Microchannel plate pixel detectors

MICROCHANNEL PLATE PIXEL DETECTORS

Looking at MCPs in a RICH Upgrade contextRichard Plackett – CERNMany thanks to colleagues from VELO, CERN Medipix and especially John Vallerga, UC Berkeley

-2-

Overview• Context

• 40Mhz readout Chip

• MCP Phototube Principle of Operation

• Photon Efficiency

• Advantages and Disadvantages

-3-

Context - Why Me?• I am currently working on proving a Medipix3/Timepix derived

hybrid pixel sensor concept for the VELO upgrade

• This chip would probably fit very nicely into an HPD….

• John Vallerga is working on a Timepix based MCP phototube as part of Medipix2/Medipix3 collaborations

• Having worked on the RICH, I thought you would like to hear about them as they have some nice (and some nasty) features both as a main RICH photon detector and possibly as a fast detector for the TORCH.

-4-

Proposed 40MHz Chip• LHCBPIX2? – currently many names

• Most probably – 55um square pixels – 256 by 256 matrix (14mm square)– 40MHz readout– On chip zero suppression– 4 side tiling with through silicon via technology – Worst case 800um inactive periphery on one side– On matrix cluster analysis a la Medipix3

• Still in design phase so some flexibility

• Well defined development path through Medipix3 and Timepix2, so there will be ‘prototypes’ available early.

Timepix readout chip

-5-

MCP Photon Detector

Quartz window and photocathode as HPD and MAPMT

200V/mm drift field

MCP cascade amplification (~1kV)

Bare readout chip array (no bump bonds)

Ceramic carrier with possible cooling built in

Photon

Photoelectron

Electron shower

0.5mm

0.5mm

Proximity focused MCP tube with 55um pixel readout

Tuning the lower drift field allows the electron shower profile to be well controlled

10um pores in an MCP

Cascade amplification similar in principle to a PMT

-6-

Photon EfficiencySimilar to current RICH HPD *IF* packing fraction is high enough

Effect HPD MCP

Photocathode ~33% ~33%

Packing Fraction ~67% ~85%

MCP acceptance ~65%

Detection efficiency ~85% ~100%

A square device is possible….85% is an 5mm gap round a 4 by 4 chip active matrix

And indeed so are COMPASS RICH / MaPMT style lenses

-7-

MCP Advantages• High Magnetic Tolerance, could operate in RICH1 (600G)

without ANY shielding – no flat mirrors?

• Pixel readout chip eliminating crosstalk, photon counting, working at 40MHz etc

• A fast (~100ps) signal could also be produced with a secondary readout system to allow ring time separation

• No ‘first dynode’ noise from capillary so 0 to 1 photons discrimination relatively easy

• ‘Chevron’ style MCPs are resistant to ion feedback

• Essentially a flat panel detector few cm deep detectors.

• Only requires ~1kV supply

-8-

Magnetic Field ToleranceMagnetic field tolerance comes primarily from the short flight path of photoelectrons and electron cascade and resultant high electric fields

“The present Burle 85011 MCP with 25 um pores works well in fields up to 0.8 T (NIM A 567 (2006) 124–128 )”

This is an order of magnitude better than we currently need (0.06T unshielded in RICH1).

The design of the Berkley groups photon counting MCP And a photo of the same

-9-

Some images from John’s tube

Images Presented at SPIE 2008 in Marseille

-10-

Fast Signal Readout• Intrinsically MCPs are fast photon detectors operating around 50-100ps

• Incorporating this into the pixel chip would incur a big power penalty, which would affect the design of the tube (cooling etc).

• A whole tube readout based on MCP current could be implemented in addition to pixel chip possibly using something like the gridpix technology developed by Harry van der Graaf to produce larger timing pixels

Taken from a presentation to the ATLAS Tracker Upgrade Workshop, Valencia, Dec 12, 2007

Here a grid is mounted over a Timepix readout chip for the InGrid project

-11-

MCP Disadvantages• Not an off-the-shelf system, but could be close, relatively

simple encapsulation with no bump bonds. John Vallerga is currently working with Hamamatsu to incorporate Timepix chips into their MCPs.

• To instrument same area as current RICHs gives far too many channels, although this compensates for limited chip occupancy and helps dead time. With a proximity focusing system 4m2 gives ~1billion channels, eg a 4 by 4 chip tube would require ~500 tubes

• MCP capillaries have a ‘recharge’ time, but this is mitigated by the overabundance of pixels and capillaries.

• Ageing (next slide)

-12-

MCP Ageing

Lifetime of the MCP coating is measured as extracted charge Approx lifetime is 1 Coulomb/cm2.  Assume a gain of 10000e- per event.  With 6x1014 events/cm2 as a lifetime, Gives18 billion events per 55 micron pixel

Assuming 150 tracks per event and 25 photoelectrons per track and 4 square meters instrumented gives a 10 year lifetime

(ok so I assumed the Cherenkov light was uniformly distributed too)

The plate itself is made from coated lead glass. The electron cascade amplification ablates away the coating giving an ageing effect.

-13-

Summary• MCP phototubes have the potential to perform with similar

photon efficiency as current HPDs but with very high tolerance to magnetic fields, reduced volume and HV requirements.

• 40MHz pixel chip being considered for VELO upgrade (Timepix or FPIX) and may allow common readout systems and development cooperation.

• A fast readout scheme could provide additional information for ring finding etc in the RICH… and even be used for the TORCH.

For more information on John Vallerga’s hybrid pixel MCP tubes…

NIM A 567 (2006) 110-113 ‘ A noiseless kilohertz frame rate imaging detector based on microchannel plates read out with the Medipix2 CMOS pixel chip’

Astrophysics and Space Science (2008), p98  ‘The current and future possibilities of MCP based UV detectors'

-14-

-15-

Backup – Dark noise

(radium dial)

Bkgd

.002 ct/pxl/s

Room Temp

Taken from John’s presentation at SPIE 2008

This gives 5x10-11 hits per pixel per event

One noise hit in the entire detector every 20 interactions