Preguntas de PMO

In § 5.1.1, we discussed the variable behaviour of pulsars on a period-by-period basis, e.g. the sub-pulse drifting shown in Figure 5.1. Of relevance to the dis- cussion of RRATs is nulling — the particular type of mode-changing where one mode shows a complete lack of radio emission. As typically observed, nulling occurs for 1 − 10 rotation periods, but the observed selection of nulling pulsars is quite biased (Wang et al., 2007). Pulsars with longer nulling fraction are less likely to be detected in a single survey pointing, and in a confirmation pointing, and hence may be discarded amongst the plethora of pulsar candidates produced in modern surveys. Also, due to a lack of sufficient signal-to-noise ratio, weaker pulsars cannot be examined on shorter timescales. Thus there may well be nulling occurring either unnoticed or undetectable in many known pulsars.

As well as the RRATs, 2006 also saw the discovery of ‘intermittent pulsars’, sources which behave as normal radio pulsars for several days before switching off entirely for days to weeks. This switching occurs in a quasi-periodic fashion with PSR B1931+24, the archetypal system, turning ‘on’ for 5 − 10 days and ‘off’ for 25 − 35 days (Kramer et al., 2006). These timescales allow the measurement of separate slow-down rates during the on and off states, ˙νonand ˙νoff. The difference

in these rates is about 50% which seems to reflect the extra energy loss due to the pulsar wind when there is radio emission, i.e. when off, the star slows down via dipole braking alone (the vacuum scenario of Chapter 2), and the magnetospheric plasma density decreases hugely. The on-off transition has been observed and lasts less than 10 seconds, indicating a massive change in magnetospheric currents on a very short timescale to a new state which is apparently stable for ∼ 106

periods before switching once more. The explanation as to why this switching is quasi-periodic is unknown. Recently Lyne et al. (2010) have shown switching between two stable states in 17 pulsars. Like in the case of PSR B1931+24, switches between two slow-down rates are observed with changes ranging from

0.3 to 13.3%, with highly-correlated pulse shape changes, i.e. moding. These results further strengthen the claim of Wang et al. (2007) and others that moding and nulling are due to the same underlying phenomenon. Also, the data are not inconsistent with such switching being a generic property of all pulsars (A. G. Lyne, private communication).

Also of relevance to transient neutron stars is the pulsar ‘death valley’, where radio emission is thought to fail. Pulsar emission requires a supply of particles from the stellar surface which can be accelerated in the pulsar magnetosphere for pair production, γ → e++ e−, to ultimately lead to coherent radio emission. The strength of the electric potential ∆V depends on B and P , e.g. in the toy model of Goldreich & Julian (1969) ∆V ∝ B/P2. An electron accelerated in this potential will acquire a Lorentz factor of e∆V /mec2. Depending on the

emission mechanism (i.e. the dependence of ∆V on B and P ) the minimum Lorentz factor sufficient for pair-production (the photon must have energy of at least 2mec2) defines a ‘death-line’, separating regions of P − B space where

radio pulsar emission is possible and regions where it is inhibited (the ‘death valley’). Detailed considerations lead to different death-lines for different field configurations, e.g. on high curvature field lines (Chen & Ruderman, 1993), and death-lines for several emission mechanisms have been proposed (see e.g. Arons (1996); Qiao & Zhang (1996); Zhang et al. (2000)). However, the various death- lines do not satisfactorily explain the observed pulsar population and there is at least one pulsar which flouts the rules in the death valley, namely the 8.5-second PSR J2144−3933 whose detection as a radio pulsar poses serious challenges to emission theories (Young et al., 1999).

The magnetars are thought to be isolated neutron stars with very strong magnetic fields of 1014− 1015 _{G, whose emission is powered by magnetic field de-}

cay (Woods & Thompson, 2004). For such strong magnetic fields pulsar emission may be even more complicated, as these fields exceed the ‘quantum critical field’ strength5_{, B}

QC = 4.4 × 1013 G, where higher order Quantum Electrodynamic

effects play a role. For example, the amplitude for photon splitting, γ → γ + γ, a third order effect, is proportional to α3_(~ω/m

ec2)5(B/BQC)6 (Adler, 1979) where

α is the fine structure constant and ~ω is the photon energy. In magnetic fields & BQC this dominates over photo-pair creation, quenching the build-up of plasma

5_{The quantum critical value corresponds to the energy gap of electron cyclotron orbits}

