glitches in radio pulsars - uni-bonn.detauris/ns2012-2/espinoza_glitches.pdf · glitches in radio...
TRANSCRIPT
Glitches in radio pulsarsCristóbal EspinozaJodrell Bank Centre for Astrophysics
42000 44000 46000 48000 50000 52000 54000MJD (Days)
6.3945
6.39452
6.39454
6.39456
6.39458
6.3946
6.39462
6.39464
6.39466Fr
eque
ncy
(Hz)
?
Monday, 22 October 2012
Glitch 0
10
20
30
!!
("H
z)
-1500 -1000 -500 0 500 1000 1500Days from MJD=53737
-7380
-7360
-7340
-7320
-7300
-7280
-7260
! (1
0-15 H
z s-1
)•
∆ν
∆ν̇
10−3 ≤ ∆ν ≤ 100µHz
Monday, 22 October 2012
What can produce a glitch?Caused by crust rearrangements; cooling, slowdown re-shaping.
Caused by rapid angular momentum exchange between the inner superfluid and the Crust; result of halted vortex migration.
Caused by magnetic field stresses on the crust, driven by vortex migration.
(Anderson & Itoh 1975; Alpar et al. 1984; many many others)
(Baym et al. 1969)
(Ruderman et al. 1998)
INTERNAL OR CRUST PROCESS
M. Ruderman (2009)
Monday, 22 October 2012
0.001 0.01 0.1 1 10Period (s)
-21
-20
-19
-18
-17
-16
-15
-14
-13
-12
-11
-10
log(
Perio
d de
rivat
ive)
Glitch detectedMagnetar
100 TG
1 TG
0.01 TG
100 MG
1 Kyr
100 Kyr
10 Myr
1 Gyr
GlitchDetections
315 glitches in plot
378 detected so far http://www.jb.man.ac.uk/pulsar/glitches.html
~130 pulsars
Monday, 22 October 2012
Glitch spin-up rate(integrated glitch activity)[Lyne, Shemar & Graham-Smith (2000)]
ν̇glitch =�
i
�j ∆νij�
k Tk
Separate pulsars according to their spindown rate and estimate their spin-up rate:
j counts the glitches on pulsar iTk is the total time for which pulsar k has been observed (and sums over ALL pulsars in the group)
|ν̇|
622 pulsars observed for more than 3 yr
Monday, 22 October 2012
-2 0 2 4
0.0001
0.001
0.01
0.1
1
log |!| (10-15 Hz s-1)•
! N
g" (
yr -1
)
•
-2 0 2 4 60.001
0.01
0.1
1
0.01 0.1 1 10 100 1000 10000 100000 1x106
log |!| (10-15 Hz s-1)•
%
of
! r
eve
rsed
by
glit
ch
es
•
(b)-2 0 2 4 6
1x10-6
0.00001
0.0001
0.001
0.01
0.1
1
10
100
1000
0.01 0.1 1 10 100 1000 10000 100000 1x106|!| (10-15 Hz s-1)•
Glit
ch
sp
in-u
p
rate
(1
0-15 H
z s-1
)
(a)
Glitch activity grows linearly with spindown rate
Glitching rateEspinoza, Lyne, Stappers & Kramer (2011)
Monday, 22 October 2012
old pulsars
young pulsars
pulsar evolution?
ν̇glitch ∝
|ν̇|
ν̇glitch ∝ |ν̇|
3 regimes
Monday, 22 October 2012
-4 -3 -2 -1 0 1 2
0
10
20
30
40
50
60
Num
ber
0.0001 0.001 0.01 0.1 1 10 100
315 glitches
315 glitches (smaller bin)
Magnetars
log(!!) ["Hz]
!! ["Hz]
Bi-modal: two types of glitches?
Large glitches: seen mostly in Vela-like pulsars
Distribution of Frequency steps
(Espinoza et al. 2011)
Monday, 22 October 2012
Giant glitchersselected for showing large frequency and spindown rate steps
0.001 0.01 0.1 1 10 100 1000 10000
0.0001
0.001
0.01
0.1
1
10
100
Positive dF1 jump
Negative dF1 jump
!! ("
Hz)
|!!| (10-15 Hz s-1)·
old pulsars
young pulsars
ν̇glitch ∝
|ν̇|
Monday, 22 October 2012
Giant glitchersNarrow size distributions.
