High Energy Detectors

Michael A. Nowak (MIT-Kavli Institute) – with a great deal of material stolen from F. Lebrun and J. Wilms – (see X-ray Astronomy Workshops, www.black-hole.eu, for original presentations with more thorough explanations) X-ray Astronomy Goals Background

I~⌫ (~x, t)

Detector

IGM, ISM, etc. Source Measurable Attributes • Position – Where is it on the Sky? • Chandra, XMM, Swift, NuSTAR, (Suzaku), ... • Energy, Part I – What range is it emitting? • INTEGRAL, Swift, NuSTAR, (Suzaku, RXTE), ... • Energy, Part II – What are the spectral features? • Chandra-HETG, XMM-RGS, (XRISM, Suzaku), ... • Time – When and how does the source vary? • MAXI, Swift, Fermi, (RXTE-ASM), ... • NICER, Astrosat, XMM, Chandra-HRC, NuSTAR, (RXTE),... • Polarization – Future: IXPE Detector Attributes

• Key Characteristics of Concern for X-ray Detectors • Field of View & Angular Resolution • Energy Bandpass & Energy Resolution • Absolute Time & Time Resolution & Duty Cycle • Effective Area (we always want more ...) • There are always tradeoffs among these abilities Current & Recent Satellites

Chandra - superb XMM - good spatial & spatial & spectral spectral resolution, resolution, limited limited bandpass, bandpass, small area large area

Swift - fast response, INTEGRAL - Wide wide field of view bandpass, large area, good spatial & wide field of view, moderate spectral moderate spatial & resolution, wide spectral resolution bandpass, small area

NICER - Superb timing, moderate NuSTAR - Wide bandpass & spectra bandpass; modest resolution, very large area, spectral and area, no spatial spatial resolution resolution

Also Fermi, Astrosat; XRISM (soon); Suzaku & RXTE (neither still operating) What We Want to Measure

• Precise determination of many photons’s: • Positions (angles where they came from) • Energies • Times of Arrival • Photons are discrete events • Poisson (i.e., “counting”) statistics will be very important N = pN What We Actually Measure • “COUNTS” (or EVENTS) which are not photons! • Detectors stop photons, convert to signals • Accumulated charge, current amplitude, ... • Charge/current is not photon energy • Energy & Positions & Times are reconstructed • Measurements are Subject to Detector Properties • “Quantum Efficiency”, Internal Background, Count Rate Limits, “Recovery Time” (e.g., deadtime), ... • High Energy Astronomers must be experimental physicists! • Understanding the detector aids data modeling Imaging Techniques

• Focusing Optics (+ Position Sensitive Detector) • Grazing incidence optics – Chandra, XMM, NuSTAR, Swift-XRT, Suzaku, (XRISM) • Collimators & Concentrators, to limit fields of view – NICER, RXTE, (STROBE-X) • Coded masks (+ Position Sensitive Detectors) • Reconstruct sky image – INTEGRAL, RXTE-ASM • Multiple-layer, position & time sensitive detectors, to reconstruct photon path – Fermi Wolter Telescopes • Soft X-rays (0.2-10 keV) • Iridium (Chandra) or Gold (Suzaku) coated optics • Hard X-rays (3-80 keV) • Multi-layer mirrors (alternating high Z/low Z coatings, to create constructive interference) • Cassegrain (Optical) Geometry: Primary

Detector Secondary

Wolter Type I Mirror XMM Mirror Geometry 1/2 1/2 ⇢ re ⇢ ✓c 60 3 ⇠ (Z/A)mu ⇡ ⇠ 1 g cm 1nm 1/2 ✓ ◆1/2 ✓ ◆ ⇢ re ⇢ ✓ 60 c ⇠ (Z/A)m ⇡ ⇠ 1 g cm 3 1nm ✓ u ◆ ✓ ◆ • Low Angle for Full Reflection • Small Effective Area • Nested Mirrors • Reflectivity exhibits atomic edge features of mirror materials • Point Spread Function has strong dependence on angle & energy

