CMOS Monolithic Active
Pixel Sensors (MAPS)
for scientific applications
Dr Renato Turchetta
CMOS Sensor Design Group Leader
CCLRC Technology
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
2
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
3
CMOS Monolithic Active Pixel Sensor (MAPS)
(Re)-invented at the beginning of ’90s: JPL, IMEC, …
¾ Standard CMOS technology
¾ all-in-one detector-connectionReadout control
readout = Monolithic
¾ small size / greater integration
¾ low power consumption
¾ radiation resistance
¾ system-level cost
¾ Increased functionality
pixel- parallel processing)
¾ random access (Region-of-Interest
ROI readout)
Column-parallel ADCs
Data processing / Output stage
I2C
control
¾ increased speed (column- or
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
CMOS sensors in digital cameras
Digital intraoral imaging
Consumer/prosumer
Digital cameras
Web
cams
Mobile phones
Digital mammography
4
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
CMOS Image Sensor Technology. 1
In general, any CMOS process have different flavours.
The basic one is digital. Then MixedMode/RF, High Voltage, …, CIS
(CMOS Image Sensor).
Transistors don’t change. Modules added: high value resistors, linear
capacitors, …
For CIS:
one/two masks/implants added to improve image quality / reduce
leakage current
5
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
CMOS Image Sensor Technology. 2
Masks for colour filters, microlenses. Special BEOL for improved oxide
transmission
Stitching for sensors larger than reticle is becoming more common due
to push for 35mm film replacement
0.18μm available, 0.13μm starting this year in different places. Pixel
transistors still at 0.35μm equivalent
Epitaxial wafers are generally used: better quality, reduced cross-talk
Thickness: depends on foundry, generally up to 20 μm thick epi layer
6
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
7
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
8
CMOS sensors for radiation detectors
Photons
Metal layers
Silicon band-gap
of 1.1 eV ↔ cutoff at 1100 nm.
Good efficiency
up to ‘low’
energy X-rays.
For higher
energy (or
neutrons), add
scintillator or
other material.
Removal of
substrate for
detection of UV,
low energy
electrons.
Polysilicon
Radiation
Dielectric
for
insulation
and
passivation
-+
N+
-+
P-Well
N-Well
P-Well
-+
-+
P-epitaxial
Potential
barriers
Charged particles
+
layer
kT N
-+
(up
to 20
V =to ln
100%
q N
-+
μm thick)
efficiency.
- +
- +
P-substrate (~100s μm thick) - +
N+
N+
sub
epi
P+
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
RAL science sensors
o Space science: Star Tracker, ESA Solar Orbiter, …
o Earth Observation: 3 μm pixel linear sensors, ..
o Particle Physics: ILC, vertex and calorimeter (CALICE), SLHC, …
o Biology: electron microscopy, neuron imaging, optical tweezers,
o Medicine: mammography, panoramic dental
o …
Detecting:
¾ Photons: visible, UV, EUV, X-ray (with scintillators), …
¾ Charged particles: MIPs, low/medium energy electrons (few keV up to 1
MeV)
¾ Voltages (!)
9
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
10
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
11
Vanilla sensor. 1
¾ Designed within the UK-MI3 consortium (http://mi3.shef.ac.uk, led by prof
N. Allinson), aiming at developing novel CMOS APS
¾ Large pixels: 25 μm, design in 0.35 μm CMOS
¾ Epitaxial layer thickness: choice of 14 and 20 μm
¾ Backthinned version will be available end of this year
¾ Format 512x512 + black pixels (520x520 full format)
¾ 3T pixel with flushed reset
¾ Noise < 25 e- rms
¾ Full well capacity > 105 e¾ DR ~ 4000 ~ 12 bits
¾ On-chip SAR ADCs, one for 4 columns. Selectable resolution: 10 or 12.
Adjustable range.
