Transcript The Development of Large-Area Thin Planar Psec Photodetectors Henry Frisch, Enrico Fermi Institute and ANL .075” ~150 20m pores INCOM glass substrate SSL SSL 11/7/2015 Herve Grabas.

The Development of Large-Area
Thin Planar Psec Photodetectors
Henry Frisch,
Enrico Fermi Institute and ANL
.075”
~150 20m pores
INCOM glass
substrate
SSL
SSL
11/7/2015
Herve Grabas
1
The Development of Large-Area Thin Planar
Psec Photodetectors
ABSTRACT
Fast, cheap and large photo-detectors1 with subnanosecond time resolution and sub-mm space resolution
would have valuable uses in high-energy physics collider
detectors, large neutrino detectors, photon
reconstruction in precision kaon and b-physics
experiments, astroparticle physics, positron-emission
tomography, and truck/container scanners, for example.
However `fast, cheap, large‘ make one think of the old engineering
adage: `Faster, Cheaper, Better- Pick Any Two'. I will describe an
effort by a group of materials scientists, chemists, surface chemists,
X-ray physicists, astrophysicists, mechanical and electrical engineers,
and a handful of high-energy physicists to achieve all three.
•We have just begun- funds from DOE arrived last August3 year project ( falls in `high risk’ category of Steve Chu)
1With
no irony intended, NASA adopted Faster Cheaper Better
as a slogan- nobody had the nerve to tell the management the
last three words.
11/7/2015
EFI Faculty Lunch, April 2010
2
The Large-Area Psec Photo-detector Collaboration
4 National Labs, 5 Divisions at
Argonne, 3 US small
companies; electronics
expertise at Universities of
Chicago and Hawaii
Goal of 3-year R&Dcommercializable modules.
11/7/2015
EFI Faculty Lunch, April 2010
3
Motivation
and
Requirements
11/7/2015
EFI Faculty Lunch, April 2010
4
Parallel Efforts on Specific Applications
PET
.
Explicit strategy for staying on task
(UC/BSD,
UCB, Lyon)
Collider
(UC,
ANL,SLAC,..
LAPD Detector
Development
ANL,Arradiance,Chicago,Fermilab,
Hawaii,Muons,Inc,SLAC,SSL/UCB,
Synkera, U. Wash.
DUSEL
K->pnn
(Matt, Mayly,
Bob, John, ..)
Drawing Not To Scale (!)
11/7/2015
(UC(?))
Security
(TBD)
EFI Faculty Lunch, April 2010
All these need work- naturally
tend to lag the reality of the
5
detector development
Application 1-Colliders
At colliders we measure the 3-momenta of hadrons, but can’t follow
the flavor-flow of quarks, the primary objects that are colliding. 2orders-of-magnitude in time resolution would all us to measure ALL
the information=>greatly enhanced discovery potential.
t-tbar -> W+bW-bbar->
e+ nu+c+sbar+b+bbar
A real top candidate
event from CDF- has
top, antitop, each
decaying into a Wboson and a b or
antib. Goal- identify
the quarks that
make the jets.
(explain why…)
Specs:
Signal: 50-10,000
photons
Space resolution: 1 mm
Time resolution 1 psec
Cost: <100K$/m2:
Application 2- Neutrino Physics
(Howard Nicholson)


Example- DUSEL detector with 100% coverage and 3D photon
vertex reconstruction.
Many questions that need work (Sacha!) e.g:


