LGAD sensors as enabling technology for low-material 4D tracker … · 2019-08-07 · LGAD sensors...

LGAD sensors as enabling technology for low-material 4D tracker systems

Iván Vila Álvarez

Instituto de Física de Cantabria (CSIC-UC)

Belle-2 VXD Open Workshop

CERN, June 9th, 2019

Outline — Precision timing basics

_ Jitter, time-walk and limit time resolution

— Low Gain Avalanche Detectors for timing.

_ LGAD vs PIN and SiPM

_ RD challenges

— LGADs architectures for 4D tracking.

_ Inverse-Low Gain Avalanche Detectors (iLGAD)

_ Resistive LGAD (RSD)

_ Trench-isolated LGAD (TI-LGAD)

— Summary

[email protected],Belle II VXD Open Workshop, CERN, July 9th 2019

Motivation… notjusttiming

— LGAD sensors with large SNR and small material budget to enable few tens of picoseconds time stamping of MIP particles (high precision ToF).

— Integrated signal amplification increases the Signal-to-Noise ratio increasing the tracking resolution:

_ Thinner detectors (reduction of the multiple scattering)

_ Improved intrinsic hit resolution.


Precision timing: basics


Timing 101: Timing resolution contributions -Jitter


Ideally, for a constant amplitude pulse, the time resolution is given by the jitter which depends on:

— Noise (dominated by the amplifier noise)

— Signal amplitude (dominated by sensor’s response)

— Rise time ( dominated by amplifier risetime)

Leading Edge Timing

Typical bandwidth of HL-LHC timing layer preamplifier of around 400 MHz tr tra 1nsthen a (modest) SNR of about 30 should provide a timing resolution of about 30 ps

Timing 101: Time resolution contributions –Time walk


In real conditions, the pulse amplitude is not always constant.

Leading edge

triggering

Option 1Correct the time shift

using the pulse amplitude

Option 2Constant Fraction

Triggering

Both methods: Amplitude-corrected LET and CFT have a similar performance as long as THE SHAPE OF THE PULSE’S LEADING EDGE IS CONSTANT.

Caveat: ToT correction is a off-line method while CDF is a real-time correction.

rt

Timing 101: Time resolution contributions – limiting systematics


Jitter induced by pulse shape changes

— Changes on the leading edge shape (i.e., different rise times or its distortion) translate into additional jitter

System aspects:

— TDC resolution.

— ADC (ToT) resolution limits time-walk correction.

— Clock distribution (jitter, slew and thermal drifts).

Those are the limiting factors and off-line data-based corrections are needed.

Time resolutionlimit

Enabling Technology:Low Gain Avalanche Sensor


Silicon sensors as enabling sensing technology for 4D tracking ?

— Silicon-based diodes provide both fast rise time and relatively large signal/noise ratio.

— Well stablished high-precision tracking technology (electrode patterning)

— Three operating modes: no signal gain (PIN), proportional (APD) and Geiger mode (SiPMT).

[email protected],Belle II VXD Open Workshop, CERN, July 9th 2019[1] A.G. Stewart et al. in Proc. of SPIE, Vol. 6119, 2006

No Gain

Moderate Gain &

Proportional response

Gain > 104

Digital response

Linear mode

Timing with PIN diodes : CMS HGCALas case of use


Scintillator + SiPM

Pin diode

PIN SILICON DIODES, active thickness 300, 200, 100 um

— Very reliable and mature mass production technology

— Main limitation: low SNR

E. Currás, Nucl.Instrum.Meth. A859 (2017) 31-36

Silicon Photomultipliers: (Geiger-mode APD) Case of use BTL detector at CMS

— Mature technology, mass produced and cheap sensing element.

— Main limitations: moderate radiation tolerance < 2x1014 neq/cm2 and concept with intrinsic poor spatial resolution, high fake pulse rate


Avalanche mode diode (Low Gain Avalanche Detector LGAD) :


— Proportional multiplication mode (impact ionization of primary carriers)

— Main advantage: custom SNR for optimal for timing and tracking

LGAD have a much larger rise-time(collecting time of primary electrons) than PIN (all carriers ballistic movement).

