How To Build Your Own Silicon Detector
How To Build Your Own Silicon Detector
How To Build Your Own Silicon Detector
Introduction
Past Presentations
Having to do with silicon detectors, but (naturally) construction methods not in scope on talk Except this one. ICs are a big part, but not the whole story This talk is chapter 2
(still not the whole story though)
15-Mar-2006 Compton Telescopes for High Energy Astrophysics (4.3 MB) by Steven Boggs, 01-Mar-2006 Specific Heat Measurements of Films and Crystals Using Si-micromachined Nano-calorimeters Frances Hellman, 15-Feb-2006 Semiconductor Radiation Detector Materials - Fact versus Fiction (7.2 MB) by Eugene Haller, 01-Feb-2006 Cancer "genomics" - Technological opportunities in cancer biology and management (7.0 MB) by Joe W. Gray, 02-Nov-2005 How to Design an Integrated Circuit (5.4 MB) by Peter Denes, Engineering Division, 19-Oct-2005 The Allen Telescope Array: A New Telescope For SETI and Radio Astronomy (13.7 MB) by Dave DeBoer, 05-Oct-2005 3D Silicon Detectors (15.3 MB) by Sherwood Parker, 21-Sep-2005 Molecular Electron Microscopy - Applications and Challenges (slides not available) by Ken Downing, 7-Sep-2005 Is Anybody Out There? Instrumentation for SETI (7.3MB) by Dan Werthimer 24-Aug-2005 Who needs better nuclear detector materials and how do we find them? (5.5 MB) by Stephen E. Derenzo, 20-Jul-2005 One-Dimensional Nanostructures as Subwavelength Optical Elements for Photonics Integration ( by Peidong Yang, 29-Jun-2005 The STAR Detector at RHIC (7.4 MB) by Jim Thomas, 15-Jun-2005 From Quarks to Quasars - Advanced Scientific CCDs (23.6 MB) by Stephen Holland, 18-May-2005 The Superconducting QUantum Interference Device: Principles and Applications (16.2 MB) by John Clarke, 4-May-2005 Biological Large Scale Integration (slides not available) by Stephen Quake, 13-Apr-2005 The ATLAS Pixel Detector (3.6 MB) by Kevin Einsweiler,
There are additionally important contributions to other experiments (for example by providing IC technology or even actual ICs)
CLEO at Cornell NA60 at CERN (actually fixed target) Phenix at BNL Others I forget or dont know about
M. Garcia-Sciveres --- How to build your own pixel detector 5
Where to begin?
Consider chips, sensors, performance specs as given How do you put them together? Isnt this the same problem as building laptops or digital cameras? (Why not just go to manufacturers that do that and have them do it?) Yes and no.
Not quite the same problem Those manufacturers wont have anything to do with us:
Vertical integration within huge company High volume and proprietary solutions Do not do work for others Big defense industry manufacturers are no different
We then make use of these things in creative ways And we insert home-made bits and pieces as needed Risk, risk risk!, but often cant be avoided (and sometimes can, but unfortunately isnt)
Technology evolution
Both industry and detector technology evolve gradually Each new detector takes established methods and adds a new twist Enough new twists lead to new ways of doing things I try to present sharp technology categories for clarity. In reality things are less black and white. There are parallel branches which are not necessarily exclusive of one another
Vertexing
Electrical components (sensors and circuits) can serve mechanical functions to reduce mass
Rigidity and stability Thermal management
The ladder model with a barrel geometry has been very successful Ideal for geometry with readout at ends only
Foam core with composite rails Glued sensors add stiffness Ceramic hybrid (next page) Machined beryllium bulkhead with active cooling
April 19, 2006 M. Garcia-Sciveres --- How to build your own pixel detector 10
Integrated circuits need mechanical support and services (external capacitors and interconnects) Thick film on ceramic technology
were merely an end user of this industrial technology Heat source Ceramic hybrid Cooling at this end LV power and signal cable to outside world
April 19, 2006
Hybrid at end of SVX ladder (1990) sensor Wire bonds (see later) integrated circuits
11
Glazes fired at 800-900C. This 5 conductor layer circuit has 15 mask steps and 10 firings. Unlike IC fabrication, customer can specify steps and materials as well as artwork
This can get one into trouble Chemical reactions happen fast at 800C. Minor incompatibilities between materials can have big effects
12
13
Sensor (green)
BOTTOM VIEW
TOP VIEW
14
A mix of thick film filled vias and thin film surface traces Manufacturing orchestrated between thick film vendor, thin film vendor, wafer dicing vendor, LBNL shops, and UCB microlab. Risk is that in full detector 300 such parts must operate for 10 years. Prototype reliability studies cannot generate the statistics needed to guarantee this.