Quasi periodic behaviour. Waiting times consistent with reaching critical lag.
Permanent spindown
-1000 -500 0 500 1000Days from Glitch
-894
0-8
900
! (1
0-15 H
z s-1
)
•-1000 -500 0 500 1000
Days from Glitch
-295
00-2
9200
02
46
8
J2021+365154177
!!
("Hz
)
-28.
7-2
8.5
! (1
0-15 H
z s-1
)
•
-1500-1000-500 0 50010001500
00.
005
0.01
J1847-013054784.449
!!
("Hz
)
-1500-1000-500 0 500 1000 1500
010
2030 J2229+6114
53064
-184
.7-1
84.4
! (1
0-15 H
z s-1
)
•
-1500-1000-500 0 50010001500
01
23
4 B1838-0453388
!!
("Hz
)
-88.
6-8
8.3
! (1
0-15 H
z s-1
)
•
-1500-1000-500 0 50010001500
00.
020.
040.
06 B1822-0952056
!!
("Hz
)
-1000 -500 0 500 1000Days from Glitch
-295
00-2
9300
-1500-1000-500 0 50010001500
05
10
J2229+611454110
-296
0-2
920
-1500-1000-500 0 50010001500
020
40
B1853+0154123
-117
7-1
176
-117
5
-1500-1000-500 0 50010001500
01
2
J1841-052453562
-89
-88.
2-1500-1000-500 0 50010001500
00.
050.
10.
15
B1822-0954114.96
-28.
45-2
8.4
-28.
35-1500-1000-500 0 50010001500
00.
10.
2 B1859+0151318
-203
.15
-203
.05-1500-1000-500 0 50010001500
00.
1
J1845-031652128
-734
0-7
260-1500-1000-500 0 50010001500
010
2030
40 B1823-1353737
-255
-250
-1500-1000-500 0 50010001500
00.
10.
20.
3 J1913+083254653.908
-203
.5-2
03.1
-1500-1000-500 0 50010001500
00.
20.
4
J1845-031654170
-805
-800
-795-1500-1000-500 0 50010001500
010
2030 J1838-0453
52162
-1500-1000-500 0 50010001500
020
40
B2334+6153642
-1000 -500 0 500 1000Days from Glitch
-800
-790
-780
0.01 0.1 1 100
0.2
0.4
0.6
0.8
1
!! ["Hz]
P(!!)
Vela
J0537-6910
B1338-62
Crab
B1737-30
J0631+1036
Modelling pulsar glitches 13
1e-13 1e-12 1e-11 1e-10 1e-09| ! |
100
1000
10000
Wai
ting
time
(day
s)
J0537-6910
J0729-1448
B2334+61
Vela
.(Hz/s)
Figure 14. We plot the approximate waiting time betweenglitches for the pulsars that have shown multiple giant glitches, asa function of the spin down rate. We also include two pulsars thathave shown only one glitch but also have a long baseline for theobservations and can thus provide us with an interesting lowerlimit on the waiting time. The data appears consistent with thenotion that giant glitches can occur once a critical lag of approxi-mately !" = 10!2 is reached. In fact the Vela like pulsars glitchevery few years, but the X-ray pulsar J0527-6910, which is spin-ning down approximately an order of magnitude faster glitchesevery few months, while the lower limits on slower pulsars indi-cate that they may glitch every decade. In fact the data appearsto be well described by a fit of the form y=A/x, as shown inthe figure, with y the waiting time in seconds, x the frequencyderivative and A = 1.082212 ! 10!3 Hz.
weak in the crust, possibly due to the fact that not all vor-tices are free, but rather that the strong pinning force givesrise to a situation in which most vortices are pinned andonly a small fraction can ’creep’ outwards. Only once themaximum unpinning lag is exceeded can the vortices moveout freely; a process which can excite Kelvin oscillations andgive rise to a strong drag and recoupling of the two compo-nents on a very short timescale, i.e. a glitch. The short termpost-glitch relaxation of the Vela, on the other hand, sug-gests that the magnitude of the drag in the core of the NS isconsistent with theoretical expectations for electron scatter-ing of magnetised vortex cores. Our model does not supportthe notion that, at least on short timescales, a significantnumber of vortices is pinned in the core (as could, for ex-ample, be the case if one has a type II superconductor andvortices cannot cross fluxtubes, e!ectively decoupling thecore and the crust). A detailed analysis of the case in whichthe core consists of a type II superconductor will be a focusof future work in order to obtain more quantitative resultsand constraints on NS interior physics. Some vortices thatcross the core may however be weakly pinned to the crust,and vortex repinning and creep (also in the core) may playa role on the longer timescales associated with the recovery.