11/7/2015
with 100 psec (1” in z) and 1-cm (in x-y) resolutions, can one separate
pizeros from electrons at the vertex (rad length ~40 cm)
In the parameter space of space resolution, time resolution, quantum
efficiency (QE), and coverage, how do fiducial volume, aspect ratio, cavern
costs, and total volume behave? Simulation effort has begun (M. Wetstein, J.
Felde, …)
EFI Faculty Lunch, April 2010
7
Neutrino Physics- cont.
(Howard Nicholson)
Range of performance is between liquid argon at 20KTons and
Super-K copy at 200 KTons: potential big savings for a given
performance for electron appearance. (but proton decay?)
 Need >10,000 square meters (!) (100 ps resolution)- this is driving
our emphasis on low cost and very large area for the glass effort.
In a parallel effort, at Space Sciences Lab, Ossy Siegmund’s group
is adapting their expertise in ceramic MCP-PMT’s to the larger 8”
size. May be more expensive, but they have long experience and a
wonderful track record.
 Simulation is critical- still in its infancy. We don’t know the
parameter space yet.
 Spec: signal single photon, 100 ps time, 1 cm space, low cost/m2
8
11/7/2015
EFI Faculty Lunch, April 2010
(5-10K$/m2)*

Application 3- Medical Imaging (PET)
TOF adds 3rd dimension to Positron-Emission Tomography
Slides from Bill Moses’s
talk at the Clermont
Workshop
(see our library page, under
workshops:
http:hep.uchicago.edu/psec)
11/7/2015
EFI Faculty Lunch, April 2010
9
Sampling Calorimetry in PET?
Alternating
radiator and
cheap 3050 psec
planar mcppmt’s on
each side
Can we solve the
depth-ofinteraction
problem and also
use cheaper
faster
radiators?
Simulations by Heejong
Kim (Chicago)
Heejong Kim
Heejong Kim
11/7/2015
Depth in crystal by timeDepth in crystal by
difference
EFI Faculty Lunch, April 2010energy- asymmetry
10
Medical Imaging (PET)-cont.
Spec: signal 10,000 photons,30 ps time resolution ,
1 mm space resolution, 30K$/m2, and
commercializable for clinical use.
SUMMARY
However- truth in advertising- there is a long way to go here
(see Bill’s talk at Clermont.) It looks promising, as it may be
possible to produce large panels with better spatial and time
resolution than possible with photomultipliers, and our initial
estimates are that MCP-PMT’s may be as much as a factor of
10 cheaper. But these may not be the limiting factors on
performance- question is whether there is another, new,
optimum design with cheaper faster radiator, cheap large
photodetector panels, and (possibly) separate energy and space
readouts. This takes a real effort on measurements and
simulation- talks are underway with Clermont, Strasbourg,
Lyon, and Chicago. (see papers by Heejong Kim et al on web).
11/7/2015
EFI Faculty Lunch, April 2010
11
Application 4Cherenkov-sensitive Sampling Calorimeters
Idea: planes on one side
read
•I both Cherenkov and
scintillation light- on other
only scintillation.
A picture of an em shower A `cartoon’ of a fixed target geometry such as for
in a cloud-chamber with
JPARC’s KL-> pizero nunubar (at UC, Yao Wah) or
½” Pb plates (Rossi,
LHCb
p215- from CY Chao)
11/7/2015
EFI Faculty Lunch, April 2010
12
Application 5- Nuclear Non-proliferation
1. MCP’s loaded with Boron or Gadolinium are used
as neutron detectors with good gamma separation
(Nova Scientific).
2. Large-area means could scan trucks, containers
3. Time resolution corresponds to space resolution
out of the detector plane IF one has a t_0– i.e can
do 3D tomography of objects
Specs: Unknown
An area for possible applications- need a
counterpart to form an application group.
(ANL an obvious place)
11/7/2015
EFI Faculty Lunch, April 2010
13
SUMMARY
Characteristics in common we need
 Small feature size << 300 microns
 Homogeneity – the ability to make
uniform large-areas (think solarpanels, floor tiles, 50”-HDTV sets)
 Intrinsic low cost: although
application specific, all need lowcost materials and robust batch
fabrication. Need to be simple.
11/7/2015
EFI Faculty Lunch, April 2010
14
Using New
Technologies
to Exploit
Fundamentally
Simple Ideas
11/7/2015
EFI Faculty Lunch, April 2010
15
Detector Development- 3 Prongs
MCP development- use modern fabrication processes to
control emissivities, resistivities, out-gassing
Use Atomic Layer Deposition for emissive material
(amplification) on cheap inert substrates (glass capillary arrays,
AAO). Scalable to large sizes; economical; pure – i.e. chemically
robust and (it seems- see below) stable
Readout: Use transmission lines and modern chip
technologies for high speed cheap low-power highdensity readout.
Anode is a 50-ohm stripline. Scalable up to many feet in length ;
readout 2 ends; CMOS sampling onto capacitors- fast, cheap,
low-power (New idea- make MCP-PMT tiles on single PC-card
readout- see below)
Use computational advances -simulation as basis for
design
Modern computing tools allow simulation at level of basic
processes- validate with data. Use for `rational design’
(Klaus Attenkofer’s phrase).
11/7/2015
EFI Faculty Lunch, April 2010
16
Micro-channel Plates PMTs
Satisfies small feature size and homogeneity
Photon and electron paths are short- few mm to
microns=>fast, uniform Planar geometry=>scalable to
large areas
11/7/2015
EFI Faculty Lunch, April 2010
17
Simplifying MCP Construction
Conventional Pb-glass MCP
Incom Glass Substrate
NEW
OLD
Chemically produced and treated
Pb-glass does 3-functions:
1.
Provide pores
2.
Resistive layer supplies
electric field in the pore
3.
Pb-oxide layer provides
secondary electron emission
11/7/2015
Separate the three functions:
1. Hard glass substrate provides
pores;
2. Tuned Resistive Layer (ALD)
provides current for electric
field (possible NTC?);
3. Specific Emitting layer
provides SEE
EFI Faculty Lunch, April 2010
18
Where we are with glass substrates
Hexagonal
bundle of
capillaries is
called a
`multi’. Each
multi has
~15,000
capillaries
Many many
multis in an
8”-square
plate.
.075”
~150 20m pores
INCOM glass
substrate
Incom, Inc
Charlton, MA


