a projected stellar obliquity of λ=−1.3±8.7∘\lambda= -1.3 \pm8.7^\circ. Residuals are shown in the lower panels. Upper right: Phase-folded TESS transit light curve of WASP-12 b together with the best-fit transit model from the joint allesfitter analysis. The lower panel shows the residuals. Lower right: Baseline-subtracted RM measurements from HIRES (blue circles), KPF1 (pink squares), and KPF2 (purple diamonds). Error bars include the fitted RV jitter added in quadrature. The solid blue and dashed purple curves show the posterior-median HIRES model with a uniform prior on vsin⁡i⋆v \sin i_\star and the joint KPF–TESS model, respectively. The shaded region spans the transit window, and the vertical dotted line marks mid-transit.

ParameterPriorValueUnit
Rb/R⋆R_b/R_\starU(0,0.2)U(0, 0.2)0.11618±0.000400.11618 \pm0.00040
(R⋆+Rb)/ab(R_\star+ R_b)/a_bU(0,0.5)U(0, 0.5)0.3689±0.00280.3689 \pm0.0028
cos⁡ib\cos i_bU(0,0.2)U(0, 0.2)0.1243−0.0083+0.00770.1243^{+0.0077}_{-0.0083}
T0;b−2457000T_{0;b} - 2457000 *U(2928.8,2929.1)U(2928.8, 2929.1)2928.964648±0.0000462928.964648 \pm0.000046BJD
PbP_b **U(1.0,1.2)U(1.0, 1.2)1.09141644±0.000000111.09141644 \pm0.00000011d
ebcos⁡ωb\sqrt{e_b}\cos\omega_bfixed0.0
ebsin⁡ωb\sqrt{e_b}\sin\omega_bfixed0.0
q1;TESSq_{1;\mathrm{TESS}}U(0,1)U(0, 1)0.247−0.038+0.0440.247^{+0.044}_{-0.038}
q2;TESSq_{2;\mathrm{TESS}}U(0,1)U(0, 1)0.236−0.056+0.0620.236^{+0.062}_{-0.056}
q1;KPF1q_{1;\mathrm{KPF1}}U(0,1)U(0, 1)0.58−0.28+0.260.58^{+0.26}_{-0.28}
q2;KPF1q_{2;\mathrm{KPF1}}U(0,1)U(0, 1)0.49±0.320.49 \pm0.32
q1;KPF2q_{1;\mathrm{KPF2}}U(0,1)U(0, 1)0.46−0.22+0.260.46^{+0.26}_{-0.22}
q2;KPF2q_{2;\mathrm{KPF2}}U(0,1)U(0, 1)0.64−0.35+0.250.64^{+0.25}_{-0.35}
ln⁡σTESS\ln\sigma_{\mathrm{TESS}}U(−10,0)U(-10, 0)−6.1760±0.0041-6.1760 \pm0.0041ln⁡(rel. flux)\ln(\mathrm{rel.\ flux})
ln⁡σjitter;KPF1\ln\sigma_{\mathrm{jitter;KPF1}}U(−10,0)U(-10, 0)−8.3±1.2-8.3 \pm1.2ln⁡(km s−1)\ln(\mathrm{km\ s}^{-1})
ln⁡σjitter;KPF2\ln\sigma_{\mathrm{jitter;KPF2}}U(−10,0)U(-10, 0)−8.58−0.96+1.1-8.58^{+1.1}_{-0.96}ln⁡(km s−1)\ln(\mathrm{km\ s}^{-1})
λ⋆\lambda_\starU(−180,180)U(-180, 180)−1.3±8.7-1.3 \pm8.7deg
vsin⁡i⋆v\sin i_\starU(0,10)U(0, 10)1.29±0.111.29 \pm0.11km s−1\mathrm{km\ s}^{-1}
VξV_\xiN(1.0,1)\mathcal{N}(1.0, 1)1.21−0.73+0.901.21^{+0.90}_{-0.73}km s−1\mathrm{km\ s}^{-1}
VζV_\zetaN(2.86,1)\mathcal{N}(2.86, 1)2.83±0.992.83 \pm0.99km s−1\mathrm{km\ s}^{-1}
Derived Parameters
ab/R⋆a_b/R_\star–3.026−0.021+0.0223.026^{+0.022}_{-0.021}
RbR_b–1.891±0.0401.891 \pm0.040RJupR_{\mathrm{Jup}}
aba_b–0.02354±0.000520.02354 \pm0.00052AU
ibi_b–82.86−0.45+0.4882.86^{+0.48}_{-0.45}deg
btra;bb_{\mathrm{tra;b}}–0.376−0.023+0.0210.376^{+0.021}_{-0.023}
Ttot;bT_{\mathrm{tot;b}}–2.9815±0.00652.9815 \pm0.0065h
Tfull;bT_{\mathrm{full;b}}–2.248−0.013+0.0142.248^{+0.014}_{-0.013}h
Teq;bT_{\mathrm{eq;b}}–2324±422324 \pm42K
δtr;b\delta_{\mathrm{tr;b}}–15.067±0.03915.067 \pm0.039ppt
ρ⋆;combined\rho_{\star;\mathrm{combined}} ***–0.4399−0.0091+0.00970.4399^{+0.0097}_{-0.0091}g cm−3\mathrm{g\ cm}^{-3}

Table 2. Joint allesfitter Model Parameters for WASP-12

For 2024 March 23, we measure a mid-transit time of T0=2460392.8165±0.0018T_0 = 2460392.8165 \pm0.0018 BJDTDB, corresponding to O − C = −3.72±2.59-3.72 \pm2.59 min. For 2026 February 2, we measure T0=2461073.8602±0.0015T_0 = 2461073.8602 \pm0.0015 BJDTDB, corresponding to O − C = −6.13±2.16-6.13 \pm2.16 min. Figure 2 compares these measurements with previously published transit and occultation timings and with the quadratic orbital-decay ephemeris of B. Akinşanmi et al. [2],

Transit and occultation timing deviations for WASP-12 b

Figure 2. Transit and occultation timing deviations (O − C) for WASP-12 b relative to the linear ephemeris. Navy points show timings compiled by K. C. Patra et al. [73] and S. W. Yee et al. [109], blue points show TESS measurements from I. Wong et al. [106], and pink points show CHEOPS measurements from B. Akinsanmi et al. [2]. The orchid curve shows the quadratic orbital-decay ephemeris from B. Akinsanmi et al. [2]. Purple stars mark the KPF RM-derived timings at E=3014E = 3014 and 36383638. Both timings are consistent with the extrapolated decay trend within their uncertainties.

T(E)=T0+PE+12dPdEE2,(1)T(E) = T_0 + PE + \frac{1}{2}\frac{dP}{dE}E^2, \tag*{(1)}

where T0=2457103.2837T_0 = 2457103.2837 BJDTDB, P=1.0914P = 1.0914 d, and dP/dE=−1.043×10−9dP/dE = -1.043 \times10^{-9} d epoch−1^{-1}. Both KPF timings are consistent with the extrapolated quadratic ephemeris within their uncertainties. Although less precise than dedicated photometric measurements, they provide two independent checks, separated by nearly two years, of the established orbital-decay trend. We did not attempt an independent measurement of the orbital-decay rate because the RM-derived timings are not sufficiently precise.

Reanalysis of the 2012 HIRES Observations

To investigate the discrepancy with the earlier projected obliquity, λ=59−10+15\lambda= 59^{+15}_{-10} [3], we reanalyzed 38 Keck/HIRES RVs from the UT 2012 January 1-2 transit using the same modified allesfitter and RM prescription as for KPF. We fixed transit geometry, including impact parameter, to the joint KPF–TESS best fit and varied the HIRES transit midpoint, quadratic limb-darkening coefficients, RV jitter, λ\lambda, and vsin⁡i⋆v \sin i_{\star}. As for KPF, a 3rd-order polynomial baseline described orbital and other slowly varying RV signals.

We performed three fits that differed only in the prior adopted for vsin⁡i⋆v \sin i_{\star}. A uniform prior, vsin⁡i⋆∼U(0,10)v \sin i_{\star} \sim U(0, 10) km s−1^{-1}, yielded λ=44.8−23.9+15.1\lambda= 44.8^{+15.1}_{-23.9}° and vsin⁡i⋆=1.3−0.4+0.5v \sin i_{\star} = 1.3^{+0.5}_{-0.4} km s−1^{-1}; the earlier spectroscopic constraint, vsin⁡i⋆=2.2±1.5v \sin i_{\star} = 2.2 \pm1.5 km s−1^{-1} [44], yielded λ=46.5−22.8+14.7\lambda= 46.5^{+14.7}_{-22.8}° and vsin⁡i⋆=1.4−0.4+0.5v \sin i_{\star} = 1.4^{+0.5}_{-0.4} km s−1^{-1}. The informative KPF posterior prior, vsin⁡i⋆=1.3±0.1v \sin i_{\star} = 1.3 \pm0.1 km s−1^{-1}, gave λ=40.8−16.3+13.1\lambda= 40.8^{+13.1}_{-16.3}°. Changing the prior shifts the obliquity by less than 6°, substantially below its statistical uncertainty. The uninformative fit independently favors low vsin⁡i⋆v \sin i_{\star} consistent with KPF; different rotation priors therefore do not primarily explain the HIRES–KPF obliquity discrepancy.