Gold Reflectivity vs. Energy & Angle

Your Spectrum will Exhibit Features Related to the Materials that Make Up the Mirrors! Chandra Point Spread Function Chandra Encircled Energy Fraction

Your Spectrum will Change, Depending Upon Where on the Detector it is, and How Wide a Region You Extract! Future - Silicon Pore Optics: Athena

https://www.cosmos.esa.int/web/athena/silicon-pore-optics Collimators

• Simple, but effective: limit the field of view • Generally used for point source studies • Contamination possible from nearby sources • “Images” possible via “pan & scan” (RXTE Galactic Bulge Survey)

Rossi Timing Explorer - Proportional Counter Array

Field of View ~ Width/Length RXTE-PCA Galactic Center

PI: Craig Markwardt (GSFC) Collimator Response

1

0 -1o 1o Offset Pointings Multiple Pointings

Or Model the Two Sources Together ... Collimator Response

1

0 -1o 1o Offset Pointings Multiple Pointings

Or Model the Two Sources Together ... Micro-Channel Plate: HRC & STROBE-X

• Chandra-HRC combines mirror & MCP for pixels • STROBE-X (Spectroscopic Time-Resolving Observatory for Broadband Energy X-rays) — Allows for large effective area with low background

(Concept for Large Observatory for Timing: LOFT) X-ray Concentrator: NICER

• 56 Foil Mirrors for 56 Individual Detectors • Kind of in between Mirror & Collimator • No Attempt to Image Coded Masks

• Hard X-rays difficult to focus (cf. NuSTAR) • Coded mask allows positioning in a field • ~25-50% of mask is open • INTEGRAL (and Swift Hard X-ray, RXTE All Sky Monitor)

(Bradt, Fig. 5.3) • Correlate Mask & Detector Images • Compare “response” of pixel x,y R(x, y)=C(x, y) C h i to that expected if there were a source at a given position using a cross-correlation:

CCF(↵, )= dxdy R(x, y)R(↵, ; x, y) Z • If a peak is found (i.e., a good match), subtract, and repeat. “Iterative Removal of Sources” • Statistics depend upon behavior of whole field of view!

ISDC/Geneva INTEGRAL: Reconstructed 20-40 keV Image Detection of X-rays

• Ultimately, photons are detected by being stopped in matter and producing a signal. E.g., photolectron(s) that are amplified and read out. • Photoelectric absorption (bound-free transitions) cross sections scale as: Z4...5/E3 • Prevalence of high-Z materials used in detectors • Enhancement of cross sections near K, L, M edges, so detected counts often reflect such features • How does one maximize the signal associated with these photoelectrons? Proportional Counters: RXTE & Astrosat

Ionization Chamber • Photons ionize the gas, producing n (proportional to photon energy!) electric

charges, which in turn nq produce a voltage pulse, nq/C E = U/D • C ~ 20 pF, nq ~ 2x105 e-, voltage spike ~ 1.6 mV • Too weak to be a useful X-ray detector Proportional Counters

• Apply a high voltage to accelerate the charges, so they themselves become ionizing. E(r) = V/(r ln(b/a)) • At low “gain”, signal remains proportional to initial photon energy. • Voltage spike: nq/C * A, with A = Amplification Factor ~ 104—106 Proportional Counters

• Inert gases typically used. Low excitation of gas atoms, and low voltage required for gain • Xenon (Z=54) used in RXTE-Proportional Counter Array • Xenon ionization =21.9 eV, therefore n=1000 electrons for ~20 keV photon • Spectral resolution will obey Poisson statistics E pN pE / / • Resolution will depend upon initial value of N (and correlations in the statistics). Can achieve ~15% at 6 keV RXTE-PCA “Response Function”

Mapping from Input Energy to 100 35 Detected Energy

30 10

25 1

20 0.1

15

0.01 Output Energy [keV] 10

3 10− 5

4 10− 5 10 15 20 25 Input Energy [keV]

