1. Workshop on 3D and p-type Sensors
Phil Allport (University of Liverpool)
26/02/10
5th Trento Workshop on Advanced Silicon Radiation Detectors
University of Manchester
Introduction
Status of 3D Sensor Technology and Recent Results
Latest results with Planar Detectors after Extreme Irradiation
New Theoretical Understanding of Signals at the Highest Doses
Conclusions

2. Possible Post-Chamonix Schedule for with 2 Major Shutdowns

3. CMS Tracker Upgrade with Trigger Layers
Both Experiments need new trackers to survive 3000fb-1 at the sLHC
CMS needs to exploit PT information from tracker to keep trigger rate at 100 kHz
major new feature for CMS tracker - ideas how to do it are still developing
Current assumption is that there will probably be dedicated PT layers, providing prompt trigger information
i.e. different from more conventional, triggered pipeline chip, layers
Several ideas for triggering layers summarised here
Longer barrel layers to match PT layers, at present locations
one possible
“strawman” layout
X section through one
quarter of tracker
Remaining end caps with present locations
Stereo layers
Stereo rings
h = 2.5
PT layers to cover full h range
Paired Trigger layers
Pixel System
Pixels: 4 barrel layers + increased size Endcap
could be 3 disks/endcap?

4. ATLAS Tracker Upgrade Layout
Barrel Pixel Tracker Layers:
Short Strip (2.4 cm) -strips (stereo layers):
Long Strip (9.6 cm) -strips (stereo layers):
r = 3.7cm, 7.5cm, 15cm, 21cm
r = 38cm, 50cm, 62cm
r = 74cm, 100cm
400 Pile-up
100 GeV Electrons

5. Radiation Background Simulations for sLHC
Completed design of neutron moderator to reduce silicon damage fluences (cyan coloured regions in layout on right).
More detailed geometry and material description implemented in FLUKA SLHC simulations for ID. In particular, investigate impact of latest service and support material information.
Providing feedback to community on maximum fluence and dose values at critical positions, for calculation of damage and signal to noise.
×2 → 1.3 × 1015 neq/cm2

6. Technology in Current Highest Dose Regions
LHC vertex detectors all used n+ implants in n- bulk:
Because of advantages after heavy irradiation from collecting electrons on n+ implants, the detectors at the LHC (ATLAS and CMS Pixels and LHCb Vertex Locator) have all adopted the n+ in n- configuration for doses of 5 – 10 × 1014 neqcm-2
Requires 2-sided lithography
ATLAS 100 million Pixels
LHCb
Vertex Locator
Z(mm)=0-990

7. Motivations for P-type Silicon Wafers
Starting with a p--type substrate offers the advantages of single-sided processing while keeping n+-side read-out:
Processing Costs (~50% cheaper).
Greater potential choice of suppliers.
High fields always on the same side.
Easy of handling during testing.
No delicate back-side implanted structures to be considered in module design or mechanical assembly.
So far, capacitively coupled, polysilicon
biased p-type devices fabricated to
ATLAS provided mask designs by:
Micron Semiconductor (UK) Ltd
(existing ATLAS barrel: 6cm×6cm and
RD50 miniatures: 1cm ×1cm),
CNM Barcelona (RD50 miniatures: 1cm×1cm),
ITC Trento (RD50 miniatures: 1cm×1cm)
CiS and HLL MPI Munich (miniatures)
Hamamatsu Photonics HPK (1cm×1cm and 10cm×10cm Full ATLAS prototypes)

8. n-in-p (p-type substrate) FZ Irradiations
900V
500V
GRad ie ~10MGy
Micro-strip Tracker
Pixel Tracker 8

9. ATLAS Pixel Module Upgrade
Current pixel module → Single chip pixel modules on stave

12. Stanford Nano-fabrication Facility
Stanford Nanofabrication Facility (SNF) is a Stanford Nano Science and Technology Lab. All the fabrication work is performed at SNF in the CIS building, which was completed in 1985.
The Stanford Nanofabrication Facility is a 10,000 sq. foot, Class 100 cleanroom housing a complete suite of tools for the micro- and nano- fabrication of devices

13. 3D process consists of > 100 steps
WAFER BONDING
(mechanical stability). After
complete processing this support
wafer will be removed.
2. PHOTOLITHOGRAPHY
3. MAKING THE HOLES
4. FILLING THE HOLES
5. DOPING THE HOLES AND ANNEALING
6. METAL DEPOSITION
Due to modest topography on the wafer surface (e.g. poly-filled holes), it makes it difficult to spin on a uniform layer of photoresist, which is a critical step repeated many times.
When plasma etching either holes, edges etc., the use of thick resist is required, however, high-viscosity resists seem to have more particulates and clumps of resist associated with them.
Mechanical fractures can occur during the plasma dicing or fusion bonding steps.
When patterning the metal, some aluminum contact strips would etch away.