Go thinner to reduce the collecting time of the primary electrons and make trs << tra

LGAD SNR better than PIN due to gain Taylor the gain for optimal jitter wrt limit time resolution

Weightfield 2

Simulation

LGAD for HL-LHC timing layers

— LGAD is the baseline technology of the endcap MIP timing detector for the high-luminosity upgrade of the Atlas and CMS experiments.

— Provide about 30ps MIP time stamping for disentangling between the different interaction vertices.

— Main challenges:

_ Radiation tolerance to neutrons and protons (primary carriers trapping, acceptor deactivation, mean-free-path reduced, electric field modification)

_ Technology long-term reliability (Safe operating voltage)

_ Large scale manufacturing yield.

_ Improve fill-factor.


Relevant to Belle-2 upgrade

RD Challenge 1: Radiation tolerance


Total area of 2.6x2.6 mm2

Active area of 1.3x1.3 mm2

Intermediate gain

Irradiated at CERN PS with24 GeV protons at 5 differentfluences.

6E13 neq/cm2

1E14 neq/cm2

3E14 neq/cm2

1E15 neq/cm2

3E15 neq/cm2

Should not be an issue for a electron-positron collider experiment

Multiplication on-set

Protons

gamma

RD Challenge 2: Defining a “safe” operating voltage

— Naïve criteria: Operated at the lowest voltage such that the electronic-induced jitter does not dominate the time resolution (try to reach the plateau)

— Constratins:

_ Noise is not constant with HV (excess noise due to the stochastic nature of avalanche).

_ Fake pulse rate induced by micro-avalanches (“pop-corn” noise).


35um LGAD200 V (1e14 neq/cm2)

1 MIP signal (Laser)

35um LGAD205 V (1e14 neq/cm2)

1 MIP signal (Laser)

RD Challenge 3: Fill-factor— Current LGAD design: multiplication layer segmented

— Reduced SNR for MIPs going through the interpad region.


New LGAD technologiesfor 4D tracking


Towards a LGAD-based 4D tracking enabling sensor

— Several technologies for improving the spatial resolution (finer electrode segmentation and increased fill factor (reducing or avoid the interpad regions with no gain:

_ Trench-isolated Low Gain Avalanche detectors (TI-LGAD)(First engineering run completed, manufacturing run in preparation at FBK)

_ Resistive AC-Coupled Silicon Detectors (RSD)

(First manufacturing run completed at FBK)

_ Inverse Low Gain Avalanche detectors (iLGAD) (Proof-of-concept demonstrated, 2nd iteration (thin iLGAD) in preparation at CNM)


Trench-isolated Low Gain Avalanche detectors (TI-LGAD)

— Pixel border region hosts structures to control E field (JTE, p-stop, etc..)

— Dimensions limited by :

_ technological (photolithographic) constraints

_ physical limits (maximum E fields) due to operational requirements (VBD , etc.)

— Trench isolation could drastically reduce inter-pixel border region down to few μm

_ Typical trench width < 1 μm (max aspect ratio: 1:20)

_ Trench filling with: SiO2, Si3N4, Polysilicon

— Trench isolation successfully used in FBK HD and UHD SiPMs

[email protected],Belle II VXD Open Workshop, CERN, July 9th 2019G. Borghi, 34th RD50 Workshop, June 2019

Resistive AC-Coupled Silicon Detectors (RSD)

— AC coupled readout:

_ Non-segmented LGAD architecture.

_ Segmented metallic contact on top of a dielectric layer.

[email protected],Belle II VXD Open Workshop, CERN, July 9th 2019Marco Mandurrino, 34th RD50 Workshop, June 2019

Inverse LGAD (iLGAD) basics: a 4D tracking sensor


Multiplication layer divided into strip Multiplication layer extended over the electrodeCollects negative carriers (e) Collects positive carriers (h)Simple single side process Complex double side process

21

Gain Spatial Uniformity: Validation with MIPs @ SPS


AIDA-2020 WP7 Setup AKA as Atlas ITK setup (2016, 2017)

Strip LGAD

I-LGAD(8533W1K05T, 45 strips 160 um, non-irradiated)

Iván Vila, 13th Trento Workshop, February 2018

Tracking performance: I-LGAD strip sensor


PIN

I-LGAD VS PIN CORRELATION

I-LGAD

PIN

I-LGAD

CLUSTER SIZE

HIT RESIDUALS


Gain Spatial Uniformity: Collected Charge

0

20

40

60

80

100

0 100 200 300 400 500

Carga en función del voltaje

Test Beam 2017

Test Beam

Q / k

e-

V / V

0

10

20

30

40

50

60

70

80

80 100 120 140 160 180 200 220

Charge dependence with voltage

for a LGAD at Test Beam

Q(Peak 1) / ke-Q(Peak 2) / ke-

Q / k

e-

V / V

LGAD

i-LGAD

Standard PIN

Set-up for timing characterization and DUT

— Time standard: constant time interval between two picosecond IR laser pulses (1060 nm)

— Fixed time interval between laser pulses generated by optical splitting and delayed recombination of a single laser pulse.

