The Pulsar That Threw A Tantrum

Millisecond pulsars (MSPs) are nature’s cosmic clocks - spinning with mind-boggling speeds, hundreds of times per second. As a sub-class of the neutron star zoo (and thus, pulsar family) MSPs distinguish themselves by their rapid rotations, subdued magnetic fields, and remarkable long-term stability. For over forty years, astronomers have been observing and tracking (timing) the pulses from these enigmatic stellar remnants, to within microseconds, and sometimes even nanoseconds accuracy. In fact, this stability makes them rival that of Earth’s atomic clocks over the long term. At the heart of this stability lies the integrated pulse profile: a unique fingerprint formed by averaging many thousands of pulses from the MSP.

This integrated profile is not just a simple parameter or observation - it is the cornerstone of highly-precise pulsar timing experiments. Pulsar astronomers rely on these integrated profiles as a reference (a ‘template’), comparing each new observation’s arrival time and shape against the established shape of the template to ensure consistency (which tells us they are stable). When the observed profiles deviate from the template, it raises red flags. These types of variations could be hinting at intriguing astrophysics, or more mundane reasons like telescope calibration issues or radio frequency interference (RFI) locally, or from satellites orbiting Earth.

Over the last few years, astronomers have started to notice some of the MSPs misbehaving. Their integrated profiles have been showing signs of profile variability. For pulsar astronomers involved in high-precision timing experiments, such as using MSPs to search for elusive low-frequency gravitational wave backgrounds, this variability is more than a curiosity. Its potential complications could ripple across the MSP population.

Astronomers rely on a rapid-alert system known as ATEL - the Astronomer’s Telegram - to broadcast their discoveries that demand immediate attention to the community. This prompts other researchers to scour their archival data or quickly swing their telescopes towards the source to collect new data on the event / object / phenomenon. 

In May 2021, an ATEL was circulated to the community by astronomers in China using the world’s biggest radio telescope - the Five-hundred-metre Aperture Spherical Telescope (FAST):

"A SUSTAINED PULSE SHAPE CHANGE IN PSR J1713+0747 …"

Within days, two other ATELs quickly spread through online networks. The CHIME telescope in Canada confirmed the anomaly and constrained the date of the event to be around 16-17 April 2021 with their ATEL:

“CONFIRMATION OF A CHANGE IN THE EMISSION PROPERTIES OF PSR J1713+0747 ….” 

And from India’s uGMRT, a low-frequency perspective added weight: 

“LOW-FREQUENCY VIEW OF A PROFILE CHANGE EVENT IN PSR J1713+0747 ….” 

Something significant had occurred on the millisecond pulsar PSR J1713+0747, and it wasn’t subtle.

Since its discovery in 1993, this MSP has remained a high-priority target in global pulsar timing array (PTA) experiments, due to its favourable sky position (all the PTAs can observe it) and its bright, narrow profile. It’s also been successfully used in several tests of General Relativity. 

The fact that several global telescopes across continents detected the same profile shape change ruled out local RFI or instrumental quirks. This was astrophysical and potentially going to be very disruptive to PTA observations.

If one of the most stable and highest priority MSPs in the PTA catalogues was undergoing profile changes, it raised several critical questions: why? Are other MSPs also doing it? How do we fix it? 

Every pulse from any pulsar that sweeps across the Earth has its own unique profile shape, and no two individual pulses are similar. For MSPs, hundreds of pulses per second are recorded, so astronomers harness this rapid-fire succession of pulses to utilise in pulsar timing work, averaging the signals over the course of the observation (for the Parkes Pulsar Timing Array (PPTA), it’s usually about an hour) and generating a time-averaged observation: a single, high signal-to-noise snapshot that serves as the sample to measure against the template. 

These profiles are more than just data - they are windows into the pulsar’s emissions, magnetosphere, and rotation. As the radio emission beam (which is aligned to the offset magnetic pole) sweeps around with the star’s rotation, we glimpse different emission regions shaped by the extreme magnetic field and distribution of plasma within this field. Each pulse and time-averaged profiles, therefore, provide insights about the complex and exotic environment localised to the MSP.

Pulsar astronomers rely on the unwavering nature of these profiles, exploiting the stability of MSP profiles in high-precision timing experiments, such as searching for nanohertz-regime gravitational wave backgrounds, or testing the limits of  General Relativity studies. Timing accuracy often reaches into the microsecond - or even nanosecond range - demanding this high degree of precision.

However, not all MSPs behave perfectly all the time. In a few rare cases, a variety of profile changes do occur. Young pulsars, for example, fresh from their supernova origins, can experience starquakes or sudden decoupling of the crust from the interior of the star, which leads to sudden timing shifts known as glitch events. Typically, these don’t affect the pulse profile; however, a few exceptions have been noted. 

MSPs, who are the elders of the pulsar family - ranging from millions to billions of years old, have long since settled from their turbulent youth, and glitching is exceedingly rare (only two have been recorded so far). Other phenomena, like moding and nulling - where the emissions can flip between two quasi-stable states or disappear altogether, have been observed in a small handful of MSPs, but these tend to occur on shorter timescales, sometimes within rotations, or after a few days. 

Some pulsars (canonical and MSPs) exhibit stochastic variability in parts of their profiles. These random fluctuations are seen in the time-averaged observations, are driven by magnetospheric processes and can persist for months or even years. Meanwhile, propagation effects - caused by the pulsar’s signal traversing the ionised interstellar medium (IISM) can also distort the profile. Luckily, the IISM effects are not intrinsic to the pulsar and are easily identified by their frequency-dependent nature. 