(Figure courtesy of John Davis) RXTE-PCA “Response Function”

Mapping from Input Energy to 100 35 Detected Energy

30 10

25 1

20 0.1

15

0.01 Output Energy [keV] 10

3 10− 5

4 10− 5 10 15 20 25 Input Energy [keV]

(Figure courtesy of John Davis) Proportional Counters

• Detector has a finite recovery time. During the time it takes to read the voltage spike from one event, a second event cannot be detected. Deadtime. • For RXTE-PCA, this was approximately 10 microseconds • This deadtime can take two forms: events during the deadtime do or do not extend the deadtime (paralyzable and non-paralyzable deadtime) • Consideration for timing of bright sources, especially on very short variability time scales. Silicon Drift Detectors • Used in NICER, proposed for STROBE-X • Allows very high count rates, moderate spectral resolutions

https://www.exvil.lt/wp-content/uploads/2012/04/SDD_Explained.pdf CCD Detectors • Semi-conductor: absorption of a photon moves electrons across the energy gap between valence and conduction bands • Number of electrons liberated go as: ⌫ N ⇠ Egap • Egap ~ 1 eV, so N ~ 1000 at 1 keV. Increased resolution compared to proportional counters! CCD Detectors • “Gate” structure on surface of CCD has pixels defined

Depletion Region by physical structure and electrodes with applied alternating currents • Applied voltage moves charge to gate, then applied voltage moves charge across pixels to readout

(See C. Peterson, http://legacy.jyi.org/volumes/volume3/issue1/features/peterson.html) CCD Detectors CCD Detectors

• Charge is typically deposited in more than one pixel. • Chandra/Suzaku: 3x3 pixel regions => Grades. Bad grades more likely to be background (cosmic rays) • XMM: Singles, Doubles CCD Detectors

• Photons can interact with the detector material! • If E > Silicon K-edge energy (1.84 keV), K-shell (n=1) electron can be liberated and reabsorbed • L-> K fluorescent line (1.74 keV) • Free-bound & higher n-shell cascade transitions, with summed energy = E - Efluorescent = Escape Peak • Similar physics occurs at Xenon L-edge in RXTE-PCA CCD Detectors

RMF @ 6 keV, 2.0664 A, sum=0.99983, moment=5.95506

Main Peak 1

0.1 Escape Peak ) E , h (

R 0.01 Flourescent Peak

3 10−

4 10− 2 3 4 5 6 Energy [keV] RXTE-PCA “Response Function”

Mapping from Input Energy to 100 35 Detected Energy

30 10

25 1

20 0.1

15

0.01 Output Energy [keV] 10

3 10− 5

4 10− 5 10 15 20 25 Input Energy [keV]

(Figure courtesy of John Davis) RXTE-PCA “Response Function”

Mapping from Input Energy to 100 35 Detected Energy

30 10

25 1

20 0.1

15

0.01 Output Energy [keV] 10

3 10− 5

4 10− 5 10 15 20 25 Input Energy [keV]

(Figure courtesy of John Davis) CCD Detectors • Gate structure can be placed either facing the X-ray source (“front side illuminated”), or away from the source (“backside illumnated”) • Backside illuminated is more sensitive to low energy photons, but has higher noise • It is possible to damage the “gate structure”. This leads to “charge transfer inefficiency” (CTI). Charge is left behind as it is moved, leading to decreased spectral resolution in the detector • Chandra & Suzaku both have CTI damage • Pixels can also go bad: either dead, or “flaring” CCD Detectors

