Abraham Gallas

V Jornadas sobre la participación española en futuros aceleradores lineales

Timepix Pixel Sensors Tracking & Timestamping for ILC

Outline

ILC environment and assumptions Detector design rationale ASIC (Timepix) Sensor thinning Low mass bump bonding Test beams with Timepix Next steps

26/10/10 Abraham Gallas 2

ILC in a nutshell

26/10/10 Abraham Gallas 3

e+ e- linear collider

Center of mass energy range 200-500 GeV

Peak luminosity 2 x 1034 cm-2 s-1

Bunch timing:• 5 pulses per second (5 Hz)• 1260, 2625, 5340 bunches per pulse separated by 180, 369, 500 ns

• power pulsing

• readout speed

14 mrad crossing angle

Background:• small bunches• create beamstrahlung → pairs

Hit density (#/mm2/BX)

Forward tracking detector

26/10/10 Abraham Gallas 4

Relevant physics processes with

particles emitted at small :

mostly e-, t, b- and c-jets

ILD's Forward Tracking Disks

The forward region 6° < < 30°: 0.1 rad < < 0.45 rad

0.9 < cos < 0.995

1.5 < | | < 3.

Detector design rationale

25x25 m pixel sensor instrumented with ROC derived from current Timepix (Timepix2-Timepix3)

Both ToT and ToA modes running simultaneously in each pixel

Time resolution of 10 ns or better ∼ (Timepix2 > 1.5ns

(25ns/16)) allowing time stamping (bunch tagging)

Full readout between pulses (5Hz) or every 100BX

Power cycling leading to a 70% reduction on the time the ROC is ON (Timepix2 45∼ W/pixel).

26/10/10 Abraham Gallas 5

The Medipix Chips

A philosophy of functionality built into the pixel matrix allows

complex behavior with a minimal inactive region

55 m square pixel matrix 256 by 256

3-side buttable

Configurable ‘shutter’ allows many different

applications

bipolar (h+ and e-)Silicon, CdZnTe, CdTe,

GaAs, Amorphous Silicon, 3D, Gas Amplification,

Microchannel Plates etc…

Timepix (2006)

sensor

Analogue amplification

Digital processing

Read-out ASIC

Timepix design requestedand funded by EUDET collaboration

Conventional Medipix2 counting mode remains.

Addition of a clock up to 100MHz allows two new modes.

Time over Threshold

Time of Arrival

Pixels can be individually programmed into one of these three modes

Time over Threshold

Threshold

Time Over Threshold counts to the falling edge of the pulse

Threshold

Time of Arrival

Time of Arrival counts to the end of the Shutter

Development history and future

26/10/10 Abraham Gallas 8

Medipix11m SCAMOS 64 by 64 pixels Photon Counting Demonstrator (1997)

Medipix2250nm IBM CMOS, 256 by 256 55m pixels Full photon counting (2002)

Analogue (ToT) and Time Stamping (ToA) (2006)Timepix

130nm IBM CMOSPhoton Counting, Spectroscopic, Charge Summing, Continuous Readout (2009)

Fast front end, Simultaneous ToT and ToA (2011)Timepix2

Medipix3

VELOpix

Timepix3CLICpix

130nm/90nm/65nmFuture LHCb readout – Data driven 40MHz ToT 12Gb/s per chip (2013)

130nm/90nm/65nmFuture Hybrid Pixel Time tagging layer for the LCD project (20??)

Timepix2 Main Requirements

Pixel size 55 µm x 55 µm

Pixel matrix array 256 x 256

Sparse readout YES

PC, TOA or TOT recorded simultaneously

YES (2 at a time)

Minimum detectable charge ≤ 500 e-

TOA resolution >1.5ns (25ns/16) 4bits (Gossipo3 style)

Peaking time < 25 ns

TOT resolution <5% channel to channel spread

Technology IBM 130nm DM 3-2-3

Power consumption <1.5W/cm2 (~45 μW/pixel) @1.2 V

Target floorplan 3 sides buttable and minimum periphery

TSVs possibility YES. Multi-dicing scheme as Medipix3

9

• Lots of different applications → Very demanding specs !