Another e!ect which will have an impact on the post-glitch recovery is the Ekman flow at the crust-core inter-face. This e!ect has been shown to be important in fittingthe post glitch recovery of the Vela and Crab pulsars byvan Eysden & Melatos (2010) and future adaptations of our
model should relax the rigid rotation assumption for thecharged component and include the e!ect of Ekman pump-ing. Further developments should also include more realisticmodels for the drag parameters in the star, as the densitydependence of the coupling strength clearly has an impacton the amount of angular momentum that can be exchangedon di!erent timescales. Truly quantitative results could thenbe obtained with the use of realistic equations of state to-gether with consistent estimates of the pinning force, suchas those of (Grill & Pizzochero 2011) and (Grill 2011).
Note that we have assumed that a giant glitch onlyoccurs when the maximum critical lag is reached. If unpin-ning could be triggered earlier, this could generate smallerglitches. In fact cellular automaton models have shown thatthe waiting time and size distributions of pulsar glitches canbe successfully explained by vortex avalanche dynamics, re-lated to random unpinning events (Warszawski & Melatos2010; Melatos & Warszawski 2009; Warszawski & Melatos2011). It would thus be of great interest to use our long-termhydrodynamical models, with realistic pinning forces, as abackground for such cellular automaton models that modelthe short-term vortex dynamics. Such a model could thenalso be extended to model not only large pulsar glitches,but more generally pulsar timing noise, an issue that is ofgreat importance for the current e!orts to detects GWs withpulsar timing arrays (Hobbs et al. 2010).
Finally, the next generation of radio telescopes, such asLOFAR and the SKA, is likely to provide much more precisetiming data for radio pulsars and is likely to set much morestringent constraints on the glitch rise time and short termrelaxation, thus allowing us to test our models and probethe coupling between the interior superfluid and the crustof the NS with unprecedented precision.
ACKNOWLEDGMENTS
This work was supported by CompStar, a Research Net-working Programme of the European Science Foundation.
BH would like to thank Cristobal Espinoza, DanaiAntonopoulou and Fabrizio Grill for stimulating discussionson pulsar glitch observations and pinning force calculation.BH also acknowledges support from the European Union viaa Marie-Curie IEF fellowship and from the European Sci-ence Foundation (ESF) for the activity entitled ”The NewPhysics of Compact Stars” (COMPSTAR) under exchangegrant 2449.
TS acknowledges support from EU FP6 Transfer ofKnowledge project “Astrophysics of Neutron Stars” (AS-TRONS, MTKD-CT-2006-042722).
REFERENCES
Adams P.W., Cieplak M., Glaberson W.I., 1984., Phys RevB, 32, 171
Alexov A. et al., 2011, A&A 530, A80Anderson P.W., Alpar M.A., Pines D., Shaham J., 1982,Pil.Mag. A, 45, 227
Anderson P.W., Itoh N., 1975., Nature, 256, 25Andersson G., Comer G.L., C.Q.G., 2006, 23, 5505
(Haskell et al. 2012) (Espinoza et al. 2011)
rate changes
Monday, 22 October 2012
PERMANENT CHANGES ( )THE CRAB AND VELA PULSARS
CRAB VELAν̇(t) ν̇(t)
RESIDUALS RESIDUALS
∆ν̇p
~36 YR ~14 YR
ν̇
Monday, 22 October 2012
0.001 0.01 0.1 1 10 100 1000 10000
0.0001
0.001
0.01
0.1
1
10
100
!! ("
Hz)
|!!| (10-15 Hz s-1)·
1 day
7 days
0.001 ph
0.0001 ph
Small glitchesMeet detection issuesPresent in all pulsars, but preferably in low spin-down rate ones
jumps appear smaller...
Do not seem to show significant permanent steps.