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
12
Vanilla sensor. 2
¾ Analogue output at 4.5 MHz
¾ Row and column address decoder
¾ Full frame readout: Frame rate > 100 fps.
¾ Region-of-interest readout: Fully programmable: any shape, any size
Example speed: six 6x6 regions of interest @ 20k fps
¾ Two-sided buttable for 2x2 mosaic
¾ Design for backthinning. Detecting capability not limited to visible light! It
will be also tested for UV and electrons.
¾ Data rate: > 40 MByte/sec in full frame mode
¾ Readout based around off-the-shelf FPGA development boards (VirtexIIPro) plus optical link (Gbit at present but already testing 10Gbit)
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Vanilla sensor. 3
¾ Sensor fully functional
¾ Already region-ofinterest readout already
tested at over 20 k fps
¾ Characterisation just
started
¾ Preliminary results
consistent with
specifications
13
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
14
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Medical X-ray detection
Project I-ImaS (http://www.i-imas.ucl.ac.uk) funded by EU.
Led by prof. R. Speller (University College London, UCL)
Application: mammography, dental (panoramic and cephalography)
Scanning system with real-time data analysis to optimised dose uptake
Step-and-shoot, and double scan
Time for 1 image: a few seconds.
Large pixels: 32 μm
Image area: 18cmx24cm covered by several sensors in several steps
Image size: 5120x7680 = 40Mpixel/image @ 14 bits, ~70MBytes
Integration time per pixel: 10 ms
1½D CMOS sensor coupled to scintillator (structured CsI)
15
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
The X-ray camera
X-ray tube
Primary slot collimator
Linear
translation
Compressed breast
Detector slot collimator
Sensor
Example of mammography application
(image courtesy of Gary Royle - UCL)
16
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Expected results
Simulated mammography scan that uses 18% lower dose
(image courtesy of Prof. R.Speller - UCL)
17
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
1½D CMOS sensor
¾ Designed in 0.35 μm CMOS
¾ 512*32 pixels at 32 μm pitch plus 4
rows and columns on both sides for edge
effects
¾ 200,000 e- full well
¾ 33 to 48 e- ENC depending on the pixel
reset technique used
¾ more than 72dB S/N ratio at full well
(equivalent to 12 bit dyn. range)
¾ possible to use hard, soft or flushed
reset schemes
¾ 14 bit digital output; one 14-bit SAR
ADC every 32 channel
¾ 20 MHz internal clock; 40 MHz digital
data rate
¾ data throughput: 40MHz·7bit =
280Mbit/s = 35 MB/sec
Sensor floorplan
18
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
1½D CMOS sensor. Architecture.
1 channel ↔
32x32 pixel
14 bit successive
approximation ADC
4 MSB on resistor string
20 MHz clock
16 cycles per conversion
↔ 1.25 MHz conversion
rate
19
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
First preliminary
images
Test at University College London
(UCL). CsI scintillator on MAPS.
Images stitched with the correct overlap,
corrected for dead pixel lines and, where
possible, corrected for intensity variations
both across the sensor and down the image
(scan position). They were not fully
corrected for dark noise nor using a true
white field image, and so the final systems
images will be superior to these.
Image of a tooth. Taken at 70 kVp, 6 mA
Image of the sensor mount card. Taken
at 75 kVp, 4 mA
20
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
21
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
CMOS Image Sensors for Particle Physics
First proposed end of 1999. Developments from a number of groups: LEPSI/IreS,
Saclay, LBL, Stanford, Perugia, Bergamo, Pisa, …
Low noise detection of MIPs first demonstrated in 2001 with a 3T pixel
Since then, with a number of technologies/epi thickness:
AMS 0.6/14, 0.35/∞, 0.35/14, 0.35/20, AMIS (former MIETEC) 0.35/4,
0.25/2,
TSMC 0.35/10, 0.25/8, 0.25/∞,
UMC 0.18/∞
Noise <~ 10 e- rms with off-chip,
off-line CDS
Spatial resolution 1.5 μm
@ 20 μm pitch, with full analogue
readout
Good radiation hardness
Low power
Speed: rolling shutter
can be a limit
IBM
22
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
23
International Linear Collider
International Linear Collider: CMOS sensors proposed for Vertex and
Electromagnetic calorimeter.
