A New Glitch on a Millisecond Pulsar

Millisecond pulsars (MSP) are well known and prized in astrophysics for their long-term rotational and integrated profile stability. These cosmic clocks are so reliable that, over long timescales, they’ve proven to be more stable in timing when compared to terrestrial atomic clocks.

This stability is extremely important and underpins a wide range of high-precision timing experiments from testing of Einstein’s General Relativity in the strong field regime, searching for low-frequency gravitational wave backgrounds, and probing the exotic states of matter inside neutron stars, where densities exceed anything achievable on Earth.

Across the MSP population, the core fundamental aspect is that this stability generally holds. Only a small handful have ever shown irregularities in their rotations. Among these rare events, glitches - which are abrupt jumps in rotation rate - are extremely unusual. Until now, only two glitches had ever been observed in MSPs since their discovery in the early 1980s. 

A new study, published on the pre-print server arXiv, has now reported the third glitch event ever seen in an MSP.

“We first noticed the anomaly while examining the Time-of-Arrival (ToA) residuals as part of the profile-shape variation project for MeerKAT Pulsar Timing Array,” said University of Manchester PhD Candidate Bhavnesh Bhat, who led the study.

“PSR J0900−3144 showed a clear, unexpected linear-slope trend in its timing residuals that could not be removed with standard noise modelling, and the form of the slope was characteristic of a glitch-like frequency jump.”

When observing pulsars, astronomers detect the signals from their radio pulses rather than directly observing the star itself, since neutron stars are only about 20 km across and are located hundreds to thousands of light-years away. These pulses can occur dozens ot times per second for young pulsars, or hundreds of times per second for MSPs. Although each individual pulse varies in shape from one rotation to the next, averaging hundreds or thousands of these pulses together produces a stable ‘integrated profile’ which becomes the key tool for precision studies, including monitoring the long-term variations of this profile to decipher any encoded information about gravitational waves, the pulsar’s magnetosphere or the interstellar medium.

Another important observable of pulsars is the period derivative, or the slowdown rate of the pulses. With each rotation, the pulsar loses a tiny amount of energy by dumping magnetic dipole radiation energy into its surrounding environment, and of course, emitting electromagnetic radiation. This gradually slows the pulsar down, and astronomers often refer to this slowdown rate as the “P-Dot”.

By plotting both the rotation period (“P) and the P-Dot value of all the known pulsars into a diagram, astronomers can create a visual representation which shows the sub-population of different types of neutron stars, including MSPs, the younger, more common canonical pulsars, magnetars and more.

Millisecond pulsars appear in the lower left of this diagram as the smaller cluster: they have faster rotation periods and smaller P-Dot values. This diagram also highlights the estimated ages and magnetic field strength of these objects, as well as indicating which are in binary systems (where they have a companion and orbit a common centre of mass).

“MSPs are older, more stable, and have interiors thought to be closer to equilibrium, making them less volatile and far less likely to undergo angular-momentum transfer events like glitches,” said Bhat.

This broader view gives astronomers clues as to the evolutionary history of these objects. In particular, MSPs are much older, have dampened magnetic fields, and nearly all of them are part of a binary system. Their rapid rotation tells us that these objects underwent a historical period of accretion, where they siphoned material from their ageing companions, causing them to speed up to millisecond rotation periods. This process evolves over millions to billions of years - enough time for the neutron star to cool and settle, smoothing out the internal stresses that might otherwise cause starquakes and glitches. This is why rotational glitches are far more commonly observed in the younger pulsar population.

Despite decades of studies and timing of pulsars, the jury is still out on what actually causes glitches. Earlier concepts suggested they were triggered by starquakes - sudden cracks in the crust of the neutron star. However, more recently, modelling has indicated that this is insufficient to match observations of glitch activity and magnitudes observed across the pulsar population.

The current leading explanation involves the transfer of angular momentum between two interior components of the neutron star - the solid crust, which gradually slows over time, and the superfluid interior, which can continue rotating at a higher rate. These two structures are thought to co-rotate, but as the crust begins to slow down as energy is lost from the system, the superfluid below it continues to rotate at the higher angular velocity.

This difference creates an increasing tension between these two layers. When the stress reaches a critical threshold, the lag breaks, allowing the superfluid interior to transfer angular momentum to the crust. This causes the crust to speed up abruptly, which is observed as a sudden glitch in the pulsar’s period and P-Dot.

To date, only two glitches have ever been detected in MSPs, and both of these events were small in magnitude compared to the glitches observed in the younger, canonical pulsar population.

Now, a new paper led by Bhat has reported the observation of a third glitch event in the timing datasets of PSR J0900-3144. 

“The event showed a clean, step-like change in spin frequency in the combined dataset, and the long observational span let us model and contain noise parameters reliably, making the glitch signature stand out clearly and ruling out timing noise,” he said.

The event occurred around late December 2022, and appears consistently across the datasets of three global pulsar timing array teams (MPTA - which uses the MeerKAT telescope; PPTA - which uses Murriyang, the CSIRO Parkes radio telescope; and the EPTA, which uses both the Nancay and Lovell telescopes).

“Multiple pulsar timing array experiments provided datasets that increased the bandwidth and observation span, giving us a better constraint on the noise model, and confirming the glitch with help of Bayesian analysis.”

To verify the signal was indeed a real glitch event, the team of astronomers also used a technique known as Bayesian model comparison, in which they tested the datasets through a model that had no glitch present, and one that did. They found that the model that did have the glitch present was much more highly in favour, strengthening confidence that this event is real. The magnitude of this glitch event on PSR J0900-3144 is also comparable to the other two glitch events that have so far been observed on the two other MSPs.

Additionally, and now using this third event, the team also reassessed the expected glitch rate within the MSP population. With the expanded timing datasets, they could better estimate how often such events should occur. Their analysis suggests that glitches should occur approximately once every 400 years in any single MSP, assuming the glitch rate is constant across all MSPs. The team also calculated the likelihood of catching future glitches with current high-cadence monitoring, and found the probability of observing at least one glitch in the next five years in an MSP to be approximately. 78%, rising to 99% over the next 15 years.

Sensitive timing experiments, such as those which are conducted by global pulsar timing array teams, rely on the ongoing rotational stability of MSPs. When rare events, such as glitches or profile changes occur, they naturally attract lots of exciting attention.

These types of events can contribute to the broader noise budget, which also includes all the other noise factors - such as intrinsic profile changes, variations in the solar wind, instrumental effects, the impact of the interstellar medium and more. To confidently detect feeble signals like the nanohertz-regime gravitational-wave background, astronomers must decouple these contributions as these noise factors contaminate the final desired result.

The challenge is that some of these events, like glitches observed in MSPs, are extremely small. There may well be additional glitches occurring all the time below the noise floor and sensitivity limits of current instruments. As datasets grow, more MSPs are added and instruments become more sensitive, it is likely that these types of events will bubble up to the surface to be detected, characterised, analysed and modelled.

Whilst one MSP (or even a handful) undergoing these types of events (be that glitches or profile changes) - is not a total breakdown of pulsar timing experiments, the risk lies in unmodelled events contributing to the overall noise budget and masking the elusive pure gravitational wave background signal. Better understanding and modelling of these behaviours is therefore essential.