(‘Landau levels’) equalling the electron rest mass. In SI units ∆E = ~eB/me so that BQC =

and hence the radio emission (Baring & Harding, 1998). For many years the known magnetars were radio-quiet and the known pulsars were radio-loud, ap- parently well separated into groups where this process was either dominant or suppressed. However the recent discovery of 3 radio magnetars (Camilo et al., 2006a, 2008; Levin et al., 2010), J1819−1458 and a handful of other radio pul- sars with B > BQC has changed this. While there is a dearth of pulsars in the

B ∼ BQCregion the existence of any is puzzling. Weise & Melrose (2006) suggest

that perhaps photon splitting may not always dominates, if single polarisation selection rules forbid it. However we might ask a more basic question: if the magnetars are powered by magnetic energy, not by rotation, then why is it fair to use the vacuum rotator expression of Equation 2.21 to estimate the magnetic field strength of magnetars? This is what is commonly done but the field strengths so derived may not be trustworthy. A related question is at what point this becomes an unreliable estimate, e.g. is it fair to use this estimate for J1819−1458? Then, in translating P − B death lines to P − ˙P space we might require a different rule than the canonical B ∝ pP ˙P , although what that might be is not known.

Conclusions

9.1

What Do We Know Now?

Here we quickly summarise the work presented in this thesis. After a review of radio transients (Chapter 1) and neutron stars (Chapter 2) we examined, in Chapter 3, what it would mean if RRATs were, as had been suggested, a distinct population of Galactic neutron stars. This led us to conclude that there would be a ‘birthrate problem’, i.e. the observed classes of neutron stars would be incompatible with the observed supernova rate1_{. However this is only the case if}

the classes are distinct and can be resolved if the various observed manifestations are in fact evolutionarily linked in some way. No such evolutionary framework exists for pulsars, which demonstrates our lack of knowledge of neutron star evolution post-supernova.

Although it seems clear without RRATs (e.g. see Figure 3.2), if they could somehow be forgotten about, then the birthrate problem would be eased some- what. So, with the RRAT estimate figuring so highly we next questioned, in Chapter 4, whether the RRAT population was in fact as large as proposed by McLaughlin et al. (2006) — perhaps, for instance, there was a huge overestima- tion. To investigate this it is necessary to find more sources. As we had developed new algorithms and software for improved searching, we decided to completely re-process the PMPS. It had been claimed that as much as 50% of the RRATs detectable in the survey had been obscured by RFI (McLaughlin et al., 2006) so we applied RFI mitigation techniques (see § 4.4) in our re-processing. Our anal- ysis, which we refer to as PMSingle, was successful and identified a further 19

1_{See Appendix C for some supplementary information for Chapter 3}

new sources, to add to the original 11 detections. Of these 19 PMSingle sources, 12 have been observed multiple times whereas 7 have been observed on only one occasion. The sources which have not been re-observed may have very low burst rates, or, may in fact be single transient events. One source in particular is of great interest due to its suggested extragalactic distance of > 50 kpc (Cordes & Lazio, 2002). These discoveries are consistent with the initial population estimate for RRATs — we removed the effects of ‘RFI blindness’, which effected ∼ 1/2 of the PMPS pointings, and (more than) doubled the known PMPS RRATs. Thus the birthrate problem does not ‘go away’ and RRATs must be explained within the context of known neutron star classes. Fortunately, as we will discuss in the following section, this is possible.

To do this we need to further characterise their properties. With this goal, we began a focused campaign of monitoring observations so as to obtain coher- ent timing solutions for as many RRATs as possible through observations at Jodrell Bank and at Parkes. The methods used and difficulties encountered in this endeavour are described in Chapter 5. These studies revealed glitches in J1819−1458, with anomalous post-glitch recovery of the slow-down rate. For the original 11 RRATs the number of coherent timing solutions is now 7, up from 3. Furthermore, of the newly discovered repeating PMSingle sources, timing so- lutions have been obtained for 7 sources. These new solutions are presented in Chapter 6. We then described, in Chapter 7, an upcoming X-ray observation of a ‘dying’ pulsar, one of the PMSingle RRATs. Finally we discussed a simultane- ous optical-radio observation of a RRAT, motivated by the possibility of a giant radio pulse association. Having reviewed the work of other authors in studies of transient neutron stars in Chapter 8, we now reflect on what it all means.