Eleuterio SpiritiILC Vertex Workshop, April On pixel sparsification architecture in 130nm STM technology ILC Vertex Workshop April 2008 Villa Vigoni 1.Our (RomeTre group) experience on MAPS Mimoroma1 chip: brief description and some results 2.The Mimoroma2 chip Main goals and constrains Pixel noise simulation Readout architecture Pixel architecture Simulation of sparsified pixel Autozero simulation 3.Conclusions Eleuterio Spiriti ( INFN RomaTre )

Eleuterio SpiritiILC Vertex Workshop, April Super-BILC Spatial resolution 10 μm2.5 μm Multiple scattering Duty cycle100 %0.5 % Occupancy~1% Monolithic pixels in vertex detector for future accelerator Layer 0

Eleuterio SpiritiILC Vertex Workshop, April Chip mimoroma2 ( STMicroelectronics 130nm technology ) Chip mimoroma1 ( TSMC 250nm technology ) Our (short) experience on MAPS at RomeTre INFN group 3 mm x 4 mm 2 mm x 3 mm 25  m x 25  m Non sparsified side Sparsified side

Eleuterio SpiritiILC Vertex Workshop, April MIMOSA ISE/TCAD simulation Pitch = 20  m Nwell diode 3x3  m  vertex Substrato  19 cm -3 Epitaxial  cm -3 pwell  17 cm -3 n+  20 cm -3 Collection time  ns Monolithic Pixel physical simulation

Eleuterio SpiritiILC Vertex Workshop, April Alfa event small pitch (17  m ) mimoroma1 chip (first MAPS designed at INFN RomaTre) Double alfa event large pitch (34  m ) 32 x 32 pixel Alfa event small pitch (17  m ) 64 x 64 pixel Mimoroma1 chip measurement results Fe55 X X ray source

Eleuterio SpiritiILC Vertex Workshop, April Only NMOS transistors: no competing n-wells 2.Small available area (pitch  m) 3.Common threshold for all pixels in a chip 4.Threshold voltage and curren mismatch in submicron CMOS 5.Noise:  Temporal noise  White and 1/f noise  Charge injection  Digital to analog cross-talk Mimoroma2 chip: main goals and constrains Characterization of the signal (epitaxial thickness and quality) level provided by the STM 0.13  m technology Implementation of an on-pixel sparsification architecture - digital and analog output

Eleuterio SpiritiILC Vertex Workshop, April Ten different matrix implemented Five 32x16 pixels arrays with 20x20  m size Five 64x32 pixels arrays with 10x10  m size Technology characterization from the signal collection point of view Different parameters for the non sparsified part of the chip ParameterValue 1Value 2 Pixel structure 3TSB Pitch 20  m10  m Diode dimension 1  m x 1  m1.5  m x 1.5  m SF transistor sizeSmall gainLarge gain Power supply2.5 V1.2 V

Eleuterio SpiritiILC Vertex Workshop, April Proposed MAPS readout architecture Column sparsified readout Sparsified data End of column ADC Column “token passing” to read only pixels over thresholds Key point: On pixel sparsification

Eleuterio SpiritiILC Vertex Workshop, April Paper submitted at the “IEEE Custum Integrated Circuit Conference” September 2008, San Jose, California ( under review ) Detailed description of the sparsified pixel circuitry

Eleuterio SpiritiILC Vertex Workshop, April Pixel architecture Size 25  m x 25  m ~ 70 transistors Sensing diode

Eleuterio SpiritiILC Vertex Workshop, April Pixel 3T noise simulation Noise level after CDS (1Hz-10THz) ~ 1.5 mV - ~ 25 elec. N-well/p-epi sensing diode model  level-0. leakage current/white noise?

Eleuterio SpiritiILC Vertex Workshop, April NAND logic Montecarlo simulation. ( used inside the Pixel ) Process and mismatch fluctuations included.

Eleuterio SpiritiILC Vertex Workshop, April Autozero correction voltage simulation. Autozero correction voltage versus external bias autozero voltage fixed at 500 mV. Autozero correction voltage and external bias autozero voltage produce a correction differential current to the discriminator to compensate the two unbalancing voltage generators 2 MHz clock

Eleuterio SpiritiILC Vertex Workshop, April Sparsification logic Montecarlo simulation (100 runs) Input signal injected on sensing diode ~22 mV (~ 350 e) Output signal after discrimination ( “100 % efficiency” ) Mismatch spread on sensing diode ~182mV (~ 2900 e) Sample and hold timing ( integration time) Treshold at the minimum value needed to have zero false trigger ( “100% purity” ) ~ 350 electron is a rough estimation of seed pixel charge in STM 0.13  m

Eleuterio SpiritiILC Vertex Workshop, April Conclusions ➢ Design effort was driven by the highly demanding performances required by the vertex detectors for future colliders ➢ The proposed architecture for the on-line data sparsification was designed and simulated ➢ Simulation results of the on-pixel signal processing seems promising, considering that this will be the first run in ST 0.13um technology. We had many unknows and we tried to be prepared for the worst case  Large room for improvements expected after feedback from measurements.  The autozero correction technique could be a key point.  Chip submitted through CMP Service November 2007, testing will start next week