First Results with the Prototype Detectors of the Si/W ECAL David Strom University of Oregon Physics Design Requirements Detector Concept Silicon Detectors - Capacitance and Trace Resistance Implications of Accelerator Technology Choice MIPS, sources and laser Si-W work personnel and responsibilities M. Breidenbach, D. Freytag, N. Graf, G. Haller,J. Deng SLAC Electronics, Mechanical Design, Simulation R. Frey, D. Strom UO Si Detectors, Mechanical Design, Simulation V. Radeka BNL Electronics This work includes contributions from Oregon students Tyler Neely and Eric Fitzgerald. LCWS 05 1 19 March 05 David Strom UO
ECAL Design Requirements Optimal contribution to the reconstruction of multijet events: Excellent separation of γ s from charged particles Efficiency > 95% for energy flow Excellent linkage of ECAL with tracker (important for SiD) Good linkage of ECAL with HCAL Good reconstruction of π ±, detection of neutral hadrons Reasonable EM energy resolution (< 15%/ E) Physics case: jet reconstruction important for many physics processes. LCWS 05 2 19 March 05 David Strom UO
Longitudinal Sampling, 30 layers needed for EM energy resolution σ EE 20% XE X is the sampling in radiation length. Useful for K 0 tracking, etc. Can tolerate small, random inefficiency See talks by Eckhard von Toerne LCWS 05 3 19 March 05 David Strom UO
Importance of Granularity Figure of merit for energy reconstruction is f E max(r M, 4d) R cal where R M is the Molière radius, d is the detector pad size and R cal is the inner radius of the calorimeter (factor of 4 somewhat arbitrary) Example (OPAL SiW luminosity monitor, 1X 0 radiator, 3mm gap) OPAL Two-cluster resolution efficiency 1 0.8 0.6 0.4 0.2 0 0 5 10 15 20 25 30 35 40 Radial cluster separation (mm) d = 2.5mm, R M 17mm LCWS 05 4 19 March 05 David Strom UO
The costs of the calorimeters, coil, and muon system have where n is 2 3. cost R n cal Thus a 10% increase in the Molière radius of the calorimeter leads to a > 20% increase in cost of the detector for constant f e. Conclusion: try and make the calorimeter as dense as possible LCWS 05 5 19 March 05 David Strom UO
Critical parameter: gap between tungsten layers. Config. Radiation Molière length Radius 100% W 3.5mm 9mm 92.5% W 3.9mm 10mm +1mm gap 5.5mm 14mm +1mmCu 6.4mm 17mm Assumes 2.5mm thick tungsten absorber plates Angle subtended by Moliere radius (mrad) 20 18 16 14 12 10 8 6 4 2 SD, Radius to calorimeter = 1.25m With copper heat sink No copper 0 0.5 1 1.5 2 2.5 3 Calice 3mm gap with 1.7m TESLA radius gives R M R Cal = 13mrad Gap width (mm) 1.0 0.8 0.6 0.4 0.2 Angle subtended by Moliere radius (deg) LCWS 05 6 19 March 05 David Strom UO
Si-W Calorimeter Concept ECAL Inner Tracker 1.25m Rolled Tungsten Circuit Board Transverse Segmentation ~5mm 30 Longitudinal Samples 1/2 Energy Resolution ~15%/E 1.1-1.3 Meters 3.6 Meters Layer Assembly Silicon Wafers LCWS 05 7 19 March 05 David Strom UO
Silicon Concept Readout each wafer with a single chip Bump bond chip to wafer To first order cost independent of pixels /wafer Hexagonal shape makes optimal use of Si wafer Channel count limited by power consumption and area of front end chip May want different pad layout in forward region Front End Chip LCWS 05 8 19 March 05 David Strom UO
Critical parameter: minimum space between tungsten layers. Evolving capacitor packaging may eliminate need for dimples. LCWS 05 9 19 March 05 David Strom UO