XAVIER LLOPART– CERN PH-ESE

Sensor thinning

26/10/10 Abraham Gallas 10

Timepix on SOI 150m

Collaboration with CNM to thin 2D-pixel (55x55 μm) sensor from 300 μm down to 200, 150, 100 μm, p-on-n & n-on-p

Read out with TimePix ASIC

Goal:• Measure resolution, efficiency, … thin sensor•Minimal guard ring design.•Production of module/ladder with 3 ASICs

55Fe

Low mass bump bonding

26/10/10 Abraham Gallas 11

Pixel detectors consist of a sensor chip and ROC which are connected with flip chip bumps. One of the current technologies (FC µ-bump):

20-30μm (SnPb, SnAgCu, SnAg, … )

40-… μm pixel pitch

For 55(25) μm pixel size adds 0.019(0.091) % Xo

Other technologies that can reduce considerably the material budget:

Solid-Liquid-Inter-Diffusion (SLID) Soldering:

55 μm pixel sensor: 0.0027 % Xo

25 μm pixel sensor: 0.013 % Xo

Carbon Nano Fiber (CNF) Interconnections (R&D stage):

55 μm pixel sensor: 0.00087 % Xo

25 μm pixel sensor: 0.0042 % Xo

• In SLID tin, indium or other metals with low melting point (MP) are capped on high melting point (MP) pads, because: 1. Creation of intermetallic compounds (IMC) with the pad metal.2. High planarity requirements of metal-metal bonding (e.g., Cu-Cu) are

compensated. • Solid-Liquid-Inter-Diffusion (SLID) soldering, AKA Transient-Liquid-Phase (TLP).

• Thin layer of solder turns completely into IMC ultra thin metallic joints!• After the first reflow the MP increases significantly and becomes thermally

very stable.• Cu-Sn-Cu structure is the most commonly used, and with an optimized

process Sn transforms to Cu3Sn in some minutes.

Solid-Liquid-Inter-Diffusion (SLID) Soldering

Beginning:

•Thick Cu pads and < 5 µm of Sn.•Bonding at 270 –

300 ̊C.

Step 1

•Sn reacts with Cu and creates IMC’s.•Cu6Sn5 phase grows in big scallops.

Step 2:•Cross-hatched Cu6Sn5 phase consumes the Sn aggressively.• Cu3Sn phase grows on at Cu

interfaces.

Step 3:•All Sn has been transformed to IMC’s. •Cu3Sn is taking over Cu6Sn5 and grows at the expense of Cu6Sn5 and Cu.

Step 4:•After long heating, the ductile Cu3Sn expands over the whole area and forms the a thin joint.

SAMI VAEHAENEN – CERN PH-ESE

Carbon Nano Fiber Interconnections

26/10/10 Abraham Gallas 13

• CERN has started a small project with Smoltek (Gothenburg, Sweden) to develop fine-pitch CNF interconnection technique for pixel chips.

• Goal is set at growing 5 µm – 10 µm long fibres on chips and joining them together.

• Electrical contacts will be tested with/without metal contacts.

• CNF’s would be ultra-low mass interconnections.

• Technology has prospects to be ultra-fine pitch capable.

• High planarity of ROC and sensor is required.

• First CNF forests have been deposited on CERN test vehicle chips.• Development plan has been made to improve the patterning resolution and to develop suitable

flip chip processes.

SAMI VAEHAENEN – CERN PH-ESE

Timepix Testbeams

26/10/10 Abraham Gallas 14

Six Testbeams with Timepix in 2009 and 2010

2009 Testbeam - Proving Timepix

26/10/10 Abraham Gallas 15

Timepix had not been used at all in a particle tracking application

We took the opportunity to run parasitically in three beam periods

Tested a 300 m standard silicon Timepix assembly and a DS3D assembly

Main Measurements:

Resolution vs. Angle

Resolution vs. Threshold

Resolution vs. Silicon Bias

Efficiency vs. Threshold

Efficiency vs. Bias

Timewalk

Three Testbeams in 2009

June 2009 : Medipix Testbeam3 days to demonstrate tracking

July 2009 : CMS SiBit beam periodTwo weeks – parasitic Timepix Telescope

2 Timepix2 Medipix~perpendicularNo DUT

2 Timepix4 Medipix~perpendicular300μm and 3D DUTsManual angle adjustment

August 2009 Timepix TelescopeAugust 2009 Timepix Telescope

4 Timepix, 2 Medipix planes in telescope

Symmetric positioning of planes around DUT

Telescope planes mounted at 9° around x & y to boost resolution

DUT position and angle controlled remotely by stepper motors

2.3m Track Reconstruction Error~100Hz track rate1 frame per second~100,000 tracks per measurement point~1.5 hours per point in SPS North Area

