GridPix/Gossip for

ATLAS SCT UpgradeILCCLIC

…….insulators in strong E- fields……….…….the frustration of innovation………..

Harry van der GraafNikhef, Amsterdam

IEEE-NSS ConferenceMPGD-Si Detector WorkshopDresden, Oct 18, 2008

Micro Strip Gas Counters: hard to operate:

- discharges, ruining electrodes- ageing

! Very strong electric field in insulator’s volume & surface !

GEMs:

- often cascade of 3 GEMs used to limit gain per GEM to ~40- ‘rim’ (dia hole Cu/kapton) critical- shape of hole wall critical- charge up effects

Atsuhiko OCHIAtsuhiko OCHIKobe University, JAPANKobe University, JAPAN

13 Oct. 2008 2nd RD51 workshop in Paris

GEM ProductionGEM Production

x750

100um

80um

Remove copper by wet etching

Irradiate CO2 laser

Remove remaining edge from the other side

RIKEN/SciEnergy GEM(thick-foil and fine-pitch) pitch 80um hole 40um thickness 100um

Recent Status of Development

GEM test setup and parameters– Thick-foil and fine-pitch GEM

(single layer)– HV supplied through a resister

chain– Ed=2.5kV/cm, Ei=4~5kV/cm,

V⊿ GEM=300~600V

– Gas: Ar+CO2(30%) flow

– Readout by 1cm x 1cm pad

Gain measurement– Gain vs applied voltage

– X-ray from Fe-55 (5.9keV)

Gain Curve (RIKEN GEM)Gain Curve (RIKEN GEM)

Fe-55 spectrum5.9keV

gain=3x104

To keep good spatial resolution and keep discharge point at high gain.Our GEM is most suitable for Cosmic X-ray Polarimeters.

Recent Status of Development

Gain instability (RIKEN GEM)Gain instability (RIKEN GEM)

F. Simon (IEEE, 2006)T. Tamagawa(IEEE,2007)

time (s)

rela

tive

gain

3 hours

No gain increase in short measurementsThis is not for a special batch of GEMs but for all GEMs we producedPossible reasons;

Less charging-up due to cylindrical hole shapeLess polarization of Liquid Crystal Polymer

No increase and decrease just after HV on.

Recent Status of Development

GemGrid 1

GemGrid 2

Bulk high-resistivity materials

hydregenated amorphous silicon

allowed gains up to 5 105

staying proportional!

Si-rich silicon nitride (Si3N4)

Measurements on Si-rich Silicon Nitride (Si3N4)

Column resistance: ρ D/OPotential surface measurable: gain drop factor 2 at dV = 17.5 VWith known current: bulk resistivity ρ measurable: ~ 1 – 50 1013 Ohm cm

Surface time constant: Column resistance x (virtual colum capacitance) =

(ρ D/O)*(ε O/D) = ρ ε (independant of layer thickness D!)

Resistive Plate Chambers (RPCs)

-essential: high-resistivity material- quenches sparks- sufficient charge compensation current

Traditional: insulator + dope (Sardinian oils…?)New: high-resistivity bulk (ceramic) material: higher counting ratesCompare graphite covered mylar foil

Micromegas on pillars

Edge discharge protection foil

discharges + vibrations

conductivity of kapton

Charge-up effects

After (rapid) ramping of HV:

- polarisation: reduction of E-field in insulator (bulk) volumeIn homogeneous field with insulator // to field: nothing

With E component perp. on insulating surface: modificationof potential by hitting e- and/or ions until E // surface

GEM hole

region ofworse insulation

equalizing with water

Stronger effects forgood insulator

Very preliminary:

Use as little as possible insulating surfaces // strong E fieldsApply high-resistivity materials instead of insulators

Even more preliminary:

As for gain: GEMs perform les than (corresponding) Micromegas

April 2004 Micromegas + MediPix 2 NIKHEF/Saclay/Univ. Twente:

MediPix2 pixel sensorBrass spacer blockPrinted circuit boardAluminum base plate

Micromegas

Cathode (drift) plane

55Fe

Baseplate

Drift space: 15 mm

MIPs

No source, 1sNo source, 1s5555Fe, 1sFe, 1s

He/Isobutane80/20Modified MediPix

δ-ray!14 mm

Integrate Micromegas and pixel sensor:

