STMicroelectronics chips
Transcript
STMicroelectronics chips
STMicroelectronicschips ATLASITKItalymee7ng,14/03/2016 AAlioAndreazza,M.Ci'erio,V.Liberali,C.Meroni,F.Ragusa,E. Zaffaroni-INFNMilano M.Biaso=,G.Darbo,G.Gariano,A.Gaudiello,C.Gemme,P. More=ni,L.Rossi,E.Ruscino,M.Sannino–INFNGenova C.Sbarra,A.SidoH,F.Fabbri–INFNBologna H.Shrimali,I.Yadav,A.Yoshi–IITMandi Overview • Reminderofpassivediodeperformance • TestofacHvepixels – amplifierdesign – injecHoncircuitry – performance • Submissionof3×4mm2array – STMpostponedsubmissionHmeto22ndApril – Hybridiza7oninAndrea’stalktoday – Simula7oncoveredinFederica’stalktomorrow Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 2 KC35A LayoutoftheSTMtestchip(KC53A): • 22I/OandVDD/GNDpads • 8pixels(50x250µm2):CollecJngdiode+Amp. • 4pixels(50x250µm2):CollecJngdiodeonly Vidyo,2febbraio2016 FirstversiondeliveredJuly2015: • shortsbetweenpowerlinesandground Fixedversiondelivered29Jan.16 A.Andreazza-HVR_CCPD Designandlayoutby: Ø H.ShrimaliandV.Liberali(MI) 3 Passivediodes:I-V INTERNAL DIODES SCHEMA • I-Vmeasurements(andC-Vaswell)needto keepintoaccountESDprotecHondiodes: – 0.8-1.4pAleakagecurrent@50V on50×250μm2pixels – wholechipleakagecurrent,dominatedby buriedlayerbelowI/Opads • “Programmed”break-downat70V. • STMwouldlikeustosetalsoprotecHons betweenSUBandGND(mulHple7V breakdownstructures) Sensor Buried layer (below pads) VDD-GND protection Other I/O pads CERN,29February2016 -3-2-10 ●D1 ▴D2 I[pA] Vbias[V] -60-50-40-30-20-10 Diode pad protections A.Andreazza-STMicroelectronics 4 Passivediodes:C-V CV MEASUREMENTS 0.20.40.60.81.01.2 • Measuredcapacitanceparameterizedas: C[pF] 1,4 1,2 ︎D1 ◆D2 ▼D3 ▲D4 1 0,8 – CP:parasiHccapacitance – Vo:built-involtage D1 (pF) D2 D3 D4 0,6 0,4 0,2 Substratedatasheet: 100-150Ωcm – -70 -60-50-40-30-20-10 -60 -50 -40 -30 -20 -10 0 0 Vbias[V] • MeasurementcompaHblewith expectaHon:k=1.8V-1pF-2 • CorrespondingdepleHonis 22-24μmat50V Q = eAdN A Outputpads0.12-0.15pF V = eN A d 2 / 2ε C= Q 2ε A = V d CERN,29February2016 A.Andreazza-STMicroelectronics 5 Passivediodes:sourcetests • Crosscheckratewith241Amsource • LowrateduetosmallacHvevolume – photoelectricpeakat~0.1Hzratewith1mCisource – rateproporHonaltodepleHon(asexpected) – peakshapedependentondepleHon(rangeof60keVelectroninSi~33μm) Counts/W Counts Energy(eV) with Am source [A E>30keV 30 E>40keV 25 20 Vbias[V] Counts E>50keV 15 10 5 0 Energy (eV) CERN,29February2016 (Time 70 min) A.Andreazza-STMicroelectronics 6 Ac7vepixels:powerconsump7on Fig. 6 Output at the NWELL terminal with applying current injecting pulse of 10 nA and 10 ns. Analog signal processing circuitry • CurrentabsorpHon:0.28mA(VDD=1.8V) The particle signal, detected at the NWELL terminal of the DNWSUB is amplified using folded cascode single ended amplifier. The MOM capacitor (MOM) is use to block DC quantity of • correspondsto35μApixel the input signal. The purely AC quantity is applied to the amplifier. The input is given to a pMOS • inagreementwithexpectaHonfromthedevicesimulaHon: device so that better noise performance can be achieved, comparing