55μm

300μm

0o

Angled Planes to Boost Resolution

Hits that only affect one pixel have limited resolution (30μm region in 55μm pixel)

Tilting the sensor means all tracks charge share and use the ToT information in centroid, CoG calculations

0o ~10μm resolution9o ~4.2μm resolution

Indicative Timepix events

55μm

300μ

m

9o

2009 Results – Resolution Vs Track Angle

Resolution result from 2009 testbeam demonstrating resolution of a Timepix assembly and the performance of the telescope

Operating point of Telescope planes

Eta distributions

26/10/10 Abraham Gallas 20

Uncorrected

0o

incidence5o

incidence8o

incidence18o

incidence

1 pixel wide clusters2 pixel wide clusters 3 pixel wide clusters

Corrected

2010 Testbeam Activity

3 beam periods as main user Added Time Tagging System May

USB2 readout 300m Timepix and PR01 fine

pitch microstrip sensor (40μm) June

USB2 readout 150m Timepix and PR01 Strip

August RELAXD readout 3D irradiated Timepix, FZ, MCZ,

BCB strip, 150m Timepix and 300m Timepix

Not all data analysed yet so not too many results to show

2010 Timepix Telescope

6 pixel telescope planes angled in 2 dimensions to optimise resolution

Device Under Testmoved and rotated viaremote controlled stepper motor

Fine pitch strip detectorwith fast electronics LHC readout

2010 Telescope in Timepix DUT Configuration

Timepix ToT Tracking

Timepix ToT Tracking

Timepix DUT

Scintillators to set shutter length to e.g. 100 tracks

beam

Shutter Generator

In this configuration the telescope was optimized for running with a Timepix DUT

The USB2 readout allowed a 7 frame per second readout rate (700Hz track rate)

The all angled six Timepix telescope gives a ~2.0μm Track Extrapolation Error

Time Resolution for LHC readouts Asynchronous SPS beam not suited to LHC systems designed for 25ns bunch structure Implemented a TDC which with Timepix ToA mode gives us ~1ns per track time stamping Able to provide and record synchronised triggers to 40MHz readout systems (TELL1) Allows software reconstruction and analysis of asynchronous tracks

PR01 DUT Timepix ToT

Tracking

Timepix ToT Tracking

Timepix ToA Track Time Tagging Plane ~100ns

Scintillator Coincidence and TDC ~1nsLogic +

TDCSynchronized

Trigger

Telescope in Time Tagging configuration for LHCb Sensor Readout

beam

150μm Sensor Results

With a 150μm sensor the optimum resolution point is at twice the angle of a 300μm

The higher data rate allows a significant number of measurements to be taken

RELAXD Readout

High Resolution Large Area X-Ray Detector RELAXD readout from NIKHEF 50 frames per second over gigabit Ethernet

August 2010 Telescope – Timepix DUT

Cooled DUT Timepix ToT

TrackingTimepix ToT Tracking

RELAXD interface

RELAXD interface

RELAXD interface

RELAXD system allowed 55 frames per second readout (~2,500 tracks per second)Each 100,000 point measurement now takes 4 minutes

Eight angled Timepix tracking planes gives a ~1.67um Track Extrapolation Error

Closer tacking planes reduce multiple scattering effects

Cooling system

26/10/10 Abraham Gallas 28

To operate irradiated assemblies its necessary to cool the sensor to below 0oC

This system achieved a steady temperature of ~-5oC

Water Block

Sensor+ROC and Pyrolytic Graphite80W Peltier

Telescope Comparisons

26/10/10 Abraham Gallas 29

Telescope Pixel Resolution Time Tag Rate

Timepix 2009 55m 2.3m 100ns 100Hz

Timepix 2010 (May) 55m 2.0m ~1ns 350Hz

Timepix 2010 (Aug) 55m 1.7m ~1ns 2.8kHz

EUDET (Low res) 18m ~2m 100us 990Hz

EUDET (High res) 10m ~1m - 300Hz?

EUDET. Telescope

Next steps

Module0 construction (3-4 ROC) Minimal guard ring design Thinning of ROC (50μm) Thinning of sensor to 80 μm Bump bonding thin sensor on thin ROC Support structures (CVD)

26/10/10 Abraham Gallas 30