Can we get the heat out? Back of the envelope calculation of change in temperature: Thermal Conductivity of W alloy 120W/(K-m) Thermal Conductivity of Cu 400W/(K-m) temperature deg. C 20 18 16 14 12 10 8 2.5 mm of W 100mW 1.0 mm of Cu 100mW 1.0 mm of Cu + 2.5 mm of W at 100mW 2.5 mm of W 40mW Need to reduce heat to below 100mW/wafer. 6 4 2 0 0 20 40 60 80 100 120 140 Length (cm) LCWS 05 10 19 March 05 David Strom UO
Silicon Detector Design DC coupled detectors (avoids bias resistor network) Two metal layers Keep Si design as simple as possible to reduce cost Cross talk looks small with current electronics design Trace capacitances (up to 30pF) are bigger than the 5pF pixel capacitance LCWS 05 11 19 March 05 David Strom UO
Ten Hamamatsu detectors are in hand LCWS 05 12 19 March 05 David Strom UO
Measurements on Silicon Detector Prototypes Leakage Current Looks Fine: Leakage current (na) 2.5 2.25 2 1.75 1.5 1.25 1 0.75 0.5 0.25 0 0 10 20 30 40 50 60 70 80 90 100 Bias Voltage (Volts) (10nA for 1µs gives only 250 electrons noise) NB: Neighboring pixels are not grounded. LCWS 05 13 19 March 05 David Strom UO
Expected contributions to detector capacitance: 5.7pF from pixel capacitance (C geom ) 20pF for sum of trace capacitance and capacitance from other traces connecting to other pixels. (C stray ) Pixels under the bump-bond array have additional stray capacitance from probing and bonding pads (currently 100pF) Expected curves Vdep +V C tot = C stray + C bi geom V bias +V V bi bias < V dep C tot = C stray + C geom V bias > V dep LCWS 05 14 19 March 05 David Strom UO
Capacitance (pf) 120 100 100 khz 1 MHz 80 60 40 20 0 0 10 20 30 40 50 60 70 80 90 100 Bias voltage (V) Typical CV curve as measured in lab LCWS 05 15 19 March 05 David Strom UO
Fraction depletion depth 1 0.8 100 khz 1 MHz 0.6 0.4 0.2 0 0 10 20 30 40 50 60 70 80 90 100 Bias voltage (V) Relative depletion depth as a function of voltage. LCWS 05 16 19 March 05 David Strom UO
Mean stray capacitance measurement obtained from a fit to the CV curve: Expected 100kHz 1 MHz 23.0 ± 0.2 pf 21 ± 1 pf 22 ± 1 pf Measurement agrees with expectation for 0.9 µm thick oxide and 6µm wide traces (3.1 pf/cm). Series resistance for 1µm by 6 µm : Expected (pure Al) Measured 47 Ω/cm (57 ± 2)Ω/cm Measurement slightly larger than nominal LCWS 05 17 19 March 05 David Strom UO
Impact of Detector Technology on Detector Design In a warm machine, exceptional pixels with large capacitance or series resistance lead to degraded time tag measurements Small impact on tagging performance since bad channels can be deweighted in determining the average time of a track In a cold machine, exceptional pixels with large capacitance or series resistance lead to a higher rate of noise events in buffers Could lead to inefficiency late in the bunch train due to buffer overflow LCWS 05 18 19 March 05 David Strom UO
Location of high resistance and capacitance pixels a.) Longest trace 10 cm b.) Radial trace 7 cm c.) Congested area near bump bond array LCWS 05 19 19 March 05 David Strom UO
For areas a and b fundamental limit to noise is given by (for e.g. correlated double sampling) ENC Rs C tot 4 KT q 2 e R s 1 2τ where R s is the series resistance, C d and τ is the shaping time of the electronics. For τ = 1µs, R s = 580 Ω and C tot = 40 pf this gives 600 electrons noise, which is not really a problem. We can slightly improve noise performance by decreasing the trace width, perhaps by a factor of 2, i.e. where w is the trace width. ENC Rs w LCWS 05 20 19 March 05 David Strom UO
