Observations of X-ray bursts Symposium on Neutron Stars, Ohio U., 2016 May Duncan Galloway Monash University
Thermonuclear X-ray bursts Occur in neutron stars accreting from lowmass binary companions; ~100 bursters known, ~10 4 bursts observed since early 1970s Understood since the `80s as resulting from unstable ignition of accreted H/He on the NS surface (e.g. Fujimoto et al. 1981, ApJ 247, 267) 2
Past & present missions BeppoSAX, wide-field Dutch-Italian mission through 90s RXTE high sensitivity & fast timing, 1995 Dec 2012 Jan INTEGRAL/JEMX; wide-field, low sensitivity, 2002 onwards Data from these last three make up the Multi-INstrument Burst ARchive (MINBAR), under assembly at Monash Swift; wide-field, rapid response to transients, new bursts etc. NUSTAR, hard X-ray sensitivity ASTROSAT, launched Sep 2015, LAXPC large-area detector + imagers 3
A new data analysis paradigm The traditional approach is to subtract the pre-burst emission, and fit with a blackbody e.g. Kuulkers et al. 2002, A&A 382, 947 Cooling may not be detectable if the ratio of the persistent to burst flux is high Linares et al. 2011, ApJL 733, L17 Analysis of all RXTE bursts suggests that the persistent emission increases 1 10 temporarily during bursts -7 Worpel et al. 2013 ApJ 772, #94 & 2015 ApJ 801, #60 Corroborated by a burst observed simultaneously with Chandra; in t Zand et al. 2013, A&A 553, A83 Poynting-Robertson drag on the disk by the burst 0 4 Bolometric Flux (erg cm -2 s -1 ) f a 8 10-8 6 10-8 4 10-8 2 10-8 20 15 10 5 Variable f a Constant f a Typical error bars 4U 1636-536 2001 Jun 15, 3:14.07 Typical error bars
Non-Planckian spectra during bursts A study of a very large (>60,000) sample of highsensitivity burst spectra indicate that they are not (en masse) consistent with blackbodies Deviations are not as predicted by atmosphere models e.g. Suleimanov et al. 2011, 2012 Including variable persistent flux doesn t help ~2000 bursts observed with RXTE 5
Burst profiles and the burst fuel Different timescales for hot-cno cycle and triple-α burning result in a diversity of burst profiles for different fuel mixes slow bursts; mixed H/He fast bursts; (almost) pure He 6
End-point of rp-process burning Dominates late-time burning in mixed H/He bursts; terminates in a Sn-Sb-Te cycle Schatz et al. (2001) As(33) Ge(32) Ga(31) Zn(30) Cu(29) Ni(28) Co(27) X-ray burst Tc(43) Mo(42) Nb(41) Zr(40) Y (39) Sr(38) Rb(37) Kr(36) Br(35) Se(34) 33343536 2223242526272829303132 Sb(51) Sn(50) In(49) Cd(48) Ag(47) Pd(46) Rh(45) Ru(44) 424344 41 37383940 45464748 Xe(54) I(53) Te(52) 53 5152 4950 Ga(31) Zn(30) Cu(29) Ni(28) Co(27) As(33) Ge(32) 56 5758 5455 Rb(37) Kr(36) Br(35) Se(34) 59 Tc(43) Mo(42) Nb(41) Zr(40) Y (39) Sr(38) 33343536 2223242526272829303132 Sb(51) Sn(50) In(49) Cd(48) Ag(47) Pd(46) Rh(45) Ru(44) 424344 41 37383940 45464748 Xe(54) I(53) Te(52) 5455 53 5152 4950 56 Steady state FIG. 1. The time integrated reaction flow above Ga during an x-ray burst and for steady-state burning. Shown are reaction flows of more than 10% (solid line) and of 1% 10% (dashed line) of the reaction flow through the 3a reaction. 5758 59 7
Label Cases of thermonuclear burning Accretion rate H-burning He-burning Notes I > 0.25 stable stable no bursts II 0.15 0.25 stable? overstable not observed? III 0.04 0.15 stable unstable mixed H/He bursts + short recurrence times III*? none present unstable fast pure-he bursts in ultracompacts IV 0.004 0.04 stable unstable pure He bursts V < 0.004 unstable unstable mixed H/He triggered by H; not observed Accretion rate given as a fraction of the Eddington rate, 1.75 10-8 M yr -1 From global linear stability analysis of Narayan & Heyl (2003) 8
Diversity of X-ray burst behaviour Intermediate duration bursts in lowaccretion rate systems, burning of large pure-he reservoirs e.g. Falanga et al. 2009, A&A 496, 333 Short recurrence time bursts occur too promptly to reach critical temperature, density Keek et al. 2010, ApJ 718, 292 Multi-peaked bursts perhaps attributable to nuclear waiting points? e.g. Fisker et al. 2004, ApJ 608, L61 Photospheric radius-expansion bursts reach the (local) Eddington limit; utility as standard candle e.g. Kuulkers et al. 2003, A&A 399, 663 Superbursts with durations of hours, likely powered by carbon Cornelisse et al. 2000, A&A 357, L21 9