Have received multiple samples of 10-micron, 20-micron, 40-micron glass
substrates from Incom in 3/4”-sq and 33 mm round formats – will show results
after ALD below
Two developments at Incom (our glass folks)- 1) 8” plates are being fabricated
and the process improved, and 2) replacement of some multis with solid
islands (`pads’) for installation of mechanical spacers. Idea is low cost
amplification section - so far so good (hesitate to quote a # yet).
19
11/7/2015
EFI Faculty Lunch, April 2010
Anodic Aluminum Oxide (AAO) Substrates
Seon Woo Lee, Hau Wang, Dmitry Routkevich (Synkera)
Alternative to glass MCP
• Possible future generation
40μm spacing
•
•
•
5μm funnel
(look close)
of MCP with Photocathode
coated on funnel via ALD
Parallel efforts at Synkera and
ANL to produce an MCP plate
with current ALD
Glass may still win, but some
unique characteristics
Potential for funnel-opening
to get large effective open
area ratio
Large Area Photodetector Development Collaboration
11/7/2015
EFI Faculty Lunch, April 2010
20
20
New MCP
Structure
(not
to
pore
scale)
1 KV
1) resistive coating (ALD)
2) emissive coating (ALD)
3) conductive coating (thermal
evaporation or sputtering)
Jeff Elam
21
11/7/2015
EFI Faculty Lunch, April 2010
21
Atomic Layer Deposition (ALD) Thin
Film Coating Technology
•Lots of possible
materials => much room
for higher performance
 Atomic level thickness control
 Deposit nearly any material
 Precise coatings on 3-D objects
(JE)
Jeff Elam pictures
11/7/2015
EFI Faculty Lunch, April 2010
22
ALD for Emissive Coating
Conventional MCP’s:
Alternative ALD Coatings: (ALD SiO2 also)
 Many material possibilities
 Tune SEE along pore (HF- possible
discrete dynode structure (speed!)
11/7/2015
EFI Faculty Lunch, April 2010
Jeff Elam ,
Zeke Insepov,
Slade Jokela
23
23
Samples S1 and S1W (Al2O3)
•Exceptionally strong characterization group in the Argonne Materials
Science Division: Igor Veryovkin, Slade Jokela, Alex Zinovev, Thomas Proslier