— The spread (std. dev) on the determination of the time interval between pulses is sqrt(2) of the sensor’s intrinsic timing resolution (Sensor’s jitter).


Laser diode Isolator Splitter

DelayLine Combiner DUT

45 strips Pitch 160 mm

25

52.23 ns

Esteban Currás, 15th VCI , February 2019Marcos Ferández, 34th RD50 workshop, June 2019

1 23

Optimum value of the k parameter for the I-LGAD in zone 1.

For the PIN k parameter is optimum in almost the full range.

1

2

3

Time resolution (Pin vs ILGAD)

[email protected],Belle II VXD Open Workshop, CERN, July 9th 2019 26

Constant Fraction Discrimination method (CFD): Simulation 300 um vs 50 um.

300 um 50 um

First part of the signal very fast and good for timing but not as good in SNR terms.

Second part still good with better SNR but worse rise time.

If we go to 50 um we can have both benefits: Good SNR and low rise time(Production run to start in September)

Attention to the axis range


Take home messages

— LGAD provides the additional boost to the SNR to achieve O(10ps) hit resolution and O(10) spatial resolution thanks to fine electrode segmentation.

— Several 4D LGAD architectures are under consideration:

_ Trench-Isolated LGAD (early stage of development)

_ Resistive AC LGADs (Promising first results)

_ Inverse-LGAD p-in-p architecture (proof-of-concept demonstrated)

— The technology is maturing (more sensor manufacturing center are jumping in) BUT still a long way to go:_ Complete proof-of-concept studies.

_ Reliability (long term stability, noise and destructive breakdown)

_ Manufacturing yield? Scalability (larger area sensors) ? Uniformity ?

_ Radiation tolerance of strip and pixel like devices?


THANK YOU FOR YOUR ATTENTION


Timing detectors: the recent past (1)

— Precise time stamping of charged particles (Dt 100 ps) by dedicated Time-of-Flight (ToF) detectors.

— Particle ID of low momentum charged particles (mostly pion/kaon disentangling in flavor physics experiments )

— ToF detector determines speed of a charged particle by measuring flight time t across a known distance L. Knowing the particle momentum p, one determines the mass.


12

22

L

tc

c

pm

Mass vs momentumCDF-II ToF detector at Tevatron

Timing detectors: the recent past (2)

— Fast detection sensing technologies:

_ Fast scintillators + conventional Photomultiplier Tubes

_ Resistive Plate Chambers (very large area detectors).

— Limitations:

_ Difficult to go below 100 ps timing resolution, severe gain reduction inside magnetic fields (PMT), bulky, scintillator and gaz radiation-induced aging, limited granularity (spatial resolution).


ALICE ToF detectorOverall time resolution

120 ps

CDF & Belle ToFDetectors

OverallTime resolution

100 ps

GigaTracker: A true 4-dimension tracker system

— Aim to measure Br(K π +νν) SM branching fraction very small ∼ 10−10

— Unstructured particle beam with ∼5 second burst every ∼42 seconds (instantaneous rate 750 MHz)


Hybrid pixel detector:• 300 um × 300 um pixels• One sensor (~6×3 cm2) bump-

bonded to 10 read-out chips

Motivation: Why do we need precise time stamping of particles? (1)

— Future hadronic colliders (HL-LHC, FCC) to increase many fold the current LHC luminosity: track multiplicity and bias events


Thousands of tracks, vertices, calorimeter clusters that must be disentangled and accurately determined


— Profit from primary vertices' time spread.


— With a time spread 200ps then 20-30 ps of PV timing resolution is required to disentagle the different primary vertices.



— Suppression of mismatched PU tracks, Jet pileup and improved of track isolation.

