Real-time observations of the transition to the quiescent state in an accreting magnetised neutron star: No propeller required?

The paper presents the first comprehensive, high-cadence monitoring campaign of the complete transition of the transient X-ray pulsar (XRP) 4U 0115+63 from the accreting regime to the quiescent state, using observations from the NICER X-ray telescope. The primary finding is that this transition, previously attributed to the 'propeller effect' (centrifugal inhibition of accretion), can be fully explained by the thermal-viscous disc instability model (DIM) without requiring the propeller effect as the primary mechanism.

Key results and arguments from the paper:

1. Observational campaign: The authors monitored 4U 0115+63 during the declining phase of its 2023 giant outburst with Swift and NICER, achieving an average interval of 2.6 hours between NICER observations during the critical transition period (MJD 60071–60074). This allowed them to resolve the transition timescale for the first time. The light curve shows that after the luminosity drops below 10 36 erg s-1, it continues to decrease gradually with an exponential timescale of 16.5 hours, with no sharp drop that would indicate a propeller onset.

2. Propeller effect analysis: The paper calculates the expected transitional luminosity for the propeller effect in 4U 0115+63, finding a broad range from several 10 34 to 10 36 erg s-1 depending on uncertain parameters (NS radius, magnetic field, factor k). It argues that two commonly cited observational signatures of the propeller—disappearance of pulsations and spectral softening—are not reliable discriminators. Pulsations are observed in quiescence for several XRPs, and spectral softening can be explained by low-level accretion physics. The paper notes that in Swift J0243.6+6124, a similar fast flux drop was observed but was definitively not caused by the propeller effect.

3. DIM framework: The paper proposes that the observed decay is driven by the viscous evolution of the accretion disc. During the decay phase, the disc consists of a hot, ionised inner region (high viscosity, α ≳ 0.1) and a cooler outer region (lower viscosity, α 0.01). The luminosity decay is governed by the inward propagation of a cooling front. The decay naturally consists of two stages: an irradiation-controlled stage at higher luminosities and a purely viscous stage at lower luminosities. The characteristic timescale shortens as the hot zone shrinks, producing the observed steepening of the light curve.

4. Numerical fits: The authors fitted the light curves of 4U 0115+63 (2015 and 2023 outbursts), SMC X-2 (2015 and 2022), Swift J0243.6+6124 (2022), and V 0332+53 (2015) using the freddi viscous-disc-evolution code. They present two sets of fits: one with fixed irradiation parameter C̃irr = 10-3 (fitting α), and one with fixed viscosity α = 0.3 (fitting C̃irr). The model reproduces the observed light curves during the final stages of all giant outbursts without invoking the propeller mechanism. The paper notes a degeneracy between α and Cirr, so uncertainties are not reported.

5. Quasi-stable low-luminosity state: The paper examines the 'meta-stable' plateau observed at 10 34 erg s-1 after the rapid decline. It considers the possibility of accretion from a 'cold' (recombined) disc, using Eq. (9) to estimate the critical luminosity for this transition. The observed plateau luminosities in 4U 0115+63, V 0332+53, and Swift J0243+6124 agree well with these estimates. However, the paper identifies difficulties: for 4U 0115+63, the inner portion of the disc adjacent to the corotation radius would be thermally unstable at the plateau accretion rate, and the magnetospheric radius would be beyond the corotation radius. The paper suggests that the disc structure might be influenced by the magnetic field or become optically thin, but a detailed model is deferred to future work.

6. Conclusions: The paper concludes that:

• The transition timescale to quiescence has been constrained for the first time.

• The observed light curve is successfully modelled using viscous disc evolution without the propeller effect, with reasonable values of viscosity and irradiation parameters.

• Low-luminosity plateaus can be interpreted as accretion from a cold disc, but this scenario is valid only if the propeller effect does not set in at a higher luminosity.

• The eventual cessation of accretion in deep quiescence may indicate either the onset of an efficient propeller regime at very low accretion rates or complete depletion of the disc, requiring further observations to clarify.

Improvements to AI Systems:

1. Time-Resolved Accretion Physics Modeling: Enhance AI systems to model transient X-ray pulsar light curves using thermal-viscous disc instability (DIM) frameworks instead of relying solely on propeller-effect paradigms. The improved system can automatically fit multi-epoch light curves (e.g., from NICER, Swift) with viscous decay parameters (α, irradiation coefficient C̃irr) and distinguish between irradiation-controlled and purely viscous decay stages, enabling real-time prediction of transition timescales and plateau luminosities.

