AQRB, April, 2006

Avalanche Diodes (APD) Detectors:

Introduction and Recent Advances*

Alfred Q.R. BaronHarima RIKEN & SPring-8/JASRI

International Symposium on the Development of Detectors for Particle, Astro-Particle, and Synchrotron Radiation Experiments

“SNIC”SLAC, Stanford, April 2006

*Focus on Fast X-Ray Detection

AQRB, April, 2006

A Change in Scale…

Compared to the detector effort in high-energy physics, theeffort in synchrotron radiation is in its childhood,

… and the APD effort in x-ray scattering is nearly a newborn.

Small efforts: ESRF, BNL, DESY, SPring-8, KEK, (SSRL)& the recent addition of a dedicated technician

at ESRF is already a major jump.

AQRB, April, 2006

Historical

Development in the 60s & 70s(Huth, McIntyre, Webb, Jones at GE & RCA)

Some discussion of a “Solid State Photomultiplier”(but gain not so high…)

Main utility is high speed (large gain-bandwidth product).Optical Communication, Laser range-finding…

& Applications where for some reason a PMT is not suitablePositron Emission Tomography (High Magnetic Field)

Introduced to the x-ray scattering community via

Nuclear Resonant Scattering (NRS) of Synchrotron Radiation

1990’s

AQRB, April, 2006

Nuclear Resonant Scattering (NRS)

PulsedX-Ray Beam

~ eV Wide

Resonant NucleiDE~ neV, t~100 ns

Detector

0 5 10 15 20

Time After Excitation (ns)

Prompt Scatteringup to 10 ordersbigger than the

Nuclear Signal

Carefully Choose the Sample (Gerdau, Rüffer)

Build the Right Optics (Faigel, Siddons, Hastings)

Use the Right Detector (Kishimoto, Baron)(1995)

Continues to provide a challenge for detectors…

Sample

AQRB, April, 2006

APD Research Applications

Fast photon counting:

Time resolved detection:

Places where a PMT is not possible…Magnetic Fields Size ConstraintsPower Requirements

1 MHz Easy, ~100 MHz Possible

~1 ns Easy~0.1 ns Possible

Note one large application (>105 devices) is for the CMS Calorimeter for scintillator readout in a 4T field (Dieters, Renker)

NRS, XIFS

Diffraction, Imaging

AQRB, April, 2006

A Diode With Gain

Drift Region102-104 V/cm

Avalanche Region~105 V/cm

eh

x-ray

Absorption10-200 um Si

Drift~ 10 ps/um

Gain 101-104

Limited Stopping Power

~50 TypicalIntrinsic (excess) noise dueto amplification of BOTH

Electrons & Holes

Depleted

10 keV X-Rays -> OK20 keV -> Losses

“McIntyre” Theory (~1970)

1 2 3

AQRB, April, 2006

Geiger vs Linear Operation

Linear Operation:Diode biased below breakdown.Well defined small-signal gain.

Diode biased above breakdown.Single electron leads to run-away

gain until quenched.Noisy: 102-105 cps/channel

Geiger Operation:

(also Poster/Abs 143, Renker)

AQRB, April, 2006

Structures

X-Ray e

h

p+p n+

pX-Ray

h

p+ pn

edepleted

depleted

X-Ray

p+

p n

p

depleted

e

hn+

(a) Reach Through (b) Beveled Edge (c) Reverse Reach Through

E E E

“Reach Through” “Reverse”“Beveled Edge”

Narrow Gain RegionMedium Voltage 50-700V

Large Drift RegionModest Gains (<200)

Wide Gain RegionHigh Voltage (1-2 kV)High Gains Possible

Larger Areas Possible

Narrow Gain RegionMedium Voltage <500V

Small Drift RegionModest Gains (<200)Also one-sided epitaxial

& diffused & back entry… Also “Planar”

AQRB, April, 2006

Some Single-Element Devices

Reliable

MostlyReliable

Not SoReliable*

Not SoReliable*

* As of ~1998Note: (1) Not extremely careful conditions.(2) Device Dependence.(3) CMS expects >99% reliability (after selection) for 10 years

AQRB, April, 2006

Electronics

ÿPhoton (Pulse) Counting & Timing

APDFast Voltage

Pre-Amp ~ GHz

Discriminator(NIM or ECL)

Logic

CounterOr

TAC/TDCOr

Autocorrelator

Not YET integrated… SM Technology & Modular (NIM) electronics

+HV

X-Ray

+12V

Signal Out

C

R

R 1

2

C

Older: ~1 GHz, 2 dBm

ASIC ?