InGrid

‘wafer post processing’by

Univ. of Twente, MESA+’

Full post-processing of a TimePixFull post-processing of a TimePix• Timepix chip + SiProt + Ingrid:

“Uniform”

MESA+

IMT Neuchatel

Charge mode

14 mm

A “long” cosmic track

Timepix +

20 μm thick Siprot

+

Ingrid

Drifttime (bin = 10 ns)

Stable operation in He iC4H10

10 mm

cathode @ - 1500 V 14 mm

Si (vertex) track detector GOSSIP

CMOS chip

Si depletion layer

Vbias

• Si strip detectors• Si pixel detectors• MAPs

Gas: 1 mm as detection medium99 % chance to have at least 1 e-

Gas amplification ~ 1000:

Single electron sensitive

All signals arrive within 16 ns

Cluster3

Cathode (drift) plane

Integrated Grid (InGrid)

Cluster2

Cluster1

Slimmed Silicon Readout chipInput pixel

1mm,100V

50um, 400V

50um

Gossip: replacement of Si tracker

Essential: thin gas layer (1.2 mm)

1.2 mm

GOSSIP-Brico: PSI-46 (CMS Pixel FE chip)First prototype of GOSSIP on a PSI46 is working:

• 1.2 mm drift gap• Grid signal used as trigger• 30 µm layer of SiProt

Tracking sensor material: gas versus Si- it is light

- primary electrons can simply be multiplied: gas amplification: low power

- no bias current: low power & simple FE circuits

- gas can be exchanged: no radiation damage of sensor

- gas has a low εr: with small voxels the source capacity can be small (10 fF) allowing fast, low-noise, and low-power preamps

- gas is usually cheap

- low sensitive for neutron and X-ray background

- δ-rays can be recognized

- [high ion & electron mobility: fast signals, high count rates are possible]

- discharges/sparks: readout system should be spark proof- ageing: must be solved and must be understood / under control- diffusion: limits max. drift length

CMOS Chip protection against - discharges- sparks- HV breakdowns- too large signals

Emperical method:Try RPC technology

Amorph Si (segmented)

Silicon Protection: SiProt

Qmax ~ 1 – 2 fC

Chip may die if Qmax > 10 fC

Ageing

Radiation damage of CMOS pixel chip is relevant- common for all tracking detectors- believed to widthstand ATLAS Upgrade Dose in 90 nm technology

Radiation damage of sensor: not relevant for Gossip sensor since this is gas being exchanged

Typical for gaseous detectors: the deposit of an (insulating) polymeron the electrodes of a detector. Decrease of signal amplitude

Little ageing expected:- little primary ionisation (~ 10 e-/track)- low gas gain (500 – 1000)- large anode surface (compare pixel anode plane with surface of thin wire)- E-field at flat anode ~3 lower than E-field at anode wire

Linear fitI = I0 + a.ta = -0.5932=> a/I2 = 0.0183

av current = 5.9 A=> total charge deposited = 5.9*3600*24*4 = 2.55 Csurface 0.49 cm2

=> 5.2 C/cm2

assume: drift distance 1 mm Ar/CH4 having 9e-/mm=> 1 mip = 9*1000*1.6*10-19

= 1.44 10-15Cdeposited charge corresponds to3.6 1015 mips/cm2

X ray irradiation at PANalytical (detail)

Time

14-M

ay-0

5

16-M

ay-0

5

18-M

ay-0

5

I cath

(A

)

0

2

4

6

8

Icath

1/x fit

3.6x1015 mips/cm2@ gain = 1000

gas: standard Ar/Methane 90/10. Deposit containing C found on anode

Irradiation with 8 keV X-rays:Irradiation with 8 keV X-rays:

No No rate effectsrate effects up to anode up to anode current density of 0.2 current density of 0.2 μA / mm / mm22

very fast track counting very fast track counting possible!possible!

After 0.3 Coulomb/mm2:

(eq. 3.7 x 1016 MIPs/cm2 !!)

deposit of carbon polymer on anode is clearly visible. Micromegas is clean (!?)Little deposit on cathode, and……Chamber still worked!

Ageing

Construction of many test chambersprototypesNext-1,2,3,4,5Next QuadNext-64 (ReNexed, ReLaXd)DICEAgeing Chambers

Next-64 / ReLaXd / ReNexd

CO2 cooling!