Small glitches with large jumps may go undetected
∆ν̇
∆ν̇
∆ν̇
Monday, 22 October 2012
10Timing no ise
Timing noise refers to unmodelled deviations from a simple spindown model, appearing as a random wandering of the phase residuals. As pictured by Lyne+2010, most timing noise could be caused by changes, with both signs, of the spindown rate.
Increasing |∆ν̇|∆ν̇ = 0.00 ∆ν̇ = −0.01 ∆ν̇ = −0.1 ∆ν̇ = −0.3 ∆ν̇ = −1.0
×10−15
Hz s−1 Increasing |∆ν̇|×10
−15Hz s
−1
fit until glitch
overall fit
∆ν̇ > 0
∆ν̇ < 0
∆ν̇ = 0.00
∆ν̇ = −0.01 ∆ν̇ = −0.1 ∆ν̇ = −0.3∆ν̇ = −1.0
∆ν̇ = 0.00
∆ν̇ = −0.01 ∆ν̇ = −0.01∆ν̇ = −0.1 ∆ν̇ = −0.1∆ν̇ = −0.3 ∆ν̇ = −0.3∆ν̇ = −1.0 ∆ν̇ = −1.0∆ν̇ = 0.00
∆ν̇ = 0.01∆ν̇ = 0.00 ∆ν̇ = 0.1 ∆ν̇ = 0.3 ∆ν̇ = 1.0∆ν̇ = 0.00 ∆ν̇ = 0.01 ∆ν̇ = 0.3∆ν̇ = 0.1 ∆ν̇ = 1.0
Glitches and timing noise can be confused.Small glitches might remain undetected because they look like timing noise and timing noise might be confused with glitches (and reported as glitches with unusual properties).
∆ν̇ > 0 & ∆ν̇ < 0
Residuals for ∆ν = 0.003µHz (ν, ν̇ fit until glitch epoch) Residuals for ∆ν = 0.003µHz (ν, ν̇ fit entire data span)
Residuals for ∆ν = 0.0µHz (ν, ν̇ fit entire data span)Residuals for ∆ν = 0.0µHz (ν, ν̇ fit until glitch epoch)
0.001 0.01 0.1 1 10 100 1000 10000
0.0001
0.001
0.01
0.1
1
10
100
All glitches (dF1 < 0)The Crab pulsar
!! ("
Hz)
|!!| (10-15 Hz s-1)·
1 day
samplin
g
0.1 milli periods RMS
(!! < 0)·
10
Characterising glitches and timing irregularities in pulsars and magnetars
Gl i tches
C. M. Espinoza1*, D. Antonopoulou2^, B. W. Stappers1, A. Watts2
Glitches are unresolved positive steps in pulsar spin frequency, normally accompanied by a change in the spindown rate. After some glitches the frequency and the spin-down rate relax back to the pre-glitch state on timescales of up to ~100 days. These events are thought to be caused by the interaction of the neutron star crust with the internal neutron superfluid.
1Jodrell Bank Centre for Astrophysics, UK2Astronomical Institute "Anton Pannekoek"*[email protected], ^[email protected]
Timing no ise
φg = −∆νp∆t−∆ν̇p∆t2
2+ τ2d |∆ν̇d|
�e∆t/τd − 1
�
Here is what a glitch looks like
Many recent detections of small events show unusual properties, like positive spindown rate jumps [Yuan+2010, Chukwude+2010, Espinoza+2011]. We focus on these small events and study, for the first time, how they are detected and measured, with the suspicion that some of them are signatures of a different physical mechanism and should instead be classified as a different phenomena, namely timing noise. After the glitch, the
phase residuals can be described by a parabola plus a decaying term.
Timing noise refers to unmodelled deviations from a simple spindown model, appearing as a random wandering of the phase residuals. As pictured by Lyne+2010, most timing noise could be caused by changes, with both signs, of the spindown rate.
By demanding at least one datapoint before and that is larger than the typical RMS of the residuals (see the first figure) we can define glitch detection limits.
These plots show all detected glitches in three particular pulsars [JBCA glitch catalogue], and the detection limits corresponding to each pulsar, reflecting the regular observations carried out at Jodrell Bank observatory.
Yuan+2010, MNRAS, 404, 289Chukwude+2010, MNRAS, 406, 1907Espinoza+2011, MNRAS, 414, 1679
-40 -20 0 20 40 60 80 100Time since glitch (Days)
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
Phas
e re
sidu
als
in a phase residuals plot, for a model that describes
the rotation up until the glitch epoch.
