Muon detection in security applications and monolithic active pixel sensors Tracking in particle physics Gaseous detectors Silicon strips Silicon pixels Monolithic active pixel sensors Cosmic Muon tomography MAPS for Xray therapy SPIDER 3D vertical integration Many thanks to those colleagues from who I borrowed borrowed pictures and results Jaap Velthuis, University of Bristol 1/19
27 km long circular accelerator 100m under ground in Geneva Protons are accelerated up to 7TeV and collided at couple of locations along beam line Here giant detectors are built Take digital pictures and decide whether they interesting or not every 25 ns Particle Physics: Very fast imaging with very precise sensors Very fast processing of large data sets LHC Jaap Velthuis, University of Bristol 2/19
Detection concept Most of the tracking detectors in particle physics exploit local ionisation A charged particle passes through the detector and due to Coulomb interaction a current pulse is generated The charge is measured on various signal wires Particle position is reconstructed by weighting charge Jaap Velthuis, University of Bristol 3/19
Gaseous wire chambers Here the detection medium is gas Charge induces signal on a couple of wires Issues: Wires are very long (here 2.4m) Need large distances between the wires Need a lot of energy to ionise an atom Charge travels slowly Jaap Velthuis, University of Bristol 4/19
Ambiguities One is only position sensitive perpendicular to the wires (electrodes) Can add sensitivity by adding new layer Problem: 2 tracks yield 4 possible positions Solutions: Place strips under smaller angle at cost of loss of precision Make strips shorter X X Jaap Velthuis, University of Bristol 5/19
Silicon strips In silicon sensors one can: Make shorter strips (~6cm per wafer) Place them closer together Additional benefits: Charge travels fast Sensors can be thin (typically 300µm) Mechanically easier to handle Jaap Velthuis, University of Bristol 6/19
Even shorter strips: pixels Every pixel has own readout circuit Connected using flipchip bump bonding ATLAS: 50x400 µm 2 CMS: 100x150 µm 2 Still separate wafers for electronics and sensor Silicon pixels Jaap Velthuis, University of Bristol 7/19
Monolithic active pixel sensors Further integration: first amplification stage integrated into sensor Can use cheap electronics standard silicon Good S/N despite very thin sensitive layer (<20 µm) Charge collection by diffusion, so slow (~0.x µs) Limitation: can only integrate NMOS electronics in-pixel Jaap Velthuis, University of Bristol 8/19
Some examples We are currently working on two applications of particle physics sensors Cosmic ray tomography MAPS for Xray therapy State-of-the-art: INMAPS Jaap Velthuis, University of Bristol 9/19
Cosmic Ray tomography Are developing high resolution Resistive Plate Chambers for Cosmic Ray tomography Planes ~1x1m 2, strips 1.5mm apart Idea: compare track of cosmic muon before entering and after exiting a volume If muon hits material it will scatter Scattering depends on the material Advantages: RPCs cheap and easy to make in large areas Muons are highly penetrating Muons are naturally occurring so impossible to trigger against Many channels but due to multiplexed readout not a problem Takes ~30 sec to find tennisball size object in a suitcase volume CH U Fe Al Jaap Velthuis, University of Bristol 10/19
MAPS for X-ray therapy Issue in Xray therapy: dose verification Currently, many calibration measurements needed Water phantom placed in position of the patient and absorbed dose measured, then patient undergoes therapy Measurements during irradiation difficult Do not want to disturb beam upstream of patient Patient scatters too much to measure downstream These measurements are getting more important as the novel trend is to use multileaf collimators to make very precise thin beams (Intensity Modulated Radiation Therapy) PTW recently introduced the DAVID, multiwire ionization chamber to monitor the dose in real time All wires parallel thus information only 1 dimensional Since many interactions, not possible to add second layer Jaap Velthuis, University of Bristol 11/19
