Light Collection Efficiency and Uniformity of Light Guides for the sPHENIXElectromagnetic Calorimeter
S. Stoll, D. Cacace, J. Huang, Z. Parsons, Z.Shi, T. Shimek, C. Woody
sPHENIX EMCalorimeterProposed for the sPHENIX Experiment at RHIC, Brookhaven Nat. Lab. Upton, NY. USA
2
7.5 cm
readout
26 cm total radial space
~14 cm absorber
(h=0)
Light Guides 24,576 total
1 Sector = 96 Blocks
= 384 towers
= 384 light guides
EMCAL Design Specs:
• Coverage: ± 1.1 in h, 2p in f
• Segmentation: Δη x Δφ ≈ 0.025 x 0.025
• Readout channels: 96x256 = 24576 (towers)
• Energy Resolution: sE/E < 15%/√E
• Provide an e/h separation > 100:1
• Approximately projective
• Compact (in order to fit inside Babar solenoid)
• Works inside a 1.5T magnetic field
1 block = 4 towersPhysics Contributions:
• Jet measurements (EM component)
• Photon measurements
• Y measurement
Spacial constraints
• The tungsten powder scintillating fiber calorimeter (SPACAL)
technology was developed by Oleg Tsai. We started with his basic
machined trapezoidal light guide design, and then tuned and optimized
it to fit our projective calorimeter design.
• We used TracePro Ray tracing software to simulate light collection in
various geometries, then produced samples of some of the designs that
gave good results.
• Lab measurements were made using a pulsed LED and a single fiber
scanned across the light guide surface to validate the ray tracing model.
• Some of the designs were produced and installed on detector
prototypes, and then evaluated in beam tests.
• We also explored other options such as diffuse reflecting cavities
• Machining and polishing ~25,000 pieces is cost prohibitive, so we
explored other manufacturing options and designs, such as injection
molding, and a “quad” light guide assembly that would combine 4 tower
light guides into a single manufactured unit. Our detector absorber
blocks are manufactured in 4-tower blocks, so a 4 tower light guide
design is attractive.
3
Overview
Light Guide Design Constraints• Needs to fit in radial space ~ 25mm
• 24,576 towers - Each tower area has ~24mm x 25mm trapezoidal shaped readout face. (22 variants)
• Readout sensors - Hamamatsu S12572-015P MPPCs each have 3mm x 3mm active area. Four SiPMs per tower, 36mm2
total active area. Optically coupled to the lightguide (Momentive RTV615 Refractive Index: 1.406)
• The SiPM pcb is mechanically attached to the LG by a screw in the center of each tower, requiring a tapped #2-56 hole
• The readout surface of each tower is tungsten powder-epoxy composite, with 667 scintillating fibers embedded in it. The fibers are 0.47mm diam. and are spaced 1mm center-center. They emit in the 420-450nm range.
• The lightguide will be epoxied to the surface (BC600 Optical Epoxy).
• So, we need to read out a 506 mm2 area surface with 36 mm2 of sensors.
• The light guide also needs to be as efficient as possible, and also uniform across the entire area.
• For the calorimeter overall, we also need to minimize the effect of gaps between towers, between blocks, and between sectors.
4
Block-block boundaries
Fiber ends in
tungsten powder-epoxy
compositeAdd sipm
photo
SiPM dimensions – 4 per tower
TracePro Raytracing simulation
5
• To simulate the light collection behavior, CAD models of light
guides were evaluated with ray tracing simulations.
• A single fiber, modeled after those in the absorber blocks, was
used to input light into the modeled light guides. The single fiber
was scanned across the input face of the light guide model in
discrete steps.
• The light was then “collected” on the readout surface. The
efficiency is the ratio of rays incident on the sipm array (or pmt) /
rays entering the light guide.
Irradiance map showing distribution of
incident rays on the SiPM array
Ray tracing diagram shows incoming and reflected
light rays
SiPM array
planeInput fiber
• Fiber Diameter 0.47mm
• Core Material – Polystyrene with index: 1.597
• Cladding Material – PMMA with index: 1.494
• 420nm Light Source
• 158mm Length
• there is an absorbing coating, on the outer surface of the fiber cladding to simulate the effect of epoxy in the block to absorb the cladding light.
• Where there is a reflector, this is referring to a perfect mirror placed on the outer surface of the far end of the fiber, core and cladding.
