Performance of MuPix8
a large scale HV-CMOS pixel sensor
Heiko Augustin on behalf of the Mu3e Pixel team
Physikalisches Institut Heidelberg
PIXEL 2018 Taipei 13. December 2018
HighRR
Motivation - The Mu3e Experiment
Searching for the charged Lepton Flavor Violating decayµ+→e+e−e+
Target Inner pixel layers
Outer pixel layers Recurl pixel layers
Scintillator tiles μ Beam
sensitivity goal of one in 1016decays, requires high muon rates of 109s−1 reconstruction of electron trajectories in a 1 T solenoidal magnetic field multiple coulomb scattering dominated (pe<53 MeV/c)
Motivation - The Mu3e Experiment
Searching for the charged Lepton Flavor Violating decayµ+→e+e−e+
Pixel Sensor Requirements
Pixel Size Time Resolution Material Budget Efficiency 80×80µm2 <20ns <1hX0 / layer ∼100%
sensitivity goal of one in 1016decays, requires high muon rates of 109s−1 reconstruction of electron trajectories in a 1 T solenoidal magnetic field multiple coulomb scattering dominated (pe<53 MeV/c)
High Voltage - Monolithic Active Pixel Sensor (HV-MAPS)
P-substrate N-well
Particle E field
I.Peric, P. Fischer et al., NIM A 582 (2007) 87 low ohmic substrate
(20Ωcm - 200Ωcm) high voltage (−120 V) 180nm HV-CMOS process
large depleted n-well diode charge collection via drift no additional readout chip thinned to 50 µm
The MuPix Concept
PixelsPeriphery
State
MuPix7 as an example active pixel matrix digital pixel cell in periphery
state machine (VCO, PLL, etc., not shown)
The MuPix Concept
PixelsPeriphery
State
Sensor
Charge sensitive amp
Source follower
Thepixel has
reverse biased diode charge sensitive amplifier
signal transmitter
The MuPix Concept
PixelsPeriphery
State
Transmission line
signal transmission: point to point connection pixel to periphery
The MuPix Concept
PixelsPeriphery
State
tune DAC amp baseline
global threshold
peripheral cell: digitisation individual pixel tuning (thr) This separation protects the analog cell from digital crosstalk.
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
detecting a signal signal generation
amplification transmission digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
detecting a signal signal generation amplification
transmission digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
detecting a signal signal generation amplification transmission
digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
detecting a signal signal generation amplification transmission digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
detecting a signal signal generation amplification transmission digitisation
timestamp sampling
column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
Time- stamp
detecting a signal signal generation amplification transmission digitisation
timestamp sampling
column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
Introduction MuPix8 Results Summary & Outlook
The MuPix Concept
PixelsPeriphery
State
Time- stamp Data
detecting a signal signal generation amplification transmission digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
The MuPix Concept
PixelsPeriphery
State
Serialiser
Data Time- stamp
detecting a signal signal generation amplification transmission digitisation
timestamp sampling column-drain readout of time and address
serialisation and stream out at 1.25 Gbit s−1
The MuPix7 prototype - Results
column direction [um]
0 20 40 60 80 100 120 140 160 180 200
row direction [um]
0 20 40 60 80 100 120 140 160 0.75
0.8 0.85 0.9 0.95 1
Efficiency Efficiency
0.88 0.9 0.92 0.94 0.96 0.98
Efficiency0.98 0.96 0.94 0.92 0.9
80×103 µm2 pixel size 3.3×3.3 mm2
20Ωcm substrate full system on-chip very well understood
The MuPix7 prototype - Results
600
− −500 −400 −300 −200 −100 0
Entries [1/run]
102 103 104
Time diffrence between hit and scintillator time [ns]
σ= 14.3 ns
99.5% efficiency
with 14.3 ns time resolution
@ 300 mW cm−2
time walk observed & signal line crosstalk
The MuPix7 prototype - Results
0 20 40 60 80 100 120
0 50 100 150 200 250
0 10 20 30 40 50 60 70 80 90
ToTtime Trigger Difference versus ToT
Latency [2,5ns]
ToT [10ns]
99.5% efficiency
with 14.3 ns time resolution
@ 300 mW cm−2
time walk observed & signal line crosstalk
Scale it up!