In region c, near the bump bonding array, we will have a large number of traces crossing a pixel. No series resistance, but amplifier FET noise similar: Possible ways to decrease capacitance in region c: Move probing pads on to pixels. Decrease trace width in area near central pixels, here EN Camp w Use a long skinny chip (e.g. 100 µm x 600 µm grid) After these three measures, worst case capacitance is 70 pf. LCWS 05 21 19 March 05 David Strom UO
Other more radical alternatives Polyimide (kapton) can be used instead of SiO 2 as insulator for traces Oxide thickness to 5µm possible. Minumum trace with probably 10µm Could reduce stray capacitances by a factor of 2 or more Hamamatsu does not currently provide metal-on-polyimide products, but we could increase the thickness of the wafer and the SiO 2. SINTEF (Norway) may be producing detectors based on 6 inch wafers with metal-on-polyimide within the next year. ( Possible collaboration with Brookhaven to produce masks.) LCWS 05 22 19 March 05 David Strom UO
Test Setup for Cosmics, Sources and Laser Modified probe station, allows laser to be target on entire detector IR microscope objective used to focus laser to 10 µm spot Bias applied to backside of detector using insulated chuck LCWS 05 23 19 March 05 David Strom UO
Contact made to test pads on bump bonding array using an AC probe Cables add 20 pf of additional capacitance, but noise performance is somewhat better than readout chip Use AMPTEK 250F preamp, shapers with τ 1µs and a digitizing oscilloscope to mockup expected electronics PC board with 1 cm 1 cm silicon pad detector used for cosmic trigger visible under chuck Test Setup detector probing LCWS 05 24 19 March 05 David Strom UO
Response of detectors to Cosmics (Single 5mm pixel) Simulate LC electronics (noise somewhat better) width = 780 electrons 10 4 Electrons 28000 26000 10 3 mean = 25750 + - 300 electrons 24000 10 2 22000 20000 10 18000 16000 LC Prototype 16 Pixel Detector 1 14000 12000 10-1 10000-5000 0 5000 10000 15000 20000 25000 30000 35000 40000 0 20 40 60 80 100 120 MIP Average vs. Voltage Voltage Electrons Errors do not include 10% calibration uncertainty (no source calibration) LCWS 05 25 19 March 05 David Strom UO
Response of Detectors to 60KeV Gamma s from Am 241 Entries/mV 1600 1400 Constant 1215. 8.126 Mean 88.65 0.2907E-01 Sigma 5.376 0.2521E-01 Entries/mV 3500 3000 Constant 2870. 14.00 Mean 89.09 0.1363E-01 Sigma 5.190 0.1649E-01 Entries/mV 4000 3500 Constant 3398. 15.22 Mean 89.55 0.1231E-01 Sigma 4.842 0.1341E-01 1200 2500 3000 1000 2000 2500 800 1500 2000 600 1500 400 1000 1000 200 500 500 0 0 0 50 60 70 80 90 100 110 120 130 140 150 50 60 70 80 90 100 110 120 130 140 150 50 60 70 80 90 100 110 120 130 140 150 Shaper Output(mV) Shaper Output(mV) Shaper Output(mV) Possible 1% wafer-wafer calibration? Width of distributions corresponds to 970 electrons noise. Pixels under test are on outer edge of wafer includes larger series resistance contribution than cosmic data. LCWS 05 26 19 March 05 David Strom UO
Laser Studies Signal (MIPS) 50 45 40 6ns Width 20ns Width λ = 1064 nm 35 IR penetrates into wafer 30 25 20 Allows controlled study of large and small pulses 15 10 5 0 0 100 200 300 400 500 600 700 800 Nominal Laser Energy (pj) LCWS 05 27 19 March 05 David Strom UO
Conclusions A narrow gap silicon tungsten detector for LC physics is attractive First round of prototype silicon detectors perform as expected Detectors can be produced with workable values of stray capacitance and series resistance some changes need for cold design LCWS 05 28 19 March 05 David Strom UO