Fiber Parameters
for simulation
Far
End
Near
End
Fiber
Cladding – Pink
Core – Blue
Light Source inside
fiber.
Light
Guide
3
With Coating and With a Reflector
Polar Angle (Degrees)
0 5 10 15 20 25 30 35 40
Co
un
t
2.5
7.5
12.5
17.5
22.5
0.0
5.0
10.0
15.0
20.0
25.0
6
Distribution of Emitted light from fiber:
Measured data:
Uniformity scan of machined
trapezoidal light guide. SiPM and
PMT readout and TracePro
simulation data.
7
“Stubby” 25mm trapezoidal Light Guide
Edge/center: 0.94 Edge/center: 0.87
25mm
machined
Acrylic LG50mm
machined
Acrylic LG
8
Comparison of Trapezoidal light guides – 50mm and 25mm on pmt
The longer, 50mm light guides had a flatter more uniform response. The shorter 25mm light guides
Had more “roll-off” near the edges. This was also consistent with the Testbeam data.
position (mm)
-30 -20 -10 0 10 20 30
am
plit
ud
e (
mV
)
0
10
20
30
40
50
60
rough surface
painted white
polished
polished
response uniformity of light guides with different surface treatments
white rough polished
white rough polished
polished
Effects of surface treatment – measured data.
9
Light guides with painted or diffuse surfaces (non-specular) show a
20-25% lower efficiency, and less uniform response, with diminished
response near the edges.
Would making the surfaces diffuse reflectors improve the uniformity?
10
Measured Data: Effect of tapped
mounting hole on LG uniformity.
The tapped hole has no apparent effect on the
uniformity. The hole is not deep enough to significantly
“shadow” the SiPMs
Tapped hole and screw
A tapped hole is required to attach the SiPM pcb.
Does it affect the light collection?
11
Winston Cone –
hybrid designs
Unlike a true Winston cone, our
application has approximately square
input and output surfaces
Y Axis (mm)
-15 -12 -9 -6 -3 0 3 6 9 12 15
Eff
iecie
ncy
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Trapezoidal
WCLG 1
WCLG 2
WCLG 3
WCLG 4
WCLG 5
WCLG 6
Read out with PMT Read out with 4 SiPMs
Y Axis (mm)
-12.5 -7.5 -2.5 2.5 7.5 12.5-15.0 -10.0 -5.0 0.0 5.0 10.0 15.0
Eff
icie
ncy
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
Trapezoidal
WCLG1
WCLG 2
WCLG 3
WCLG 5
WCLG 6
In simulation, these
light guides were more
uniform when read out
with a PMT (full
coverage), but were
less uniform when
read out with 4 SiPMs
(21% coverage).
WCLGx = Winston Cone Light Guide
Winston cone – hybrid design – cast acrylic
Measured with pmt (full coverage)
Measured data –
with 4 SiPMs (summed output)(Geometric efficiency: SiPM active area / end
area = 36mm2 / 174mm2 = 0.21)
12
For both the measurement
and simulations, with full
coverage of the readout
surface using a pmt, the
response is very uniform…
But with 4 SiPMs , the
response is significantly
less uniform.
Simulation with pmt and SiPMs
pmt
4 SiPMs
13
Measured Uniformity of Quad, injection molded
acrylic light guide on pmt
Measured Uniformity of block + quad light
guide on pmt
…combined response is flatter
“collared” fiber block - read out directly
with pmt (no light guide)
14
0
50
100
150
200
250
0
10
20
30
40
50
-10
0
10
20
30
40
ampl
itude
X position (mm)
Y p
osi
tion (m
m)
apr03 uniformity map of UI block w/fiber "collar" on 7cm pmt. RTVcoupled. LED fiber @ 5mm
X profile at Y= 30mm
X position (mm)
0 10 20 30 40 50 60
am
plit
ude
0
50
100
150
200
250
Y profile at X=15mm
Y position (mm)
-10 0 10 20 30 40
am
plit
ude
0
50
100
150
200
250
• “Collaring brings” the fibers in from the block edges at the
readout end. This improves the light collection near the
block edges
• It also allows for a single light guide shape to be used for
all 25k towers.
