Simulation of signal in irradiated silicon detectors Gregor Kramberger , DESY Hamburg Devis Contarato, University of Hamburg G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Outline Motivation Basics of simulations calculation of induced current ATLAS silicon detectors simulation micro-strip detectors (trapping induced charge sharing) pixel detectors More radiation hard detectors (towards SLHC) thin pixel detectors novel semi-3D design Summary G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Motivation CERN: RD48 RD39 RD50 All LHC experiments will use silicon (diamond?) for vertex detectors! Degradation of performance of irradiated silicon detectors (bulk): increase of |Neff| Significant improvement in last years to improve performance: Material: DOFZ, Czochralski, epitaxial material Geometry: semi-3D, 3D and thin detectors Operational conditions: cryogenic operation, current induced devices Increase of leakage current (independent on material) loss of drifting charge – trapping (TCT studies – systematic measurements of trapping times) CERN: RD48 RD39 RD50 Determine the signal formation in irradiated detectors! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Questions How does the geometry of the electrodes influence the performance? Can silicon detectors be successfully operated at fluences around 1016 cm-2 ? G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Basics of simulation point charge “bucket”: electric and magnetic field trapping weighting field constant 1/D if pad or strip dimension>>D In general complex - highest close to collecting electrodes Irradiation: Calculation of electric and weighting potential performed by custom made software and ISE-TCAD package! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Point charge track y x y x holes electrons ionizing particle track electron-hole pair holes electrons Point charge x y ionizing particle track buckets 1 mm apart track G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

What is not considered in simulation! Charge generation – a uniform charge generation along the track was assumed (of course a full GEANT simulation for non-uniform charge generation would be more appropriate for dealing with delta electrons …) Homogenous effective dopant concentration – (so called effect of double-junction is not taken into account – how important is it really?) No further electronic processing of induced current G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Atlas – silicon strip detectors Detector: thickness=280 mm, strip pitch=80 mm , strip width=18 mm weighting potential electric potential Uw Neff=-6x1012 cm-3 y[mm] y[mm] x[mm] x[mm] x[mm] y[mm] carriers drifting towards the strips contribute to a large part of the induced charge far from constant as it is in a diode G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

NOTE THAT DIODE SIGNAL IS ALWAYS THE SAME! p+-n-n+ p+-p-n+ irrad. Feq=5x1013 cm-2 p+,n+ n+,p+ n,p n+-n-p+ n+-p-p+ irrad. Feq=5x1013 cm-2 NOTE THAT DIODE SIGNAL IS ALWAYS THE SAME! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

 n+ strips p+ strips just good enough U>Vfd are simulated – detectors will be always fully depleted (|Neff| = 0.02 cm-1x Feq) CCE for detectors with n+ strips is higher than for p+ strips (LHC~10%) S/N~9 after 10 years of operation (U=450 V) p+,n+ n+,p+ p bulk just good enough  G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

p+ strips Small difference in CCE for detectors with n+ and p+ strips! Loss of charge high U: trapping low U: diffusion p+,n+ n+,p+ p bulk ±U x Average over all strips yields <4% lower signal than for central strip at 450 V! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Trapping induced charge sharing measuring p bulk x ±U At 450 V around 1000 e are induced on left and right neighbors for central strip! After Feq=2x1014 cm-2 (F=3.3x1014 p cm-2) p+ strips n+ strips diode G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

the amount of charge induced depends also on strip geometry measuring p bulk x ±U trapped charge trapping absence of trapping this effect is far more important in irradiated detectors with p+ strips the amount of charge induced depends also on strip geometry This effect is present also in other devices – not unique to silicon! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Constant weighting field 1/D equal charge measured in the front and in the back electrode electrode hit by ionizing particle p+ - induced charge on neighboring electrodes has the same polarity as for the hit electrode n+- induced charge on neighboring electrodes has the opposite polarity as for the hit electrode n+ - higher signal in hit electrode p+ - wider clusters G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

“Segmentation” in terms of charge collection means how much weighting field deviates from constant (diode) G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Atlas – pixel detectors Pixel: 400 mm x 50 mm, 23 mm implant width, 250-280 mm thick! Electric potential (linear electric field – diode-like) Weighting potential (far from constant – not diode-like) G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Trapping times: te~1.8 ns, th~1.3 ns Feq=1x1015 cm-2 U=Vfd~350 V, D=250 mm n-type p-type trapping switched off in simulation Trapping times: te~1.8 ns, th~1.3 ns Neff = 0.0071 cm-1 x Feq (DOFZ silicon used, k=0.62) Current integrated over 25 ns! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Overdepletion becomes less important at high Feq d/D=62% but Q(d)/Q(D)=73% Shape of the weighting field (small at x~D) in combination with trapping results in a smaller contribution to the charge from the region at x~D. Overdepletion becomes less important at high Feq Around 10000 e at Feq=1015 cm-2 (most probable – not mean) Even at 200V more than 6000 e, U>400V seems enough with DOFZ G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

experimental data U [V] good agreement with measured data Vfd experimental data U [V] good agreement with measured data only small increase of charge at U>Vfd (saturation of vdr ) clear deviation from: (more linear) at higher fluences it is difficult to extract depletion voltage G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

signal on neighbors is below typical cuts applied (2000 - 3000 e) particle track p+-pixels would perform better than n+-pixels even if operated at U>Vfd. signal on neighbors is below typical cuts applied (2000 - 3000 e) Charge sharing caused by trapping is a very strong argument for using n+ pixels instead of p+! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Thin Pixel Detectors   The way to cope with high fluences (1016 cm-2): - at high Neff detectors can be fully depleted  short collection times i.e. collection distance  small signal: need for radiation hard low noise electronics ATLAS pixel ~ 150 eo, can sustain 2x1015 cm-2 Other issues: low mass – small X0 small pixel dimensions ~50mm – low capacitance – noise fast read-out high series noise power consumption as low as possible G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

