Cole Miller University of Maryland 1 Collaborators Romain Artigue Didier Barret Sudip Bhattacharyya Stratos Boutloukos Novarah Kazmi Fred Lamb Ka Ho Lo Outline ID: 315023
Download Presentation The PPT/PDF document "Challenges in the Measurement of Neutron..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Challenges in the Measurement of Neutron Star Radii
Cole MillerUniversity of Maryland
1
Collaborators:
Romain
Artigue
, Didier
Barret
,
Sudip
Bhattacharyya,
Stratos
Boutloukos
,
Novarah
Kazmi
, Fred Lamb,
Ka
Ho LoSlide2
Outline
Radii from X-ray burstsRadii from cooling neutron stars
Radii from X-ray light curvesThe promise of gravitational waves
2
NS masses are known up to 2
M
sun
. What about radii?
Key point:
all
current NS radius estimates
are dominated by systematics.
None
are
reliable. But hope exists for the future.Slide3
Measuring stellar radii
Ordinary star, like the SunToo far for angular resolutionBut can get luminosity L
If we assume blackbody, R2=L/(4ps
T
4
)
But for NS, usually gives ~5 km!
Why? Spectral shape is ~Planck, but inefficient emissionNeed good spectral modelsBut data usually insufficient to test3Slide4
M and R from X-ray Bursts
van Paradijs
(1979) methodXRB: thermonuclear explosions on accreting NSAssume known spectrum, emission over whole surf.
Only with RXTE (1995-2011) is there enough data
http://cococubed.asu.edu/images/binaries/images/xray_burst3_web.jpg
4Slide5
4U 1820 Bursts: Soft EOS?
Fits of good spectral models to hours-long bursts
show that fraction of emitting area changes!
Guver et al. 2010; known dist (globular)
Uses most optimistic
assumption: no systematics,
only statistical uncertainties
But small errors are
misleading; only ~10
-8
of prior prob. space gives
M, R in real numbers!
(
Guver
et al., Steiner et al.)
S
pectral model is
terrible
fit to best data!
5Slide6
6
Inferred relative emitting areas,
for
102 16-s segments near
the peak
of the 1820
superburst
: Miller et al., in prepSlide7
Emission from Cooling NS
Old, transiently accreting NSDeep crustal heating (e.g., e capture)
If know average accretion rate, emission provides probe of cooling; can we use to fit radius?
Predictions of simple model:
Minimum level of emission Spectrum should be
thermal
No
variability: steady, slow decay7Slide8
Cooling NS Observations
Oops!All the predictions fail
L sometimes below minimum
Large
power law component
Significant
variability
Excuses exist, but failure of basic model means we can’t use these observations to get RAlso: is surface mainly H? He? C? Makes 10s of percent difference to R
Magnetic field can also alter spectrum
Again, wide variety of models fit data, thus can’t use data to say which model is correct
8Slide9
RXJ
1856.5–3754
Specific isolated NS
Argument: BB most efficient emitter, thus R>=R
BB
True for bolometric but
not
for given band
Example: Ho et al. condensed surface fit
Different R constraints for different models
9
Klähn
et al. 2006Slide10
RXJ
1856.5–3754
Specific isolated NS
Argument: BB most efficient emitter, thus R>=R
BB
True for bolometric but
not
for given band
Example: Ho et al. condensed surface fit
Different R constraints for different models
10
Klähn
et al. 2006Slide11
Baryonic vs. Grav. Mass
Pulsar B in the double pulsar systemMgrav
=1.249+-0.001 MsunIf this came from e capture on Mg and Ne,
M
bary
=1.366-1.375
M
sun for coreBut what about fallback?Or could mass be lost after collapse?11Slide12
Ray Tracing and Light Curves
Rapidly rotating star 300-600 Hz v
surf~0.1-0.2c
SR+GR effects
Light curve informative about M, R
Bogdanov
2012; MSP Must deal carefully with degeneraciesLo et al., arXiv:1304.2330 (synth data); no systematic that gives good fit, tight constraints, and large bias
Weinberg, Miller, and Lamb 2001
12Slide13
Phase Accumulation from GWs
aLIGO/Virgo: >=2015
Deviation from point mass in NS-NS inspiral: accumulated tidal effectsFor
a
LIGO
,
can measure tidal
param (Del Pozzo+ 2013: distinguish R~11, 13 km?)Recent analytics confirmed by numerical relativity (
Bernuzzi
et al. 2012)
High-
freq
sensitivity key
Damour
et al., arXiv:1203.4352
13
High-
freq
modeling, tooSlide14
Conclusions
Current radius estimates are all dominated by systematics
Light curve fitting shows promise:
No deviations we have tried from our models produce significant biases while fitting well and also giving apparently strong constraints.
LOFT, AXTAR, NICER
Future measurements of M and R using gravitational waves
may be competitive in their precision with X-ray based
estimates, and will have very different systematics
Open question: how can we best combine astronomical
information with laboratory measurements (e.g.,
208
Pb skin
thickness)?Slide15
Ray Tracing from MSP
S.
Bogdanov
2012
Binary millisecond pulsar J0437-4715
Two spots, H
atm
Multitemp
plus
Comptonized
spect
Qs about beaming, spectrum; intriguing results, though!
15
Bogdanov
2012Slide16
High inclinations allow tight constraints on M and R
Spot and observer inclinations = 90°, high background
16Slide17
Low inclinations produce looser constraints
Amplitude similar to the previous slide, but low spot and
observer inclinations, low background
17Slide18
Independent knowledge of the
observer
’
s
inclination can increase the precision
Observer inclination unknown
spot and observer inclinations = 90°, high background
18Slide19
Observer inclination known to be 90°
Independent knowledge of the
observer
’
s
inclination can increase the precision
spot and observer inclinations = 90°, high background
19Slide20
Incorrect modeling of the spot shape
increases the uncertainties
Actual spot elongated E-W by 45°
spot and observer inclinations = 90°, medium background
20Slide21
21
Fits Using New Models
64-second segment at peak
temperature
This model has F=0.95F
Edd
Best fit:
2
/
dof
=42.3/48
Best B-E fit:
2
/
dof
=55.6/
50
For full 102-segment data set,
best fit has
2/dof=5238/5098B-E best: 2/dof=5770/4998Fits are spectacularly good!Much better than B-E, so further info can be derived
Pure He, log g = 14.3, F=0.95F
Edd
Model from Suleimanov et al. 2010
Yes! New models from
Suleimanov
et al. 2010 do seem
to fit the data quite well. Slide22
Keplerian Constraints
Suppose we observe periodic variations in the
radial velocity of star 1, with period P
b
and
amplitude v
rad
. Then we can construct themass function
This is a lower limit to the mass of star 2, but
depends on the unknown inclination i and the
unknown mass m
1
of the observed star.
22Slide23
Post-Keplerian Parameters
With high-precision timing, can break degeneracies:
If both objects are pulsars, also get mass ratio.
Allows mass measurements, GR tests
23Slide24
Artigue
et al. 2013
24
c
2
/
dof
for all five bursts combined: 1859/1850 (44%)
c
2
/
dof
for far left burst only: 401.8/372 (14%)
Hot spot model fits very well
Analysis of bursts from 4U 1636-536; previously
claimed
to contradict rotating spot model