Figure 1 (lower right) compares baseline-subtracted HIRES, KPF1, and KPF2 observations at common orbital phases. The jointly modeled KPF nights share one RM solution. HIRES’s preference for moderate misalignment appears driven mainly by several first-half-transit measurements favoring a more asymmetric RM waveform than the joint KPF model. The obliquity difference is nevertheless modest, approximately 2σ2\sigma. Given the lower precision of the HIRES observations, the apparent asymmetry may reflect a statistical fluctuation or an unmodeled systematic; our reanalysis does not distinguish between these possibilities.

Stellar Evolution Model

Non-rotating Models

To better constrain the internal structure of WASP-12, we compute a grid of stellar evolution models using the Modules for Experiments in Stellar Astrophysics (MESA; [74, 75, 76, 77, 78, 54]), version r25.12.1. Each model is evolved from the pre-main-sequence (with central temperature Tc=9.9×105T_{\mathrm{c}} = 9.9 \times10^{5} K) through the main sequence and subgiant phase, until either the surface temperature drops below Teff=5500T_{\mathrm{eff}} = 5500 K or the photospheric radius exceeds R⋆=3R⊙R_{\star} = 3R_{\odot}.

We adopt the AGSS09 solar abundance mixture [5] with corresponding opacity tables (kap_file_prefix = ‘a09’), and the pp_and_cno_extras nuclear-reaction network. Convection is treated using the [46] formulation of mixing-length theory (MLT), with the Ledoux criterion for convective stability and semiconvection and thermohaline coefficients following [21]. Predictive mixing is applied at convective-core boundaries to avoid composition reversals. Exponential convective overshooting is included at both the core and envelope base. Atomic diffusion and gravitational settling of H, He, C, N, O, Ne, and Mg are enabled following [98], with diffusion suppressed where the convective diffusion coefficient is Dmix>100D_{\mathrm{mix}} > 100 cm2^{2} s−1^{-1}. The atmosphere is attached using the [25] TT–τ\tau relation with iterated opacities. A minimum diffusion coefficient of Dmin⁡=10D_{\min} = 10 cm2^{2} s−1^{-1} is imposed throughout to account for other mixing processes.

The stellar models are constrained by four observables derived from spectroscopic or transit analyses of WASP-12. The effective temperature Teff=6155±100T_{\mathrm{eff}} = 6155 \pm100 K and bulk metallicity [M/H] =0.15±0.06= 0.15 \pm0.06 come from SpecMatch-Syn [80] analysis of a high-S/N iodine-free Keck/HIRES spectrum obtained UT 2012 Jan 5. Our transit analysis gives the mean stellar density ρˉ=0.4399±0.0097\bar{\rho} = 0.4399 \pm0.0097 g cm−3^{-3}. The luminosity L=3.75±0.13L⊙L = 3.75 \pm0.13L_{\odot} comes from isoclassify [52], using the Gaia parallax, 2MASS KsK_{\mathrm{s}} magnitude with bolometric correction, and AKs≈0.3A_{K_{\mathrm{s}}} \approx0.3 mag extinction from the [38] dust map. For each grid model, L∝exp⁡(−χ2/2)\mathcal{L} \propto\exp(-\chi^{2}/2), where χ2=∑i[(Oi−Mi)/σi]2\chi^{2} = \sum_i [(O_i - M_i)/\sigma_i]^2 over these four observables. Marginalizing over all grid models yields posterior distributions for every output quantity.

From the constrained ensemble, we infer global and internal properties including M⋆M_{\star}, R⋆R_{\star}, age tt, and surface gravity log⁡g\log g. The models favor three layers: a central convective core, overlying radiative layer, and thin outer convective envelope. WASP-12 retains a convective core in 99% of the posterior, strongly favoring a star not yet evolved to a post-main-sequence structure lacking one. Throughout, “convective core” denotes the central convective region, while “stellar interior below the convective envelope” collectively denotes the radiative layer and convective core.

Of particular interest are the convective-zone-base radius rbczr_{\mathrm{bcz}}, fractional mass coordinate mbcz/M⋆m_{\mathrm{bcz}}/M_{\star}, and moments of inertia of the convective envelope and underlying stellar interior. The envelope is very shallow in mass: mbcz/M⋆=(99.51±0.61)%m_{\mathrm{bcz}}/M_\star= (99.51 \pm0.61)\%, implying ∼0.5%\sim0.5\% of the stellar mass, yet it contributes approximately 5% of the total moment of inertia, with log⁡Ienv=52.92±0.56\log I_{\mathrm{env}} = 52.92 \pm0.56 versus log⁡I⋆=54.22±0.07\log I_\star= 54.22 \pm0.07 in cgs units. These quantities determine how strongly the observable surface responds to angular momentum transferred from the decaying orbit and redistributed within the star, and are central to the tidal angular-momentum budget in Section 4.2. Table 3 summarizes the inferred parameters and adopted uncertainties.

The inferred evolutionary state of WASP-12 bears directly on the origin of its rapid orbital decay. N. N. Weinberg et al. [104] showed that the observed dissipation rate can be explained by nonlinear breaking of tidally excited internal gravity waves if WASP-12 is a subgiant that has lost its central convective core. In this configuration, inward-propagating waves can reach the stellar center, where geometric focusing drives wave breaking.

wave breaking. Within our MESA-based stellar interior model, a convective core, rather than a radiative core, is found in 99% of posterior samples. However, unmodeled physics and systematic uncertainties in the MESA calculations could alter this conclusion and potentially favor a radiative core.

This does not imply that the orbital decay is incompatible with a main-sequence host; rather, under this interpretation, a different dissipation mechanism must operate. In main-sequence models, the convective core prevents gravity waves from reaching the stellar center, while linear radiative and convective damping appears insufficient to explain the measured decay rate without fine-tuning [6, 97]. Mechanisms that operate in the presence of a convective core, such as absorption near an interior corotation layer [6] or magnetic conversion and subsequent dissipation of internal gravity waves [24], therefore warrant particular attention. Our proposed angular-momentum-transfer scenario does not identify the mechanism responsible for the observed tidal dissipation; it requires only that a fraction of the lost orbital angular momentum is temporarily deposited in the surface layers. The grid uses Sobol quasi-random sampling ([91, 10]) over initial mass M∈[0.9,1.6]M⊙M \in[0.9, 1.6]M_\odot, metallicity [M/H]∈[−0.1,0.5][\mathrm{M}/\mathrm{H}] \in[-0.1, 0.5], mixing length αMLT∈[1.5,2.6]\alpha_{\mathrm{MLT}} \in[1.5, 2.6], and initial helium fraction Yinit∈[0.22,0.32]Y_{\mathrm{init}} \in[0.22, 0.32]. We predict rotational evolution following T. S. Metcalfe et al. [67, 68],5 with three modifications: (1) initial rotation periods are drawn from a Gaussian centered on 8 days with a 2-day standard deviation, with disk locking released at 1 Myr; (2) mean density, but not rotation period, is included with luminosity, effective temperature, and surface metallicity as an observational constraint; and (3) systematic model errors are omitted.

Magnetic Braking Models

To estimate WASP-12’s expected rotation rate at its current evolutionary stage, we used the rotevol tracer code ([81]) on a YREC model grid (described in [67]), adopting the standard magnetic-braking law and N. Saunders et al. [86] calibration based on S. Matt & R. E. Pudritz [65] and J. L. van Saders & M. H. Pinsonneault (2013). We obtain veq=14−1+11v_{\mathrm{eq}} = 14^{+11}_{-1} km s−1^{-1} and Prot=5.9−2.5+3.1P_{\mathrm{rot}} = 5.9^{+3.1}_{-2.5} d, with mass, mixing length, and age agreeing within 1σ1\sigma with the MESA models above. The broad period posterior reflects weak magnetic braking from the shallow convective envelope and sensitivity to the assumed initial rotation. Nevertheless, the median equatorial velocities from both estimates are an order of magnitude above the measured vsin⁡i⋆v\sin i_\star, ruling out the observed value at 2.5σ2.5\sigma.

Turning the question around, WASP-12’s thin convection zone means that we expect it to undergo very little magnetic braking on the main sequence. If we ask what initial rotation period would be required to match the observed slow surface rotation at the observed age and inferred stellar parameters, it would require a period of ∼1000\sim1000 days for a 1 Myr disk-locking timescale and ∼60\sim60 days for a long, 20 Myr disk-locking timescale (an early end to disk locking results in a model with more initial angular momentum at fixed period). Rotation periods of this length have not been observed in young open clusters and associations ([47, 93]), although most existing efforts to measure

Note.— U(a,b)U(a, b) denotes a uniform prior between aa and bb. Gaussian priors N(μ,σ)\mathcal{N}(\mu, \sigma) with σ=1 km s−1\sigma= 1\ \mathrm{km\ s}^{-1} were adopted for the micro- and macroturbulent velocities. KPF1 and KPF2 denote the two independent Rossiter–McLaughlin observing sequences obtained on UT 2024 March 23 and UT 2026 February 2, respectively. Separate limb-darkening coefficients and radial-velocity jitter parameters were fitted for each dataset.

  • allesfitter automatically shifts the input prior epoch to the center of the data time span.

** The listed period is the best-fit linear-ephemeris period used as a reference; the system departs measurably from a strictly linear ephemeris because of orbital decay (§ 2).

*** Combined host density from all orbits.

such periods would have been insensitive to periods this long. The observed slow rotation of WASP-12 is therefore difficult to reproduce through standard rotational evolution alone and motivates the angular-momentum considerations discussed in § 4.