No difference between n and p pixels is expected for PW/D<1! Geometries considered (3x3 pixels array was simulated): 70 x 70 mm (50 mm implant width) thicknesses: 25,50,75,100 mm Only central hits were considered: diode-like electric field! Weighting potential along the central line! D=100 mm No difference between n and p pixels is expected for PW/D<1! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Simulated current at Vfd ! D = 50 mm Neff = 0.0071 cm-1x Feq (DOFZ, no donors left after 1015 cm-2 ) New materials can reduce the increase of |Neff|: 50 mm thick epi-diodes (small donor removal) Czochralski material Simulated current at Vfd ! The charge collection times are short – so are trapping times (at Feq=1016 cm-2 of order 0.15 ns ) What are the consequences? G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

p-type pixels n-type pixels At best only 1000-2000 e at high fluences Small difference between different pixel thicknesses at 1016 cm-2 Much better performance of n-type pixels for PW/D<1 almost no difference between U=VFD and U=VFD+100 V around 800 e more at 5·1015 cm-2 Very high VFD (<E>=30000 V/cm for 50 mm thick detector) G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Trapping induced charge sharing D=100 mm, Pitch=70 mm, Width=50 mm. operated at Vfd particle track Diffusion is negligible due to the short collection times Very beneficial n-type pixels (possible use of signals of opposite polarity to enhance S/N) Up to 30% of the signal is induced on neighbors for p-type pixels, which is usually not enough to reach the threshold for detection – it is lost!!!! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

What if we make a device that has ideal Neff~0 ? visible light - hole injection High resistivity EPI material or Current Induced Devices The signal that we can get out of thin pixel detectors after Feq=1016 cm-2 is between 1000 e – 1600 e ! Is this enough? G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Novel semi-3D silicon strip detectors Junction grows from both sides – reduction of Vfd for up to 40%! Can they stand Feq=1015 cm-2 – also in terms of CCE ? p+ implants 200 mm asymmetric device symmetric device 200 mm n+ implants read-out Detectors studied: thickness=200 mm, strip pitch=120 mm Neff=-8x1012 cm-3 electric potential U=280 V U=250 V G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

How much over-depletion is needed to obtain sufficiently large CCE? Neff=-1.2x1013 cm-3 also detectors with p+ Smaller Vfd of semi-3D detectors for wide strips! How much over-depletion is needed to obtain sufficiently large CCE? G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Drift paths of electrons and holes in asymmetric detector U=250 V , Feq~1.1x1015 cm-2, Neff=-8x1012 cm-3 regions with low E central hit 20 mm from center 40 mm from center G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Red: strip left of hit strip Blue: strip right of hit strip n+ strips Semi 3D n+ strips (U=280 V) U=250 V much larger charge spread, but also larger cluster signal as in detector with n+ strips! U=250 V D=200 mm, pitch/width=120/60 mm Neff=-8x1012 cm-3, Feq~1.1x1015 cm-2 p+ strips Black: hit strip Red: strip left of hit strip Blue: strip right of hit strip G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Cluster charge (3 strips) as a function of impact position Neff=-8x1012 cm-3, Feq~1.1x1015 cm-2 Can large charge sharing in semi-3D detectors be used in non-irradiated detectors to improve position resolution? Only Q>0 considered Completely different picture as for conventional strip detectors. The device can be efficiently operated also near the depletion voltage. A large signal with respect to conventional strip detectors is anyway gained at higher operational voltages. G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

highest cluster charge of all in maximum – but large variations The charge spread in symmetric detector is highly position dependent! sensing electrodes connected together U=280 V Neff=-8x1012 cm-3, Feq~1.1x1015 cm-2 Higher noise if both electrodes are connected to same channel? highest cluster charge of all in maximum – but large variations Charge collection is poor for central hit! G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii

Conclusions In irradiated segmented detectors it is beneficial to collect electrons (n+-strips, pixels): charge sharing mechanism. ATLAS strip detectors: Q~15500 e, improvement at U>Vfd , significant signal induced also on neighbors, n+-strips would be a better option. ATLAS pixel detector: a good agreement with measured values was found: CCE~60% after Feq=1x1015 cm-2 ; if operated at U<Vfd, the collected charge loss due to partial depletion is smaller than predicted by 1-d/D. Thin pixel detector: no advantage of n+-type pixels for PW/D>1, even if detectors are operated at Neff~0 expected signals are ~1000-1600 e after Feq=1x1016 cm-2. Novel semi 3D design: interesting properties, reduction of Vfd, large charge sharing, cluster signal comparable to n+-strip detectors. G. Kramberger, Simulation of signal in irradiated silicon detectors Vertex 2002, Nov. 3-8, 2002, Kailua-Kona, Hawaii