3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 1 Surface Characteristics of ATLAS07 n-on-p FZ Silicon Detectors J. G. Wright, S. Lindgren, C. Betancourt, N. Dawson, N. Ptak, H. F.-W. Sadrozinski, J. von Wilpert Santa Cruz Institute for Particle Physics, Univ. of California Santa Cruz, CA USA Y. Unno, S. Terada, Y. Ikegami, T. Kohriki Institute of Particle and Nuclear Study, KEK, Oho 1-1, Tsukuba, Ibaraki , Japan K. Hara, H. Hatano, S. Mitsui, M. Yamada University of Tsukuba, Institute of Pure and Applied Sciences, Tsukuba, Ibaraki , Japan A. Chilingarov, H. Fox Physics Department, Lancaster University, Lancaster LA1 4YB, United Kingdom 1

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 2 Large-Scale Production -> Involve HPK 2 After much work done within ATLAS and RD50, what is left to know? – Can adequate strip isolation be achieved after irradiation? – Does it depend on specific surface treatment? (p-spray, p-stop) – Do detectors with better strip isolation have lower breakdown? – Are complicated punch- through structures needed for adequate punch-through protection? Silicon Detectors for the High Luminosity Upgrade Signal collection requires - High voltage operation Higher rate of particles requires - higher segmentation of detecting electrodes - acceptable data transfer rate N-strips in P-bulk wafer (n-in-p) - always depleting from strip side - lower cost than n-in-n - collecting electrons

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 3 3 History of ATLAS07 HPK X1: p-stop, p-spray+p-stop Weak spots identified Mask modification X2 ~ with modified mask X3 many doping densities S2 p-stop, p-spray 90 wafers S1 p-stop, 30 wafers

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 4 4 n-in-p Sensor Past Japanese studies – 4 inch (100 mm) wafers FZ (~6k Ωcm) MCZ (~900 Ωcm) – 6 inch (150 mm) wafers FZ1 (~6.7k Ωcm) FZ2 (~6.2k Ωcm) MCZ (~2.3k Ωcm) – FZ, MCZ available at HPK ATLAS07 submission – 6 inch (150 mm) wafers FZ1 (~6.7k Ωcm) (FZ2 (~6.2k Ωcm) Miniature sensors – 1cm x 1cm Irradiation studies Full size prototype sensors for Stave program – 9.75 cm x 9.75 cm 4 segments: two "axial" and two "stereo" (inclined) strips Short strips

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz n-in-p Miniature Sensors Radiation damage study – Strip Isolation (Zone1, Zone2, Zone3) Structure: p-stop, p-spray, p-stop+p-spray Density: 1x, 2x, 4x, 10x10 12 ions/cm 2,... – "Punch-through Protection" structures (Zone4) – Narrow metal effect (Zone5) – Wide pitch effect (Zone6)

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 6 6 Evaluations Irradiations – 70 MeV protons at CYRIC (Tohoku Univ., Japan) – Reactor neutrons at Ljubljana (Slovenia) Measurements – Production Testing of full-size sensors (see Nobu’s talk) – Onset of Microdischarge I-V Hot electron (IR camera) – Surface: Interstrip resistance Interstrip capacitance – Punch-through Protection Dynamic resistance with a constant bias voltage to the backplane

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 77 Using the detectors DC-pads The setup purpose is to measure the current in the test strip due to the applied voltage on the neighbors using a parameter analyzer. Measurement was done at different bias voltages. Data taken is plotted as test current against the neighbor voltage Interstrip Resistance Measurements

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 8 Interstrip Resistance Results 8 To first order, interstrip resistance depends not on the specific zone, but depends on the total p-dose applied. Higher total p-dose means better strip isolation after irradiation. All Series 1 detectors exhibit a good post-rad Rint (>10^8 Ohms) behavior, even after being irradiated with protons up to 1e13 neq

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 9 Interstrip Capacitance 9 AC pads are used. 5 probes are used; one on the test strip, one on each neighbor, and two more to ground the next neighbors.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 10 R int vs. C int 10 Helpful to make scatter plot between R int and C int Strip isolation is best in the upper left corner, and worst in the lower right. Dependence of post-rad C int on specific zone is seen. Zone 5 (narrow metal) has the highest Post-rad C int without providing better breakdown performance.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 11 Punch-Through Protection PTP : (against large strip voltages due to field collapse in high fluences) 11 Zone 4 detectors include an additional punch-through structures. Apply a voltage to DC pad and measure the induced current between DC pad and bias resistor. The effective resistance is then Punch-through resistance is defined as where R bias is the resistance of the bias resistor. “Base line” Z3 is similar, but with 70  m distance

