Spectral-timing the emission from close to black holes Phil Uttley - - PowerPoint PPT Presentation

Spectral-timing the emission from close to black holes

Phil Uttley

University of Amsterdam with thanks to: Abigail Stevens, Jakob van den Eijnden, Adam Ingram and Julien Malzac

Accretion rate Corona/disk luminosity

Spectral states: evolution of the central engine in BHXRBs

Hard State Soft State Jets Winds

We can understand the evolving structure broadly in terms of model-able spectral components

Dependence of variability on state: the rms amplitude and power spectrum

Hard Intermediate Soft Broadband noise dominates

Strong quasi- periodic

• scillations

(QPOs)

Broadband noise dominates

SPECTRAL-TIMING WITH BROADBAND NOISE

Constraining accretion and the coronal geometry

Power-law component: time-lags vs. frequency and energy

Kotov et al. 2001

• Variations of power-law emission in hard bands lag behind variations
• f power-law emission at softer energies
• Time-lags increase towards lower frequencies (longer time-scales)
• Lag vs energy dependence is approximately log-linear

Lag vs frequency Lag vs energy

Models for the broadband noise lags

• Compton scattering (light-travel delays) in

the corona/jet (e.g. Kazanas & Hua 1999, Reig

et al. 2003):

• Can explain log-linear energy dependence

(time-delay proportional to number of scatterings which scales with log(E))

• Cannot explain the large size of the lags:

requires X-ray emitting region which is much too big (>103 Rg).

• Propagation through an accretion flow with

radially-dependent temperature profile (Kotov

et al. 2001, Arévalo & Uttley 2006):

• Explains large lags and frequency

dependence (longer time-scale variations travel from further out).

• Cannot easily explain log-linear energy

dependence (must hard-wire into model)

• Requires a hot accretion flow: truncated disk

with large inner radius???

þ ý þ ☐ ☐

Disk photons show very different lag behaviour! power spectrum lag vs frequency

GX 339-4 energy spectrum: disk dominates below 1 keV

Probing the disk lags with XMM-Newton

GX 339-4 hard state: the causal connection between disk and power-law variability

The disk varies substantially before the power-law: no immediate upscattering of disk photons by the corona. Therefore the corona must be central and compact (also consistent with reverberation measurements and microlensing of X-rays in AGN). Uttley et

• al. 2011

But we still need to explain the lags between different power-law energies!

Modelling power-law variability driven by mdot fluctuations propagating through the disk

Time Intensity

Impulse response:

Observer sees slow disk photon response, rising before the power-law responds Corona sees a later, more rapid rise in seed photons from the disk, causing cooling of corona Finally, the corona is heated by the mdot fluctuation reaching it Power-law softens then hardens: hard lags!

Γ ∝ ✓lseed lheat ◆1/6

(Beloborodov 2000)

Large hard lags from a compact (few Rg) corona

GX 339-4 hard state, De Marco et al. 2015 2-9 vs. 0.5-1.5 keV 10-30 vs 2-9 keV Lag of PL vs. dissipated (disk) flux Lag of 4, 16 and 64 keV photons wrt 1 keV (Uttley & Malzac in prep)

lag ∝ log(E2/E1)

We can now explain the power-law lags in terms of a varying disk, and log-linear energy dependence! We still need to explain the lag ‘steps’: multiple coronal components?

SPECTRAL-TIMING WITH QPOS

Revealing GR effects with Doppler tomography of the accretion flow

Low-frequency QPOs

• Low-frequency QPOs (~1-10’s Hz) are seen in the X-ray

light curves of X-ray binaries

• General relativistic precession of the central corona/hot

flow?(Stella & Vietri 1998; Ingram, Done & Fragile 2009)

• Use LF QPOs to probe strong-field general relativity --

How does matter behave in strong gravitational fields?

• Basic prediction: should see quasiperiodic heating of the

inner disk by the precessing flow (“lighthouse” effect)

Origin of the technique (or why it’s good to think about new missions…):

E (keV) → 0 5 10 15 Time (ms) →

The method we use to study low-frequency QPOs was developed for the LOFT M3 mission proposal, to carry out Doppler tomography of high-frequency QPOs, associated with an orbiting hotspot in the inner disk.

