1. Oct. 2004Intro to pixel detectors1 Introduction to silicon pixel detectors M. Garcia-Sciveres October 2004

2. Oct. 2004Intro to pixel detectors2 Charged Particle Tracking Charged particle shooting through space detector Negligible change to energy and direction Record of path taken by particle (:trajectory”). Not the same thing as a “picture” of the particle.

3. Oct. 2004Intro to pixel detectors3 Flying back and forth to CERN Think of jet airplanes. Easy to see where they have been by streaks left in the sky An insignificant bit of the energy and mass of the airplane (of the fuel) are used to produce the streak Atmospheric conditions affect sharpness, persistence, and amplification (natural cloud formation) of the streak. The earliest charged particle tracking detectors were cloud chambers

4. Oct. 2004Intro to pixel detectors4 Cloud and Bubble Chambers A gas-liquid (cloud chamber) or liquid-gas (bubble chamber) is exploited to greatly amplify microscopic perturbations due to a charged particle passing through gas or liquid. Both require taking a picture of the cloud streak or bubble trail left behind by the particle. See 5 th floor hallway “art”. The detection is a 3 step process: –Particle goes by, interacting with a few gas or liquid atoms- losing a small amount of energy in each interaction –A phase change is triggered by the interaction, which eventually involves huge numbers of near- by atoms in a chain reaction (domino effect). Cloud streaks or bubbles form. –A picture is taken (this is the “raw” data that is recorded and must then be analyzed). Bubble chambers are faster and more precise than cloud chambers, but both are painfully slow by today’s standards. The gas or liquid must be brought to its critical point for every event, it takes time for bubbles/clouds to form, and taking pictures is also slow. (Gargamel bubble chamber on display at CERN)

5. Oct. 2004Intro to pixel detectors5 Drift Chambers: much faster Cut out the middleman. Forget bubbles, clouds, and pictures. Wire chambers, time projection chambers. Measure the ionization left behind by a passing particle (through a gas). –Two step process: use avalanche in gas to amplify the primary signal (obviously faster than bubbles + pictures) Wire at –ve voltage Wire at +ve voltage + + + + - - - - Ions created by passing charged particle. Too small for electronic detection- just a few atoms When ions get close to HV wire they trigger an avalanche- each ion leads to ~10,000 new ions. Now you have a charge you can measure with an electronic circuit.

6. Oct. 2004Intro to pixel detectors6 Why Does This Work? (1) An avalanche is a natural phenomenon in gasses (think lightning) (2) Gasses consist 100% of neutral molecules- there is no natural contamination of ions an any level. –The ions needed to start an avalanche must be externally introduced (drift chambers) or are created when the electric field is high enough to rip atoms apart. –Until ions are introduced the gas will happily ignore the electric field. What’s wrong with drift chambers? –Drift: it takes time for ions to move towards the HV wires –Rate limit: must wait for all ions to clear away before device is sensitive to new particles. –Resolution: The diffusion of ions in the gas and the ionization statistics (randomness in location of primary ions) limit the ultimate resolution (~100um).

7. Oct. 2004Intro to pixel detectors7 Silicon Detectors: still faster and more accurate Cut out the middleman again! Detect the primary ionization directly Silicon (a solid) is much denser than gases => more primary ions are produced (this also means that the charged particle loses more energy in order to be tracked). –Just enough charge for direct measurement with electronic circuits. Silicon (a solid) has less diffusion than a gas => higher resolution (~10um). The catch: solids in general are not 100% made or neutral atoms, free of ions like gasses. –Need a solid that one can prepare to be 100% ion-free –Has to conduct electricity (so ions can flow to an electronic circuit)

8. Oct. 2004Intro to pixel detectors8 Silicon Detector Basics Metals: Atoms arranged in lattice that shares valence electrons. Number of free charge carriers ~10 22 /cc (~1,000 Coulomb of charge per cc). Impossible to remove this free charge. Pure silicon: atoms arranged in metal-like lattice, but number of “charge carriers” is ~10 10 /cc (~1nC/cc). –A modest electric field can remove this free charge. Impure silicon can have varying number and kind (+ or -) of charge carriers, depending on impurity type and concentration. –N-type has –ve carriers –P-type has +ve carriers –Typical concentration in range 10 12 – 10 18 /cc After free carriers are removed by electric field, silicon looks electrically like a gas- 100% neutral. This state is called “depleted”. A passing charged particle creates 22,000 (avg.) charge carrier pairs per 300  m. (~4fC). –If my laptop ran on 4fC/s the battery would last ~10 10 years (the age of the universe).

9. Oct. 2004Intro to pixel detectors9 Actual ATLAS Pixel Sensor A diode junction forms wherever p-doped and n-doped regions touch. Depletion always begins at the diode junction as reverse external voltage is applied. Hadron irradiation introduced p-type defects. Eventually this will cause the bulk to “type invert” and become p-type. At this point the diode junction shifts to the top. This was chosen on purpose because it allows to operate without fully depleting the bulk. Lightly n-doped bulk Heavily n-doped pixel implants (doping too heavy to deplete) Heavily p-doped back side contact Guard rings P-spray doping to isolate individual pixels Diode junction Bumps connect to implants

10. Oct. 2004Intro to pixel detectors10 Basic Charge Amplifier Q=CV Gain = 1/C(fF) V/fC Pixel chip input stage gain is ~300mV/fC C = 3.5fF Q 1 /C 1 = Q 2 /C 2 Pixel capacitance + parasitics ~ 100x inverse gain! Active amplifier. Increases effective capacitance by factor of “open loop gain” at the expense of adding a rise-time (active takes time) With Leakage current Compensation DC “current source” compensates for detector leakage

11. Oct. 2004Intro to pixel detectors11 Pixel Chip Front End comparator Input from Pixel sensor (bump goes here) preamp

12. Oct. 2004Intro to pixel detectors12 Front End Output Threshold (adjustable up and down). Comparator output is on when red line is above threshold, off if below.

13. Oct. 2004Intro to pixel detectors13 Front End Features Programmable threshold = Global Threshold + Pixel Threshold Calibration charge injection Ability to measure leakage current Time over Threshold (TOT) charge measurement –How long the red curve says above threshold depends on the size of the input charge Can easily change threshold for whole chip Can fine tune each pixel to compensate for response differences (Tuning) V1 V2 switch Input from detector Good old charge amplifier Injection capacitor (must be small)

14. Oct. 2004Intro to pixel detectors14 CMOS integrated circuits CMOS = combination metal oxide silicon Combination = p-type and n-type implants on the same wafer. silicon photoresist + ion implantation p-implant Gate oxide n-implant source gate drainsource gate drain

15. Oct. 2004Intro to pixel detectors15 The MOS transistor Transistor Diode ~Resistor Capacitor S D G S D S D

16. Oct. 2004Intro to pixel detectors16 Complexity in Numbers Complex circuits are possible using a large number of MOS transistors (~3M in FE-I3). In practice there are some special circuit elements also used in small numbers, such as metal-insulator-metal (MIM) capacitors and polysilicon resistors.

17. Oct. 2004Intro to pixel detectors17 The MOS transistor schematic of the FE-I3 charge amplifier

18. Oct. 2004Intro to pixel detectors18 Pixel Chip Readout Architecture Data driven Trigger Trigger driven Time stamp