Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” The complex and puzzling phenomenology of thermonuclear X- ray bursts APS April 2008, St. Louis Duncan Galloway Monash University Andrew Cumming McGill Jean in ‘t Zand SRON Deepto Chakrabarty & Jake Hartman MIT Mike Muno Caltech Dimitrios Psaltis Arizona

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Motivation Thermonuclear bursts are a key observational phenomenon which uniquely allow us to derive information about the neutron stars upon which they occur Burst oscillations -> NS spin Peak flux of radius-expansion bursts -> distance Spectrum in the burst tail -> NS radius… … and hence (in principle) constrain the neutron star EOS Of course, there are also the details of the thermonuclear burning, how it spreads over the star, the balance between stability and instability…

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” What are neutron stars made of?

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst theory: 3 ignition regimes 3 cases, in order of increasing accretion rate (e.g. Fujimoto et al. 1981): 3) H-burning is unstable, ignition is from H in mixed H/He fuel; 2) H-burning stable, H is exhausted prior to unstable He-ignition, pure He burst; 1) H is not exhausted prior to He-ignition, mixed burst; Case 1 Case 2 Case 3 ignition curves stable burning accretion rate

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” In general, the behavior of bursts from individual sources do NOT match our expectations from theory

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” The exceptions are the rule Bursts tend to decrease in frequency at higher accretion rates, rather than increasing, as expected theoretically Such bursts are often much fainter than might be expected (given the wait time and inferred accretion rate), suggesting an “energy leak” A (possibly) related issue is that “normal” thermonuclear bursts don’t produce enough carbon to power “super” bursts We also see bursts with inexplicably short recurrence times, down to a few minutes, sometimes in groups of three or four

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Low-mass X-ray binaries compact objects accreting material from a low-mass stellar companion LMXBs are thought to be “old” systems with weak magnetic fields allowing the accreting material to fall directly onto the surface Approx. 100 are known within our Galaxy, with orbital periods between 10 min and 16.6 d Unstable nuclear burning of accumulated fuel on the neutron star surface leads to thermonuclear (type-I) bursts The Rossi X-ray Timing Explorer (RXTE) has observed more than 1200 bursts from 40 of these systems over it’s 10-year mission lifetime, with high time resolution and signal-to- noise

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Important scales Neutron star mass ≈ 1.4 M , & radius ≈ 10 km Distance to Galactic LMXB systems ≈ 8 kpc Typical persistent intensity erg cm -2 s -1 … corresponding to an accretion rate of ≈10% of the Eddington rate (8.8  10 4 g cm -2 s -1 or 1.3  M  yr -1 ) Typical burst peak intensity erg cm -2 s -1 Characteristic burst fluence (integrated flux) of erg Ratio  of integrated persistent flux to burst flux is  40 (i.e. accretion is much more efficient than thermonuclear burning!)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Examples of X-ray bursts from RXTE

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Superbursts: carbon burning? 1000x more energetic than typical thermonuclear bursts (10 42 erg) 1000x less frequent (recurrence times of months, instead of hours) 10 4 s Thought to arise from unstable ignition of carbon produced as a by-product of burning during “normal” thermonuclear bursts 4U

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” The modeling revolution Accretion onto the neutron star occurs preferentially at the equator, although accretion may be more even at depths relevant for thermonuclear burning (Inogamov & Sunyaev 1999, AstL 25, 269) End-point of rp-process burning in the closed Sn-Sb-Te cycle (Schatz et al. 2001, PRL 86, 3471) Role of rotation on the spread of the burning region (Spitkovsky et al. 2002, ApJ 566, 1018) Time-dependent models which correctly treat thermal & compositional intertia in H-rich bursts (Woosley et al. 2004, ApJS 151, 75) Occurrence of “delayed mixed bursts” in the transition to stable burning (Narayan & Heyl 2003, ApJ 599, 419) Prediction of mHz oscillations at the transition to stable He-burning (Heger et al. 2007, ApJ 665, 1311) Role of sedimentation in suppressing H-ignited bursts at very low accretion rates (Peng et al., astro-ph/ ) Shear-mediated mixing of the ashes and fuel (Piro & Bildsten, astro-ph/ ) … and more

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” A well-behaved burster: GS This source, discovered in the late 80s by the Ginga satellite, is unique in that it consistently exhibits highly regular bursts Lightcurves are extremely consistent, and recurrence times exhibit very little scatter within an observation epoch We infer “ideal” burst conditions: steady accretion, complete coverage of fuel, complete burning etc. -> unique opportunity to test theoretical models

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” … aka the textbook burster In RXTE observations spanning several years, the persistent X-ray flux increased by almost a factor of two The burst recurrence time decreased by a similar factor, exactly as expected for constant accreted mass at ignition A ~10% change in  over this time suggests solar fuel composition (Galloway et al. 2004, ApJ 601, 466; see also Thompson et al. 2007, ApJ accepted, arXiv: )