Radiation hardness: moderate (~100s kRad, 1012 n/cm2)
Large area (stitched) sensor required in both cases.
Total covered area: ~m2 for Vertex (FAPS design), a few 103 m2 for Electromagnetic
calorimeter
Anti-electron
(positron) e+
(~TeV)
Electron e(~TeV)
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
24
Calorimeter for the Linear Collider
Experiment (CALICE)
UK collaboration (Birmingham, Cambridge, Imperial College, Manchester,
RAL, RHUL and UCL) is developping also a MAPS version of the
calorimeter. Sensor designed by RAL.
Present specification: large (50x50 μm) pixels, measurement of hit timing
with BCO precision (>~ 150 ns).
Present architecture: groups of pixels with ‘common’ digital area. Device
simulation helps layout definition.
Designed in an advanced 0.18 μm CIS process, fully stitchable.
R&D just started. First test structures expected spring 2007.
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
25
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
26
Radiation hardness
1E3
Transistors.
0.25 micron
Threshold shift: reduces with shrinking feature size
1E2
eff
ec
t
1E-2
Radiation damage increases leakage current
diffusion distance is reduced
1E-3
1
With tunn
Diodes.
el effect
~tox2
1E-1
Radiation damage reduces minority carrier lifetime Æ
nn
el
1E0
ith
ou
t tu
separate transistors
1E1
W
Transistor leakage current: use guard-rings to
- Vfb/ Mrad(Si)
Bird’s beak effect: use enclosed geometry transistors
10
Oxide thickness (nm)
100
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
27
Imaging devices. 1
Co60 irradiation
From: E.-S. Eid et al, “Design and characterization of ionizing radiation-tolerant CMOS APS image sensors up to
30 Mrd(si) total dose,” IEEE Trans. Nucl. Sci., vol. 48, pp. 1796–1806, Dec. 2001
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Imaging devices. 2
From: Jan Bogaerts, Bart Dierickx, Guy Meynants, and Dirk
Uwaerts, Total Dose and Displacement Damage Effects in a
Radiation-Hardened CMOS APS, IEEE Trans. On Elec.
Dev., vol. 50, n. 1, January 2003, 84-90
28
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
RAL Sensors for Particle Physics
Parametric test sensor: RAL_HEPAPS family
RAL_HEPAPS2: 0.25 μm CIS, 8 μm epitaxial layer, 384*224
pixels, 15 μm pitch: 3MOS, 4MOS, ChargePreAmplifier
(CPA), Flexible APS (FAPS, 20 μm pitch)
RAL_HEPAPS3: 0.25 μm MM, no epitaxial layer, 192*192
pixels, 15 μm pitch: 3MOS, 4MOS, Deep N-well diodes
Test sensors
RAL_HEPAPS4: 0.35 μm CIS, 20 μm epitaxial layer,
1026*384 pixels, 15 μm pitch. 3 versions: 1, 2 or 4
diodes per pixel. Rad-hard, 5MHz row rate (now in
manufacturing)
29
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
30
Signal from individual particles
Beta source (Ru106) test results. Sensors HEPAPS2.
Cluster in S/N
Signal spread
Number of pixels
in a “3x3” cluster
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
31
Distribution of signals
From different types of pixels.
HEPAPS2
Type
Specs
S
N
S/N
3MOS E
4 diodes
99
4.94
20.1
3MOS C
GAA
87
4.85
18.0
3MOS B
Diode 1.2x1.2
92
3.87
23.8
3MOS A
Diode 3x3
67
3.31
20.3
4MOS C
Lower VT
101
4.14
24.4
4MOS B
Higher VT
114
4.70
24.2
4MOS A
Reference
111
4.45
25.0
Typical ‘Landau distribution
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Noise (ADU counts)
Noise (ADU counts)
Radiation test. Source results
•
increase slightly with
dose.
