The Thousand-Pulsar-Array programme on MeerKAT -- XVIII. Complex sub-pulse modulation patterns, bi-drifting and mode changing of nine radio pulsars

TL;DR

This study analyzes nine bi-drifting pulsars from the MeerTime Thousand-Pulsar-Array survey using MeerKAT and complementary GMRT data. By applying baseline subtraction, longitude-resolved fluctuation spectra, sub-pulse phase tracking, and $P_3$ folding across multiple bands, the authors identify persistent bi-drifting and frequent mode changes, even in narrow-profile pulsars with small $P_3$ values. The results challenge the classical circular carousel model and motivate exploration of alternative geometries such as elliptical or spiral spark tracks, including non-dipolar magnetic-field effects. The findings place bi-drifting pulsars within the broader drifting-pulsar population and highlight the need for diverse theoretical models to explain complex drift and mode-changing behavior across frequency and time.

Abstract

We present a detailed analysis of sub-pulse modulations in nine pulsars which show evidence of changes in sub-pulse drift direction as a function of pulse longitude in the Thousand Pulsar Array single pulse survey with MeerKAT. We confirm that all of these are consistent with persistent drift direction changes. These 'bi-drifting' pulsars present a challenge to the classical carousel model for sub-pulse drifting. In general, bi-drifting in this expanded sample is less clear than some of the previously published cases, which we attribute to narrower profile widths ($<20$ degrees) or smaller $P_3$ values (close to 2$P$). However, given the broad variety of pulse shapes and drift behaviours across the pulsar population, it is unsurprising that the phenomenon is not limited to only those where it can most easily be detected. Four of our samples show at least two emission modes with different profile shapes and drift properties, which seems to be a relatively common feature of bi-drifting pulsars. We also find jumps in sub-pulse phase between adjacent components in two pulsars. In addition to our MeerKAT L-band data, we used GMRT observations for four, and MeerKAT UHF observations for two of these pulsars to investigate the frequency dependence of sub-pulse drift. We find subtle changes in the drift as a function of frequency, but no clear overall pattern. Looking at the distribution of bi-drifting pulsars over $P$, $\dot{P}$ and $P_3$ suggests they are consistent with the underlying population of all drifting pulsars.

Figures (25)

• Figure 1: MK-L ($\sim\!1284$ MHz) pulse stacks, LRFS and mean profiles of our sample of pulsars that show bi-drifting. The bottom panels show the pulse stacks of up to 1000 pulses from these observations of each pulsar. For PSRs J1537$-$4921 and J1921+2003, the power is clipped at 1/5 the maximum power to bring out the weak single pulse features. The top and middle panels show the mean profiles and LRFS of the full pulse stack respectively, with the power clipped at 0.1 of the maximum power to show weaker spectral features clearer.
• Figure 2: As Fig. \ref{['fig:PulseStacksTPA']}, but showing pulse stacks of the GMRT observations ($\sim650$ MHz). The pulse stacks show 1000 pulses, starting from the 1000th pulse of the observations to show mode changes.
• Figure 3: The figure shows the LRFS of the two modes from the long MK-L observation of PSR J1418$-$3921. The left plot shows the bright mode and the right plot shows the weak mode. The plots have three panels: the central panel shows the LRFS; the top panel shows the standard deviation of the pulse stack (black solid) which indicates the modulation intensity, and the mean profile of the mode separated pulse stack (blue dashed). The left side-panel shows the integrated fluctuation power, which emphasises the strength of the fluctuations as function of fluctuation frequency. The FFT lengths used are 82 for the the weak mode and 512 for the bright mode.
• Figure 4: The $P_3$ fold plots of J1418$-$3921 MK-L (top row) and GMRT (bottom row) observations in the bright (left column) and weak (right column) modes. The $P_3$ used for folding is 2.5$P$, the smoothing factor used is 20. The top panel shows the mean pulse profile of the pulse stack. The bottom panel shows the $P_3$ fold together with sub-pulse phases. The sub-pulse phases of the full pulse stack are represented by orange dots with error bars, the phases of the first and second halves of the pulse stack are marked with blue and violet horizontal bars. The plotted phases are limited to those with error bars smaller than 10% of the $P_3$. The sub-pulse phases are vertically aligned to the highest intensity of the $P_3$ fold features (see the main text for further details). Since the $P_3$-fold is cyclic, two cycles are shown to ensure continuity around the wrap boundary. To make the plot clearer, only one complete phase track is shown. The colours represent the intensity fractions relative to the mean. For more details, see Section \ref{['sec: P3Folding']}.
• Figure 5: LRFS plots of the different modes of PSR J1534$-$4428 with the MK-L observation. The panel arrangement is the same as described in Fig. \ref{['fig:J1418-3921TPAModesLRFS']}. The FFT lengths used are 138 for the double component and 114 for the triple component modes.