Each scan takes 4min
20s with 30s between
scans
Comparable yield to
Arradiance samples
Slight decrease in yield
over time is observed
•From Slade Jokelaexample of new
measurements from
their new dedicated
characterization
facility
•Closely tied to
simulation of Zeke
Insepov and
Valentin Ivanov
4.0
Secondary Electron Yield (per primary)

Al2O3 Samples
3.5
S1W
3.0
1st Scan
2nd Scan
5th Scan
2.5
2.0
S1
1st Scan
2nd Scan
5th Scan
1.5
1.0
0.5
0.0
0
200
400
600
Primary Electron Energy (eV)
800
ALD-Functionalized substrates
Picture
is seam
between
blocks
Jeff Elam, Thomas Prolier (ESD)
11/7/2015
EFI Faculty Lunch, April 2010
25
MCP and Photocathode Testing
Jeff Elam, Anil Mane, Qing Peng, Neal Sullivan
(Arradiance), Bernhard Adams, Matt Wetstein, Slade
Jokela, Igor Veryovkin, Alex Zinovev,, Ossy Siegmund
SSL
basic scrub period

ALD allows separate control of resistive and
emissive layers
separately optimize each layer for best overall
performance
Precise control over composition;
tunable resistance




40μm pore, L/D=40 MCP
Arradiance ALD coating
Functionalized in H-furnace requiring long
“scrubbing” time (removal of volatiles)


No Vacuum Bake
Conventional lead-oxide MCPs have single
composition for resistive/emissive material
Arradiance coatings on Incom plate- Scrub
time reduced by up to ×10 (!) (SSL)
Have functionalized several pairs with
newly developed resistive layer plus Al2O3
secondary emissive layer (ANL)
image credit: J. McPhate
ANL
Signal from MCP pair coated
with
new
image credit:
M. Wetstein
resistive layer Al2O3 emissive layer
Large Area Photodetector Development Collaboration
11/7/2015
EFI Faculty Lunch, April 2010
26
26
MCP and Photocathode Testing
Testing Group: Bernhard Adams, Matthieu Cholet, and Matt Wetstein at the APS, Ossy
Siegmund’s group at SSL
N. B.!
LAPPD
Preliminary
(very)
First measurements of gain in an ALD SEE layer at the APS laser
test setup (Bernhard Adams, Matthieu Cholet, and Matt Wetstein)
11/7/2015
EFI Faculty Lunch, April 2010
27
First-ever test of an ALD pair (Ossy,
SSL)
Note- at
high gain
the
boundaries
of the
multi’s go
away
Electron pattern (not a picture of the plate!)- SSL test, Incom
substrate, Arradiance ALD. Note you can see the multi’s in
both plates => ~50 micron resolution
28
11/7/2015
EFI Faculty Lunch, April 2010
Simulation (crosses all groups)
Valentin Ivanov, Zeke Insepov, Zeke Yusof, Sergey Antipov
10μm pore
40μm spacing
Funnel
(!)
Large Area Photodetector Development Collaboration
11/7/2015
EFI Faculty Lunch, April 2010
29
29
Advanced Photocathode Group
This is the hardest intellectual task
Klaus Attenkofer, Sasha Paramonov, Zikri Yusof




III-V have the potential for high
QE, shifting toward the blue, and
robustness i.e. they age well, hightemp)
Opaque PC’s have much higher QE
than transmission PC’s- we have
the geometry
Many small factors to be gained in
absorption, anti-reflection- see
papers by Townsend and talk by
Fontaine on our web site
Quantum Effic. Of 60% have been
achieved in bialkalis
Big payoff if we can get >60% QE robust
Photocathodes, and/or more robust (assembly). Also want to
30
get away from `cooking
to rational design.
11/7/2015
EFI Facultyrecipes’
Lunch, April 2010
SSL Photocathodes: Large Process
Chamber – our backup (Ossy Siegmund)
UV Transmissive
Window
Manipulators
Glass Window
18” ID Chamber
UHV valves
Photo-Cathode
Forming Well Flange
16.5” Detector
Loading Flange
Ion Pump supply
Ion Pumps
11/7/2015
EFI Faculty Lunch, April 2010
31 31
Readout- Transmission Line Anodes
Anode Design and Simulation(Fukun Tang)
•Our first
prototypes
were matched
to Burle 2”square
Planacons
with 1024
anode padsthis is Tang’s
card