MAPS for Xray therapy MAPS can be very thin (<25µm) with still very good S/N Can use this to measure the dose in real 2D and really in real-time. Frame rates of 40 per second with 6x6cm 2 sensors (16Mpix) have been demonstrated Can go up to ~10Gpix/sec Here collimated beam and sensor read out 8 frames per second Added 15 cm Pb Slice parallel to X-axis shows know where beam is better than 60µm Improving cuts yields 50µm Looking for funding and/or industrial partner to take this forward Jaap Velthuis, University of Bristol 12/19
Towards intelligent pixels Limitation: can only integrate NMOS electronics in-pixel PMOS transistors need Nwell. This Nwell competes with diode for charge. If can live with 25-50% signal can now make intelligent pixels Pixel size: 30x30µm 2 Per pixel: two 8-bit DRAM cells, one 8-bit ROM cell and one 1-bit DRAM Forms ADC in-pixel with hit flag Can store one event while processing another Jaap Velthuis, University of Bristol 13/19
SPIDER: taking MAPS to the next level In SPIDER we are taking MAPS to the next level using INMAPS process Problem in MAPS: PMOS electronics need Nwell Nwell acts as charge collection diode So can t make PMOS without losing huge amount of Q New development: make deep pwell with Nwell inside can do CMOS with no charge loss Road to data processing in pixel for small signals Jaap Velthuis, University of Bristol 14/19
The SPIDER future: Cherwell Uses INMAPS plus 4T to achieve: 100% fill factor with integrated sensor and readout electronics Incorporation of complex logic within a in series of pixels Investigation of data reduction/clustering Low noise using transfer gate, CDS and in-pixel amplification Low power using rolling shutter readout First prototype submitted. Has ADC in islands This can be the next generation intelligent detector for Particle Physics SELECT Jaap Velthuis, University of Bristol 15/19 RESET Pixel Boundary 0 um pixel boundary BIAS VTH COL 4 T COL Control logic COL 4 T TRIM 1x SRAM MEMORY [0] MEMORY [1] MEMORY [2] MEMORY [3] MEMORY [4] MEMORY [5] COL 4 T (CDS circuit) C S + - MEMORY [6] WrEn MEMORY [7] C R
3D integration Technology for the day after tomorrow Multi-layer devices new trend chip industry Make chips faster by minimizing distance between components Processing each layer can be optimised Technologies can be mixed (e.g. 0.09 CMOS on 0.25 CMOS mixed mode) Concept simple : make chip for 1 layer Thin down to remove excess Si Place on top and make connections Optical In Power In Photo from MIT LL Opto Electronics and/or Voltage Regulation Digital Layer Sensor Layer Physicist s Dream Jaap Velthuis, University of Bristol 16/19 Optical Out Six inch wafer thinned to 6 microns Analog Layer 50 um and mounted on 3 mil kapton.
3D Laser rader imager 3D integration example 64x64 array, 30µm pixels 3 tiers 0.18µm SOI 0.35 µm SOI High resistivity substrate diodes Oxide to oxide wafer bonding 1.5µm vias dry etch 6 3D vias/pixel Jaap Velthuis, University of Bristol 17/19
Summary In Particle Physics the challenges are to make very fast, very large detector arrays of tiny detector elements with very fast processing The technology developed can also be used for various other applications like: Security Medicine INMAPS the way forward. Watch this space, exciting developments ahead Jaap Velthuis, University of Bristol 18/19
Many thanks go to Acknowledgements My colleagues at the University of Bristol My colleagues of the SPIDER collaboration (slide 14&15) My colleagues of the CMOS Sensor Design group at RAL (slide 13&14) My colleagues at AWE (slide 10) My colleagues at Bristol Hospital and Dr Hugtenburg (Swansea) (slide 11&12) I borrowed pictures from CERN/ATLAS/CMS (slide 2&7) ZEUS collaboration (slide 4&6) LEPSI/Strasbourg on MIMOSA (slide 8) Ray Yarema (Fermi lab) on 3D integration (slide 16&17) Jaap Velthuis, University of Bristol 19/19