an nMOS input stage. The amplifier composed of DC biasing circuit [3] [4] and the gain stage [2]. The complete circuitry – is 9μAfortheamplifier including biasing is shown in Fig. 7. – current 25μAforthebiasingcircuitry Fig. 7 The folded cascode amplifier with the supporting circuitry. CERN,29February2016 A.Andreazza-STMicroelectronics 7 Ac7vepixels:testsetup • 8 active pixels • 7 with injection enabled CERN,29February2016 A.Andreazza-STMicroelectronics 8 Fig. 4 A pixel detector circuit under test, to me Ac7vepixels:injec7on Fig. 4 has two blocks at the hierarchical level 0. mirror circuit, used to bias the pmosll4 connected to PBIA The dark current injection simulating environ • InjecHonthroughacurrentmirror INJECTION WITH RESISTOR – nominal1/103reducHonfactor – operaHngbelowsaturaHon:1/70-1/94 200 ns pulser – nottoodifferentfromsimulaHon… Fig. 5 Particle injection s inj. padbe written using calculator. *The skill function can Vin *The calculator can be evoked from ADE –> Tools –> Ca i M1 imirror M2 Figure 2.5: Schematic diagram of current gener Vidyo,2febbraio2016 is a current pulse whose pulsewidth is nearly 10 ns. If the tion of charges in this period of time is nearly 600 e , the has to generate a 10 nA current. It’s impossible to gene current. A possible solution is to inject a higher current ratio of mirrored and injected current. So, M2 width cha than M1. In our design M2 width channel is 103 times la mirrored current is 103 times smaller than injected one. S current is 10 nA, the injected one is nearly 1 µA, which c by a current generator. So, M2 is a NMOS transistor whos A.Andreazza-HVR_CCPDand whose length is 400 nm. Instead, M1 length is the9sa Ac7vepixels:pulseshape OUTPUT PULSE InjecHon: 3μA×200ns Averagedpulse peak 60mV Probe pickup returntobaseline ~100ns Singlepulse: noise~10mVRMS Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 10 Ac7vepixels:linearity OUTPUT PULSE 450,0 ◼︎Iinj=3 µA (changing pulse length) ▲Iinj=5 µA (changing pulse length) ▲Δt = 100 ns (changing Iinj) 400,0 350,0 OUTPUT PULSE 200,0 180,0 300,0 160,0 140,0 120,0 I=3uA T=100ns I=5uA 200,0 V [mV] V [mV] 250,0 150,0 100,0 80,0 60,0 100,0 40,0 50,0 20,0 0,0 0,0 0 100000 200000 0300000 10000 20000 400000 30000 Q [e] Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 40000 500000 50000 60000600000 70000 80000 Q [e] 11 Ac7vepixels:comparisonwithsimula7on Comparison with Simulation (200ns) 140 mV Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 263 mV 12 Ac7vepixels:comparisonwithsimula7on Output on KC53AB ACTIVE Pixel Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 13 ● We measured the output noise of the active Ac7vepixel:noise pixels (no input current) OutBuf Noise ( 6.6 mV rms) The noise RMS without nothing connected to the oscilloscope is ~6 mV 10kHz Vidyo,2febbraio2016 1MHz 100MHz A.Andreazza-HVR_CCPD 10kHz 1MHz 100MHz 14 TPM1layout 1.18 mm pads and general purpose space 2.5 mm one Pixel array 2.8 mm o specifiche 1 3 mm 4 mm • Inspiredfromdemonstratorlayout (picturebelow) • Lateralspacetobetunedtomatch alignmentwithFE-I4 • Letoverspaceadbo'omofpixelregion isonly20μm,maybeincreasedin mulHpleofthe50μmpixelsize. tà di 2-8 mA 0.25 mm available space out Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 15 Pixelcell Pixel block-diagram Genova + Mandi + Milano Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 16 HV-CMOSinATLAS CPPM • Twolargesizedemonstrators: Monday , February 22, 2016 2 – AMS180nm(Genève-Karlsruhe),available – LFoundry(CPPM-Bonn),submi'ed MLM3_RUN-1(CPPM) - Fev 2016 ,505mm 10mm 10mm SEAT 1 SEAT 2 5,84mm Othertechnologiesbeing tested: • XFAB(SOI) • TowerJazz Test Features • Seat 1 : LF-CPIX ( see details) • Seat 2 : LF-CPIX with new guards rings strategy • Seat 3 : test features Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 17 Testbeams Possible (?) time slots for hardware test Vidyo,2febbraio2016 André Rummler | ITK Pixel Testbeam Activities | CERN, 15.02.2016 A.Andreazza-HVR_CCPD 15 18 Summaryandshort-termplans • Detectedsignalfrompassivediodes – sHlltofullyunderstanddepleHondepthandrate – TCAD/Geant4simulaHons(helpwelcome) – tryingtoorganizeirradiaHontests (setupindevelopmentwithPolitecnico) ATLAS TCAD workshop CPPM, 11-12 May. • FixedKC53ABreceived: – allcomponentsareoperaHonal – gainlowerthanexpectedandlargedispersion – sHlltounderstandmismatchbetweenmeasurementsand simulaHon • Largepixelmatrixtobesubmi'edbymid-April – canbematchedtoFE-I4chips – possiblyalsotoRD53prototype Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 19 BACKUP BCD8TechnologySTMicroelectronics BCD8 Technology (STMicroelectronics) BCD = Bipolar + CMOS + DMOS http://www.st.com/web/en/about st/bcd.html • D. AmongthecompeHngtechnologiesBCD8hasseveralappealingfeatures: Riccardi, A. Causio, I. Filippi, A. Paleari, L. Vecchi, A. Pregnolato, P. Galbiati, C. Contiero, BCD8 from 7V to 70V: a new 0.18 µm technology platform to address the evolution of – availabilityofdifferentdevicesintegratedinthesameprocess applications towards smart power ICs with high logic contents, in Proc. 19th Int. Symp. Power – epitaxialprocess:caneasilygrowondifferentsubstrates. Semiconductor Devices & ICs, 2007. http://dx.doi.org/10.1109/ISPSD.2007.4294935 – possibletoreachthickdepleHonlayers:30μmlooksanopHmalcompromisebetween R. Roggero, G. Croce, P. Gattari, E. Castellana, A. Molfese, G. Marchesi, L. Atzeni, C. Buran, signalandmaterialthicknessofthedetector A. Paleari, G. Ballarin, S. Manzini, F. Alagi, G. Pizzo, BCD8sP: An advanced 0.16 µm – long-termavailability:itisoneofthemajorproducHonlineforSTautomoHveproducts. technology platform with state of the art power devices, in Proc. 25th Int. Symp. Power Semiconductor Devices & ICs, 2013. http://dx.doi.org/10.1109/ISPSD.2013.6694422 HVR_CCPD-LNSPAC Catania,17novembre2015 21 Noiseperformance S/N~2for600esignal Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 22 CMP cmp.imag.fr • BCD8nowavailablealso throughCMP – easiertogetnewpeople started – increaseofcostswith respecttoiniHalinformal contacts: • 2600Euro/mm2 ifArea≤5mm2 • 13000Euro +[(Area-5)*2100Euro] if5mm2<Area<15mm2 Vidyo,2febbraio2016 A.Andreazza-HVR_CCPD 23