Transmission Line- readout both ends=> pos and time
Cover large areas with much reduced channel account.
11/7/2015
EFI Faculty Lunch, April 2010
32
Photonis Planicon on Transmission Line Board
Couple 1024 pads to strip-lines with silver-loaded epoxy (Greg
Sellberg, Fermilab).
11/7/2015
EFI Faculty Lunch, April 2010
33
Comparison of measurements (Ed May and JeanFrancois Genat and simulation (Fukun Tang)


Transmission Line- simulation shows 3.5GHz
bandwidth- 100 psec rise (well-matched to MCP)
Measurements in Bld362 laser teststand match
velocity and time/space resolution very well
11/7/2015
EFI Faculty Lunch, April 2010
34
Scaling Performance to Large Area
Anode Simulation(Fukun Tang)
48-inch Transmission Line- simulation shows 1.1 GHz
bandwidth- still better than present electronics.
KEY POINT- READOUT FOR A 4-FOOT-WIDE
DETECTOR IS THE SAME AS FOR A LITTLE ONEHAS POTENTIAL…

11/7/2015
EFI Faculty Lunch, April 2010
35
ANL Test-stand Measurements
Jean-Francois Genat, Ed May, Eugene Yurtsev
Sample both ends of transmission line
with Photonis MCP (not optimum)
2 picoseconds; 100 microns measured
11/7/2015
EFI Faculty Lunch, April 2010
36
Put it all together- the `Frugal’ MCP


Put all ingredients
together- flat glass
case (think TV’s),
capillary/ALD
amplification,
transmission line
anodes, waveform
sampling
Glass is cheap, and
they make vacuum
tubes out of itwhy not MCP’s?
11/7/2015
EFI Faculty Lunch, April 2010
37
Mechanical Assembly
Rich Northrop, Dean Walters, Joe Gregar Bob Wagner, Michael Minot,
Jason McPhate, Ossy Siegmund
8” proto-type stack
Design sketch
11/7/2015
8” proto-type mock-up
EFI Faculty Lunch, April 2010
38
Glass Hermetic Packaging Group
Basic `proof-of-concept’ module- 8” by 8” MCP
Model assembly of pre-production real parts:
glass/silver anode, glass sidewall, glass window
 Bottom-seal trials at ANL (Joe Gregar) and Minotech Eng.looks promising.
 Top-seal trials at ANL; promising--metallurgists at UIC and
ANL working on understanding why Ferenc and Ossy recipes
work.
39
11/7/2015
EFI Faculty Lunch, April 2010

The Inside-Out Solution- can we
ground all the anode traces together?
The evolution of an idea