The first large scale prototype - MuPix8
Group Ivan Peric @KIT
AMS aH18 process (180 nm) 80×81 µm2 pixel size 16×10 mm2 active area 128×200 pixels
MuPix8 Design Features
submatrix A
submatrix B
sub- matrix
C
digital periphery pixel matrix
bias blocks & pads 10.8 mm
19.5mm
128 columns
48 48 32
200 rows
2×1 cm2 chip size
→ full column length radiation hard design
time walk correction possible increased signal charge
(20→80−200Ωcm substrate) thinned to 62.5 µm + 100 µm bandgap reference, voltage DACs, ...
MuPix8 Architecture
Pixel Periphery State Machine
Pixel Periphery State Machine
Pixel Periphery State Machine
readout state machine
VCO
&
PLL
8b/10b
encoder serializer LVDS ...
other pixels
LVDSLVDS
MUX LVDS
sensor CSA
comparator 1&2
tune DAC
threshold baseline
test-pulse injection
readout
x3
integrate charge
amplification
line driver per pixel digital output
threshold adjustment
3 sub-matrices with individual data output additional merged/mirrored data output
→ 4 differential links @ 1.25 Gbit/s
MuPix8 - Signal Transmission
Bias Voltage
Bias Voltage Bias Voltage
Comparator CSA
Pixel
Periphery Res
very dense routing:
2 metal layers,
200 signal lines per column sub-matrix A: source follower
→ prone to crosstalk
sub-matrix B&C: current driven
→ aiming for crosstalk reduction
MuPix8 - Pixel Peripheral Cell
∆t
Amplitude
short ToT Time long ToT
Threshold
Time walk
2 comparators
3 time walk correction approaches 5 tune bits + pixel switch
10 timestamp bits 6 bit ToT
MuPix8 - Pixel Peripheral Cell
Comp Thr1 Signal
Thr2 Comp Signal
Voltage
Time Threshold1
Threshold2 Signal
Hold TimeStamp
Store Timestamp
& Hit Flag
2-threshold On-Chip suppression
2 comparators
3 time walk correction approaches 5 tune bits + pixel switch
10 timestamp bits 6 bit ToT
MuPix8 - Pixel Peripheral Cell
Voltage
Time Dynamic Threshold
ToT Threshold
Baseline
Thr Comp Signal
Dynamic Threshold
Start Ramp
& Store Timestamp
Dyn Thr Comp Signal
Store ToT Timestamp
& Hit Flag
voltage ramp Off-Chip correction
2 comparators
3 time walk correction approaches 5 tune bits + pixel switch
10 timestamp bits 6 bit ToT
Setup
Characterisation Setup
motherboard with insertable PCB versatile tool
integration for MuPix-like sensors:
MuPix8, AtlasPix, MuPix9, MuPix7 lab characterisation
MuPix Telescope
4-8 layers of sensors one sensor as DUT
integration test for the Mu3e readout system
MuPix Telescope
extensive testbeam campaigns: DESY, PSI and MAMI online efficiency
fast analysis framework
MuPix Telescope
All-AtlasPix telescope proof of principle
performed exemplary threshold scans
Results
Commissioning
submatrix A
submatrix B
sub- matrix
C
digital periphery pixel matrix
bias blocks & pads 10.8 mm
19.5mm
128 columns
48 48 32
200 rows
early breakdown @−60 V (120V design)
powering issue in the digital part
→ tuning possible but not effective pixel switch is working
configuring with 6 Mbit s−1 (100 ms per chip)
more than 10 MHits/s per matrix
Efficiency
Matrix A - Efficiency & Noise
efficiency
0.4 0.5 0.6 0.7 0.8 0.9 1
col / pixel
0 10 20 30 40
row / pixel
0 20 40 60 80 100 120 140 160 180
noise rate per pixel / Hz
−1
10 1 10 102
103
104
not matched col / pixel
0 10 20 30 40
not matched row / pixel
0 20 40 60 80 100 120 140 160 180
Matrix A - Efficiency & Noise
threshold / mV 40 60 80 100 120 140 160 180
efficiency
0.8 0.85 0.9 0.95
1 efficiency
noise
noiserate per pixel / Hz
1
10−
1 10