AQRB, April, 2006

Typical Scope Traces

-150-100

-500

mV 5343 LC, f1mm (~2 pF) (a)

-60-40-20

0

5343, f1mm, (~5 pF)

mV (b)

-200

-100

0

mV

SPL 2625 3x5 mm2 (~15 pF) (c)

-300-200-100

0

mV C30626 5x5 mm2 (~30 pF) (d)

-150-100

-500

-10 -5 0 5 10 15 20 25 30

mV

C30703 10x10 mm2 (~120 pF)

Time (ns)

(e)

Leading edge (rise time) isgoverned by carrier transportin the device

Trailing edge by diodecapacitance and amplifierimpedance

Best case ~ 0.6 to 0.9 ns FWHM

AQRB, April, 2006

X-Ray Time Resolution

5343 f1mm x ~10 mm

82 ps FWHM

5344 LC f3mm x ~20 mm

160 ps FWHM

30719

5x5 mm2 x <10 mm

~170 ps FWHM

B

C30626

5x5 mm2 x ~100 mm

~ 0.8 ns

SPL 2625f3 mm x ~120 mm

~ 1.3 ns

C30703 (prot)

10x10 mm2 x ~180 mm

1.6 ns FWHM

Bevelled EdgeLAAPD

f16 mm x ~40 mm

0.4ns FWHM (Tail to 5 ns)

~ 10 ps/um near saturation

As x-rays penetrate and tend touniformly illuminate a device,the FWHM of the timeresolution is mostly determinedby the thickness of the driftregion

Tails determined by the fieldprofiles near the surface

5 ns Full Scale

Note use of leading edge discriminator.

AQRB, April, 2006

Best Time Resolution

100

101

102

103

104

105

106

-2 -1 0 1 2

Coun

ts

Time (ns)

1.95 nsFW M/105

0

5 105

1 106

-0.5 0 0.5 1

5343 (f=1mm)15.8 keV

Coun

ts

Time (ns)

82 ps FWHM

Best FWHM: 75 ps~ 10 um Device, Hamamatsu

Good Tail: < 2 ns at 1/105

Best Tail: ~1.4 ns at 1/105

~30 um Device, PKI

(Work in progress, Deschaux, et al)

Note: Geiger Mode operation ofsmall area devices has shown~20 to 30 ps nominal resolution(Cova, Lacaita)

AQRB, April, 2006

Fast Counting

104

105

106

107

108

109

Measured RateNon-Paralyzable ModelIdeal Linear Detector

Mea

sure

d Ra

te (c

ps) (a) 5x5 mm2 , C30626

104

105

106

107

108

Measured RateNon-ParalyzableParalyzableIdeal Linear Detector

Mea

sure

Rat

e (c

ps)

Nominal (Input) Rate (cps)

(b) f3mm SPL262516.5 keV

104

105

106

107

108

109

Measured RateParalyzable Model (t=6ns)Ideal Linear Detector

104 105 106 107 108 109

Mea

sure

d Ra

te (

cps) (c) 5x5 mm2 C30626

ESRF Package

Nominal (Input) Rate (cps)

Simplest “Non-Paralyzable” Model

†

n = m1- mT

n = True Rate, m = Measured Rate

T = System Dead TimeT = 3 to 20 ns

mT = 0.1 (20 MHz for T=5 ns) -> Can correct to ~1%

Source + APD + Discriminator + Counter

For mT > 0.1 -> MUST Calibrate!

AQRB, April, 2006

Pulse Height Distribution(X-Rays and Fast Electronics)

6 keV 14.4 keV(a) C30626, 5x5 mm2, <1 Mcps

Coun

ts (S

caled

)

1 Mcps5 Mcps10 Mcps

Coun

ts (s

caled

)

(b) C30626, 5x5 mm2, 14.4 keV

6 Mcps12 Mcps47 Mcps

Coun

ts (

Scal

ed) (c) SPL2625, f3mm, 16.5 keV

280V310V340V360V380V

0 0.4 0.8 1.2 1.6 2Co

unts

(Sca

led) (d) C30626, 5x5 mm2, 6 keV

Discriminator Window Position (Scaled)

Typically 20-30% for thicker devices.

