2004A&A...426..229G

Pulsars with drifting subpulses are thought to be an important key to unlocking the mystery of how radio pulsars work. We present new results from high sensitivity GMRT observations of PSR B0826-34 - a wide profile pulsar that exhibits an interesting but complicated drifting pattern. We provide a model to explain the observed subpulse drift properties of this pulsar, including the apparent reversals of the drift direction. In this model, PSR B0826-34 is close to being an aligned rotator. Using information about the polarization and frequency evolution of the pulse profile, we solve for the emission geometry of this pulsar and show that the angle between the rotation axis and the dipole magnetic axis is less than 5°. As a result, our line of sight samples a circular path that is entirely within the emission beam. We see evidence for as many as 6 to 7 drifting bands in the main pulse at 318MHz, which are all part of a circulating system of about 15 spark-associated subpulse emission beams that form, upon averaging, one conal ring of the mean emission. We also see evidence for a second ring of emission, which becomes dominant at higher frequencies (above 1GHz) due to the nature of the emission geometry. We model the subpulse drift behaviour of this pulsar in detail, providing quantitative treatments of the aliasing problem and various effects of geometry which play an important role. The observed drift rate is an aliased version of the true drift rate which is such that a subpulse drifts to the location of the adjacent subpulse (or a multiple thereof) in about one pulsar period. We show that small variations, of the order of 3-8%, in the mean drift rate are then enough to explain the apparent reversals of drift direction seen in the data. We find the mean circulation time of the drift pattern to be significantly longer than the predictions of the original (Ruderman & Sutherland, 1975ApJ...196...51R) model and propose an explanation for this, which relates to modified models with temperature regulated partial ion flow in the polar vacuum gap. The small variations in drift rate are then explained by very small heating and cooling effects - less than 3500K change in the ∼2.5x10⁶K surface temperature of the neutron star polar cap. From a detailed consideration of the variation of the mean subpulse separation across the main pulse window, we show that the circulating spark pattern is not centred around the dipole axis, but around a point much closer (within a degree or so) to the rotation axis. This is an indicator of the presence of a ``local pole'' corresponding to the non-dipolar magnetic fields that are expected to be present close to the neutron star surface. PSR B0826-34 thus provides a very rich and powerful system in which to explore important aspects of the physics of pulsar radio emission and neutron star magnetospheres.