50 mV threshold = 650e− highly efficient(>99.9%), low noise plateau further increase by masking hot pixels increase with higher HV and resistivity
Matrix A - Efficiency & Noise
threshold / mV 40 60 80 100 120 140 160 180
efficiency
0.8 0.85 0.9 0.95
1 efficiency
noise
noiserate per pixel / Hz
1
10−
1 10
50 mV threshold = 650e− highly efficient(>99.9%), low noise plateau further increase by masking hot pixels increase with higher HV and resistivity
Matrix A - Efficiency & Noise
threshold / mV
60 80 100 120 140
hit efficiency
0 0.2 0.4 0.6 0.8 1 1.2
50 V 15 V 05 V
50 mV threshold = 650e− highly efficient(>99.9%), low noise plateau further increase by masking hot pixels increase with higher HV and resistivity
Matrix A - Efficiency & Noise
threshold / mV
60 80 100 120 140
hit efficiency
0 0.2 0.4 0.6 0.8 1 1.2
m 50V µ cm 100 Ω 200
m 50V µ cm 725 Ω 80
50 mV threshold = 650e− highly efficient(>99.9%), low noise plateau further increase by masking hot pixels increase with higher HV and resistivity
Introduction MuPix8 Results Summary & Outlook
Matrix A - Sub-Pixel Studies
efficiency
0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
HV = -50V, optimal settings utilising the EUDET-telescope at DESY position resolution ≈6 µm
folded map to 2×2 pixel
→ inefficiency at the pixel edges
Matrix A - Sub-Pixel Studies
efficiency
0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
HV = -15V, high threshold utilising the EUDET-telescope at DESY position resolution ≈6 µm
folded map to 2×2 pixel
Introduction MuPix8 Results Summary & Outlook
Matrix A - Sub-Pixel Studies - Clustering
0 2 4 6 8 10 12 14 16
cluster size
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10 1
Entries
15 V: mean 1.176 50 V: mean 1.186
Cluster sizes very low average cluster size
charge sharing at edges and corners
Introduction MuPix8 Results Summary & Outlook
Matrix A - Sub-Pixel Studies - Clustering
matched
0 10 20 30 40 50
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
Single Hits very low average cluster size
charge sharing at edges and corners
Matrix A - Sub-Pixel Studies - Clustering
matched
0 2 4 6 8 10 12 14
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
Double Cluster very low average cluster size
Matrix A - Sub-Pixel Studies - Clustering
matched
0 0.5 1 1.5 2 2.5 3 3.5 4
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
Triple Cluster very low average cluster size
Matrix A - Sub-Pixel Studies - Clustering
matched
0 0.5 1 1.5 2 2.5 3
µm x / 0 20 40 60 80 100 120 140 160
mµ y /
0 20 40 60 80 100 120 140 160
Quadrupel Cluster very low average cluster size
Time Resolution
Matrix A - Timing Studies
/ ndf
χ2 1.96e+05 / 22
Constant 7.056e+05 ± 4.765e+02 Mean −0.02349 ± 0.00972 Sigma 16.8 ± 0.0
time difference / ns
−300 −200 −100 0 100 200 300 400
entries
0 0.1 0.2 0.3 0.4 0.5 0.6 106
× χ2 / ndf 1.379e+05 / 22
Constant 6.005e+05 ± 4.015e+02 Mean 18.65 ± 0.01 Sigma 20.02 ± 0.01
Sensor time resolution: ≈20 ns
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
Time-Walk corrected Hit TS - Trigger TS [8ns]
−6 −4 −2 0 2 4
entries
0 10 20 30 40 50 60
/ ndf
χ2 6.506 / 5
scale 58.17 ± 7.13 mean [bins] −1.811 ± 0.091
[bins]
σ 0.7552 ± 0.0679
BG slope 0.003091 ± 0.000832
0.544 [ns]
± = 6.042 σ
Single pixel(0/1) time resolution
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
entries
10 20 30 40 50 103
×
Hit TS - Trigger TS [8ns]
−10 −5 0 5 10
ToT [64ns]
10 15 20 25 30 35 40
Time walk: raw data
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
Peak position [bins]
−2
−1 0 1 2 3 4 5