DICE

DICE

Nuclear ReactorWater Bassin10 x 10 x 10 m3

ReactorInstituteDelftRID

Upgraded SCT: Gossip could replace:

- Pixel vertex detector: Gossip- Si Strip detectors: replace by Gossip Strixel detectors- TRT: use Gossip as tracker/TR X-ray detector

Essentials:

- power dissipation: 60 mW/cm2

- intrinsic mass: 0.1 % radiation length- low cost: 10 $ / cm2

Barrel SCT unit

EndCap SCT unit

Semiconductor (pixel, strip) Semiconductor (pixel, strip) detectorsdetectors

Depleted Si, 300 μm

(pixel) chip withpreamps, shapers,discriminators

Vbias = 150 V

electron-holepairs

ATLAS pixel: basic elementATLAS pixel: basic element

C-C support

sensor

Flex Hybrid

bumps

MCC Side view

not to scale

Wire-bonding FE’sWire-bonding MCC

FE chipFE chip

Flex module 2.xFlex module 2.x

- Ladder strings fixed to end cones- Integration of beam pipe, end cones & pixel vertex detector- 5 double layers seems feasible

Gossip in ATLAS (Goat-1) Stave TimePix-2 chipSiNProt layerInGrid (Si3N4)Gas Cover

‘G’(round)String

‘P’(ositive)Stringcarrying 1.3 V

‘Road’: C-fibre reinforced databus + aux services

Stiff, light Stave formed by G-stringP-stringRoad triangle

StainlessSteel tubefor CO2 cooling

casted aluminiumpower line

gasmanifold

Ø60mm Beampipe

Inner Layer: 7 double Goat strings

CO2 cooling channels

P-string conductor (+voltage)

G-string conductor (+voltage)

Gossip detector unit

Gossip readout

GOAT: GOssip in ATlas

Upgraded SCT: Gossip/GridPix could replace:

- Pixel vertex detector: Gossip- Si Strip detectors: replace by Gossip Strixel detectors- TRT: use GridPix as tracker/TR X-ray detector

strixels/strips

preamp channels

Essentials:

- power dissipation: 1/16 x 60 mW/cm2 = 4 mW/cm2

now:25 mW/cm2

- intrinsic mass: 0.1 % radiation length- low cost: 10 $ / cm2

~ 20 mm

Upgraded Tracker: Gossip could replace:

- Pixel vertex detector: Gossip- Si Strip detectors: replace by Gossip Strixel detectors- TRT: use Gossip with 17 mm Xe layer

as tracker/TR X-ray detector

Essential:- high position-resolution tracker throughout tracker- low mass, low cost detector- Efficient TRD possible

L=30 mm

0.05 mm

V0 V1

Anatoli Romaniouk, Serguei Morozov, Serguei KonovalovMartin Fransen, Fred Hartjes, Max Chefdeville, Victor Blanco Carballo

Transition Radiator

Testbeam Nov 5 – 12, 2007PS/T9: electrons and pions, 1 – 15 GeV/c

Samples pions (left) and electrons (right)

Particle Identification Particle Identification

6 GeV/c

5 (double) layer Gossip Pixel

4 layer Gossip Strixel

3 layers Gossip TRT

radiator

Gas instead of SiPro:- no radiation damage in sensor: gas is exchanged- modest pixel (analog) input circuitry: low power, little space- no bias current: simple input circuit- CMOS pixel chip main task: data storage & communication (rad hard)- low detector material budget: 0.06 % radiation length/layer

typical: Si foil. New mechanical concepts- low power dissipation : little FE power (2 μW/pixel); no bias dissipation- operates at room temperature (but other temperatures are OK)- less sensitive for neutron and X-ray background- 3D track info per layer if drift time is measured

Con:- Gaseous chamber: discharges (sparks): destroy CMOS chip- gas-filled proportional chamber: ‘chamber ageing’- Needs gas flow- Parallax error: 1 ns drift time measurement may be required

New mechanics + cooling concepts for Gossip

- As little as possible material- detector consists of foil!- less power required ( less cooling) w.r.t. Si

string: power, chip support, cooling

in 2030….