•
Dose (log10(p/cm2))
Signal (ADU counts)
Signal (ADU counts)
Dose (log10(p/cm2))
Dose (log10(p/cm2))
Noise seems to
Dose (log10(p/cm2))
Signal decreases with
dose.
3MOSA
3x3 μm2
3MOSB
1.2x1.2 μm2
3MOSC
GAA
3MOSE
4 diodes
4MOSA
Reference
4MOSB
Higher VT
4MOSC
Lower VT
32
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
33
S/N
S/N
Radiation test. Summary
Usable
sensor
Dose (log10(p/cm2))
•
•
•
•
Dose (log10(p/cm2))
Sensors yield reasonable S/N up to 1014 p/cm2
0.35 μm technology in the pixel transistors. Enclosed layout in 3MOS_E
S/N reduction seems to be dominated by charge collection
Especially 3MOS_E (4 diodes) looks interesting
-– Larger capacitance yields larger noise
+– Four diodes: less dependence of S/N on impact point
+– After irradiation remains a larger “sensitive area”
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
34
Single pixel S/N dependence on impact point. 1
Device simulation.
Single diode 15 μm pixel
No rad
1015
•
•
S/N varies over pixel between 12 and 4 before irradiation.
S drops to zero at edges after 1014 p/cm2.
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
35
Single pixel S/N dependence on impact point. 2
Device simulation.
Single diode 15 μm pixel
Device simulation.
4-diode 15 μm pixel
No rad
No rad
1015
•
•
Less variation in S/N varies over pixel before and after irradiation.
S at edges still usable after 1015 p/cm2.
1014
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
36
Outline
¾ Introduction.
¾ CMOS MAPS for science
Visible light
X-ray
UV
Charged particles
¾ A few words on radiation hardness
¾ Future and conclusions
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Future
CMOS Image Sensors (CIS) technology become more and more
widespread. Continuous improvement in noise: reduced leakage
current, in-pixel double correlated sampling (4T+pinned diodes)
More foundries proposing stitching Æ size of sensor limited by wafer
size (200mm diameter!)
Continuous improvement in noise: reset noise reduction by design.
New architectures and ways of modelling the reset noise.
Single figure (< 10 e- rms) noise achieved, approaching sub-electron
noise (one group already declares it for a linear, non-cooled sensor!)
For increased rad-hard: high-voltage, triple well technology, SOI, …
37
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Conclusion
CMOS Image Sensors can be used to detect photons from IR down
to low energy X-rays (direct detection), X-rays (indirect detection)
and charged particles (direct detection with 100% efficiency) … and
voltages
Large demonstrators built.
Working towards delivery of CMOS Image Sensors-based for
scientific instruments for space-science, particle physics and biomedical applications, also with large stitched sensors
And last but not least …
38
6th International Conference on Radiation Effects on
Semiconductor Materials, Detectors and Devices
10-13 October 2006, Florence (Italy)
Acknowledgements
The CMOS Sensor Design Group: A. Clark, J. Crooks, A. Fant, P.
Gąsiorek, N. Guerrini
Other designers and device simulation expert: M. Prydderch, G. Villani
DAQ developers: R. Halsall, M. Key-Charriere, S. Martin, J. Edwards, I.
Brawn, E. Freeman
RAL scientists: N. Waltham, M. Towrie, A. Ward, M. Pollard, M. Tyndel
This presentation shows results obtained within:
the MI3 collaboration led by prof. N. Allinson (Sheffield U)
the I-ImaS collaboration led by prof. R. Speller (UCL)
the MAPS for CALICE collaboration led by prof. P. Dauncey (IC)
the MAPS for Particle Physics and Space Science collaboration
39