For empirical comparison, we also use the gyrochronological model of K. C. Schlaufman [88], whose mass–age–rotation relation follows Skumanich Prot∝t1/2P_{\mathrm{rot}} \propto t^{1/2} spin-down (A. Skumanich [89]) calibrated on young clusters. Assuming spin–orbit alignment, he found veq=vsin⁡i⋆=14.4±2.2 km s−1v_{\mathrm{eq}} = v\sin i_\star= 14.4 \pm2.2\ \mathrm{km\ s}^{-1} for WASP-12 (rotation period ∼5.8 d\sim5.8\ \mathrm{d}). Given the agreement between empirical and theoretical vsin⁡i⋆v\sin i_\star estimates, we adopt fiducial values veq=14 km s−1v_{\mathrm{eq}} = 14\ \mathrm{km\ s}^{-1} and Prot=6 daysP_{\mathrm{rot}} = 6\ \mathrm{days} for a WASP-12-like star hereafter.

Discussion

WASP-12 is aligned but rotates anomalously slowly

Our KPF observations yield a low projected obliquity of λ=−1.3±8.7∘\lambda= -1.3 \pm8.7^\circ. However, the RM effect measures only the angle between the sky projections of the stellar spin and orbital angular-momentum vectors. The true three-dimensional obliquity is given by (D. C. Fabrycky & J. N. Winn 2009):

cos⁡ψ=cos⁡i⋆cos⁡ib+sin⁡i⋆sin⁡ibcos⁡λ,(2)\cos\psi= \cos i_{\star} \cos i_b + \sin i_{\star} \sin i_b \cos\lambda, \tag*{(2)}

where i⋆i_{\star} and ibi_b are stellar and orbital inclinations to the line of sight. Small λ\lambda therefore does not guarantee small true obliquity ψ\psi. Transiting planets have ib≈90∘i_b \approx90^\circ, but stellar inclination is generally unknown unless both veqv_{\mathrm{eq}} and vsin⁡i⋆v \sin i_{\star} are precisely measured (S. H. Albrecht et al. [4]). Asteroseismology can also constrain inclination (D. Huber et al. [51]; J. M. J. Ong [72]), but current photometry is unsuitable for WASP-12 (§ 4.4).

The observed low λ\lambda and low vsin⁡i⋆v \sin i_{\star} of WASP-12 have two qualitatively different interpretations:

  1. WASP-12 could have a typical equatorial rotation velocity of veq∼14v_{\mathrm{eq}} \sim14 km s−1^{-1} at its current evolutionary stage, but is viewed nearly pole-on, with sin⁡i⋆≲0.1\sin i_{\star} \lesssim0.1, thereby producing the observed low vsin⁡i⋆≃1.3v \sin i_{\star} \simeq1.3 km s−1^{-1}. In this case, the true three-dimensional obliquity would be nearly polar. However, as shown shortly, the stellar-spin axis must occupy a restricted region of orientation space to reproduce both the low vsin⁡i⋆v \sin i_{\star} and the measured projected obliquity λ\lambda (Figure 3).

  1. WASP-12 could instead be intrinsically slowly rotating, with veq≃1.3v_{\mathrm{eq}} \simeq1.3 km s−1^{-1}. An approximately equator-on stellar inclination would then naturally reproduce the measured vsin⁡i⋆v \sin i_{\star}. Combined with the low projected obliquity, this interpretation favors a nearly aligned system.

We compared the probabilities of these two interpretations by marginalizing over an isotropic distribution of stellar-spin orientations. With D={vsin⁡i⋆obs,λobs}D = \{v \sin i_{\star\mathrm{obs}}, \lambda_{\mathrm{obs}}\} denoting the measured quantities, we define the velocity residual as Δv=vsin⁡i⋆pred(ψ,ϕ,veq,ib)−vsin⁡i⋆obs\Delta v = v \sin i_{\star\mathrm{pred}}(\psi,\phi,v_{\mathrm{eq}},i_b) - v \sin i_{\star\mathrm{obs}} and the angular residual as Δλ=λpred(ψ,ϕ,ib)−λobs\Delta\lambda= \lambda_{\mathrm{pred}}(\psi,\phi,i_b) - \lambda_{\mathrm{obs}}, wrapped onto [−180∘,180∘)[-180^\circ, 180^\circ). The corresponding chi square is

χ2(ψ,ϕ;veq,ib)=(Δvσv)2+(Δλσλ)2(3)\chi^2(\psi,\phi;v_{\mathrm{eq}},i_b) = \left(\frac{\Delta v}{\sigma_v}\right)^2 + \left(\frac{\Delta\lambda}{\sigma_{\lambda}}\right)^2 \tag*{(3)}

The orientation-marginalized likelihood is then

Lmarg(D∣veq,ib)∝∫02π∫0πexp⁡(−χ22)sin⁡ψ4π dψ dϕ.(4)\mathcal{L}_{\mathrm{marg}}(D \mid v_{\mathrm{eq}},i_b) \propto\int_{0}^{2\pi} \int_{0}^{\pi} \exp\left(-\frac{\chi^2}{2}\right) \frac{\sin\psi}{4\pi}\,d\psi\,d\phi. \tag*{(4)}

Here, ψ\psi is the three-dimensional angle between the stellar-spin and orbital angular-momentum vectors. The angle ϕ\phi is the azimuth of the stellar-spin vector about the orbital angular-momentum axis, measured from the plane containing the orbital angular-momentum vector and the observer’s line of sight. We fix the orbital inclination at ib=82.9∘i_b = 82.9^\circ, whose uncertainty is negligible here. The predictions vsin⁡i⋆predv \sin i_{\star\mathrm{pred}} and λpred\lambda_{\mathrm{pred}} depend on (ψ,ϕ,veq,ib)(\psi,\phi,v_{\mathrm{eq}},i_b) and (ψ,ϕ,ib)(\psi,\phi,i_b), respectively. We approximate the integration on a regular 720×720720 \times720 grid spanning 0∘≤ψ≤180∘0^\circ\le\psi\le180^\circ and 0∘≤ϕ<360∘0^\circ\le\phi< 360^\circ. The factor sin⁡ψ/4π\sin\psi/4\pi is the normalized joint isotropic prior on (ψ,ϕ)(\psi,\phi), and σv\sigma_v and σλ\sigma_{\lambda} are the observational uncertainties in vsin⁡i⋆v \sin i_{\star} and λ\lambda, respectively. The measured λ\lambda constrains the joint (ψ,ϕ)(\psi,\phi) posterior and thus remains important for inferring true obliquity.

As noted earlier, scenario 2, corresponding to the intrinsically slow-rotation interpretation, is strongly preferred. The top panels of Figure 3 illustrate the origin of this preference. Under the fast-rotation model, the solution is confined to a very limited region of the ψ\psi–ϕ\phi parameter space with nearly polar true obliquities and a narrow range of azimuths. The slow-rotation model, on the other hand, permits a much broader range of azimuths and shifts the posterior support toward lower true obliquities.

The middle panel of Figure 3 shows that the orientation-marginalized likelihood peaks at veq≃1.35 km s−1v_{\rm eq} \simeq1.35\ {\rm km\ s}^{-1}. This value is close to the value implied by alignment, veq,align=vsin⁡i⋆/sin⁡ib≃1.30 km s−1v_{\rm eq,align} = v\sin i_{\star}/\sin i_b \simeq1.30\ {\rm km\ s}^{-1}. For a quantitative comparison of the two scenarios, we now include an ad hoc prior on the rotational velocity: a slow-rotation model with veq∣Hslow∼U(0,2.5 km s−1)v_{\rm eq}\mid H_{\rm slow} \sim U(0, 2.5\ {\rm km\ s}^{-1}) and a fast-rotation model based on the K. C. Schlaufman [88] prediction, veq∣Hfast∼N(14.4,2.2) km s−1v_{\rm eq}\mid H_{\rm fast} \sim N(14.4, 2.2)\ {\rm km\ s}^{-1}. For each hypothesis HH, we combined the marginalized likelihood with the corresponding veqv_{\rm eq} prior:

p(veq∣D,H)=Lmarg(D∣veq,ib)p(veq∣H)ZH,p(v_{\rm eq}\mid D,H) = \frac{\mathcal{L}_{\rm marg}(D\mid v_{\rm eq},i_b)p(v_{\rm eq}\mid H)}{\mathcal{Z}_H},
ZH=∫Lmarg(D∣veq,ib)p(veq∣H) dveq.(5)\mathcal{Z}_H = \int\mathcal{L}_{\rm marg}(D\mid v_{\rm eq},i_b)p(v_{\rm eq}\mid H)\,dv_{\rm eq}. \tag*{(5)}

Here, ZH\mathcal{Z}_H is the Bayesian evidence for hypothesis HH, and Bslow,fast=Zslow/ZfastB_{\rm slow,fast} = \mathcal{Z}_{\rm slow}/\mathcal{Z}_{\rm fast}. We obtain ln⁡Bslow,fast=4.1\ln B_{\rm slow,fast} = 4.1, corresponding to Bslow,fast=60.5B_{\rm slow,fast} = 60.5. Admittedly, the adopted veqv_{\rm eq} prior is arbitrary, but our conclusion is qualitatively unchanged over the range of uniform priors tested below. As a sensitivity test, changing the upper bound of a uniform slow-rotation prior from 3 to 10 km s−1{\rm km\ s}^{-1} changes the Bayes factor from 55.3 to 21.2, all favoring scenario 2, or the slow-rotation interpretation.