2. Discriminator for Physical Mechanisms: Build an AI classifier that evaluates competing hypotheses (propeller effect vs. DIM) for flux drops in accreting systems. The system can ingest high-cadence light curves, spectral hardness ratios, and pulsation detections, then output probabilistic assessments of whether a drop is due to centrifugal inhibition or viscous cooling-front propagation, using the paper’s criteria (e.g., absence of sharp drops, exponential decay timescales, spectral softening explained by low-level accretion).

3. Parameter Degeneracy Handling: Implement AI algorithms that explicitly account for degeneracies between viscosity (α) and irradiation (C̃irr) when fitting disc-evolution models. The improved system can produce posterior distributions or confidence intervals that reflect this degeneracy, rather than reporting single best-fit values, and can flag when data are insufficient to break the degeneracy—improving robustness of astrophysical inferences.

4. Predictive Quiescent-State Modeling: Develop AI that predicts the onset and duration of quasi-stable low-luminosity plateaus (10 34 erg s-1) in XRPs. Using the paper’s Eq. (9) for critical luminosity and thermal stability criteria, the system can forecast whether a cold disc can sustain accretion, and identify cases where the magnetospheric radius exceeds corotation—prompting follow-up observations to test propeller onset or disc depletion.

5. Automated Observational Strategy Optimizer: Create an AI that plans high-cadence monitoring campaigns for future outbursts. Given the paper’s success with 2.6-hour intervals, the system can optimize telescope scheduling (e.g., NICER, Swift) to resolve transition timescales, prioritizing sources with predicted giant outbursts and dynamically adjusting cadence based on real-time luminosity decay rates.

6. Cross-Source Generalization Engine: Train a transfer-learning model on the fitted light curves of 4U 0115+63, SMC X-2, Swift J0243.6+6124, and V 0332+53 to generalize to other transient XRPs. The improved system can predict decay shapes and plateau properties for new sources with minimal data, using shared physical parameters (e.g., magnetic field, spin period) as priors, and can flag outliers that may require new physics.

7. Uncertainty-Aware Spectral/Pulsation Analysis: Upgrade AI pipelines that analyze pulsation disappearance and spectral softening to incorporate the paper’s finding that these are not unique propeller signatures. The system can model low-level accretion spectra and pulsation visibility under DIM, reducing false positives for propeller detection and improving classification of quiescent states.

8. Real-Time Disc Evolution Simulator: Integrate the freddi viscous-disc-evolution code into an AI-driven simulator that can be run interactively. The system can take user-specified parameters (α, C̃irr, NS properties) and generate synthetic light curves, allowing rapid hypothesis testing and comparison with observed data, including the two-stage decay and plateau features described.

The final stages of outbursts in transient X-ray pulsars (XRPs), which are characterised by a significant decline in the mass accretion rate, provide valuable insight into the physics of the accretion disc and its interaction with the strong magnetic field of the neutron star (NS). In particular, the `propeller effect', or centrifugal inhibition of accretion, has been proposed as a key mechanism governing both the onset luminosity and the timescale of the rapid transition to the quiescent state. In addition, it offers an independent method for estimating the magnetic field strength of the NS. On the other hand, the decrease in the mass accretion rate itself is driven by processes occurring in the accretion flow at larger distances from the NS. Recovering the information encoded in the light curve therefore requires sensitive high-cadence X-ray monitoring capable of capturing the rapid and often unpredictable transition from the accreting regime to the quiescent regime. In this study, we present the results of the first comprehensive monitoring campaign that tracks the entire transition to quiescence in the transient XRP 4U 0115+63 utilising observations by the NICER X-ray telescope. We show that the observed behaviour can be explained by the thermal-viscous disc instability model (DIM), with the emission observed immediately after an outburst possibly arising from the ongoing accretion from the recombined (`cold') disc and the subsequent quiescent emission being produced by the cooling NS. We further applied this model to a larger sample of XRPs encompassing a broad range of physical parameters. Ultimately, our findings indicate that the temporal behaviour of XRPs, including the quiescent state, can be consistently explained within the DIM framework without requiring the propeller effect as the primary mechanism governing the observed transition.