col block
0 1 2 3 4 5 6 7
row block
0 5 10 15 20
Pixel position delay
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
entries
10 20 30 40 50 60 103
×
Delay corrected Hit TS - Trigger TS [8ns]
−10 −5 0 5 10
ToT [64ns]
10 15 20 25 30 35 40
Time walk: delay corrected
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
entries
10 20 30 40 50 60 103
×
Time-Walk corrected Hit TS - Trigger TS [8ns]
−10 −5 0 5 10
ToT [64ns]
10 15 20 25 30 35 40
Delay + Time walk correction
lab measurement with Sr90 and scintillating tile
time walk observed pixel position delay (routing and power distribution)
Matrix A - Timing Studies
2Thr ThLow 520mV 2Thr ThLow 525mV 2Thr ThLow 530mV 2Thr ThLow 535mV ToT
Time Resolution [ns]
4 6 8 10 12 14 16 18 20
complete chip row ≤ 20
uncorrected uncorrected
delay corrected delay corrected delay + tw corrected delay + tw corrected
Time resolution: σ = 6.5 ns
Crosstalk
Introduction MuPix8 Results Summary & Outlook
Matrix A - Signal Line Crosstalk
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Introduction MuPix8 Results Summary & Outlook
Matrix A - Signal Line Crosstalk
charge sharing
charge sharing / crosstalk
crosstalk
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Introduction MuPix8 Results Summary & Outlook
Matrix A - Signal Line Crosstalk
row / pixel
20 40 60 80 100 120 140 160 180
triple crosstalk
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
/ ndf
χ2 2933 / 172
threshold 72.69 ± 0.109 offset 0.0002191 ± 1.914e−05 slope 0.002792 ± 8.422e−06
/ ndf
χ2 2933 / 172
threshold 72.69 ± 0.109 offset 0.0002191 ± 1.914e−05 slope 0.002792 ± 8.422e−06
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Introduction MuPix8 Results Summary & Outlook
Matrix A - Signal Line Crosstalk
threshold / mV 40 60 80 100 120 140 160 180
crosstalk / %
0 2 4 6 8 10
Threshold dependence
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Introduction MuPix8 Results Summary & Outlook
Matrix A - Signal Line Crosstalk
threshold / mV 40 60 80 100 120 140 160 180
clustersize / pixel
1.03 1.04 1.05 1.06 1.07 1.08
Crosstalk removed
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Matrix A - Signal Line Crosstalk
threshold / mV 40 60 80 100 120 140 160 180
efficiency
0.8 0.85 0.9 0.95 1
= 235.75 µ Row 0 :
= 228.89 µ Row 100 :
= 221.92 µ Row 199 :
Signal reduction
capacitive coupling row dependence
→ increased readout load
→ signal reduction
→ degrade rising edge
Matrix B - current driver
Matrix B - Efficiency
efficiency
0.4 0.5 0.6 0.7 0.8 0.9 1
50 60 70 80 90
row / pixel
0 20 40 60 80 100 120 140 160 180
recent testbeam measurement preliminary result
highly efficient >99%
further analysis ongoing
Matrix B - Time Resolution
Time-Walk corrected Hit TS - Trigger TS [8ns]
−20 −15 −10 −5 0 5 10 15 20
entries
0 0.2 0.4 0.6 0.8 1 1.2 1.4 106
×
/ ndf
χ2 2.7e+04 / 7
scale 1.367e+06 ± 7.022e+02 mean [bins] 0.4557 ± 0.0008
[bins]
σ 1.925 ± 0.001
BG slope 42.51 ± 0.30
0.009 [ns]
± = 15.401 σ
investigation with coincidence setup delay compensation + time walk correction
→ time resolution: σ <15 ns
Matrix B - Time Resolution
ThLow
520 mV 530 mV 540 mV 550 mV 560 mV
Time Resolution [ns]
5 10 15 20 25 30 35 40
uncorrected delay corrected delay + tw corrected
investigation with coincidence setup delay compensation + time walk correction
→ time resolution: σ <15 ns
Matrix B - Crosstalk
row / pixel
0 20 40 60 80 100 120 140 160 180
triple crosstalk
−0.02 0 0.02 0.04 0.06 0.08 0.1
no crosstalk observed!