We therefore favor intrinsic slow rotation over a typical rapidly rotating F star viewed nearly pole-on. Under this interpretation, the small λ\lambda is consistent with a low true obliquity. Setting sin⁡i⋆=1\sin i_{\star} = 1 in Prot=2πR⋆sin⁡i⋆/vsin⁡i⋆P_{\rm rot} = 2\pi R_{\star}\sin i_{\star}/v\sin i_{\star} gives a maximum rotation-period estimate of 64 d for WASP-12.

WASP-12’s low chromospheric activity provides additional context for evaluating the slow-rotation interpretation. The bottom panel of Figure 3 compares it with main-sequence LAMOST–Kepler stars (H. Yang et al. [108]) in the activity–Rossby plane. For WASP-12, we calculate Ro=Prot/τcz{\rm Ro} = P_{\rm rot}/\tau_{\rm cz} with model convective-overturn time τcz=2.6−1.2+1.3\tau_{\rm cz} = 2.6^{+1.3}_{-1.2} d (§ 3.2), normalizing to the Sun with τcz,⊙=10.9\tau_{\rm cz,\odot} = 10.9 d and Prot,⊙=25.4P_{\rm rot,\odot} = 25.4 d. For comparison stars, we use global convective-overturn times τg\tau_g from H. Yang et al. [108] and the corresponding solar normalization, τg,⊙=34.9\tau_{g,\odot} = 34.9 d. At the predicted Prot=5.9P_{\rm rot} = 5.9 d, WASP-12 has near-solar Rossby number but weaker observed emission than most stars at similar Rossby numbers: log⁡RHK′=−5.50\log R'_{\rm HK} = -5.50 (H. A. Knutson et al. [57]) and unusually weak Ca II H&K and Mg II h&k cores (C. A. Haswell et al. [43]; L. Fossati et al. [30]). The inferred maximum period of 64 d instead places it at a much higher Rossby number, beyond this comparison sample. If intrinsic, the low chromospheric emission would be qualitatively consistent with slow ro-

tation. However, the Ca II H&K and Mg II h&k emission cores are formed in the chromosphere and may be suppressed by absorption from circumstellar gas, possibly lost from WASP-12 b [43, 30]. We therefore do not use the activity measurement as an independent constraint on the stellar rotation or viewing geometry. We also searched for photometric variability of WASP-12 in the TESS data, and found no compelling rotational modulation.

Where did the orbital angular momentum go?

WASP-12 b is aligned, at least with the surface convective layer, while its orbit is rapidly decaying and losing angular momentum. If the lost orbital angular momentum is transferred to the surface convective layer, the surface should have spun up, contrary to the observed slow rotation.

To quantify this discrepancy, we compare the angular-momentum reservoirs of the system. Assuming a circular orbit and Mp≪M⋆M_p \ll M_\star, the planet’s orbital angular momentum is

Lorb≃MpGM⋆a,(6)L_{\rm orb} \simeq M_p\sqrt{G M_\star a}, \tag*{(6)}

We next estimate the whole-star rotational angular momentum. Using the stellar models described in § 3, we obtain a total stellar moment of inertia of Itot≃1.67×1054 g cm2I_{\rm tot} \simeq1.67 \times10^{54}\ {\rm g\ cm^2} and a convective-envelope moment of inertia of Ienv≃8.32×1052 g cm2I_{\rm env} \simeq8.32 \times10^{52}\ {\rm g\ cm^2}. The outer convective envelope contributes ∼5%\sim5\% of the total stellar moment of inertia. We define Iint=Itot−IenvI_{\rm int} = I_{\rm tot} - I_{\rm env} as the moment of inertia of the stellar interior beneath the convective envelope, including both the radiative layer and central convective core. Table 4 summarizes the angular momenta and moments of inertia of the planet, surface convective layer, and stellar interior.

To estimate the orbital angular momentum available for stellar spin evolution, we consider two high-eccentricity migration histories for WASP-12, focusing on the orbital period just after circularization (e.g., [83, 26, 70, 29, 64] (Y. Wu & Y. Lithwick 2011a)). During circularization, tides raised primarily in the planet remove orbital energy while approximately conserving orbital angular momentum. In our simplified accounting, the angular-momentum transfer relevant to stellar spin therefore occurs during the subsequent orbital-decay stage. The post-circularization orbit is important because the difference between its angular momentum and the planet’s present orbital angular momentum determines the amount available for transfer to the star. Such a migration history is plausible because

WASP-12 is a hierarchical triple system: the planet-hosting primary, WASP-12 A, has a distant companion, WASP-12 BC, itself a close pair of M dwarfs. The BC pair lies at a projected separation of approximately 1′′1'' from WASP-12 A, corresponding to roughly 400 AU [11, 8].

For a fluid planet, the Roche-limit orbital period scales approximately with the planetary mean density as [82]

PRoche≃12.6 hr(ρp1 g cm−3)−1/2.(7)P_{\rm Roche} \simeq12.6\ {\rm hr}\left(\frac{\rho_p}{1\ {\rm g\ cm^{-3}}}\right)^{-1/2}. \tag*{(7)}

Because tides raised in the planet approximately conserve orbital angular momentum during circularization, the final circular semimajor axis is roughly twice the minimum pericenter distance:

afinal≃2rperi.(8)a_{\rm final} \simeq2r_{\rm peri}. \tag*{(8)}

Traditionally, the minimum pericenter distance is assumed to be at least 2.7rt≡Rp(M⋆Mp)1/32.7r_t \equiv R_p\left(\frac{M_\star}{M_p}\right)^{1/3} to avoid tidal disruption [39]. More recent work by Q. Fan & S.-F. Liu [28] revised the critical value to approximately 2rt2r_t. We therefore adopt an intermediate value of 2.5rt2.5r_t and assume:

afinal≃5rt.(9)a_{\rm final} \simeq5r_t. \tag*{(9)}

Thus, the planet’s mean density during circularization determines the orbital period at which it emerges from circularization. This post-circularization period—rather than the inflation mechanism itself—sets the orbital angular momentum subsequently available for transfer to the star. We consider two limiting possibilities:

  1. Scenario 1: Inflation by circularization. The planet has already been inflated by eccentricity tides to nearly its current radius, Rp≃1.89 RJupR_p \simeq1.89\ R_{\rm Jup}, during circularization. For Mp≃1.47 MJupM_p \simeq1.47\ M_{\rm Jup} [20], this gives ρp,infl≃0.269 g cm−3\rho_{p,\mathrm{infl}} \simeq0.269\ {\rm g\ cm^{-3}} and Pinit≃2.97 dP_{\rm init} \simeq2.97\ {\rm d}.

  1. Scenario 2: Inflation by irradiation. We assume that eccentricity tidal heating is quickly shed through radiative cooling or mass loss [110], so the planet is not inflated during circularization. We adopt a typical planetary radius of Rp≃1.1 RJupR_p \simeq1.1\ R_{\rm Jup} and ρp,comp≃1.37 g cm−3\rho_{p,\mathrm{comp}} \simeq1.37\ {\rm g\ cm^{-3}}, giving Pinit≃1.32 dP_{\rm init} \simeq1.32\ {\rm d}. In this case, WASP-12 b is inflated by strong irradiation or other processes after circularization.

Figure 4 illustrates both scenarios. Subsequent tidal evolution brings the planet to its present 1.09 d orbit, but the key difference is the total orbital angular momentum shed by WASP-12 b after circularization. In Scenario 1, the orbital angular momentum decreased from 3.05×10493.05 \times10^{49} to 2.18×10492.18 \times10^{49} g cm2^{2} s−1^{-1}, giving ΔLorb≃8.65×1048\Delta L_{\rm orb} \simeq8.65 \times10^{48} g cm2^{2} s−1^{-1}. In Scenario 2, it decreased only from 2.33×10492.33 \times10^{49} to 2.18×10492.18 \times10^{49} g cm2^{2} s−1^{-1}, giving ΔLorb≃1.42×1048\Delta L_{\rm orb} \simeq1.42 \times10^{48} g cm2^{2} s−1^{-1}. Thus, Scenario 1 provides approximately 6 times more angular momentum than Scenario 2 for modifying the stellar spin after circularization.

Schematic illustration of the two post-circularization scenarios for WASP-12 b

Figure 4. Schematic illustration of the two post-circularization scenarios for WASP-12 b. In both scenarios, the planet undergoes high-eccentricity migration and reaches a highly eccentric orbit with pericenter distance rperir_{\mathrm{peri}}. Tidal circularization approximately conserves orbital angular momentum, producing a circular orbit with semimajor axis afinal≃2rperia_{\mathrm{final}} \simeq2r_{\mathrm{peri}}. Left—Scenario 1: Inflation by circularization. The planet has reached approximately its current radius (Rp≃1.89RJupR_{\mathrm{p}} \simeq1.89 R_{\mathrm{Jup}}) by circularization. Its low mean density places the Roche limit farther from the star, yielding a post-circularization period of approximately 2.97 d. Right—Scenario 2: Inflation by irradiation after circularization. The planet remains relatively compact (Rp≃1.10RJupR_{\mathrm{p}} \simeq1.10 R_{\mathrm{Jup}}) during circularization. Its higher density gives a smaller Roche limit and a post-circularization period of approximately 1.32 d. The planet subsequently inflates as tidal evolution brings it to the present-day orbital period of 1.09 d and radius of Rp≃1.89RJupR_{\mathrm{p}} \simeq1.89 R_{\mathrm{Jup}}. The two scenarios therefore provide substantially different amounts of orbital angular momentum for transfer during subsequent inspiral.