Rarest of all are the long-term, discrete profile changes that unfold over months to years. To date, there are only two MSPs that have shown this behaviour: PSR J0437-4715, which has shown long-term profile variations, and PSR J1643-1224, whose profile underwent a dramatic and permanent reconfiguration.

Discovered using the iconic 300m Arecibo Telescope, PSR J1713+0747 quickly earned its place as one of the most favourable MSPs in the pulsar timing community. Its bright, stable profile and favourable sky position made it a perfect candidate for all global PTA experiments. 

When our team turned our attention to the April 2021 profile change event on this MSP, we had a unique advantage over other global telescopes. Murriyang (the CSIRO Parkes radio telescope) has been equipped with the ultra-wideband low-frequency (UWL) receiver since the end of 2018. This powerful instrument spans a colossal bandwidth of 3.3 GHz (ranging from 704 MHz - 4032 MHz), allowing us to observe the MSP across a vast stretch of radio frequencies. With this, we could dissect the profile change event in detail, teasing out how it behaved across the spectrum. 

We also had another edge up our sleeve. Like all MSPs in the PPTA suite, PSR J1713+0747 had been tracked since the installation of the UWL (and indeed further back to 2004 with other instruments). This gave us a rare opportunity to also study the temporal evolution of the profile change -  not just as a snapshot, but rather as a dynamic story that is unfolding in time. 

The UWL also enabled us to measure the polarisation of the pulsar’s radio light - an essential tool for probing the magnetosphere. Polarisation refers to the orientation plane of light, which can be linear or circular. Pulsar emissions, in general, are intrinsically highly polarised, with linear polarisation being more dominant than circular polarisation. When light passes through a magnetised plasma, such as the IISM, the plane of the linear polarisation rotates - a phenomenon known as Faraday Rotation. We hoped to decipher clues about this event and its impact on this system by studying how these polarisations changed as a function of time and frequency in response to the April 2021 profile change.

What we found was striking. The entire 3.3 GHz bandwidth was affected - but not uniformly. Different frequency ranges responded differently, revealing a non-monotonic frequency dependence in the event. We also found that the change persisted for years, with lingering effects still detectable today. 

But the exciting revelations came from the polarisation data. Some components of the linear polarisation profile were significantly affected, whilst the other remained largely untouched. More specifically, one of the orthogonal polarisation modes (OPMs - two plasma wave modes in the magnetosphere) was much more significantly affected as a result of this profile change, relative to the other. This asymmetry was unprecedented. No other MSP had ever shown such behaviour.

This was our smoking gun. 

It pointed directly to the MSP magnetosphere being the origin of the profile change, and not the IISM. The fact that only one of the OPMs was affected ruled out propagation effects. No such event or phenomenon has ever been documented in any MSP. How can these supposedly stable integrated profiles and OPMs suddenly be changed?

We began to explore possible causes. 

One hypothesis we had: the magnetic field lines within the MSP magnetosphere had suddenly snapped and reconfigured. Another: the plasma density inside the magnetosphere had redistributed, creating a temporal alteration in the way the emissions pass through this birefringent medium before slowly settling back. 

Yet, despite our years of data, modelling and analysis - we still don’t know.

We tested two scenarios: one where the profile was permanently reconfigured, and another where in the very long term, the profiles would re-settle back into their pre-event configuration. However, neither of these models was statistically conclusive. If the April 2021 profile change event on PSR J1713+0747 does indeed turn out to be a permanent reconfiguration, then it would only mark the second time such a phenomenon has been observed in any MSP, as well as the most significant.

But for now, the mystery remains. We don’t know what triggered the change. We don’t know if it is permanent.

While the 2021 profile change event in PSR J1713+0747 stands as the most significant event of its kind ever observed on any MSP to date, it wasn’t the first time this favourable cosmic clock showed signs of misbehaviour. Back in 2008, then again in 2016, astromers recorded similar disturbances - albeit smaller in scale, but still notable. In both of these cases, the profiles recovered between 100 - 200 days post their events, returning to their usual configurations. 

Initially, these two anomalies were chalked up to being passing blobs of ionised materials in the IISM, drifting between us and the MSP and distorting the signal. However, by the second event, the IISM explanation began to unravel, and for at least the 2016 event, this was determined by astronomers as also being magnetospheric in origin. 

PSR J1713+0747 isn’t alone in this aspect, either. Other MSPs have shown signs of stochastic profile variations, and two cases in the MSP population (PSR J643-1224 and PSR J0437-4715) have both exhibited long-term discrete profile changes. Holistically, all of these cases suggest an underlying truth: perhaps MSP profiles aren’t quite as stable as we once believed.

They are definitely more stable than their younger, slower cousins, but when our experiments demand timing precision down to the nanosecond regime, then even the smallest deviations matter. 

This has major implications for the future of pulsar timing science - especially in PTA experiments which are hunting nanohertz-regime gravitational waves. It’s likely that these profile changes have always been occurring, just on timescales and magnitudes too subtle for our current instruments to detect. They’ve probably been hiding below the noise floor all this time.

With next-generation telescopes (e.g., SKA, NGVLA, DSA-2000) soon to come online over the next few years, we’re likely about to uncover more of these elusive events, which we will then need to account for in our MSP models. 

To that end, at the PPTA, we are now conducting a comprehensive overview of all of our MSPs in our long-term dataset. We’re searching for signs, however faint, that other MSPs also have profile change features.

If we find more, then it will reinforce a growing realisation: MSPs, whilst remarkably stable, may not be the perfect timekeepers we once thought they were.