Signal-to to-noise with SiGe 7 th RD50 Workshop CERN SCIPP UC Santa Cruz Signal-to-Noise, SiGe 1
Technical (Practical) Issues The ATLAS-ID upgrade will put large constraints on power. Can we meet power and shaping time requirements with deep submicron CMOS? Achieving sufficient transconductance of the frontend transistor typically requires large bias currents. The changes that make SiGe Bipolar technology operate at 100 GHz for the wireless industry coincide with the features that enhance performance for our application. Small feature size increases radiation tolerance Extremely small base resistance (of order 10-100 Ω) affords low noise designs at very low bias currents. Can these features help us save power? Will the SiGe technologies meet rad-hard requirements? Signal-to-Noise, SiGe 2
Evaluation of SiGe Radiation Hardness The Team D.E. Dorfan, A. A. Grillo, J. Metcalfe, M Rogers, H. F.-W. Sadrozinski, A. Seiden, E. N. Spencer, M. Wilder SCIPP-UC Sanat Cruz Collaborators: A. Sutton, J.D. Cressler Georgia Tech, Atlanta, GA 30332-0250, USA M. Ullan, M. Lozano CNM, Barcelona S. Rescia et al. BNL Signal-to-Noise, SiGe 3
Thanks, Michael & Maurice! Irradiated Samples Pre-rad ATLAS Upgrade Outer Radius 4.15 x 10 13 1.15 x 10 14 3.50 x 10 14 Mid Radius Inner Radius 1.34 x 10 15 3.58 x 10 15 1.05 x 10 16 Signal-to-Noise, SiGe 4
Radiation Damage Mechanism Forward Gummel Plot for 0.5x2.5 µm 2 I c,i b vs. V be Pre-rad and After 1x10 15 p/cm 2 & Anneal Steps 10-4 10-6 IC (pre-rad) IB (pre-rad) IC (1e15, anneal) IB (1e15, anneal) Collector current remains the same Radiation damage increases base current causing the gain of the device to degrade. Gain=I c /I b (collector current/base current) I c, I b [A] 10-10 10-12 10-14 Ionization Damage (in the spacer oxide layers) VB [V] The charged nature of the particle creates oxide trapped charges and interface states in the emitter-base spacer increasing the base current. Displacement Damage (in the oxide and bulk) The incident mass of the particle knocks out atoms in the lattice structure shortening hole lifetime, which is inversely proportional to the base current. Signal-to-Noise, SiGe 5 10-8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 V be [V] Base current increases after irradiation
Annealing Effects Annealing of 0.5 um x 2.5 um: Current Gain beta vs. Ic pre-rad and after 1*10 15 p/cm 2 and anneal steps 1000 Before Irradiation 100 After Irradiation & Full Annealing Current Gain, β 10 1 After Irradiation 0.1 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 I c [A] We studied the effects of annealing. The performance improves appreciably. In the case above, the gain is now over 50 at 10µA entering into the region where an efficient chip design may be implemented with this technology. The annealing effects are expected to be sensitive to the biasing conditions. We plan to study this in the future. Signal-to-Noise, SiGe 6 pre-rad 1e15, no anneal 1e15, 5 days RT 1e15, +6 days RT+1 day 60deg C 1e15, +1 day 100deg C 1e15, +6 days 100deg C
Initial Results Current Gain beta vs. Ic for 0.5 um x 10um pre-rad and for all Fluences including full annealing Before Irradiation 1000 100 Increasing Fluence Lowest Fluence Current Gain, β 10 1 Pre-rad 3e13 1e14 3e14 1e15 3e15 1e16 Highest Fluence 0.1 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 IC [A] After irradiation, the gain decreases as the fluence level increases. Performance is still very good at a fluence level of 1x10 15 p/cm 2. A typical I c for transistor operation might be around 10 µa where a β of around 50 is required for a chip design. At 3x10 15, operation is still acceptable for certain applications. Signal-to-Noise, SiGe 7
Feasibility for ATLAS ID Upgrade Qualifications for a good transistor: A gain of 50 is a good figure of merit for a transistor to use in a front-end circuit design. Low currents translate into increased power savings. Fluence: 3.50E14 p/cm 2 (2.17x10 14 n eq /c m β=50 Transistor Size µm 2 Ι c irrad I c anneal 0.5x1 2.E-06 0.5x2.5 4.E-06 5.E-08 0.5x10 3.E-05 8.E-07 0.5x20 5.E-05 2.E-06 4x5 9.E-06 5.E-07 At 3.5x10 14 in the outer region (60 cm), where long (10 cm) silicon strip detectors with capacitances around 15pF will be used, the collector current I c is low enough for substantial power savings over CMOS. Fluence: 1.34E15 p/cm 2 (8.32x10 14 n eq /cm 2 ) β=50 Transistor Size µm 2 Ι c irrad I c anneal 0.5x1 3.E-05 1.E-07 0.5x2.5 7.E-05 4.E-06 0.5x10 4.E-04 9.E-06 0.5x20 6.E-05 4x5 1.E-04 1.E-05 At 1.34x10 15 closer to the mid radius (20 cm), where short (3 cm) silicon strip detectors with capacitance around 6pF will be used, the required collector current I c is still only 5 10 µa. We expect even better results from 3rd generation IBM SiGe HBTs. Signal-to-Noise, SiGe 8
