Tom Hassall

Mapping pulsar emission regions with LOFAR

1. ? Mapping pulsar emission regions with LOFAR Tom Hassall (University of Southampton) & the LOFAR PWG
2. Wide-band Simultaneous Observations of Pulsars • Ultra wide-band observations using 3 telescopes simultaneously • Aims: • Probe of the ISM • Probe of the pulsar magnetosphere Hassall et al. (accepted by A&A)
3. LOFAR LBA (DE601) 60 MHz 36 MHz band The Effelsberg 100-m Telescope 8.350GHz 1GHz band Frequency (MHz) LOFAR HBA (CS302) 163 MHz 48 MHz band The Lovell Telescope 1524 MHz 512 MHz band Observations
4. • The refractive index of ionised gas: • So group velocity is given by: • And travel time over some distance is given by: DM = ∫nedl • High frequency emission arrives before low frequency 2 Dispersion
5. Dispersion • First detection of Giant Pulse from Crab Pulsar in LBAs • Emission at 80MHz arrives ~50s before same pulse at 50MHz
6. Scattering • Light is scattered in random directions by ISM • More scattering => longer path length • Takes longer to arrive • Also less likely to happen
7. • LBA GP ~ 2 seconds long • HBA GP ~ 0.1 seconds long Scattering
8. Other Effects? • Reports of “Super Dispersion” by Shitov & Malofeev 1985, Kuz’min 1986 and Shitov et al 1988 • Disputed by Phillips and Wolszczan 1998 • Recent paper by Cordes & Shannon gives a list of possible delays caused by ISM with steep frequency dependences...
9. Refraction
10. Multi-path DM Averaging
11. Extra DM Terms • DM law is a Taylor series expansion • Next term : • This can be used to determine the Emission Measure (EM) • A probe of the “clumpiness” of the ISM Δt2DM = 0.25 EM ν-4
12. Results (i) • Pulses at all frequencies arrive at the same time (apart from the dispersive delay) • Largest second-order ISM delay @40 MHz <4ms • Extrapolating to pulsar timing frequencies <50ns (~100ns precision necessary for first generation of PTAs) • ISM is relatively smooth • No large “clumps” (at least along the sight-lines we probed)
13. Not to scale! Magnetosphere Rotation axis Magnetic axis Pulsar Image credit: Hessels Radius-to-Frequency Mapping
14. Not to scale! Magnetosphere Rotation axis Magnetic axis Pulsar Radio Frequency Image credit: Hessels Radius-to-Frequency Mapping
15. PSR B1133+16 Fig. 13. Linear polarisation profiles of PSR B0809+74 between 925 and 328 MHz (Gould & Lyne 1998; Edwards & Stappers 2003b), and LOFAR polarisation profile at 136 MHz (black lines) plotted along with the Stokes I profiles at each frequency (grey lines). The polarised com- ponent moves from the position of the leading component of the profile towards later pulse longitudes, tracing the broad component of the pulse profile. The width of the conal components as a function of fre- quency has been a subject of interest in the past. Mitra & Rankin (2002) found that the component widths remain constant be- tween 40 and 3000 MHz. We also see evidence of this in our model above 80 MHz, although the component widths begin to broaden below this value. Mitra & Rankin (2002) also no- ticed this broadening, and attributed it to dispersive smearing across a frequency channel or scattering from the ISM. In our observations, the dispersive smearing at 48 MHz (across a sin- gle 12 kHz channel) is 1.5o, which is enough to explain the observed broadening in the profile10. However, even disregard- ing this low frequency broadening Mitra & Rankin (2002) found that the spacing between the components between 100 MHz and 10 GHz changes too rapidly to be caused by a dipolar magnetic field. PSR B1133+16 is the pulsar which is most consistent with radius-to-frequency mapping of all the pulsars in our sample. However, in Section 6 we showed that its emission is confined to a very narrow region in the magnetosphere (<59 km) which is incompatible with the standard radius-to-frequency model. Radius-to-frequency mapping assumes that the emission at a given emission height traces the last open field line in the pulsar magnetosphere. From geometrical arguments (see for example Lorimer & Kramer 2005) it is possible to write the opening an- 10 The half power width of component 1 increases from 1.9o at 72 MHz to 3.6o at 48 MHz, and component 2 increases from 2.1o at 72 MHz to 3.4o at 48 MHz.