‘balloon’

‘laundry line’

Minimum Material Budget(% rad length)

Z = 0 mm Z = +/-600 mm

Gossip detector (50 μm Si) 0.06 0.06

Cooling (stainless steel tube) 0.001 0.001

Power (max 0.28 mm aluminium) 0.0 0.3

Data transfer (max 1.7 mm kapton)0.0 0.6

total 0.06 1

angle correction x √2 0.09 x 2 x √2 3

- Ladder strings fixed to end cones- Integration of beam pipe, end cones & pixel vertex detector- 5 double layers seems feasible

Virtual goal: ATLAS pixel vertex

Gossip in ATLAS (Goat-1) Stave TimePix-2 chipSiNProt layerInGrid (Si3N4)Gas Cover

‘G’(round)String

‘P’(ositive)Stringcarrying 1.3 V

‘Road’: C-fibre reinforced databus + aux services

Stiff, light Stave formed by G-stringP-stringRoad triangle

StainlessSteel tubefor CO2 cooling

casted aluminiumpower line

gasmanifold

Ø60mm Beampipe

Inner Layer: 7 double Goat strings

CO2 cooling channels

P-string conductor (+voltage)

G-string conductor (+voltage)

Gossip detector unit

Gossip readout

GOAT: GOssip in ATlas

The future:

Electron Emission Foil

MEMS made MicroChannelPlates: 200 ps time resolution: CLIC

electron emission foil

CMOS pixel chip

electron avalanche in gasEE-Foil replaces InGridParallel Plate Chamber

electron emission foil

CMOS pixel chip

replace gas by vacuumMicro Channel Platesub-ns time resolutionNote CLIC experiments

July 2005: ATLAS Upgrade Worlshop GenuaContribution Gossip was refused:

……………..pixel B-layer replacement too soon for new R&D projects…..

Sept 2006: ATLAS Upgrade Workshop Liverpool:Contribution Gossip was granted (after repeated requests)

Dec 2007: ATLAS Upgrade Workshop Valencia:Invited to present Gossip R&D project

Sept 2008: PSD8, Glasgow:keynote: Norbert Wermes: Pixel detectors for charges particlesmentions 3D, Diamond, MAPs, but NOT Gossip

Si tracking detectors

- the ultimate detector in HEP- associated with Moore’s Law, chip technology

Proposal:

(imaginary) separation of

Interaction Matter and

readout electronics

Integration (MAPs) maybe profitable later

Si (vertex) track detector GOSSIP

CMOS chip

Si depletion layer

Vbias

• Si strip detectors• Si pixel detectors• MAPs

Gas: 1 mm as detection medium99 % chance to have at least 1 e-

Gas amplification ~ 1000:

Single electron sensitive

All signals arrive within 16 ns

Cluster3

Cathode (drift) plane

Integrated Grid (InGrid)

Cluster2

Cluster1

Slimmed Silicon Readout chipInput pixel

1mm,100V

50um, 400V

50um

NIKHEFNIKHEFHarry van der Graaf, Max Chefdeville, Fred Hartjes, Jan Timmermans, Harry van der Graaf, Max Chefdeville, Fred Hartjes, Jan Timmermans, Jan Visschers, Marten Bosma, Martin Fransen, Yevgen Bilevych,Jan Visschers, Marten Bosma, Martin Fransen, Yevgen Bilevych,Wim Gotink, Joop RovekampWim Gotink, Joop Rovekamp

University of TwenteUniversity of TwenteCora Salm, Joost Melai, Jurriaan Schmitz, Sander Smits,Cora Salm, Joost Melai, Jurriaan Schmitz, Sander Smits,Victor Blanco CarballoVictor Blanco Carballo

University of NijmegenUniversity of Nijmegen Michael Rogers, Thei Wijnen, Adriaan Konig, Jan Dijkema,Michael Rogers, Thei Wijnen, Adriaan Konig, Jan Dijkema,

Nicolo de GrootNicolo de Groot

CEA/DAPNIA SaclayCEA/DAPNIA SaclayD. Attié, P. Colas, I. GiomatarisD. Attié, P. Colas, I. Giomataris

CERNCERNM. Campbell, X. LlopartM. Campbell, X. Llopart

University of Neuchatel/MTIUniversity of Neuchatel/MTINicolas WyrschNicolas Wyrsch

Czech Tech. Univ. Prague, PrahaCzech Tech. Univ. Prague, PrahaPixelman: T. Holy et al.Pixelman: T. Holy et al.