Frontend Simulation Results Signal-to-Noise, SiGe 9
First Guess at Potential Power Savings Using simulation of first SiGe frontend, an estimate for power can be obtained. Compare to 0.25 µm CMOS design of J. Kaplon et al., 2004 IEEE. FEATURE CHIP TECHNOLOGY 0.25 µm CMOS ABCDS/FE J. Kaplon et al., (IEEE Rome Oct 2004) IHP SG25H1 SCT-FE Preliminary design Power: Bias for all but front transistor 330 µa 0.8 mw = 30 µa (conservative).06 mw Power: Front bias for 25 pf load 300 µa 0.75 mw 150 µa 0.30 mw Power: Front bias for 7 pf load 120 µa 0.3 mw 50 µa 0.10 mw Total Power (7 pf) 2x10 15 Total Power (7 pf) 1x10 15 1.1 mw 0.16mW Total Power (25 pf) 2x10 14 1.5 mw 0.36 mw 0.34mW Signal-to-Noise, SiGe 10
Noise in Bipolar Transistors (H. G. (H. G. Spieler): 2 n F 2 2 2 v = in FT i s en Ctot tot in Ts Q + C C + = det C i 2 n Ic = 2qe + Ibias β 2 (k T) e 4 2 2 B n = + ktr bb qe Ie Temperature dependence of bias current ties the noise to the temperature management of the slhc tracker, which is already complicated by annealing of leakage current and depletion voltage. Noise depends on product of leakage current I Bias and shaping time T s (applies to CMOS as well) Signal-to-Noise, SiGe 11
Short Strips Signal-to to-noise using SiGe Frontend Electronic noise small, leakage current important Trade shaping time against operating temperature 1500 Temperature: -10 o C vs.- 20 o C Noise vs. Shaping time, Short Strips Fixed Fluence: 2.2 10 15 neq/cm 2 (short strips) The maximum bias voltage is 600 V S/N vs. Temperature, Short Strips 20 RMS Noise [e-] 1000 500 c=6, f=0-10 o C c=6, f=2e14-10 o C c=6, f=2e15-10 o C c=6, f=2e15, -20deg o C S/N 15 10 5 C = 6, 10 ns C = 6, 15 ns C = 6, 20 ns 0 5 10 15 20 25 Shaping Time τ [ns] Noise(20 ns & -20 o C) = Noise(15 ns & -10 o C)! 0-35 -30-25 -20-15 -10-5 Temperature [ o C] No advantage to very low temps! Signal-to-Noise, SiGe 12
Long Strips Signal-to to-noise using SiGe Frontend Electronic noise dominant, leakage current not so important Expect no sensitivity to shaping time or operating temperature 1500 Temperature: -10 o C Noise vs. Shaping time, Long Strips 20 Fixed Fluence: 2.2 10 14 neq/cm 2 (long strips) The maximum bias voltage is 600 V S/N vs. Temperature, Long Strips RMS Noise [e-] 1000 500 C=15, f=0 C=15, f=2e14 S/N 15 10 5 C = 15, 10 ns C = 15, 15 ns C = 15, 20 ns 0 5 10 15 20 25 Shaping Time τ [ns] 0-35 -30-25 -20-15 -10-5 Temperature [ o C] Noise at -10 o C and 20 ns acceptable! No advantage to very low temps! Signal-to-Noise, SiGe 13
Inter-strip Capacitance One of the most important sensor parameters contributing to the S/N ratio Depends on the width/pitch ratio of the strips and on the isolation technique (p-stops, p- spray). SMART reported large bias dependence on p-type detectors, due to accumulation layer. Cint [F] 2.0E-11 1.8E-11 1.6E-11 1.4E-11 1.2E-11 1.0E-11 8.0E-12 6.0E-12 4.0E-12 2.0E-12 0.0E+00 Interstrip Capacitance 0 100 200 300 400 500 Bias Voltage [V] 100 µm pitch 14-5 250krad Pre-rad Cint = 1.5 pf/cm Irradiation with 60 Co (250 krad) reduces the bias dependence, as expected (c.f. talk by C. Piemonte) SMART 14-5 p-type FZ low-dose spray w/p = 15/50 V dep = 85 V (I. Henderson, J. Wray, D. Larson, SCIPP) Signal-to-Noise, SiGe 14
Expected Performance for p-type p SSD Details in : Operation of Short-Strip Silicon Detectors based on p-type Wafers in the ATLAS Upgrade ID M. Bruzzi, H.F.-W. Sadrozinski, A. Seiden, SCIPP 05/09 Conservative Assumptions: α p = 2.5 10-17 A/cm (only partial anneal) C total = 2 pf/cm V dep = 160V + β Φ ( with 2.7* 10-13 V/cm 2 ) (no anneal) (= 600V @ Φ = 10 16 neq/ cm 2 ) σ 2 Noise = (A + B C)2 + (2 I τ s )/q A = 500, B = 60 S/N for short strips vs. fluence for different bias voltages: S/N 35.0 30.0 25.0 20.0 15.0 300um, -20deg, 400V 300um, -20deg, 600V 300um, -20deg, 800V S/N 35.0 30.0 25.0 20.0 15.0 200um, -20deg, 400V 200um, -20deg, 600V 200um, -20deg, 800V no need for thin detectors, unless n-type: depletion vs. trapping 600V seems to be sufficient 10.0 5.0 0.0 10.0 300 µm 5.0 200 µm 1.E+12 1.E+13 1.E+14 1.E+15 1.E+16 Fluence [neq/cm 2 ] 0.0 1.E+12 1.E+13 1.E+14 1.E+15 1.E+16 Fluence [neq/cm 2 ] do need update on fluences Signal-to-Noise, SiGe 15