14. 3D Detector – Fabrication Steps
SiO2 1μm
Si Support Wafer
1μm SiO2 Al
Si Substrate
SiO2 1μm
Si Support Wafer
Si Substrate
N+ poly N+
N+ poly
1μm SiO2
Si Support Wafer
1μm SiO2
Si Substrate N+
Support Wafer
SiO2 1μm
SINTEF attempted to reproduce the Stanford process in a more industrial/ production enviroment in 2007
Wafer Bonding
DRIE 250µm deep, 14µm diameter
Fill Holes with n-doped poly
Etching of polysilicon
Oxide cap
SiO2
SiO2 1μm
Si Support Wafer
Si Substrate N+
Trench 7μm
600Å oxide Al
p-hole
14μm
Oxide cap
SiO2
SiO2 1μm
Si Support Wafer N+
Trench 7μm
p-spray P+
Trench 7μm
SiO2 p+ N+ p+ P+
SiO2 1μm
Si Support Wafer
P+ poly
Etching of polysilicon and oxide etch
DRIE, holes and trenches
Filling holes and trenches with p-doped poly

15. 1. 3D lateral cell size can be smaller than wafer thickness, so
2. in 3D, field lines end on electrodes of larger area, so
3. most of the signal is induced when the charge is close to the electrode, where the electrode solid angle is large, so planar signals are spread out in time as the charge arrives, and
4. Landau fluctuations along track arrive sequentially and may cause secondary peaks
5. if readout has inputs from both n+ and p+ electrodes,
shorter collection distance
2. higher average fields for any given maximum field (price: larger electrode capacitance)
3. 3D signals are concentrated in time as the track arrives
4. Landau fluctuations (delta ray ionization) arrive nearly simultaneously
5. drift time corrections can be made
Speed: planar D 4.

20. Medipix

21. CNM: Leakage current after irradiation CV Extracted Vdepletion
Irradiated p-type, no annealing, -10ºC
Annealing time
Charge multiplication?
Leakage current, -10ºC, 1016 neq/cm2
Leakage current increases with irradiation dose
Two competing effects in annealing curves:
Annealing of leakage current at low V
Charge multiplication?
More pronunced and earlier for longer annealing times
CV Extracted Vdepletion

26. New Hamamatsu R&D Wafer
Strips (30-89)
10 mm x 10 mm
Pixel (1-29)
10.5 mm x 10 mm
PTP study
Strips mini
58-71
Slim edge study
Diodes
1-20
Guard ring study
21-32
Diodes (1-32)
4mm x 4mm
HPK 6 inch (150 mm) wafers, p-type and n-type
Y. Unno, 5th "Trento" Workshop, Manchester, UK, Feb., 2010 26

27. Slim Edge - Measurement
Square root of V_bias is linearly dependent on the edge distance
Reflecting the depletion along the surface
Distance can be ≤500 µm for the bias voltage up to 1 kV
(Almost) No safety margin for a thickness of 320 µm
More than a factor of 2 safety margin for a thickness of µm
One of FE-I3 pixel sensor
Thinned Pixel Sensors
Wafer thickness
P-type: 320 µm
N-type: thinned to 200 µm
150 µm possible
Excellent I-V performance
≤1,000 V
Both in the p-type and n-type

28. Studies With Micron Semiconductor Miniature Strip Detectors on Enhanced Signal After Extreme Fluences
Expected Signal
No Multiplication 28

29. Manufacturer Comparisons
At the lowest fluence, there may be some differences between Micron and Hamamatsu
Most likely systematic error
At higher fluence, no trend seen
900 V
500 V
Appears once multiplication dominates, no strong differences between manufacturer, isolation and pitch’s tested. Weird!! 29
A. Affolder – 5th Trento Workshop, Feb 2010, Manchester, UK

30. A. Affolder – 5th Trento Workshop, 24-26 Feb 2010, Manchester, UK
Thickness Overview
Below 5×1014 neqcm-2, charge advantage to being thicker
Above 3×1015 neqcm-2, charge advantage to being thinner 30
A. Affolder – 5th Trento Workshop, Feb 2010, Manchester, UK

31. Annealing of S/N, 1E15 n cm-2 0oC -25oC
Noise is the sum in quadrature of shot noise and parallel noise (taken from the Beetle chip specs, and estimated as 600ENC)
0oC
-25oC 31

32. CCE Measurements Thinned Detectors (MPI HLL)
Expected before irradiation
Measurements performed at T=-30 oC (for sensors irradiated at F=1x1015 neq/cm2 ) and at T=-40oC (for the sensor irradiated at F=3x1015 neq/cm2), before any intentional annealing
The error bands correspond to 500 e- uncertainty estimated on each measured point.
At F=(1-3)x1015 neq/cm2 signal sizes close to the ones measured before irradiations are obtained
Further measurements at F=(3-10)x1015 neq/cm2 are presently ongoing

46. Conclusions
The schedule for upgrades at LHC already start to put significant pressure on the sensor community to provide more radiation hard pixel and strip designs
Studies with both 3D and planar p-type (and n-in-n as current pixels) now all show results which promise adequate signal/noise even at the most extreme doses
Planar p-type being processes with 3 mainstream sensor companies with excellent pre-irradiation results and very consistent post-irradiation performance
Commercialisation of 3D is progressing with SINTEF, but results mainly with research laboratory production: Stanford, CNM Barcelona ICT Trento
It seems the understanding of charge multiplication is needed to understand recent results with both sensor types
At first, avalanche multiplication seemed not to fit all effects observed but huge progress has been made is terms of modelling including trapping and device studies with the post-irradiation sensors can explain the data with modest 3-4 multiplication

47. Our understanding is getting better and better but there is still a challenge to predict the avalanches