16. PSR B1133+16 Fig. 13. Linear polarisation profiles of PSR B0809+74 between 925 and 328 MHz (Gould & Lyne 1998; Edwards & Stappers 2003b), and LOFAR polarisation profile at 136 MHz (black lines) plotted along with the Stokes I profiles at each frequency (grey lines). The polarised com- ponent moves from the position of the leading component of the profile towards later pulse longitudes, tracing the broad component of the pulse profile. The width of the conal components as a function of fre- quency has been a subject of interest in the past. Mitra & Rankin (2002) found that the component widths remain constant be- tween 40 and 3000 MHz. We also see evidence of this in our model above 80 MHz, although the component widths begin to broaden below this value. Mitra & Rankin (2002) also no- ticed this broadening, and attributed it to dispersive smearing across a frequency channel or scattering from the ISM. In our observations, the dispersive smearing at 48 MHz (across a sin- gle 12 kHz channel) is 1.5o, which is enough to explain the observed broadening in the profile10. However, even disregard- ing this low frequency broadening Mitra & Rankin (2002) found that the spacing between the components between 100 MHz and 10 GHz changes too rapidly to be caused by a dipolar magnetic field. PSR B1133+16 is the pulsar which is most consistent with radius-to-frequency mapping of all the pulsars in our sample. However, in Section 6 we showed that its emission is confined to a very narrow region in the magnetosphere (<59 km) which is incompatible with the standard radius-to-frequency model. Radius-to-frequency mapping assumes that the emission at a given emission height traces the last open field line in the pulsar magnetosphere. From geometrical arguments (see for example Lorimer & Kramer 2005) it is possible to write the opening an- 10 The half power width of component 1 increases from 1.9o at 72 MHz to 3.6o at 48 MHz, and component 2 increases from 2.1o at 72 MHz to 3.4o at 48 MHz. Starts narrow + gets broader
17. PSR B1133+16 Fig. 13. Linear polarisation profiles of PSR B0809+74 between 925 and 328 MHz (Gould & Lyne 1998; Edwards & Stappers 2003b), and LOFAR polarisation profile at 136 MHz (black lines) plotted along with the Stokes I profiles at each frequency (grey lines). The polarised com- ponent moves from the position of the leading component of the profile towards later pulse longitudes, tracing the broad component of the pulse profile. The width of the conal components as a function of fre- quency has been a subject of interest in the past. Mitra & Rankin (2002) found that the component widths remain constant be- tween 40 and 3000 MHz. We also see evidence of this in our model above 80 MHz, although the component widths begin to broaden below this value. Mitra & Rankin (2002) also no- ticed this broadening, and attributed it to dispersive smearing across a frequency channel or scattering from the ISM. In our observations, the dispersive smearing at 48 MHz (across a sin- gle 12 kHz channel) is 1.5o, which is enough to explain the observed broadening in the profile10. However, even disregard- ing this low frequency broadening Mitra & Rankin (2002) found that the spacing between the components between 100 MHz and 10 GHz changes too rapidly to be caused by a dipolar magnetic field. PSR B1133+16 is the pulsar which is most consistent with radius-to-frequency mapping of all the pulsars in our sample. However, in Section 6 we showed that its emission is confined to a very narrow region in the magnetosphere (<59 km) which is incompatible with the standard radius-to-frequency model. Radius-to-frequency mapping assumes that the emission at a given emission height traces the last open field line in the pulsar magnetosphere. From geometrical arguments (see for example Lorimer & Kramer 2005) it is possible to write the opening an- 10 The half power width of component 1 increases from 1.9o at 72 MHz to 3.6o at 48 MHz, and component 2 increases from 2.1o at 72 MHz to 3.4o at 48 MHz. Starts narrow + gets broader Smooth, roughly symmetrical evolution
18. PSR B1133+16 Fig. 13. Linear polarisation profiles of PSR B0809+74 between 925 and 328 MHz (Gould & Lyne 1998; Edwards & Stappers 2003b), and LOFAR polarisation profile at 136 MHz (black lines) plotted along with the Stokes I profiles at each frequency (grey lines). The polarised com- ponent moves from the position of the leading component of the profile towards later pulse longitudes, tracing the broad component of the pulse profile. The width of the conal components as a function of fre- quency has been a subject of interest in the past. Mitra & Rankin (2002) found that the component widths remain constant be- tween 40 and 3000 MHz. We also see evidence of this in our model above 80 MHz, although the component widths begin to broaden below this value. Mitra & Rankin (2002) also no- ticed this broadening, and attributed it to dispersive smearing across a frequency channel or scattering from the ISM. In our observations, the dispersive smearing at 48 MHz (across a sin- gle 12 kHz channel) is 1.5o, which is enough to explain the observed broadening in the profile10. However, even disregard- ing this low frequency broadening Mitra & Rankin (2002) found that the spacing between the components between 100 MHz and 10 GHz changes too rapidly to