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 12 Punch-Through Voltage 12 Punch-through Voltage is defined as the voltage where For Pre-rad detectors, the punch-through voltage is dependent on the wafer number (i.e. the total p-dose). Dependence on the total p-dose is seen after irradiation, higher p-dose means lower PT Voltage. Zone 3 exhibits similar PT voltage to Zone 4, without the having a complicated PT structure. Further, Zone 3 shows adequate protection even at high fluences.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 13 DC Punch-Through with Neighboring Implants at test Voltage DC to bias punch-through measurements have been performed keeping the neighboring implants to the test DC implant at the test voltage Initial results show that in the negative test voltage regime, which corresponds to real world voltages on implant during charge collection for p-type detectors tested, the punch-through voltage is minimally changed by voltages on neighboring strips: Punch-through happens to the bias ring.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 14 Conclusions I As usual, well-behaving sensors are boring; …. but PTP comes to the rescue! After irradiation, Zone 3 detectors (Gap = 70  m) have a similar punch-through voltage as Zone 4 detectors, which are made with a specific punch-through protection structure (Gap = 20  m). Needs explanation! The acceptable punch-through voltage of the Zone 3 sensors (No PTP!) with p-stops of 4*10 12 cm -2 extends to proton fluences beyond p/cm 2. Is this relevant for our application? Need to test PTP dynamically by flooding sensors with charge. HPK:N-side isolation with p-stops works well ! All HPK detectors have a breakdown voltage > 900V, exceeding the specifications. To first order, the interstrip resistance does not depend on the specific zone, but instead depends on the total p-dose of p-impurities on the surface (p-stop + p-spray). Reasonable post-rad interstirp isolation (R int >> R input ), better than RD50 p-spray sensors. The interstrip capacitance shows little change after irradiation and is dependent on the specific zones. Zone 5 (narrow metal) has the highest interstrip capacitance after irradiation (Don’t use narrow metal!). The punch-through voltage depends on the total p-dose in all configurations (p-stop only, p- stop+spray, p-spray only). Wafers with the highest total p-dose have a higher punch-through voltage, which holds true even after irradiation. Initial results indicate that DC punch-through happens to the bias rail, and not to the nearest neighboring strips.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 15 Punch-Through Protection and Collapse of the E-Field in SSDs J. Wright, C. Betancourt, N. Ptak, H. F. –W. Sadrozinski, J. von Wilpert, Santa Cruz Institute for Particle Physics SCIPP, Univ. of California Santa Cruz, CA USA

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 16 Ac-coupled sensors are reverse biased through a filter network (series R and parallel C) connected to the backplane. Bias ring is held to ground, and the voltage on a DC pad and AC pad are read out on an oscilloscope. Any charge deposited in the detector will induce a charge on the DC pad, and also to the AC pad which is coupled to the DC pad through the coupling capacitor. Termination of AC strip is ~1.5k  via the amplifier, here we use 50  to see the fast AC signal. Testing Dynamics of E-Field Collapse and High Implant Voltages T. Dubbs, M. Harms, H. F.-W. Sadrozinski, A. Seiden, M. Wilson, “Voltages on Silicon Microstrip Detectors in High Radiation Fields”,, IEEE Trans. Nuclear Science, Volume 47, Issue 6, Dec 2000:1902 – Alessi IR cutting laser is used to deposit huge amounts of charge (>10 10 electron/hole pairs ~ 10 6 MIPs ~ 1 Rad) inside the detector, which collapses the field inside.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 17 Alessi IR (1064 nm) Cutting Laser Signal Laser sends pulses every 3-4 μs. Number of pulses in train is linearly proportional to laser power LP. No pulses are sent below LP=505. Increasing laser power causes the integral of the signal to be more spread out. This is presumably caused by the E- field being lower in the detector as more charge is pumped in, leading to larger effects from diffusion. DC AC AC signal (yellow) is prompt signal from the IR. DC voltage on implant (green) is much slower and indicates the potential changes in the sensors. Changing field modifies the response of the sensors to IR.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 18 E-Field Breakdown with IR Cutting Laser Breakdown of E-Field End of Pulse Train Breakdown of the Electric field is local, i.e. the field in the whole detector does not collapse, just around the area where the laser is focused. Far from the laser, field is still present and strips are able to read out the signal pulses. Closer to the laser source, more charge is deposited, leading to a collapse of the field in that region. End of Pulse Train Once field is collapsed, subsequent pulses are swept from the bulk slowly, a time which can be controlled by the RC network attached to the backplane.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 19 Signal Parameterization after E-Field Breakdown tbtb tltl tctc The signal can be broken up into four regions; t b – The time at which the field breaks down. t l – The time at which the pulse train ends. t c – The time at which all mobile charges are cleared from the bulk (“stable potential?”) t n – The time at which the normal electric field is restored. tntn