ComponentMoment of inertia II (g cm2^{2})Angular frequency Ω\Omega (s−1^{-1})Angular momentum J=IΩJ = I\Omega (g cm2^{2} s−1^{-1})
Whole star (slow rotator)1.67×10541.67 \times10^{54}1.14×10−61.14 \times10^{-6}1.90×10481.90 \times10^{48}
Convective envelope (slow rotator)8.32×10528.32 \times10^{52}1.14×10−61.14 \times10^{-6}9.45×10469.45 \times10^{46}
Whole star (fast rotator)1.67×10541.67 \times10^{54}1.23×10−51.23 \times10^{-5}2.06×10492.06 \times10^{49}
Convective envelope (fast rotator)8.32×10528.32 \times10^{52}1.23×10−51.23 \times10^{-5}1.02×10481.02 \times10^{48}
Stellar interior below envelope1.59×10541.59 \times10^{54}UnknownUnknown
Planet orbit (current; Porb=1.09P_{\rm orb} = 1.09 d)3.28×10533.28 \times10^{53}6.66×10−56.66 \times10^{-5}2.18×10492.18 \times10^{49}
Planet orbit (Scenario 1; Porb=2.97P_{\rm orb} = 2.97 d)1.25×10541.25 \times10^{54}2.45×10−52.45 \times10^{-5}3.05×10493.05 \times10^{49}
Planet orbit (Scenario 2; Porb=1.32P_{\rm orb} = 1.32 d)4.21×10534.21 \times10^{53}5.52×10−55.52 \times10^{-5}2.33×10492.33 \times10^{49}

Table 4. Angular-momentum reservoirs and stellar spin-reversal budget. Upper section: Stellar reservoirs assume rigid rotation with Prot=64P_{\rm rot} = 64 d (slow) or 5.9 d (fast, corresponding to vrot=14.4v_{\rm rot} = 14.4 km s−1^{-1}). The rotation rate of the interior below the convective envelope is observationally unconstrained. Orbital angular momenta are calculated as Lorb=MpGM⋆aL_{\rm orb} = M_{\rm p}\sqrt{G M_{\star}a}, with an effective moment of inertia Iorb=Mpa2I_{\rm orb} = M_{\rm p}a^{2} such that Lorb=IorbΩorbL_{\rm orb} = I_{\rm orb}\Omega_{\rm orb} and Ωorb=2π/Porb\Omega_{\rm orb} = 2\pi/P_{\rm orb}. The present semimajor axis, a=0.02295a = 0.02295 AU, follows from Kepler’s third law using M⋆=1.35 M⊙M_{\star} = 1.35\,M_{\odot} and Porb=1.09P_{\rm orb} = 1.09 d. Lower section: Required angular momentum for evolution from Prot,i=−5.9P_{\rm rot,i} = -5.9 d (retrograde) to Prot,f=+64P_{\rm rot,f} = +64 d (prograde), calculated as ΔJrev=I(Ωf−Ωi)\Delta J_{\rm rev} = I(\Omega_{\rm f} - \Omega_{\rm i}). The available orbital angular momentum is ΔLorb=8.65×1048\Delta L_{\rm orb} = 8.65 \times10^{48} g cm2^{2} s−1^{-1} for Scenario 1 and 1.42×10481.42 \times10^{48} g cm2^{2} s−1^{-1} for Scenario 2. The ratio Rrev=ΔLorb/ΔJrevR_{\rm rev} = \Delta L_{\rm orb}/\Delta J_{\rm rev} indicates whether the angular-momentum budget is sufficient (Rrev≥1R_{\rm rev} \ge1).

Stellar reservoirΔJrev\Delta J_{\rm rev} (g cm2^{2} s−1^{-1})Scenario 1 Feasible? (RrevR_{\rm rev})Scenario 2 Feasible? (RrevR_{\rm rev})
Convective envelope1.12×10481.12 \times10^{48}Yes (7.7)Marginally yes (1.3)
Whole star2.25×10492.25 \times10^{49}No (0.38)No (0.06)

Table 4. Angular-momentum budget for stellar spin reversal

Adopting the empirical slow-rotation period of Prot=64P_{\rm rot} = 64 d, the present whole-star spin angular momentum is J⋆=I⋆Ω⋆≃1.90×1048J_{\star} = I_{\star}\Omega_{\star} \simeq1.90 \times10^{48} g cm2^{2} s−1^{-1}, while the convective-envelope angular momentum is Jenv≃9.45×1046J_{\rm env} \simeq9.45 \times10^{46} g cm2^{2} s−1^{-1}. In both Scenarios 1 and 2, the available orbital angular momentum of WASP-12 b is comparable to that of the whole star and much larger than that in the surface convective layer. The surface rotation of WASP-12 could therefore have been significantly spun up by the decaying planet.

Stellar spin-up and internal angular-momentum transport

We estimate the stellar response by adding the planet’s lost orbital angular momentum to an initially prograde star with Prot,i=5.9P_{\rm rot,i} = 5.9 d, the YREC+rotevol prediction without tidal spin evolution. We calculate final periods for absorption by either the whole star or convective envelope, assuming complete transfer and fixed stellar moments of inertia.

The magnetic-braking timescale may be written as τmb=J⋆/∣J˙mb∣\tau_{\rm mb} = J_{\star}/\lvert\dot{J}_{\rm mb}\rvert [55]. Under the favored slow-rotation interpretation, however, WASP-12 lies in the weakened-magnetic-braking regime. In this regime, the wind torque is strongly suppressed and is set to zero in the step-function prescription of J. L. van Saders et al. (2016). Magnetic braking is therefore expected to be weak at the present surface rotation rate and throughout the modeled main-sequence evolution because of the star’s thin surface convection zone. The post-circularization budgets below neglect wind losses.

If the transferred angular momentum is redistributed throughout the whole star, Scenario 1 changes the rotation period from 5.9 d to Prot,f≃4.2P_{\rm rot,f} \simeq4.2 d, whereas Scenario 2 gives Prot,f≃5.5P_{\rm rot,f} \simeq5.5 d. Despite providing ΔLorb≃8.65×1048\Delta L_{\rm orb} \simeq8.65 \times10^{48} and 1.42×10481.42 \times10^{48} g cm2^{2} s−1^{-1}, respectively, both scenarios produce only modest whole-star spin-up because of the star’s large moment of inertia.

The effect is more pronounced if transferred angular momentum remains in the thin convective envelope. Scenario 1 spins the envelope from 5.9 d to Penv,f≃0.6P_{\mathrm{env,f}} \simeq0.6 d, while Scenario 2 gives Penv,f≃2.5P_{\mathrm{env,f}} \simeq2.5 d. Both are far shorter than the present Prot≃64P_{\mathrm{rot}} \simeq64 d inferred from vsin⁡i⋆v\sin i_{\star}. The observed slow photospheric rotation rules out long-term storage of transferred angular momentum solely in the convective envelope.

Figure 5 summarizes angular-momentum exchange among the planetary orbit (LorbL_{\mathrm{orb}}), convective envelope (JenvJ_{\mathrm{env}}), and stellar interior (JintJ_{\mathrm{int}}). Tidal torques transfer orbital angular momentum to stellar spin, while tidal dissipation removes orbital energy. Inefficient internal transport rapidly spins up the low-inertia envelope; efficient transport redistributes angular momentum internally, allowing slow photospheric rotation. Alternatively, tidally excited internal gravity waves may transport angular momentum directly into the interior [104, 6].

Schematic illustration of angular-momentum transfer between WASP-12 b and its host star using a hydraulic analogy

Figure 5. Schematic illustration of angular-momentum transfer between WASP-12 b and its host star using a hydraulic analogy. The upper and lower panels show an illustrative initially retrograde configuration and the present-day prograde configuration, respectively. Reservoir widths represent effective moments of inertia, and fluid levels represent angular frequencies. Here, Ω>0\Omega> 0 and Ω<0\Omega< 0 denote prograde and retrograde rotation relative to the planetary orbit, respectively. The three reservoirs represent the planetary orbit (LorbL_{\mathrm{orb}}), the stellar convective envelope (JenvJ_{\mathrm{env}}), and the stellar interior beneath the convective envelope (JintJ_{\mathrm{int}}). The stellar interior and planetary orbit are much larger angular-momentum reservoirs than the convective envelope. Orbital angular momentum may be deposited into the convective envelope at a rate J˙pl→env\dot{J}_{\mathrm{pl}\to\mathrm{env}} and subsequently redistributed into the stellar interior at a rate J˙env→int\dot{J}_{\mathrm{env}\to\mathrm{int}}. Direct deposition into the stellar interior is also possible. In the retrograde-to-prograde scenario discussed in § 4.3, tidal transfer first cancels the initially retrograde stellar spin and then drives the star into the present prograde configuration while the planetary orbit loses angular momentum.

However, orbit–surface alignment suggests that some orbital angular momentum passed through the surface. Whether the photosphere remains slowly rotating depends on how rapidly the envelope transfers angular momentum to the stellar interior. The relevant orbital angular-momentum-loss timescale is Lorb/∣L˙orb∣=3P/∣P˙∣≃9.8L_{\mathrm{orb}}/|\dot{L}_{\mathrm{orb}}| = 3P/|\dot{P}| \simeq9.8 Myr, broadly comparable to both the observed period-decay timescale and our illustrative core–envelope coupling timescale.