Base-line shift at high rates (AC-Coupling).Better for lower capacitance devices.

Contributions:Amplifier NoiseAPD Gain NoisePenetration into gain region

Note: Slow Electronics, Low Gains,Cooling (-20C) -> resolution ~5% possible(Kishimoto)

AQRB, April, 2006

Other Points

Low T operation is possible - increased gain and T-sensitivity ~ 40K with API Beveled Edge (Yang)~ 100K with Ham. Rev. APD (Dorokhov)

Electron Detection is possible - note radiation damage!Hamamatsu Devices with special surface (Kishimoto)With Gain - VAPD (Kushman)

Progress toward a more realistic PMT replacement?Many (103/mm2) Geiger mode devices to keep

dynamic range & high gain (Buzhan, Sadygov)(Next Talk, Otte and poster/ abs 126: Yokoyama)

AQRB, April, 2006

APD Arrays

Note: Still relatively small numbers of pixels

M=Monolithic A=Assembled R=Replaceable

AQRB, April, 2006

ns Linear Array(Webb & McIntyre, 1984 - PKI C30985)

-0.2 0 0.2Horiz Pos (mm)

45 mm

125 mm

Coun

ts (

a.u.

)

Segmented cathode with bump bonding.

Linear Array, 25 elements, 0.3 mm pitch~ 100 um thick, ~1 ns resolution

101

102

103

104

-4 -2 0 2 4Time (ns)

~1 ns resolution

AQRB, April, 2006

Fast Linear ArrayHow can one keep good (200 ps) time resolution and high efficiency?

-> Grazing Incidence

16 Elements, 1.1 mm pitch 1x2.5 mm2 x 20 um thick

2 Degree grazing angle -> 0.6 x 2.5 mm Acceptance, 0.6 mm ThickEfficiency: ~1% -> 17 %

MagneticRelaxation in Spin-Ice(Sutter et al)

Note: Device stability is an issue.

AQRB, April, 2006

Modular ArrayNow Under Construction - SPring-8 - ESRF Collaboration

Similar concept to the “Fast Array”Grazing incidence gets you good (200 ps) time resolutionand good x-ray stopping power

But Modular: Element replacement possible.Amplifier design simplified.

8 Elements, f3mm x 20 um, incline at 2 degrees -> ~ 0.8 x 2.0 mm2 x 0.6 mm thick

AQRB, April, 2006

The Next Step: Integration

A Properly Instrumented APD ArrayPixel Size: ~ 0.3 x 0.3 mm2 x 0.2 mm thick

(Efficient and ~ 2 ns time resolution)~103 channels (1 cm2) at first -> 105

First Goal (?): Fast (us) framing detector.

Utility: Fast imaging, Stroboscopic Measurements, XPCS With different electronics, NRS: NSAXS, SRPAC

Discussion/Collaboration: ESRF, APS, SPring-8, DESY, KEK, &..

AQRB, April, 2006

Fast Framing Detector1. The Array Device -> Not so hard.2. Electronics -> ?

APD~2 GHz

1 PhotonAmp

>508 MHz

CounterDisc.

2 PhotonDisc.

3,4 Photon…?

12 bit(30 bit slow)

Question: On board histogramming for stroboscopic work?

Gating,Control

Readout

AQRB, April, 2006

Modified Electronic for NRS

APD102-103

Chan

AmpPromptCounter

Disc. TimingAdjust

20 ps Res.

Fast Switch

Fan inTDC

TAC/ADC?

508 MHz

100 kHz

Channel Info

AQRB, April, 2006

APD Collaborators

KEK: S. Kishimoto

ESRF: T. Deschaux, R. Rüffer

SPring-8: T. Ishikawa (& T. Kudo)

AQRB, April, 2006

Packaged Fast Counting Systems

ESRFRigal, Morse, et al.

BNLKuczewski, Siddons