current driver allows for dense rounting
Summary & Outlook
Summary
Pixel Sensor Performance
Matrix A Matrix B AtlasPix
Efficiency >99.9% >99% >99.9%
Time res 6.5 ns 15 ns 10 ns
Crosstalk yes no no
Power 200 mW cm−2 210 mW cm−2 300 mW cm−2 exceptional performance →scaling successful
obvious improvements at hand: breakdown, powering, ...
current driver scheme can be further improved
Outlook
4.5mm
3.5mm
MuPix9
MuPix9: shunt, serial powering, mu3e slowcontrol interface ...
AMS aH18: long delivery times qualification of new foundry:
TSI Semiconductors (H18) resubmitted MuPix7 to TSI testbeam analysis ongoing
submission of fullscale sensor in 2019
Acknowledgments
Many important test beam campaigns have been performed at the Test Beam Facility at DESY Hamburg (Germany), a member of the Helmholtz Association (HGF).
We would like to thank the PSI for providing high rate test beams under excellent conditions.
We thank the Institut für Kernphysik at the JGU Mainz for giving us the opportunity to take data at MAMI.
Thank you
Back-up
trigger_ts_diff
Entries 399254
Mean −0.6394
RMS 8.511
/ ndf
χ2 213.9 / 3
p0 4253 ± 62.7 p1 −4.175 ± 0.016 p2 1.136 ± 0.020 p3 1216 ± 31.8
−15 −10 −5 0 5 10 15
1000 1500 2000 2500 3000 3500 4000 4500 5000
5500 trigger_ts_diff
Entries 399254
Mean −0.6394
RMS 8.511
/ ndf
χ2 213.9 / 3
p0 4253 ± 62.7 p1 −4.175 ± 0.016 p2 1.136 ± 0.020 p3 1216 ± 31.8
Trigger TimeStamp Difference Distribution
Time resolution:
MuPix7_TSI
Sensor hit rate [kHz]
1 10 102 103 104
Average readout load
−4
10
−3
10
−2
10
−1
10 1 10
Sensor 0 Sensor 1 Sensor 2 Sensor 3 Sensor 4 Sensor 5 Sensor 6 Sensor 7
Rate test
col / pixel
50 60 70 80 90
row / pixel
0 20 40 60 80 100 120 140 160
180 efficiency
0.95 0.955 0.96 0.965 0.97 0.975 0.98 0.985 0.99 0.995 1
MuPix8 MuPixX
Bias Voltage
Bias Voltage Bias Voltage
Comparator CSA
Pixel
Periphery Res
Pixel Matrix
Digital Cells Signal lines
103 µm
80 µm
row / pixel
20 40 60 80 100 120 140 160 180
double crosstalk
0 0.1 0.2 0.3 0.4
0.5χ2 / ndf 7609 / 177 scale 0.06542 ± 0.0002717
µ 95.04 ± 0.2851
σ 22.57 ± 0.218 / ndf
χ2 7609 / 177
scale 0.06542 ± 0.0002717
µ 95.04 ± 0.2851
σ 22.57 ± 0.218
row / pixel
20 40 60 80 100 120 140 160 180
triple crosstalk
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
0.4χ2 / ndf 2933 / 172
threshold 72.69 ± 0.109 offset 0.0002191 ± 1.914e−05 slope 0.002792 ± 8.422e−06
/ ndf
χ2 2933 / 172
threshold 72.69 ± 0.109 offset 0.0002191 ± 1.914e−05 slope 0.002792 ± 8.422e−06
threshold / mV 40 60 80 100 120 140 160 180
clustersize / pixel
1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2 1.22
w/ crosstalk
threshold / mV 40 60 80 100 120 140 160 180
clustersize / pixel
1.03 1.04 1.05 1.06 1.07 1.08
w/o crosstalk