Independent evidence from stellar rotation, lithium evolution, and asteroseismology indicates that angular momentum can be exchanged efficiently between the stellar surface layers and the deeper stellar interior. The coupling generally becomes faster with increasing stellar mass ([92, 19] (G. Li et al. 2020)). A semi-empirical estimate follows from the two-zone rotational-evolution model of F. Spada & A. C. Lanzafame [94], which parameterizes the core–envelope coupling timescale as

τce=τce,⊙(M⋆M⊙)−α(10)\tau_{\mathrm{ce}} = \tau_{\mathrm{ce},\odot}\left(\frac{M_{\star}}{M_{\odot}}\right)^{-\alpha} \tag*{(10)}

with τce,⊙≃22\tau_{\mathrm{ce},\odot} \simeq22 Myr and α≃5.6\alpha\simeq5.6. Extrapolating this relation to M⋆=1.353M⊙M_{\star} = 1.353 M_{\odot} gives

τce≃22 Myr(1.353)−5.6≃4 Myr.(11)\tau_{\mathrm{ce}} \simeq22\ \mathrm{Myr}(1.353)^{-5.6} \simeq4\ \mathrm{Myr}. \tag*{(11)}

The estimated core–envelope coupling timescale is intriguingly similar to the observed period-decay timescale of 3.25±0.233.25 \pm0.23 Myr. Rapid envelope–interior angular-momentum transfer may therefore prevent angular momentum deposited near the surface from accumulating solely in the convective envelope. We caution that the F. Spada & A. C. Lanzafame [94] relation is extrapolated beyond its calibrated mass range and was developed for magnetic-braking-dominated evolution. The effective coupling timescale may further depend on the shear or ΔΩ\Delta\Omega between the envelope and interior. Stronger shear may further speed up angular-momentum transport. Our estimate of τce≃4\tau_{\mathrm{ce}} \simeq4 Myr, which was calibrated on non-planet-host stars, is therefore likely an underestimate of the real coupling rate. Moreover, the core–envelope coupling timescale is also model-dependent (see also [33, 59]).

An initially retrograde WASP-12 b

We propose that WASP-12 b might have started on a retrograde orbit immediately following the circularization stage of its high-eccentricity migration. The envelope and interior of the host star were rotating rapidly with a period of approximately 6 days, typical for a star of its mass and evolutionary stage. The orbital angular momentum of the decaying planet quickly aligned the surface layer. However, due to rapid core–envelope angular-momentum coupling, the orbital angular momentum was eventually deposited in the deep interior of the star. The surface layer therefore currently has slow, prograde rotation. The total moment of inertia of the star is too large, and the planet probably has not yet flipped the angular-momentum axis of the whole star. Such an evolutionary pathway has also been suggested by T. Hallatt et al. [42], although Roche-lobe overflow is the mechanism for angular-momentum transfer in their analysis.

We quantitatively test an initially retrograde orbit for WASP-12 b. We adopt the YREC+rotevol prediction as the initial stellar rotation period Prot,i=−5.9P_{\mathrm{rot,i}} = -5.9 d, and the rotation inferred from the present vsin⁡i⋆v \sin i_{\star} as the final state, Prot,f=+64P_{\mathrm{rot,f}} = +64 d. The signs denote retrograde and prograde rotation, respectively.

For a stellar reservoir with moment of inertia II, the angular momentum required to evolve between these two signed rotation states is

ΔJrev=I(Ωf−Ωi)=2πI(1Prot,f+1∣Prot,i∣).(12)\Delta J_{\mathrm{rev}} = I(\Omega_{\mathrm{f}}-\Omega_{\mathrm{i}}) = 2\pi I\left(\frac{1}{P_{\mathrm{rot,f}}}+\frac{1}{|P_{\mathrm{rot,i}}|}\right). \tag*{(12)}

This expression accounts both for canceling the initial retrograde spin and for producing the present prograde spin.

We define the available-to-required angular-momentum ratio as

Rrev=ΔLorbΔJrev.(13)\mathcal{R}_{\mathrm{rev}} = \frac{\Delta L_{\mathrm{orb}}}{\Delta J_{\mathrm{rev}}}. \tag*{(13)}

A value of Rrev≥1\mathcal{R}_{\mathrm{rev}} \ge1 indicates that the orbital angular momentum lost after circularization is sufficient for the specified spin evolution. Table 4 (lower section) summarizes the results for the whole star and convective envelope.

Scenario 1 clearly distinguishes envelope from whole-star evolution. Its orbital-angular-momentum budget supplies only 38% of that required to reverse the whole star, but only 7.7 times that required to evolve the convective envelope from −5.9-5.9 d to +64+64 d. Scenario 1 can therefore reverse the envelope but not the whole star.

Scenario 2 supplies only 6% of the whole-star reversal requirement but 1.3 times the envelope requirement. It permits the signed envelope evolution in the maximum-transfer calculation, with much less margin than Scenario 1, but cannot reverse the whole star.

The signed final rotation periods show the same physical distinction. If each reservoir absorbs the full available orbital angular momentum starting from Prot,i=P_{\mathrm{rot,i}} =

−5.9-5.9 d, Scenario 1 leaves the whole star and stellar interior retrograde at −10.2-10.2 and −10.6-10.6 d. Scenario 2 leaves the whole star and interior near their initial retrograde states, at −6.3-6.3 and −6.4-6.4 d, respectively. Thus, neither scenario can reverse the whole star, but both provide enough orbital angular momentum for the convective envelope to flip to prograde and reach the representative +64+64 d rotation state; Scenario 2 does so only marginally.

Our two-zone stellar interior model is of course only a toy model. We note that the strong shear between a rapidly rotating, retrograde interior and a slowly rotating, prograde envelope may be susceptible to various instabilities such as the Rayleigh criterion for centrifugal instability (e.g., A. J. Barker et al. [7]) or the Goldreich–Schubert–Fricke instability (P. Goldreich & G. Schubert [37]; K. Fricke [32]). We suggest that these instabilities may destroy a sharp interior–envelope boundary adopted in our toy model and instead set up a gentle transition zone with a smoother shear layer.

Prospects for testing internal stellar rotation

Our scenario raises a testable possibility: a retrograde WASP-12 interior whose strong surface–interior shear could rapidly transfer orbital angular momentum inward. Future asteroseismic rotational splittings could test differential interior–surface rotation (C. Aerts et al. [1]; P. G. Beck et al. [9]; S. Deheuvels et al. [22]; D. Huber et al. [51]). A related configuration may already be suggested for Kepler-56: mixed modes reveal evidence for distinct core and envelope rotation axes, with the envelope potentially closer to orbital alignment (J. M. J. Ong [72]). Surface alignment therefore need not imply interior alignment. Although the J. M. J. Ong [72] mixed-mode method is not directly applicable to a late-main-sequence star, future constraints on WASP-12’s interior and envelope rotation rates and orientations could test whether tidal angular-momentum transfer reversed the whole star or only its envelope. WASP-12 lies outside PLATO’s first long-duration field (R. Heller et al. [45]); even with future coverage, its broad F-star mode linewidths would hinder rotation and inclination measurements unless rotational splittings were sufficiently large.

Summary and Conclusions

We present new KPF Rossiter–McLaughlin observations of WASP-12 b and interpret its spin–orbit architecture, orbital decay, and slow stellar rotation in a scenario in which the planet started with a retrograde orbit immediately after the circularization stage of its high-eccentricity migration. Our main conclusions are:

  • Joint modeling of two KPF RM datasets and TESS photometry yields a low sky-projected obliquity of λ=−1.3±8.7∘\lambda= -1.3 \pm8.7^\circ. A reanalysis of the 2012 HIRES observations shows that the discrepancy is not primarily driven by the different stellar-rotation priors.

  • The measured wsin⁡i⋆=1.3±0.1w \sin i_\star= 1.3 \pm0.1 km s−1^{-1} is much smaller than the ∼14\sim14 km s−1^{-1} expected for a star of similar mass and evolutionary stage. Our Bayesian model comparison favors a slowly rotating WASP-12 with a low true obliquity ψ\psi rather than a rapidly rotating star with a high true obliquity and pathological sin⁡i⋆\sin i_\star values. Weak Ca and Mg chromospheric emissions of WASP-12 (log⁡RHK′=−5.5\log R'_{\mathrm{HK}} = -5.5) also favor a slowly rotating star.

  • Transit midpoints from both KPF RM observations agree with the established quadratic orbital-decay ephemeris, supporting continued rapid decay.

  • Stellar evolution modeling favors a late-main-sequence host with a central convective core, an overlying radiative layer, and a thin outer convective envelope containing only ≃5%\simeq5\% of the total stellar moment of inertia.

  • We calculated that the net orbital angular momentum change from the 2.97-day post-circularization orbit to the current orbit is ΔLorb≃8.65×1048\Delta L_{\mathrm{orb}} \simeq8.65 \times10^{48} g cm2^2 s−1^{-1}.

  • Our YREC+rotevol self-consistent stellar-evolution and magnetic-braking model predicts a rotation period of ∼6\sim6 d for WASP-12 in the absence of tidal spin evolution.

  • The estimated envelope–interior angular-momentum coupling timescale, τce≃4\tau_{\mathrm{ce}} \simeq4 Myr, is similar to the observed tidal-decay timescale and may represent an upper limit because the large shear between the tidally aligned surface and initially retrograde interior could enhance angular-momentum transport. Orbital angular momentum can readily erase the surface–orbit misalignment and subsequently be redistributed into the stellar interior. This scenario could therefore reconcile the aligned orbit with the observed slow surface rotation.