• These blocks will be beam tested in a new prototype (Feb
2018)
Simulation:Locus of points where rays hit the narrow end of the light guide for different fiber positions (indicted by yellow dot)
Square boxes
indicate locations of
SiPMs 15
Measured data:
Uniformity scan of trapezoidal
light guide – 3/4 SiPMs masked
Light distribution –
dependence on location of input. Four SiPMs with Air Gap Flux Efficiency (X Axis)
Position (mm)
-12.5 -7.5 -2.5 2.5 7.5 12.5-15.0 -10.0 -5.0 0.0 5.0 10.0 15.0
Eff
icie
ncy
0.00
0.03
0.06
0.09
0.12
0.15
0.18
0.21 All SiPMs
SiPM 1
SiPM 2
SiPM 3
SiPM 4
16
50mm
machined
lightguides
25mm inj.
molded
lightguides
50mm
25mm
Test beam results A prototype was tested in 8 GeV electron beam at FermiLab Test Beam Facility (FTBF) /T1044 Jan/2017 This 8x8 tower prototype was instrumented with 50mm machined light guides on the top half and injection molded 25mm light guides on the lower half.
50mm LGs have
more uniform
response, less
significant loss
between towers
25mm LGs have
peaked response,
drops off near
edges, with more
loss between towers
Test beam results A prototype was tested in 8 GeV electron beam at FermiLab Test Beam Facility (FTBF) /T1044 Jan/2017
17
0 Degree -
Normal incidence
10 Degree
rotation
Rotating the detector 10deg
relative to the incident
beam smooths out the
block boundary effects and
reduces “channeling”
This tilt angle was
incorporated into the
calorimeter design in eta
and phi
50mm machined LGs
Inter-block
boundary
Intra-block
tower
boundary
Projection
18
Scintillation
Emission
peakScintillation
Emission
peak
Based on
measurements
and simulations of
the experimental
hall, the sPHENIX
EMCal is expected
to receive
~ 10 krad per run
year.
So cumulative
effects of radiation
should be minimal
- 2-3% over
detector lifetime
Acrylic samples
irradiated with 60Co gamma rays
at BNL SSGRIF
facility.
Conclusions
19
• The short radial space constrains us to a short, “stubby” trapezoidal light guide design. The
longer 50mm light guides were more uniform, but radial space constraints preclude their use.
• Although the Winston cone hybrids had a slightly higher efficiency when their full surface was
read out, they had a less uniform response with the 4 sipm readout.
• Attempts to make the LG surface more uniform by making it a diffuse reflector rather than a
specular reflector made the response less uniform and less efficient.
• We have attempted to mitigate the effect of block boundaries by “collaring” the fibers at the
ends of the absorber blocks to move them in from the light guide edges. This design change will
be evaluated in an upcoming beam test.
• We will also use a single tower, 25mm injection molded light guide with a 1mm “step” in the next
prototype. Simulations suggest that the “step” improves the light collection near the edges of the
light guide.
• Given the proximity of the scintillation emission to the transmission edge in acrylic, it would be
prudent to use UVT acrylic.
References
20
Tsai, O. et al. Results of R&D on a new construction technique for W/ScFi Calorimeters.
J. Phys Conf Ser, 404:012023, 2012.
Tsai, O. et al. Development of a forward Calorimeter system for the STAR experiment.
J. Phys Conf Ser, 587(1):012053, 2015.
Aidala, C. et al, Design and Beam Test Results for the sPHENIX Electromagnetic
and Hadronic Calorimeter Prototypes. arxiv.org/abs/1704.01461
sPHENIX - 2017 Testbeam Analysis Note – Osborne, Huang
Simon, Hertzog, Jones, Rhodes, Yairi. Modeling and scanning lightguides for Pb/SCIFI calorimeters
Nucl Inst & Meth in Phy Res A 335 (1993) 86-101.
TracePro Lambda Research Corp, Littleton, MA USA www.lambdares.com
Momentive Performance Materials Inc. Waterford, NY www.momentive.com (RTV615)
Hamamatsu Corp. Bridgewater, NJ. www.Hamamatsu.com
Other sPHENIX talks/posters at IEEE NSS 2017:
• N-38-1 Test Beam Results and Status of the sPHENIX Calorimeter System – M.E. Connors• N-38-2 Design of a Compact TPC for the sPHENIX Experiment – K. Dehmelt• N-38-3 Design and Performance of the Readout Electronics for the sPHENIX Calorimeters – E. Mannel• N-35-6 The Readout of the sPHENIX Tracking System – M. Purschke• N-03-049 Design Studies for a TPC Readout Plane – B. Azmoun