4D tracking with Silicon: Choice matrix

TECHNOLOGYRISE – TIME (CAPACITANCE DRIVEN)

SNRRADIATION TOLERANCE

RELIABLE MASS PRODUCTION(AFFORDABLE)

LIMITATIONS / CHALLENGES

4D – TKPROPOSALS

PIN (NO GAIN) FAST < 1ns LOW HIGH POSSIBLE LOW SIGNAL 3D SENSORS

LGAD (PROP. GAIN) FAST <1ns MEDIUM-HIGH MEDIUMPOSSIBLE (TO BE DEMONSTRATED)

RADIATION TOLERANCE, RELIABILITY, FILL-FACTOR

ILGADAC-LGAD

SiPM (GEIGER) FAST <1ns VERY HIGHMEDIUM/POOR

POSSIBLERADIATION TOLERANCE, HIGH RATE SPURIOUS SIGNALS, POOR SPATIAL RESOLUTION

-


Radiation tolerance (pad LGADs)

Two radiation damage mechanism present: acceptor removal & Bulk Space Charge Inversion (BSCI).


PoS(Vertex 2017)041

Moritz WieheTredi 2019

Radiation tolerance (pad LGADs) (2)


— Due to the BSCI of the p bulk, the LGAD becomes effectively a Shockleyfour-layer diode (aka Thyristor with floating gate).

— Can we explain the CV characteristic based on a Shockley four-layerdiode?

— Slew rate significantly better for the case of thinner LGADs.

I-LGAD basics: multiplication footprint

— Distinct signature of signal amplification: cathode illumination with red laser

— Injections of electron into the cathode, resulting transient current is a sequential contribution of primary electrons reaching the amplification layer and secondary holes drifting towards the anode.


Cu

rren

t (a

.u)

Primary electrons current

Secondary holes

current

Multiplication onset

PAD-like LGAD or ILGAD (reverse

polarity)

Strip-like I-LGAD

I-LG

ADba

sics

Gain Spatial Uniformity: Gain vs. Hit position (1)— Dependence of the Signal-to-Noise Ratio with the “fractional

position” for cluster of size two.


I-LG

ADfill

fact

or


Gain Spatial Uniformity: Gain vs. Hit position (2)

LGAD Strip

i-LGAD Strip

100 Volt bias

400 Volt bias

Artifact due to front-endsaturation

<SNR> 30

<SNR> 43

I-LG

ADfill

fact

or

Tracking performance: Reference strip sensor

— Challenges:_ Difficult synchronization between Alibava daq (x3) and Eudaq.

_ Saturation of the Alibava ADC.


I-LG

ADtra

ckiin

g

Time resolved transient currents (PIN vs ILGAD)

— Acquired averaged waveform from I-LGAD and PIN sensors with a time interval of 52.23 ns between pulses

— Laser pulse creates an amount of e-h pairs of similar to the one created by a few hundreds of MIP (usual configuration for TCT study).

— Signal amplified (60db, Miteq 1660) & digitized (25GSa/s)

— Transient current shaped by weighting field and multiplication.

— Is the fast leading edge of the I-LGAD fully dominated by the primary carriers? Or Is there any contribution from multiplication? Let’s compare the timing performance.


Split the input signal in two copies

Delay one copy by td

Attenuate (-kVmax) and invert the other copy.

Add the two signals together

The zero crossing point of the bipolar signal (tk) does not depend of the amplitude (assuming pulse shape invariance)

Emulate by software a real CFD electronic circuit

Emulation of the CDF


PIN vs ILGAD: comparison of fast leading edges.

— How the amplitude of the fast leading edge depends on bias voltage.


2nd peak

1st peak

… concerning radiation tolerance.

— The layout of the proposed LHC timing detectors: a mosaic of mini-pads (elements with area of few mm2)

— A pad-like LGAD is also a pad-like I-LGAD therefore they exhibit the same radiation tolerance

— For a strip-like geometry things are different (optimal geometry where the maximum of both the electric field and weighting field overlap)

— Advise againg: go thinner.


285 um

LGAD

I-LGAD

LGAD sensors as enabling technology for low-material 4D tracker … · 2019-08-07 · LGAD sensors...

Documents

Transcript of LGAD sensors as enabling technology for low-material 4D tracker … · 2019-08-07 · LGAD sensors...