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 20 Breakdown of the E-Field Breakdown of field starts here Far from the laser source, the field is not broken down, with the largest implant voltage seen at the end of the pulse train. Field breakdown is seen at about 3mm from the laser source, so that a large portion of the field in the whole detector is broken down. In region near laser focus, breakdown voltage exceeds punch-through voltage measured in DC tests For region where field does not break down, voltages remains below measured punch-through voltage Field Breakdown

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 21 Dependence of Field Collapse on Backplane Capacitor Both the voltage when the field collapses, and the time at which it happens depends on the value of the capacitor connected in parallel to our detector. There is a logarithmic dependence of Vb as a function of capacitance, with higher capacitance leading to a larger Vb, presumably due to larger amount of charge being able to be dumped into the sensor.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 22 Grounding Nearest Neighbor DC Pads Since the field collapse extends over a finite area of the sensor, neighboring strips are at about the same potential Grounding the nearest neighbor DC pads significantly lowers Vb and Vc. This presumably is closer to what is happening in our DC punch- through measurements, as then the DC pads are tied to ground via the bias rail. Initial results show that Vc seems to saturate at about the DC punch- through voltage, and Vb saturates at about 80V. Several detectors broke when using a large capacitor and at high bias voltages (>200V) using this method. DC PT bias

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 23 Conclusions II An infrared cutting laser is used to simulate the real beam accident conditions, which creates large amount of charge in the bulk so that the implants float to a large voltage after the E-field has collapsed. Using the infrared laser creates a situation where the neighbors of a given test strip will be at about the same voltage, while in DC measurements the neighbors are held to ground via the bias rail (since there is no bulk current). Laser tests involving grounding the nearest neighbors might reproduce the DC measurements, but are unrealistic and causes detectors to be damaged easily. Scanning the profile of signal indicate that large areas of the detector exhibit field breakdown, ~ 5 mm. The dynamics of the breakdown involves the charge stored on the filter capacitor. During the initial collapse of the field, very large voltages appear at the implants, up to close to the bias voltage. A dependence on the size of the filter cap is observed. After the radiation stops, the voltage on the implants revert to DC-like smaller voltages, and might be governed by the PTP mechanism. Our measurements indicate that during breakdown of the E-field dynamic voltages appear on the implants much larger than expected from the DC PTP measurements. The magnitude of these voltages question the merit of the PTP structures and of the DC method to evaluate them.

3D-Manchester-Feb-2010 H. F.-W. Sadrozinski, UC Santa Cruz 24 We acknowledge the good collaboration within the ATLAS Upgrade Sensor Collaboration and with the Hamamatsu Photonics team. The invaluable assistance of CYRIC and the Triga at Ljubljana and their staff in carrying out the radiation is acknowledged. ATLAS Upgrade Silicon Strip Detector Collaboration: H. Chen, J. Kierstead, Z. Li, D. Lynn, Brookhaven National Laboratory J.R. Carter, L.B.A. Hommels, D. Robinson, University of Cambridge K. Jakobs, M. Köhler, U. Parzefall, Universitat Freiburg A. Clark, D. Ferrere, S. Gonzalez Sevilla, University of Geneva R. Bates, C. Buttar, L. Eklund, V. O'Shea, University of Glasgow Y. Unno, S. Terada, Y. Ikegami, T. Kohriki, KEK A. Chilingarov, H. Fox, Lancaster University A. A. Affolder, P. P. Allport, H. Brown,G. Casse, A. Greenall, M. Wormald, University of Liverpool V. Cindro, G. Kramberger, I. Mandic, M. Mikuz, Josef Stefan Institute and University of Ljubljana I. Gorelov, M. Hoeferkamp, J. Metcalfe, S. Seidel, K. Toms, University of New Mexico Z. Dolezal, P. Kodys, Charles University in Prague J.Bohm, M.Mikestikova, Academy of Sciences of the Czech Republic C. Betancourt, G. Bredeson, N. Dawson, V. Fadeyev, M. Gerling, A. A. Grillo, S. Lindgren, P. Maddock, F. Martinez-McKinney, H. F.- W. Sadrozinski, S. Sattari, A. Seiden, J. Von Wilpert, J. Wright, UC Santa Cruz R. French, S. Paganis, D. Tsionou, The University of Sheffield B. DeWilde, R. Maunu, D. Puldon, R. McCarthy, D. Schamberger, Stony Brook University K. Hara, H. Hatano, S. Mitsui, M. Yamada, N. Hamasaki, University of Tsukuba M. Minano, C. Garcia, C. Lacasta, S. Marti i Garcia, IFIC (Centro Mixto CSIC-UVEG), and K. Yamamura, S. Kamada, Hamamatsu Photonics K.K. Acknowledgments