Outstanding questions What causes burst oscillations? See Chakraborty & Mahmoodifar talks on Fri Can we use bursts to unambiguously measure neutron star mass and radius? See talks by Guillot, Heinke, Nattila, Steiner & Miller session 5, Wed What ignites in superbursts? See Keek talk on Fri Can we use bursts to constrain (or measure) nuclear reactions? See Deibel talk, next Why do many bursts short, intermediateduration, and super seem to ignite at columns well below theoretical expectations? What causes the decrease in burst rate, observed for most sources at accretion rates above ~5% Eddington? 10
Sub-critical ignition thresholds? A wide range of bursts seem to ignite at columns below what is predicted theoretically Column can be inferred from the burst energetics, e.g. in 4U 1728-34 Alternatively, we can fit the burst decay curve, in longer bursts e.g. Cumming et al. 2006 Extra heating must be present? Sure, but instead we get extra cooling Schatz et al. 2014 More shallow heating? Misanovic et al. 2010 mn depth (calculated by multiplying the local accret 11
Global burst behaviour Analysis of 6 years of BeppoSAX data; Cornelisse et al. 2003 Burst ignition models predict increasing burst rate with increasing accretion rate, up to the stable burning threshold Observations instead show a peak burst rate achieved at much lower accretion rate, and then decreasing burst rate at higher accretion rates 12
Analysis of the MINBAR sample Burst behaviour varies from source to source and broadly does not match models 1+z bolometric correction Preliminary MINBAR data (v0.6) +KEPLER models [Lampe et al. 2016] KEPLER models Inefficient, weak short bursts; an additional burst regime not predicted by theory? 13
An exception: Terzan 5 X2 The other source that shows bursts becoming more frequent up to high accretion rates is MXB 1730-335 perhaps also a slow rotoator? Bagnoli et al. (2013) MNRAS 431, 1947 A new transient outburst of a previously unknown globular cluster LMXB Atel #2919 11 Hz pulsations (<< typical freq); 21-hr orbit Atel #2919 Bursts occurred more frequently as the luminosity approached Eddington -> quasi-stable burning First time this transition has been observed, although details differ from models e.g. Chakraborty et al. 2012, MNRAS 422, 2351; Motta et al., 2011, MNRAS 414, 1508; Linares et al., 2012, ApJ 748, 82L; Cavecchi et al., 2011, ApJ 740, 8; etc. etc. 14
Theory-compliant bursts Inferred posterior distributions for model-observation comparison; p1 is the inverse redshift (i.e. 1/1+z) p2 is the relative scaling (prop. To distance) Model-observation comparison for regular, consistent bursts from GS 1826-24, the clocked or textbook burster; Heger et al. 2007, ApJ 671, 141L A few systems show trains of regular, consistent bursts amenable to comparisons with models (e.g. KEPLER) An ongoing JINA-CEE + ISSI project seeks to improve the comparisons that are possible via sharing of software tools, and observed & predicted bursts 15
New databases of model results We have assembled, published, a large sample of KEPLER simulations for comparison with observations Lampe et al. 2016, ApJ 819, 46 16
Calibrating burst models Another important activity is quantifying the uncertainty in our predictions of burst lightcurves (beyond that introduced from nuclear physics) To this end we are working on a set of test cases for numerical models Numericists with different codes will be encouraged to test their codes against our best estimates for the system parameters (accretion rate etc.) of these objects Can compare code results directly against each other Will be out soon (preliminary results available now) 17
Summary and future prospects There remain some fundamental shortcomings in our understanding of the various burst phenomena At the same time we have access to a substantial accumulated dataset to analyse, as well as detailed models Prospects for future model-observation comparisons are excellent, and incorporating nuclear physics may allow us to constrain reaction rates etc. Longer term we have the prospect of exciting new data coming in from ASTROSAT and NICER 18