Glitch
Increasing |∆ν̇|∆ν̇ = 0.00 ∆ν̇ = −0.01 ∆ν̇ = −0.1 ∆ν̇ = −0.3 ∆ν̇ = −1.0
×10−15
Hz s−1
The panels below show how the glitch signature is lost as the size of the spindown rate jump increases.
tup
φm
Increasing |∆ν̇|×10−15
Hz s−1
Is the time at which the residuals become positive.
Is the minimum value reached by the phase residuals.
tup
φm
fit until glitch
overall fit
∆ν̇ > 0
∆ν̇ < 0
∆ν̇ = 0.00
∆ν̇ = −0.01 ∆ν̇ = −0.1 ∆ν̇ = −0.3∆ν̇ = −1.0
∆ν̇ = 0.00
∆ν̇ = −0.01 ∆ν̇ = −0.01∆ν̇ = −0.1 ∆ν̇ = −0.1∆ν̇ = −0.3 ∆ν̇ = −0.3∆ν̇ = −1.0 ∆ν̇ = −1.0∆ν̇ = 0.00
∆ν̇ = 0.01∆ν̇ = 0.00 ∆ν̇ = 0.1 ∆ν̇ = 0.3 ∆ν̇ = 1.0∆ν̇ = 0.00 ∆ν̇ = 0.01 ∆ν̇ = 0.3∆ν̇ = 0.1 ∆ν̇ = 1.0
Glitches and timing noise can be confused.Small glitches might remain undetected because they look like timing noise and timing noise might be confused with glitches (and reported as glitches with unusual properties).
What can we detect/have detected ? ∆νmin ←→ ∆ν̇min
tup
φm
The plots also show all detected glitches, in more than 100 pulsars.
The distribution of glitches in the space is determined by our ability to detect them.Glitches below the lines wouldn’t be detected.
∆ν̇ — ∆ν
Lyne+2010, Science, 329, 408Glitch catalogue: http://www.jb.man.ac.uk/pulsar/glitches.html
∆ν̇ > 0 & ∆ν̇ < 0
IAU Symposium 291| Beijing, August 2012
Residuals for ∆ν = 0.003µHz (ν, ν̇ fit until glitch epoch) Residuals for ∆ν = 0.003µHz (ν, ν̇ fit entire data span)
Residuals for ∆ν = 0.0µHz (ν, ν̇ fit entire data span)Residuals for ∆ν = 0.0µHz (ν, ν̇ fit until glitch epoch)
The two segmented lines represent lower limits determined by observing cadence and the RMS of the timing residuals
∆ν > 0 & ∆ν̇ < 0
0.001 0.01 0.1 1 10 100 1000 10000
0.0001
0.001
0.01
0.1
1
10
100All glitches (dF1 < 0)J0631+1036 (dF1<0.0)J0631+1036 (dF1>0.0)
!! ("
Hz)
|!!| (10-15 Hz s-1)·
3 milli periods RMS
10 day
samplin
g
(!! < 0)(!! < 0)(!! > 0)
··
·
0.001 0.01 0.1 1 10 100 1000 10000
0.0001
0.001
0.01
0.1
1
10
100All glitches (dF1 < 0)B1737-30 (dF1<0.0)B1737-30 (dF1>0.0)
!! ("
Hz)
|!!| (10-15 Hz s-1)·
10 day
samplin
g
0.8 milli periods RMS
(!! < 0)(!! < 0)(!! > 0)
·
··
Study of the small glitches population is complicated because of contamination by other phenomena
UNDERSTANDING OF JUMPS BECOMES IMPORTANT∆ν̇
Monday, 22 October 2012
Summary: what we know about glitches
All pulsars can glitch. But different pulsars glitch different.
Pulsar with faster spindown rates glitch more often.
The cumulative effect over time is larger for pulsars with large (linearly) and it represents ~%1 of the spindown rate.
-distribution is bimodal. Some pulsars only exhibit large glitches.
∆νν̇
∆ν
Monday, 22 October 2012
Future work...or in progress
Any trend involving -jumps? Re-measure and improve error determinations.
Are all glitches followed by a permanent change?
How is the size distribution towards lower end?
ν̇
ν̇
Monday, 22 October 2012
FIN
Monday, 22 October 2012