After folding
CDF SVX-II L0 ladder Atlas SCT module
16
hybrids
Tricky to get away with this Detector has noise issues It has taken years to produce software and calibrations capable of making the data useful Nevertheless, Bs mixing is a very big payoff
April 19, 2006 M. Garcia-Sciveres --- How to build your own pixel detector 17
Si sensors
SVX-II z side L00
For any given radial position there is an optimum ladder geometry. A detector spanning a large radial range wants to have many types of ladders
ISL
18
The Module
As channel density and size increase further the ladder model breaks down. The module is conceptually a tile.
A passive mechanical skeleton is tiled over with modules.
6cm
April 19, 2006 M. Garcia-Sciveres --- How to build your own pixel detector 19
2.2cm
20
Material Penalty
Ionizing energy loss for particles passing through ATLAS pixel detector (in Radiation Lengths)
65% due to support structures and cables (worse at shallow angles) Only 15% due to silicon sensors => Only 15% of energy lost by particles is used for detection.
April 19, 2006 M. Garcia-Sciveres --- How to build your own pixel detector 21
Connecting them up
23
Wire boning
Wire bonding is the link between integrated circuits and the real world. It is a vast subject that could take up many 1hr talks. We are purely an end user of this industrial technology
25m round wire of just the right Al alloy Not a traditional weld. Not a solder joint. A special kind of metal-metal bond formed over only a part of the contact area IC bond pad of just the right Al alloy (a different one than the wire)
wire
Bottom view of wire in wedge
wedge
Bonding head
25
Silicon detectors
26
Bump bonding
Why the wire bond pitch stopped decreasing: Why not use the entire chip area for interconnection then? =area bump bonding
Bump bonding makes it possible for us to build hybrid pixel detectors. But it is a much more high maintenance technology than wire bonding. Not just me saying this:
27
28
Recall area bump pitch in industry roadmap is 70m even by 2018 ATLAS modules use 50m pitch! => non-standard process (expensive and low volume)
Sensor (below)
6.3cm
April 19, 2006
ICs
29
Future possibilities
Plain finer pitch
The silicon detectors built in 1990 were ahead of industry using 96m single row pitch and dual row wire bonding Exactly the same parameters are being used in todays detectors, significantly behind what industry is doing There are gains to be made by reducing bond pitch
High reliability ONLY when properly designed. If cables will bend repeatedly often need several design cycles to fix weak spots Flex occasionally used for hybrids as well as cables. Even more iterations typical in this case.
31
A constant of nature
32
A necessary evil
Sometimes flex PCB cant be avoided ATLAS pixel detector geometry leaves no space for anything thicker
Flex hybrid on PCB frame (if we had found a rigid circuit board technology thin enough we would have gladly used it instead) Combination of FlexMCC + Bare module
Rigid PCB frame used to manage flex handling This flex circuit is never bent. PCB frame provides mechanical support and temporary packaging for stand-alone operation
April 19, 2006
The flex hybrid is glued to the bare module without leaving the frame Flex module is detached only at time of loading on detector support structure.
33
ALICE
Done to reduce mass NOT done for industrial applications. Commonly done for shielding planes (low risk) Occasionally attempted for critical elements such as power distribution
Note that resistivity of Al is higher than Cu => reduction is modest Performance Impact of such modest reduction had better be significant before resorting to this!
34
Quality control
Recall one difference with industry is that we build only one detector and it has to work the first time. Different quality control problem Must track and store information individually for each component Only this permits investigating any single failure of any type as soon as observed Statistics of prototyping and construction are very much lower than detector operation- a very prominent effect in operation may only show up once in production!
A picture of every ATLAS pixel bump is viewable online This has helped understand failures during flip chipv
Each ATLAS pixel module has 16 wire bonds whose sole purpose is to be pulled This has helped understand and correct subtly problems
35
2000
CDF
Not just bigger, but at the same time faster. Bigger => automated assembly Faster => power, cooling and interconnect advances
2007
CMS
36
Robotic Assembly
ATLAS tracker: sensor tile alignment on robot, followed by manual full module assembly. Custom made robot CMS tracker: Full module assembly on robot (except wire bonding). Modified industrial pick-and-place robot
37
Assemble modules
Assemble rods
April 19, 2006
Rod burn-in
Ship to CERN
38
THE END
(Left over for Chapter 3)
How to decide on a detector layout.
Barrels, disks, number measurements / track Speed, occupancy, granularity Point resolution, 2-hit separation, stereo angle Iteration with construction constraints
CDF-II layout
Mechanical structures
The carbon composite era Cooling
Other problems
Reduction of service plant Optical readout Etc., etc.
CDF-II ISL frame
39