Now we apply the method for the first time to real (RXTE) data!

GX 339-4: Type B vs. Type C QPOs

Here we use Type B LF QPOs from GX 339-4, to give a stronger, cleaner signal

No counts in this channel

8.15 ms )

(Quasi-)phase-resolved spectroscopy with the 2-D (energy-time) cross-correlation function

Stevens & Uttley, MNRAS submitted

QPO-Phase-Resolved Energy Spectra

10 5 20 0.5 keV2 (Photons cm−2 s−1 keV−1) Energy (keV)

Deviations from mean spectrum, “fluxed”

• Spectral

shape is varying with QPO phase

Spectral fitting the phase-resolved spectra

Used variations on:

absorption × (power law ★ blackbody + iron line)

Parameters that are required by the data to vary:

• 1. PL normalization

(scattering fraction)

• 2. PL index
• 3. BB temperature

Same 4 phases, as previous, added to mean

Parameter Variations with QPO Phase

Parameter variations with relative QPO phase

• Blackbody variation is

~0.3 out of phase with power law

• Power law variation:

~25% fractional rms

• Blackbody variation:

~1.4% fractional rms

• Small BB variation

implies only small variability of illumination

• f inner disk: large scale-

height corona (jet-like?)

QPO amplitude depends on binary system inclination

(Motta et al. 2015, see also Heil, Uttley & Klein-Wolt 2015)

Type B Type C High inclination: more edge-on disks Low inclination: more face-on disks

• Type C higher rms at higher inclination: disk-like power-law emitting region
• Type B higher rms at lower inclination: jet-like power-law emitting region

Precession Model Interpretation

a b c d

SHORT TERM QPO PHASE-LAG EVOLUTION

Revealing a possible physical origin for coherence/ decoherence of QPOs

Qu et al. (2010)

GRS 1915+105: QPO frequency depends

• n energy

2 possibilities:

• Same intrinsic frequency. The frequency and energy-dependent

amplitude causes apparent difference in frequency. Phase difference is constant.

• Different intrinsic frequencies. Phase difference increases with time.

(but it must reset?)

Optimal filtering to track QPO cycles

Coherent intervals

After filtering, it is clear that the QPO amplitudes are themselves modulated

• n a time-scale we call the ‘coherent interval’ since it corresponds also to

the coherence-time of the QPO. We can use the CCF to measure the phase lag between hard and soft bands as a function of position in this interval….

Coherent Interval

Coherent Interval Start End

Phase lag [rad] Thus different energy bands do show an intrinsically different QPO frequency!

Result 1: phase lag ‘runs away’ during a coherent interval, then resets

(van den Eijnden, Ingram & Uttley 2016)

QPO cycles since start of coherent interval

Phase lag [rad]

4 cycles 6 cycles 8 cycles 10 cycles

The faster the run-away, the faster the QPO decoheres: the ‘quasi-period’ mechanism may be intrinsically linked to the run-away effect!

Result 2: the ‘run-away speed’ of phase-lag depends on the length of the coherent interval

(van den Eijnden, Ingram & Uttley 2016)

Interpretation: differential precession

The runaway soft phase lags can be explained if the hotter, inner part

• f the flow has a higher precession frequency than the cooler, outer

part: differential precession! The outcome of differential precession is a warped or ‘twisted’ disk: QPO rise and decay linked to growth of ‘twist’ in hot flow?

Figure from Armitage & Natarajan 1999

Summary

• Spectral-timing offers a unique probe of the innermost regions of

accreting black holes, allowing us to resolve scales which are nano-arcseconds on the sky.

• We now have a ~complete, self-consistent model for the X-ray

lags seen from the broadband noise, which can be explained by mdot variations propagating through the accretion disk which illuminates a central, compact corona.

• Our technique for QPO phase-resolved spectroscopy reveals

evidence for variable heating of the disk by a more vertically extended, precessing corona.

• Short-time-scale measurements of type C QPO phase lag

evolution in GRS 1915+105 shows evidence for differential precession of the inner hot-flow/corona, which may explain the formation mechanism of visible QPOs in terms of a ‘twisted’ flow.