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Data & model comparisons We also compared lightcurves with predictions by the time-dependent model of Woosley et al Confirms that the bursts occur via ignition of H/He in fuel with approximately solar metallicity (i.e. CNO mass fraction) In addition, we obtained stunning agreement between the observed and predicted lightcurves (this is not a fit!) Except for a “bump” during the burst rise, which may be an artifact of the finite time for the burning to spread, or something arising from a particular nuclear reaction (Heger et al. 2007, ApJL 671, L141)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Simpler models are sometimes OK The time-independent models are still OK where steady H-burning is not the dominant heating process For example, at low accretion rates where the recurrence time is long enough to exhaust all the hydrogen prior to ignition For the bursts observed during the 2002 outburst of the millisecond pulsar SAX J , we obtained excellent agreement between model predictions and observations Can derive the distance of kpc (Galloway et al. 2006, ApJ 652, 559)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” What about all the other burst sources?

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” No shortage of data

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst ignition: case 3 At lowest accretion rate, unstable ignition of H in a mixed H/He environment; perhaps the least well understood case Short-recurrence time bursts (doublets and even triplets) characteristic of this bursting regime Possibly arising from ignition of unburnt or partially burnt fuel (Boirin et al. 2007, A&A 465, 559)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” A comparable sample of almost 1200 bursts has been assembled from observations by the Rossi X-ray Timing Explorer (Galloway et al. 2007, astro- ph/ ) Although RXTE has a relatively narrow field of view, the large effective area allows us to make precision measurements of burst properties and the persistent emission Long-duration X-ray satellite missions have accumulated large samples of bursts from many sources Analysis can identify global trends, as well as revealing important exceptions (e.g. Cornelisse et al. 2003, A&A 405, 1033) Studies of large burst samples H-ignition He-ignition H-poor fuel He-ignition H-rich fuel Note that these are carefully-selected subsamples!

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” X-ray colors  accretion rate? This is about where we observe mHz oscillations; transition to stable He- burning? (Heger et al. 2007, ApJ 665, 1311)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Diverse bursts at moderate M-dot The bursts in the accretion- rate range where the burst rate is decreasing are highly inhomogeneous

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst rate vs. X-ray colors short long

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Steady He-burning at 10% Eddington? Observations at 10-30% Eddington, including Infrequent, He-like bursts with  ~1000 mHz oscillations The occurrence of superbursts requiring efficent carbon production All suggest that steady He-burning occurs in this range of accretion rates HOWEVER It is far from clear how steady and unsteady (i.e. bursts) He- burning can happen simultaneously!

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” The (a) next step: MINBAR The Multi- INstrument Burst Archive (MINBAR) project

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Summary and future work Studies of thermonuclear bursts are experiencing a “renaissance” with much recent theoretical and observational activity Wide-field observations by INTEGRAL and Swift are a promising new source of observations, as well as existing samples from BeppoSAX and the growing RXTE sample Detailed comparisons with predictions by state-of-the-art models confirm our present understanding of the nuclear physics Studies of larger burst samples can help to resolve some of the discrepancies with the observational properties

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” The modeling revolution Nuclear burning during thermonuclear bursts involves thousands of isotopes and reactions; some rates are poorly known This has led to uncertainty for the burst models, in particular for rp-process burning Early studies with limited reaction networks could not identify the end-point of rp-process burning, but it is now thought to terminate in a closed Sn-Sb-Te cycle (Schatz et al. 2001, PRL 86, 3471) Modeling suggests burning may not reach this cycle unless the temperature and density become sufficiently high (e.g. Woosley et al. 2004, ApJS 151, 75)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Time dependent models Ignition by unstable He burning in a mixed H/He environment Includes a realistic adaptive reaction network Initial burst is the largest, illustrating the “inertia” characteristic of this case low Z 0 ) For 2nd and later bursts, ignition takes place in the layer of ashes produced from the previous burst (simulations from Woosley et al. ‘04)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Resulting composition Lack of convection after the start of the burst prevents most of the ashes being mixed back up into the burning layer for further burning Don’t reach the endpoint of the rp-process (cf. with Schatz et al. 2001) Very little 12 C - not enough to power a superburst

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst ignition: case 2 Lower accretion rate, so that enough time passes between bursts to exhaust H at the base Ignition of He in a pure He-layer, with freshly accreted H on top Burst is much shorter and intense, and convection plays a bigger role

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Lighter products, & more C In case 2 ignition, more carbon is left which may accumulate and subsequently power a superburst Otherwise the ashes are much lighter than in case 1 burning Radial composition gradient results in additional density- driven (“thermohaline”) convection Ashes are thus well-mixed