11/7/2015
Would like a simple surface
for the bottom seal- led to
the idea that since the
signals are differential one
can put the ground INSIDE
Andy Axtel (Ferro Corp)
silk-screened some anodesfirst ones were singlesided. We put one on the
top the other to get a pair
of transmission line stripsi.e. the metal was on two
separate planes
Then Herve’ Grabas had the
sweet idea that the bottom
strips could be on a PC
card, making the connection
to the front-end chips high
analog band-width
EFI Faculty Lunch, April 2010
40
High (multi-GHz) ABW readout
New Idea (Herve’ Grabas)-Put bottom
traced on PC-card (not on the glass).
Note signal is differential between ground
(inside, top), and PC traces (outside)
11/7/2015
EFI Faculty Lunch, April 2010
41
Microstrip to Coplanar E-Field transition 42
Herve Grabas
?
Simplest
More complicated
0.15 millimeters
3.76 millimeters
50
Ohms
50
Ohms
Many configurations are possible from very simple to complicated. Simulation
and test can only tell which one will be the best.
11/7/2015
EFI Faculty Lunch, April 2010
42
High (multi-GHz) ABW readout- cont.
Led to another new
idea- multiple
standard 8” glass
boxes `tile’ a PC
card-modular
construction.
• PC card has the anode traces and electronics and can be tested and
serviced separately- only the simplest of electrical connections between
the tray and the tiles.
•Allows a `factory’ for mass-production 8” square modules that are then
used to make application-specific larger super-modules (DUSEL, PET).
11/7/2015
EFI Faculty Lunch, April 2010
43
Front-end Electronics/Readout
Waveform sampling ASIC
Electronics Group: Jean-Francois Genat, Gary Varner, Mircea Bogdan, Michael Baumer,
Michael Cooney, Zhongtian Dai, Herve Grabas, Mary Heintz, James Kennedy, Sam
Meehan, Kurtis Nishimura, Eric Oberla, Larry Ruckman, Fukun Tang
First have to understand signal and noise in the
frequency domain
A typical MCP
signal (Planacon)
Frequency spectra of signal and noise
(JF Genat)
11/7/2015
EFI Faculty Lunch, April 2010
44
Time Resolution- few psec?
Resolution depends on 3 parameters: Number of
PhotoElectrons, Analog Bandwidth, and Signal-to-Noise
See J-F Genat, G.
Varner, F. Tang, and
HF
arXiv: 0810.5590v1
(Oct. 2008)- now
published in Nucl.
Instr. Meth.
Wave-form sampling is best, and can be implemented in low-power widely
available CMOS processes (e.g. IBM 8RF). Low cost per channel.
11/7/2015
EFI Faculty Lunch, April 2010
45
Front-end Electronics/Readout
Waveform sampling ASIC prototype

Varner, Ritt, DeLanges, and Breton have pioneered waveform–
sampling onto an array of CMOS capacitors.
11/7/2015
EFI Faculty Lunch, April 2010
46
Electronics Group
J.F. Genat, Gary VarnerHerve Grabas, Eric Oberla, Larry Ruckman,
Kurtis Nishimura

11/7/2015
text
EFI Faculty Lunch, April 2010
47
Second frontend chip- at MOSIS
Eric Oberla, Herve Grabas
(lots of help from Larry and
Gary)
2nd prototype- fix problems,
improve performance, add
features
One 2nd-yr goal is to use
these chips in our MCP
test facilities instead of
expensive scopes
11/7/2015
EFI Faculty Lunch, April 2010
48
Next Year’s Goals
(draft- may be optimistic)






Hermetic 8” assemblies in ceramic and glass
Demonstration of 8”-ALD-functionalized glasscapillary chevron with gain > 106
Bialkali photocathodes at SSL; some bialkali
capability at ANL
Photocathode characterization facility at ANL
(hope joint with APS, MSD, BNL, …)
Integrated chip/transmission-line/readout tray for
multiple 8”tile assembly
Decision on where to make production facility.
11/7/2015
EFI Faculty Lunch, April 2010
49
The End-
11/7/2015
EFI Faculty Lunch, April 2010
50
ANL-UC Glass Hermetic Packaging Group

Proceed in 3 steps: 1) hermetic box; 2)
Add MCP’s, readout, (Au cathode); 3)
Add photocathode
Box
Box+ 8” MCPs
Possible Au anode
Box+MCP+PC
Yr 1
11/7/2015
Yr 2
EFI Faculty Lunch, April 2010
Yr 3
51
SSL Tube Processing Facilities
Sealed tube facilities and oven
UHV detector/cathode
processing station
SSL Sealed tube detectors Pre-process assembly
Planacon, with fiber optic window and
cross strip anode (signal vias straight
through substrate), in assembly with MCPs
installed (above) ready to process.
SSL: Alkali Photocathodes
•Jean-Francois Genat
Emission spectrum of Cherenkov in
water compared with bialkali response.
11/7/2015
UCB SSL cathode compared
with commercial product.
EFI Faculty Lunch, April 2010
54