1

The pixel readout of TPCs

Max Chefdeville, NIKHEF, Amsterdam

2

Overview

• Motivations

• 2D readout of small gas volume by means of 55 µm2 pixels in combination with Micromegas & GEMs

• Time information (and more) with the new TimePix chip

3

Motivations for pixel readout of TPCs

• Spatial resolution:– Narrow charge distribution (RMS ~15 μm)– No c.o.g calculation possible– σxy limited by the pad size (pitch/√12)

• Fine granularity– δ-ray recognition/suppression in TPC– Direction of low-energy e- for X-ray polarimetry– Directionality of nuclear recoils in WIMP or

neutrino interactions– Energy & direction of 2 e- from double beta decay– Energy of photo-electrons from axion conversions

wire

Cathode pads

GEMMicromegas

4

• High granularity pitch/√12 ~ 15 µm– Potentially better spatial resolution

• Smaller input noise (Cin ~ 15 fF)– Lower gain, less aging, smaller ion backflow– Small charge sensitivity

Single electron detection with an efficiency depending on the gain and the amplification structures

• May be possible to count primary clusters– dE/dx for TPC

• Possible to count primary electrons– Accurate energy measurement of low energy recoils & electrons

The pixel readout of TPCs

5

The Medipix2 chip

Use the “naked” chip as the detector anode

• Developed by the Medipix consortium, CERN

• Chip layout:– 1.4 x 1.6 cm2 area– 256 x 256 pixel matrix– 55 x 55 µm2 pixels

• On each pixel:– Preamp. + shaper– 2 discri. (thresholds)– 14 bit counter

6

MediPix2 pixel sensorBrass spacer blockPrinted circuit boardAluminum base plate

Micromegas

Cathode (drift) plane

Baseplate

Drift space: 15 mm

Medipix2 & MicromegasNIKHEF

SACLAY

TWENTE

7

He/Isobutane80/20

δ-ray!

Efficiency fordetecting single

electrons:> 90 %

8

Helium VS Argon mixtures

• Argon: larger primary statistic & transverse diffusion

He 20% iC4H10 Ar 20% iC4H10

Larger diffusion

Top of track

Interesting tool to study ionization statistic of photons & charged particles …

9

Integrate amplification and readout structures:

InGrid

Walls Pillars

Wafer post-processing

10

InGrid, an integrated Micromegas• Low temperature process:

spin coating, wet etching;

• Perfect alignment between grid holes and pixel pads;

• No dead areas due to pillars;

• Flexible design.

13 % FWHM @ 5.9 keV

30 µm Ø pillar

11

InGrid on a Medipix2 chip

• Goal: post-process full wafers

• Post-processing of individual chips @ Twente University, Netherlands

InGrid on top of pixel matrix

pillarspixels

Grid hole centered between 4 pixels

First trials promising (using rejected chips)

12

The spark issue

Detector very sensitive to gas discharges

Pixel InGrid mesh (Micromegas OK)

13

Discharge protections

• destructive discharge

• Proposals:– multi-stage amplification;– high R coating of anode.

Discharge signals of 2 Micromegas detectors

e- multiplication @ high E

e- extraction @ low E

Current attenuated by the high R layer

Maybe enough to protect the chip… Both are being applied on Medipix2

Un-coated anode

Coated anode

TwinGrid

14

Micromegas & GEMs

With GEM, decoupling of amplification and readout:– Smaller charge per pixel, smaller single

electron detection efficiency– Low field above the chip, no gas discharges

involving the chip

80 kV/cm 50 µm

40 kV/cm 50 µm

1 kV/cm 1000 µm

15

GEMs & MediPix2 @ DESYBONN

FREIBURG

6 mm

2 mm

2 mm

1 mm

5 GeV/c electron beam

Si telescope

Si-telescope serves as track reconstruction in the drift volume, e. g. for the determination of the drift velocity and the σ0 near top GEM

16

GEMs & MediPix2 @ DESY• Gas mixture: Ar CO2 70/30

~30 clusters per cm created, ~10 cluster/cm reconstructedLess primary information due to diffusion in the amplification gap

• Tracks parallel to the pixel plane: same diffusion along the track

17

The TimePix chip

• Based on MediPix2 design– Same dimensions and readout protocol– Replace 14 bit hit counter by TDC– Only low threshold– Fired pixels count clock pulses 100 MHz

• Counting modes (can be mixed)– “Time over threshold” Charge info.– “Common stop” Time

info.

• Characteristics:– Dynamic range, 160 µs @ 100 MHz– Integration time, 200 ns

11 22 44

55

66

11 22 44

55

66

33

55μm

55μm

prea

mp/

shap

er

Low

th

resh

old

Hig

h th

resh

old

Con

trol

logi

c

MediPix2 → TimePix

18

Counting modes (simulations)

17

 time over thresholdfrom hit till end of shutter time

100 MHz clock

not detected

detected

Charge summed

19

Time over threshold events

TOT linear above ~ 4 ke-

Better calibration to come…

> 30 ke- per pixels

• Ar CO2 70/30

• No time information (2D)

• Charge information should improve spatial resolution

0

10

20

30

40

0 200 400 600TOT-Counts

En

erg

y [K

e]

TOT counts

N e

-

Freiburg Bonn

20

“Common stop” (Time) mode• Every fired pixel counts till the end of a 12 μs shutter window• Tracks parallel to the pixel plane (same color)

The larger the number of counts, the shorther the drift time

21

Mixed Mode operation

• Consecutive pixels have Time and TOT assignment and are here separated via mapping onto a 181x181 matrix

• Benefit from charge & time information

• Interesting for double track separation

Ar

CO

2 70

/30

He

CO

2 70

/30

22

Outlook• Pixel readout

– Proof of principle of pixel readout with MediPix2 demonstrated– TimePix works and opens the way to 3D high granularity tracking

• Amplification structures– GEMs: very stable but low detection efficiency, nice tracks with TimePix;– Micromegas: spark issue to be solved soon, provide almost all information

on event spatial structure.

• Future plans– Reduce the number of GEMs (lower threshold with TimePix)– Integrate/place a Micromegas on MediPix2/TimePix

high R protection or not– Study tracking capabilities / gas ionization statistic– Build a small endplate by chip tilling

• Low energy event detection– TimePix + Micromegas is a good candidate

Energy of low recoil by electron counting

23

NIKHEF Harry van der GraafMartin FransenJan TimmermansJan VisschersSipho van der PuttenArno Aarts

Saclay CEA DAPNIA David AttiePaul ColasArnaud GiganonYannis Giomataris

Univ. Twente/Mesa+ Jurriaan SchmitzVictor Blanco CarballoCora SalmSander Smits

FREIBURG A. BambergerK. DeschU. RenzM. TitovN. VlasovA. ZwergerP. Wienemann

CERN Erik HeijneXavier LlopartMedipix Consortium