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Type-I bursts: nuclear H/He burning Ratio of integrated burst flux (fluence) to integrated persistent X-ray emission (arising from accretion) constrains the burst energetics - the  -value Compactness of the neutron star means that accretion liberates roughly 50% of the rest-mass energy of the accreted material Nuclear burning is much less efficient, at around 1%; expected ratio is then ~50 or more, in agreement with measurements Rapid (type-II) bursts, on the other hand, are thought to be transient accretion events, confused the issue early on, and remain rather poorly understood

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst physics in a nutshell Properties of bursts depend on the composition of the fuel - pure He- bursts are fast (3  reaction), H/He bursts are slow (H burns via the  p and rp process;  -decay limited) Burst ignition can arise via unstable H-ignition, at very low accretion rates, or unstable He-ignition, once H-burning stabilizes Accretion rate largely sets the temperature of the fuel layer, as well as the time to ignition accretion rate stability H-burning He-burning H exhausted prior to ignition He-ignition in mixed H/He no bursts H-ignition in mixed H/He

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Recent milestones Already a key observational tool - black holes don’t exhibit bursts, and bursts allow us to measure the neutron star distance Some new aspects: Burst oscillations observed only during bursts trace the neutron-star spin (e.g. Chakrabarty et al. 2003, Nature 424, 42) Observation of discrete emission/absorption lines during bursts allow measurement of the surface gravitational redshift (e.g. Cottam et al. 2004, Nature 420, 51) “Superbursts” are extremely long bursts with recurrence times and energies 1000x greater, thought to burn C rather than H/He

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Burst oscillations In bursts from 14 sources we find coherent oscillations around frequencies in the range Hz, characteristic of the source

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” LMXB neutron star spins Majority of NS don’t ever exhibit pulsations! Can detect pulsations (and hence measure the spin) by –Burst oscillations –Persistent pulsations in 7 sources –Intermittent pulsations in up to three more LMXBs are the evolutionary progenitors to rotation-powered MSPs And so cluster at high spin frequencies (in the accreting aka “recycling” phase) Despite this, the fastest-spinning NS known is a rotation powered pulsar at 716 Hz

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” A neutron star spinning at 1122 Hz? A burst from XTE J revealed evidence for oscillations at 1122 Hz (Kaaret et al. 2007; astro- ph/ ) Maximum Leahy power Significance from M-C simulations is 3.97  ; taking into account the number of trials, at most 3.5  Signal was present at high significance only in a single time bin of a single burst -> needs confirmation

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Discovery of thermonuclear bursts Instruments like the Netherlands/USA satellite ANS first observed bright X-ray flashes from around the Galactic center and elsewhere in the early ‘70s Multiple bursts from some sources Ratio of integrated burst flux (fluence) to integrated persistent X-ray emission (arising from accretion) constrains the burst energetics - the  -value Compactness of the neutron star means that accretion liberates roughly 50% of the rest-mass energy; nuclear burning is much less efficient, at around 1% Expected  -ratio is then ~50 or more, in agreement with measurements

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” … and probe the global burning physics  (i.e. the composition of the burst fuel) depends only upon the burst rate -> dominated by the steady (H) burning Possible role of steady He- burning?

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Observed ignition column Carbon Ignition curve Superbursts and the core composition C ignition is a plausible explanation for the superburst properties BUT Models don’t produce enough C to power them Furthermore, ignition occurs at too low a column (Such “premature” ignition also occurs in H/He-burning thermonuclear bursts) The crust is too hot; cooling mechanisms are inefficient

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” A possible solution: No crust! The compact object is composed of “strange quark matter”, overlaid by a layer of normal matter supplied by accretion The fuel layer can be electrostatically supported, and bursts may yet occur (Cumming et al. ApJ 646, 429; See also Page & Cumming 2005, ApJL 635, 157) Superbursts are difficult to study because of their scarcity (only a few have been observed; estimated recurrence times are ~months) H/He bursts on the other hand, also depend on the crust cooling, and are more frequent -> an alternative test for strange stars

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” “Late” ignition As the accretion rate increases, we expect that the burst rate also increases, up to the point when He burning stabilises Instead we observe that bursts become less frequent, and exhibit profiles more consistent with He-rich bursts One explanation is that the area over which accretion is occuring is changing in a sufficiently diabolical way to give exactly the opposite trend we want… Alternatively some have proposed yet another phase of bursting, “delayed mixed bursts”, which burn much of their fuel stably prior to ignition (Narayan & Heyl 2003, ApJ 599, 419)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” Onset of stable burning Drop in burst rate well below the Eddington accretion rate is not expected From modelling the onset of steady H-burning is expected at around 0.92 of that level Around the transition models exhibit erratic switching between bursting and oscillatory behaviour “flickering burning” This may be observable as quasi-periodic X-ray oscillations (e.g. Revnitsev et al. 2001)

Galloway, “The complex and puzzling phenomenology of thermonuclear X-ray bursts” … can similarly study energetics… H-ignition He-ignition H-poor fuel He-ignition H-rich fuel