• Number of readouts could be: one (a row at a time), one per column, or one per pixel. Slow (Chandra, Suzaku), or Fast (XMM). • More readouts = greater power requirement, greater complexity (translates to greater cost) • Fewer readouts = longer time to read chip image. Chandra: 3.2 sec, Suzaku: 8 sec, XMM: 0.074 sec • Readout faster by sacrificing exposure efficiency, image area, or spatial dimension. Chandra: 2.85 msec, Suzaku: 1/8 sec, XMM: 0.03 msec CCD Detectors • One or more photons landing in the same region in the same frame = pileup. • Piled photons are either read as a “bad grade”, or as an event with the summed energy! • Pileup can be avoided by limiting the exposure: • Filters, i.e., inserting gratings • Reading only a fraction of the CCD imaging area to reduce readout times • Sacrificing efficiency (short exposures, with long readouts: efficiency ~ exposure/readout time) • “continuous readout” modes which sacrifice one spatial dimension of information XMM Readout Modes

Full Frame, Large Window, 376x384, 198x384, 73 msec 48 msec

Small Window, Timing, 63x64, 64x200, 6 msec 0.03 msec

Also Burst Mode: 64x180, 7 usec, but 3% Efficiency NuSTAR - CdZnTe Detector

• Different detector material for higher energy photons • Active readout per pixel for very high rates (have looked at the Sun!)

https://www.nustar.caltech.edu/page/instrumentation Gratings

Reflection Gratings

Transmission Gratings

Gratings Equation: hc m = m = p sin p E ⇡ Chandra-HETG

HEG

MEG

hc m = m = p sin p E ⇡ LETG

HETG z

Greater Distance = Higher Resolution Resolution Limited by CCDs & Gratings Accuracy hc m = m = p sin p E ⇡ LETG

HETG HEG

MEG

z MEG

HEG Greater Distance = Higher Resolution Resolution Limited by CCDs & Gratings Accuracy hc m = m = p sin p E ⇡ LETG

, , ,... 2 3 HETG E, 2E, 3E,... HEG

MEG

z MEG

HEG Greater Distance = Higher Resolution Resolution Limited by CCDs & Gratings Accuracy hc m = m = p sin p E ⇡ LETG

HETG HEG

MEG

z MEG

HEG Greater Distance = Higher Resolution Resolution Limited by CCDs & Gratings Accuracy hc m = m = p sin p E ⇡ LETG

HETG HEG

MEG MEG

HEG = z 2 MEG

HEG Greater Distance = Higher Resolution Resolution Limited by CCDs & Gratings Accuracy Incoming Calorimeter Photon dT C = P G(T T )+E (t) dt b

C ⌧ = 0 G E C Temperature

Time E P T (t)= exp( t/⌧ )+ + T C 0 G b ✓ ◆

(See http://faculty.wcas.northwestern.edu/enectali-figueroa-feliciano/research/detectors.html Incoming Calorimeter Photon dT C = P G(T T )+E (t) dt b

C ⌧ = 0 G E C Temperature

Time E P T (t)= exp( t/⌧ )+ + T C 0 G b ✓ ◆

Transition Edge Sensor Resistance

Temperature (See http://faculty.wcas.northwestern.edu/enectali-figueroa-feliciano/research/detectors.html Environment & Background • Detector performance is affected by spacecraft environment • What was the solar activity? • What is the orientation of the detectors relative to the earth? • Particle contamination (sun or earth) - often mitigated with shielding or anti-coincidence detectors (or both!) • What was the spacecraft temperature? • Where has it recently been in the spacecraft orbit? • South Atlantic Anomaly (SAA) 1.5 6 200 650 1500 3000

RXTE became “radioactive”, with 24 minute & 240 minute decay time components, each time it passed through the SAA Detector Aging

• Chandra & Suzaku - CTI Damage increases with time • Contaminants build up on detectors & filters. I.e., outgasing from spacecraft components settles on cold components • NuSTAR - Thermal shielding degraded with time, detector response temperature-dependent. • RXTE-PCA over the course of it’s lifetime changed its voltage settings, to avoid breakdowns of the Proportional Counter Array Summary

• This is only scratching the surface of the physics of these detectors • How the detectors operate determine the fidelity with which X-ray photon positions, energies, and times can be measured • The properties of these detectors are “imprinted” on the information that we record from X-ray sources. Learning how the detectors work will help you understand your data better.