  • For the 2.97-day post-circularization scenario, ΔLorb\Delta L_{\mathrm{orb}} is 7.7 times the angular momentum required to reverse the surface convective layer from retrograde to the observed prograde rotation. However, ΔLorb\Delta L_{\mathrm{orb}} is only 38% of that required to reverse the whole star, and thus the stellar interior

could still be rotating retrograde relative to the planet, provided that ongoing tidal forcing maintains differential rotation against envelope–interior coupling. This could be tested with future asteroseismic data.

Acknowledgments

This work made use of the Exoplanet Follow-up Observing Program for TESS (ExoFOP-TESS), available at 10.26134/ExoFOP3.

We thank Caleb Lammers, Tim Hallatt, Sarah Millholland, Jeremy Goodman, Avery Bailey, Joel Ong, and Meng Sun for thoughtful comments and helpful discussions.

R.L. is funded by the European Union (ERC, THIRSTEE, 101164189) and acknowledges financial support from the Severo Ochoa grant CEX2021-001131-S funded by MCIN/AEI/10.13039/501100011033. Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them.

This work was performed, in part, at the Jet Propulsion Laboratory, California Institute of Technology, sponsored by the United States Government under the Prime Contract 80NM0018D0004 between Caltech and NASA.

References

  1. [1]Aerts, C., Christensen-Dalsgaard, J., & Kurtz, D. W. 2010, Asteroseismology (Springer), doi: 10.1007/978-1-4020-5803-5DOI
  2. [2]Akinsanmi, B., Barros, S. C. C., Lendl, M., et al. 2024, A&A, 685, A63, doi: 10.1051/0004-6361/202348502DOI
  3. [3]Albrecht, S., Winn, J. N., Johnson, J. A., et al. 2012, ApJ, 757, 18, doi: 10.1088/0004-637X/757/1/18DOI
  4. [4]Albrecht, S. H., Marcussen, M. L., Winn, J. N., Dawson, R. I., & Knudstrup, E. 2021, ApJL, 916, L1, doi: 10.3847/2041-8213/ac0f03DOI
  5. [5]Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481, doi: 10.1146/annurev.astro.46.060407.145222DOI
  6. [6]Bailey, A., & Goodman, J. 2019, MNRAS, 482, 1872, doi: 10.1093/mnras/stzy805
  7. [7]Barker, A. J., Jones, C. A., & Tobias, S. M. 2019, MNRAS, 487, 1777, doi: 10.1093/mnras/stz1386DOI
  8. [8]Bechter, E. B., Crepp, J. R., Ngo, H., et al. 2014, The Astrophysical Journal, 788, 2, doi: 10.1088/0004-637x/788/1/2
  9. [9]Beck, P. G., Montalban, J., Kallinger, T., et al. 2012, Nature, 481, 55, doi: 10.1038/nature10612
  10. [10]Bellinger, E. P., Angelou, G. C., Hekker, S., et al. 2016, ApJ, 830, 31, doi: 10.3847/0004-637X/830/1/31DOI
  11. [11]Bergfors, C., Brandner, W., Daemgen, S., et al. 2012, Monthly Notices of the Royal Astronomical Society, 428, 182–189, doi: 10.1093/mnras/sts019
  12. [12]Biswas, S., Jiang, I.-G., Yeh, L.-C., et al. 2026, AJ, 171, 246, doi: 10.3847/1538-3881/ae4c86DOI
  13. [13]Blanco-Cuaresma, S. 2019, MNRAS, 486, 2075, doi: 10.1093/mnras/stz549
  14. [15]Blanco-Cuaresma, S., Soubiran, C., Jofré, P., & Heiter, U. 2014b, A&A, 566, A98, doi: 10.1051/0004-6361/201323153
  15. [16]Bonomo, A. S., Desidera, S., Benatti, S., et al. 2017, A&A, 602, A107, doi: 10.1051/0004-6361/201629882DOI
  16. [17]Bouma, L. G., Winn, J. N., Howard, A. W., et al. 2020, ApJL, 893, L29, doi: 10.3847/2041-8213/ab8563DOI
  17. [18]Bruntt, H., Bedding, T. R., Quirion, P. O., et al. 2010, MNRAS, 405, 1907, doi: 10.1111/j.1365-2966.2010.16575.xDOI
  18. [19]Cao, L., Pinsonneault, M. H., & van Saders, J. L. 2023, ApJL, 951, L49, doi: 10.3847/2041-8213/acd780DOI
  19. [20]Chakrabarty, A., & Sengupta, S. 2019, AJ, 158, 39, doi: 10.3847/1538-3881/ab24ddDOI
  20. [21]Choi, J., Dotter, A., Conroy, C., et al. 2016, ApJ, 823, 102, doi: 10.3847/0004-637X/823/2/102DOI
  21. [22]Deheuvels, S., García, R. A., Chaplin, W. J., et al. 2012, ApJ, 756, 19, doi: 10.1088/0004-637X/756/1/19DOI
  22. [23]Doyle, A. P., Davies, G. R., Smalley, B., Chaplin, W. J., & Elsworth, Y. 2014, MNRAS, 444, 3592, doi: 10.1093/mnras/stu1692DOI
  23. [24]Duguid, C. D., de Vries, N. B., Leconet, D., & Barker, A. J. 2024, ApJL, 966, L14, doi: 10.3847/2041-8213/ad3c40DOI
  24. [25]Eddington, A. S. 1926, The Internal Constitution of the Stars (Cambridge: Cambridge University Press)
  25. [26]Fabrycky, D., & Tremaine, S. 2007, ApJ, 669, 1298, doi: 10.1086/521702DOI
  26. [28]Fan, Q., & Liu, S.-F. 2026, ApJL, 1003, L7, doi: 10.3847/2041-8213/ae6659DOI
  27. [29]Ford, E. B., & Rasio, F. A. 2006, ApJL, 638, L45, doi: 10.1086/500734DOI
  28. [30]Fossati, L., Ayres, T. R., Haswell, C. A., et al. 2013, ApJL, 766, L20, doi: 10.1088/2041-8205/766/2/L20DOI
  29. [31]Fossati, L., Haswell, C. A., Froning, C. S., et al. 2010, ApJL, 714, L222, doi: 10.1088/2041-8205/714/2/L222DOI
  30. [32]Fricke, K. 1968, ZA, 68, 317
  31. [33]Gallet, F., & Bouvier, J. 2013, A&A, 556, A36, doi: 10.1051/0004-6361/201321302DOI
  32. [34]Gibson, S. R., Howard, A. W., Marcy, G. W., et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, ed. C. J. Evans, L. Simard, & H. Takami, 990870, doi: 10.1117/12.2233334
  33. [35]Gibson, S. R., Howard, A. W., Roy, A., et al. 2018, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 10702, Ground-based and Airborne Instrumentation for Astronomy VII, ed. C. J. Evans, L. Simard, & H. Takami, 107025X, doi: 10.1117/12.2311565
  34. [36]Gibson, S. R., Howard, A. W., Rider, K., et al. 2020, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 11447, Ground-based and Airborne Instrumentation for Astronomy VIII, ed. C. J. Evans, J. J. Bryant, & K. Motohara, 1144742, doi: 10.1117/12.2561783
  35. [37]Goldreich, P., & Schubert, G. 1967, ApJ, 150, 571, doi: 10.1086/149360DOI
  36. [38]Green, G. M., Schlafly, E., Zucker, C., Speagle, J. S., & Finkbeiner, D. 2019, ApJ, 887, 93, doi: 10.3847/1538-4357/ab5362DOI
  37. [39]Guillochon, J., Ramirez-Ruiz, E., & Lin, D. 2011, ApJ, 732, 74, doi: 10.1088/0004-637X/732/2/74DOI
  38. [40]Günther, M. N., & Daylan, T. 2019, Allesfilter: Flexible Star and Exoplanet Interference From Photometry and Radial Velocity,, Astrophysics Source Code Library http://ascl.net/1903.003DOI
  39. [41]Günther, M. N., & Daylan, T. 2021, ApJS, 254, 13, doi: 10.3847/1538-4365/abe70eDOI
  40. [42]Hallatt, T., Owen, J. E., & Millholland, S. 2026, arXiv e-prints, arXiv:2606.11292. https://arxiv.org/abs/2606.11292
  41. [43]Haswell, C. A., Fossati, L., Ayres, T., et al. 2012, ApJ, 760, 79, doi: 10.1088/0004-637X/760/1/79DOI
  42. [44]Hebb, L., Collier-Cameron, A., Loeillet, B., et al. 2009, ApJ, 693, 1920, doi: 10.1088/0004-637X/693/2/1920DOI
  43. [45]Heller, R., Jiang, C., Bluhm, P., et al. 2026, arXiv e-prints, arXiv:2604.02437, doi: 10.48550/arXiv.2604.02437DOI
  44. [46]Henyey, L., Vardya, M. S., & Bodenheimer, P. 1965, ApJ, 142, 841, doi: 10.1086/148357DOI