be caused by a dipolar magnetic field. PSR B1133+16 is the pulsar which is most consistent with radius-to-frequency mapping of all the pulsars in our sample. However, in Section 6 we showed that its emission is confined to a very narrow region in the magnetosphere (<59 km) which is incompatible with the standard radius-to-frequency model. Radius-to-frequency mapping assumes that the emission at a given emission height traces the last open field line in the pulsar magnetosphere. From geometrical arguments (see for example Lorimer & Kramer 2005) it is possible to write the opening an- 10 The half power width of component 1 increases from 1.9o at 72 MHz to 3.6o at 48 MHz, and component 2 increases from 2.1o at 72 MHz to 3.4o at 48 MHz. Starts narrow + gets broader Smooth, roughly symmetrical evolution All pulses arrive at the same time
19. PSR B1133+16 Fig. 13. Linear polarisation profiles of PSR B0809+74 between 925 and 328 MHz (Gould & Lyne 1998; Edwards & Stappers 2003b), and LOFAR polarisation profile at 136 MHz (black lines) plotted along with the Stokes I profiles at each frequency (grey lines). The polarised com- ponent moves from the position of the leading component of the profile towards later pulse longitudes, tracing the broad component of the pulse profile. The width of the conal components as a function of fre- quency has been a subject of interest in the past. Mitra & Rankin (2002) found that the component widths remain constant be- tween 40 and 3000 MHz. We also see evidence of this in our model above 80 MHz, although the component widths begin to broaden below this value. Mitra & Rankin (2002) also no- ticed this broadening, and attributed it to dispersive smearing across a frequency channel or scattering from the ISM. In our observations, the dispersive smearing at 48 MHz (across a sin- gle 12 kHz channel) is 1.5o, which is enough to explain the observed broadening in the profile10. However, even disregard- ing this low frequency broadening Mitra & Rankin (2002) found that the spacing between the components between 100 MHz and 10 GHz changes too rapidly to be caused by a dipolar magnetic field. PSR B1133+16 is the pulsar which is most consistent with radius-to-frequency mapping of all the pulsars in our sample. However, in Section 6 we showed that its emission is confined to a very narrow region in the magnetosphere (<59 km) which is incompatible with the standard radius-to-frequency model. Radius-to-frequency mapping assumes that the emission at a given emission height traces the last open field line in the pulsar magnetosphere. From geometrical arguments (see for example Lorimer & Kramer 2005) it is possible to write the opening an- 10 The half power width of component 1 increases from 1.9o at 72 MHz to 3.6o at 48 MHz, and component 2 increases from 2.1o at 72 MHz to 3.4o at 48 MHz. Starts narrow + gets broader Smooth, roughly symmetrical evolution All pulses arrive at the same time Too much broadening?
20. PSR B0329+54 Different sides of the same cone:
21. PSR B0329+54 Different sides of the same cone:
22. PSR B0329+54 Different sides of the same cone:
23. PSR B0329+54 Different sides of the same cone: Move out at different rates?
24. PSR B0329+54 Different sides of the same cone: Move out at different rates?
25. PSR B0329+54 Different sides of the same cone: Have different spectral index? Move out at different rates?
26. PSR B0329+54 Different sides of the same cone: Have different spectral index? Move out at different rates?
27. PSR B0329+54 Different sides of the same cone: Have different spectral index? Move out at different rates? -10 -5 0 5 10 15 100 1000 Position (o) Frequency (MHz) 5 4 3 2 1
28. PSR B1919+21
29. PSR B1919+21 Starts broad + gets narrower?!
30. PSR B0809+74
31. PSR B0809+74 Starts broad
32. PSR B0809+74 Starts broad Gets narrower
33. PSR B0809+74 Starts broad Gets narrower Broadens again
34. PSR B0809+74 Starts broad Gets narrower Broadens again But this happens in a very smooth way
35. PSR B0809+74 Starts broad Gets narrower Broadens again But this happens in a very smooth way Looks like one component is moving and one is fixed
36. PSR B0809+74 Starts broad Gets narrower Broadens again But this happens in a very smooth way Looks like one component is moving and one is fixed -20 -10 0 10 20 30 40 50 60 70 10 100 1000 Component Separation (o) Frequency (MHz)
37. Results (ii) • In each case, pulses arrived at the same time • Emission is confined to a narrow range of heights in the magnetosphere • For B1133+16 this height range is < 110km • In each case (particularly B0809+74), profile evolution cannot be explained by simple RFM • What is the pulse profile? Why does it change with frequency?
38. Results (ii) • In each case, pulses arrived at the same time • Emission is confined to a narrow range of heights in the magnetosphere • For B1133+16 this height range is < 110km • In each case (particularly B0809+74), profile evolution cannot be explained by simple RFM • What is the pulse profile? Why does it change with frequency?
39. • Best theory (so far): Birefringence • Different propagation modes through the magnetosphere • Emission all comes from a narrow height range and propagation effects cause the profile to change shape (Beskin & Philippov 2011) • E.g. For B0809+74: One component is refracted, and one is not If not RFM, then what?