Orbital Period Changes in Recurrent Nova T Corona Borealis Prove That It Is Not a Type Ia Supernovae Progenitor

TL;DR

This work tackles the Type Ia supernova progenitor problem by testing whether the white dwarf in T CrB gains mass over eruption cycles. The authors combine long-baseline radial-velocity data (1946–2025) with ellipsoidal light-curve timings (1866–1946) to measure the orbital-period changes across the 1946 eruption, obtaining a large positive jump $\Delta P=+0.146\pm0.019$ d and a post-eruption period $P_{\rm post}=227.6043$ d. From $\Delta P$ and system parameters, they derive an ejecta mass $M_{\rm ejecta}=0.00074\pm0.00009\,M_\odot$ for the 1946 event, and compare it to the accreted mass $M_{\rm accreted}=1.38\times10^{-6}\,M_\odot$, finding $M_{\rm ejecta}/M_{\rm accreted}\approx540$. Neon-line ratios $F_{3869}/F_{5007}\approx1.4$ indicate an ONe WD, supporting the conclusion that T CrB cannot reach the Chandrasekhar mass and thus cannot be a Type Ia supernova progenitor. Together, these results challenge the single-degenerate channel for this system and emphasize the need for advanced modeling of nova ejecta in binary environments.

Abstract

T Corona Borealis (T CrB) is a recurrent nova and a symbiotic star that is commonly highlighted as the best case for being a progenitor of a Type Ia supernova (SNIa) within the framework of single-degenerate models. This exemplar can be tested by measuring whether the white dwarf (WD) mass ($M_{\rm WD}$) is increasing over each eruption cycle. This is a balance between the mass ejected during each nova event ($M_{\rm ejecta}$) and the mass accreted onto the WD between the nova events ($M_{\rm accreted}$). I have used all 206 radial velocities from 1946--2024 to measure the orbital period just after the 1946 eruption to be $P_{\rm post}$=227.6043 days, while the steady orbital period change ($\dot{P}$) is ($-$3.1$\pm$1.6)$\times$10$^{-6}$. I have used my full 213,730 magnitude $B$ and $V$ light curve from 1842--2025 to measure the times of maximum brightness in the ellipsoidal modulations to construct the $O-C$ from 1866--1946. I fit the broken parabola shape, to find the orbital period immediately before the 1946 eruption to be $P_{\rm pre}$=227.4586 days. The orbital period changed by $ΔP$=$+$0.146$\pm$0.019 days. With Kepler's Law, conservation of angular momentum, and the well-measured binary properties, the ejecta mass in 1946 is 0.00074$\pm$0.00009 M$_{\odot}$. $M_{\rm accreted}$ is reliably measured to be 1.38$\times$10$^{-6}$ M$_{\odot}$ from the accretion luminosity. $M_{\rm ejecta}$ is larger than $M_{\rm accreted}$ by 540$\times$, so $M_{\rm WD}$ is {\it decreasing} every eruption cycle. T CrB can never become a SNIa.

Figures (3)

• Figure 1: Ellipsoidal modulations in $V$-band after the 1946 eruption. These light curves are folded with period 227.5687 days and an epoch of the best fit $T_{\rm max}$. Phases 0.0, 1.0, and 2.0 correspond to the times of elongation in the orbit, when the red giant is maximally to one side of the white dwarf, and when the radial velocity is at its maximum value. The zero phase is the time of peak brightness in the light curve (with the smallest magnitude). Phases 0.5 and 1.5 are for the times of elongation with the red giant maximally on the other side of the WD. Phases 0.25 and 1.25 are for the times of superior conjunctions in the orbit, with the red giant behind the WD, when the radial velocity is decreasing through the $\gamma$-velocity. Phases 0.75 and 1.75 are for the times of inferior conjunctions in the orbit, with the red giant in front of the WD, when the radial velocity is increasing through the $\gamma$-velocity. The $V$-band magnitudes are shown as green dots, many of which are averages over 0.01 year bins, representing 2 to 200 input magnitudes. These individual points are phase averaged into bins each 0.05 wide in phase, as shown by the red squares. My full light curve with 213,730 fully-modern Johnson $B$ and $V$ magnitudes from 1842--2022 are explicitly listed in Schaefer (2023a) and this is supplemented by recent data collected by the AAVSO. The sinewave fit is shown by the black curve. A large amount of the scatter is due to the usual 'flickering' intrinsic to T CrB, whose variability has large power on all timescales (see Figure 13 of Schaefer 2023a). With a chi-square fit to a sinewave, the 1-$\sigma$ uncertainty in the derived $T_{\rm max}$ is $\pm$1.0, $\pm$2.2, $\pm$0.8, and $\pm$0.8 days for the four intervals in time order. A major point in showing these folded light curves is so that the ellipsoidal modulation can be seen in detail. Another major point is to illustrate the sinewave fits from which the $T_{\rm max}$ are measured for the $O-C$ curve.
• Figure 2: Ellipsoidal modulations in $V$-band before the 1946 eruption. The details of this figure are the same as for Figure 1. The interval 1890--1910 has relatively few points, so the phase binned points (in red squares) have relatively large scatter, and $T_{\rm max}$ has a formal uncertainty of $\pm$5.5 days. For the earliest two intervals, the formal uncertainty in $T_{\rm max}$ is $\pm$2.8 and $\pm$2.1 days. This is important because those two points have the largest lever arm for proving that the $O-C$ curve had a large upward kink in 1946. Critically, these first two points are around 20 days late compared to that required for the $\Delta P$=0 case.
• Figure 3: $O-C$ curve for T CrB. The blue and green dots are the final $O-C$ values with one-sigma error bars for the $B$ and $V$ measures respectively (see Table 1), with smaller dots for measures with larger error bars. The best-fitting curve for the radial velocity fit is the thick black curve from 1946--2025. The dotted black curves show the extensions to earlier than 1946 for the $\Delta P$=0 case, with the thick dotted curve for the best fit $\dot{P}_{\rm post}$, and the flanking thin dotted curves representing the $\Delta P$=0 case with the 1-$\sigma$ range in the curvature. All of the $O-C$ measures before 1910 are significantly and substantially above the $\Delta P$=0 case. That is, T CrB certainly does not have a small or near-zero $\Delta P$. The best fit model must be a broken parabola, where the steady period changes between eruptions make for parabolic segments with a sharp kink in 1946. The parabola from 1866--1946 is accurately anchored in the year 1946 by the radial velocity data. The only adjustable fit parameters for the 1866--1946 parabola are then the orbital period immediately before the 1946 eruption ($P_{\rm pre}$) and the curvature ($\dot{P}_{\rm pre}$). With this, the best-fitting model with $\dot{P}_{\rm pre}$ equal to the best-fit $\dot{P}_{\rm post}$ is represented by the thick black curve from 1866 to 2025. The flanking magenta curves represent the cases where the 1-$\sigma$ variations in the curvature are followed. The nearly straight magenta curve represents the maximum possible curvature. In all cases, there must be a sharp kink in 1946. Visually, this can be seen because all the measures before 1910 are up to 20 days late as compared to any $\Delta P$=0 solution, and this is highly significant. The kink is upward, so the sudden period change in 1946 must be positive, meaning that the period $\it increased$ due to the nova. Quantitatively, the orbital period change is $+$0.146$\pm$0.019 days, and certainly $>$0.121 days. This $\Delta P$ is huge, being two orders-of-magnitude larger than for any other measured nova. This huge $\Delta P$ can only arise from a huge mass of ejecta.