  45. [47]Herbst, W., Bailer-Jones, C. A. L., & Mundt, R. 2001, ApJL, 554, L197, doi: 10.1086/321701
  46. [48]Higson, E., Handley, W., Hobson, M., & Lasenby, A. 2019, Statistics and Computing, 29, 891, doi: 10.1007/s11222-018-9844-0
  47. [49]Hirano, T., Suto, Y., Taruya, A., et al. 2010, ApJ, 709, 458, doi: 10.1088/0004-637X/709/1/458DOI
  48. [50]Hirano, T., Suto, Y., Winn, J. N., et al. 2011, ApJ, 742, 69, doi: 10.1088/0004-637X/742/2/69DOI
  49. [51]Huber, D., Carter, J. A., Barbieri, M., et al. 2013, Science, 342, 331, doi: 10.1126/science.1242066DOI
  50. [52]Huber, D., Zinn, J., Bojsen-Hansen, M., et al. 2017, ApJ, 844, 102, doi: 10.3847/1538-4357/aa75caDOI
  51. [53]Jenkins, J. M., Twicken, J. D., McCauliff, S., et al. 2016, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9913, Software and Cyberinfrastructure for Astronomy IV, ed. G. Chiozzi & J. C. Guzman, 99133E, doi: 10.1117/12.2233418
  52. [54]Jermyn, A. S., Bauer, E. B., Schwab, J., et al. 2023, ApJS, 265, 15, doi: 10.3847/1538-4365/acae8dDOI
  53. [55]Kawaler, S. D. 1988, ApJ, 333, 236, doi: 10.1086/166740DOI
  54. [56]Kipping, D. M. 2013, MNRAS, 435, 2152, doi: 10.1093/mnras/stt1435DOI
  55. [57]Knutson, H. A., Howard, A. W., & Isaacson, H. 2010, ApJ, 720, 1569, doi: 10.1088/0004-637X/720/2/1569DOI
  56. [58]Lammers, C., Su, Y., & Winn, J. N. 2026,
  57. [59]Lanzafame, A. C., & Spada, F. 2015, A&A, 584, A30, doi: 10.1051/0004-6361/201526770DOI
  58. [62]Lightkurve Collaboration, Cardoso, J. V. d. M., Hedges, C., et al. 2018, Lightkurve: Kepler and TESS time series analysis in Python,, Astrophysics Source Code Library, record ascl:1812.013
  59. [63]Maciejewski, G., Dimitrov, D., Fernández, M., et al. 2016, A&A, 588, L6, doi: 10.1051/0004-6361/201628312DOI
  60. [64]Matsakos, T., & Königl, A. 2016, ApJL, 820, L8, doi: 10.3847/2041-8205/820/1/L8DOI
  61. [65]Matt, S., & Pudritz, R. E. 2008, ApJ, 678, 1109, doi: 10.1086/533428DOI
  62. [66]McLaughlin, D. B. 1924, ApJ, 60, 22, doi: 10.1086/142826DOI
  63. [67]Metcalfe, T. S., van Saders, J. L., Basu, S., et al. 2020, ApJ, 900, 154, doi: 10.3847/1538-4357/aba963DOI
  64. [68]Metcalfe, T. S., Petit, P., van Saders, J. L., et al. 2025, ApJ, 986, 120, doi: 10.3847/1538-4357/add40aDOI
  65. [69]Millholland, S., & Laughlin, G. 2018, ApJL, 869, L15, doi: 10.3847/2041-8213/aaedb1DOI
  66. [70]Nagasawa, M., Ida, S., & Bessho, T. 2008, ApJ, 678, 498, doi: 10.1086/529369DOI
  67. [71]Ohta, Y., Taruya, A., & Suto, Y. 2005, ApJ, 622, 1118, doi: 10.1086/428344DOI
  68. [72]Ong, J. M. J. 2025, ApJ, 980, 40, doi: 10.3847/1538-4357/ada4a3DOI
  69. [73]Patra, K. C., Winn, J. N., Holman, M. J., et al. 2017, AJ, 154, 4, doi: 10.3847/1538-3881/aa6d75DOI
  70. [74]Paxton, B., Bildsten, L., Dotter, A., et al. 2011, ApJS, 192, 3, doi: 10.1088/0067-0049/192/1/3DOI
  71. [75]Paxton, B., Cantiello, M., Arras, P., et al. 2013, ApJS, 208, 4, doi: 10.1088/0067-0049/208/1/4DOI
  72. [76]Paxton, B., Marchant, P., Schwab, J., et al. 2015, ApJS, 220, 15, doi: 10.1088/0067-0049/220/1/15DOI
  73. [77]Paxton, B., Schwab, J., Bauer, E. B., et al. 2018, ApJS, 234, 34, doi: 10.3847/1538-4365/aaa5a8DOI
  74. [78]Paxton, B., Smolec, R., Schwab, J., et al. 2019, ApJS, 243, 10, doi: 10.3847/1538-4365/ab2241DOI
  75. [79]Pepe, F., Cristiani, S., Rebelo, R., et al. 2013, The Messenger, 153, 6
  76. [80]Petigura, E. A., Howard, A. W., Marcy, G. W., et al. 2017, AJ, 154, 107, doi: 10.3847/1538-3881/aa80deDOI
  77. [81]Pinsonneault, M. H., van Saders, J. L., Cao, L., et al. 2026, arXiv e-prints, arXiv:2603.25792, doi: 10.48550/arXiv.2603.25792DOI
  78. [82]Rappaport, S., Sanchis-Ojeda, R., Rogers, L. A., Levine, A., & Winn, J. N. 2013, ApJL, 773, L15, doi: 10.1088/2041-8205/773/1/L15DOI
  79. [83]Rasio, F. A., & Ford, E. B. 1996, Science, 274, 954, doi: 10.1126/science.274.5289.954DOI
  80. [84]Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, Journal of Astronomical Telescopes, Instruments, and Systems, 1, 014003, doi: 10.1117/1.JATIS.1.1.014003DOI
  81. [85]Rossiter, R. A. 1924, ApJ, 60, 15, doi: 10.1086/142825DOI
  82. [86]Saunders, N., van Saders, J. L., Lyttle, A. J., et al. 2024, ApJ, 962, 138, doi: 10.3847/1538-4357/ad1516DOI
  83. [87]Savitzky, A., & Golay, M. J. E. 1964, Analytical Chemistry, 36, 1627, doi: 10.1021/ac60214a047DOI
  84. [88]Schlaufman, K. C. 2010, ApJ, 719, 602, doi: 10.1088/0004-637X/719/1/602DOI
  85. [89]Skumanich, A. 1972, ApJ, 171, 565, doi: 10.1086/151310DOI
  86. [90]Smith, J. C., Stumpe, M. C., Cleve, J. E. V., et al. 2012, Publications of the Astronomical Society of the Pacific, 124, 1000, doi: 10.1086/667697DOI
  87. [91]Sobol’, I. 1967, USSR Computational Mathematics and Mathematical Physics, 7, 86, doi: https://doi.org/10.1016/0041-5553(67)90144-9
  88. [92]Somers, G., & Pinsonneault, M. H. 2016, ApJ, 829, 32, doi: 10.3847/0004-637X/829/1/32DOI
  89. [93]Somers, G., Stauffer, J., Rebull, L., Cody, A. M., & Pinsonneault, M. H. 2017, ApJ, 850, 134, doi: 10.3847/1538-4357/aa93edDOI
  90. [94]Spada, F., & Lanzafame, A. C. 2020, A&A, 636, A76, doi: 10.1051/0004-6361/201936384DOI
  91. [96]Stumpe, M. C., Smith, J. C., Catanzarite, J. H., et al. 2014, Publications of the Astronomical Society of the Pacific, 126, 100, doi: 10.1086/674989DOI
  92. [97]Sun, M., Townsend, R. H. D., Xia, H., & Liu, J. 2025, ApJL, 995, L29, doi: 10.3847/2041-8213/ae23cfDOI
  93. [98]Thoul, A. A., Bahcall, J. N., & Loeb, A. 1994, ApJ, 421, 828, doi: 10.1086/173695DOI
  94. [99]Turner, J. D., Ridden-Harper, A., & Jayawardhana, R. 2021, AJ, 161, 72, doi: 10.3847/1538-3881/abd178DOI
  95. [102]Wang, X.-Y. 2026, Allesfitters: a modified version of allesfitter for advanced Rossiter-McLaughlin effect modeling, 1.0.0 Zenodo, doi: 10.5281/zenodo.21792490DOI
  96. [103]Wang, X.-Y., Rice, M., Wang, S., et al. 2024, ApJL, 973, L21, doi: 10.3847/2041-8213/ad7469DOI
  97. [104]Weinberg, N. N., Sun, M., Arras, P., & Essick, R. 2017, ApJL, 849, L11, doi: 10.3847/2041-8213/aa9113DOI
  98. [105]Winn, J. N., & Stefánsson, G. 2025, PSJ, 6, 300, doi: 10.3847/PSJ/ae21dbDOI
  99. [106]Wong, I., Shporer, A., Vissapragada, S., et al. 2022, AJ, 163, 175, doi: 10.3847/1538-3881/ac5680DOI
  100. [108]Yang, H., Liu, J., Soria, R., et al. 2025, A&A, 699, A251, doi: 10.1051/0004-6361/202554379DOI
  101. [109]Yee, S. W., Winn, J. N., Knutson, H. A., et al. 2020, ApJL, 888, L5, doi: 10.3847/2041-8213/ab5c16DOI
  102. [110]Zanazzi, J. J., MacLeod, M., Bryan, M. L., & Mahadevan, S. 2026, arXiv e-prints, arXiv:2606.20789, doi: 10.48550/arXiv.2606.20789DOI

Paper details

vlnc:W_8dkqskw7
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents