Epitaxial detectors

1
Epitaxial detectors

• Gregor Kramberger
• Jozef Stefan Institute, Ljubjana, Slovenia
• (on behalf of RD50 collaboration)

2
Outline

• Motivation
• Characteristics of epitaxial wafers
• Damage parameters
• Leakage current
• Vfd evolution after neutron and 24 GeV proton
  irradiations
• stable damage
• P-type
• N-type
• Annealing studies
• Warm SLHC scenario
• Trapping parameters
• Modeling of the damage
• CCE
• Pads
• Strips
• Conclusions

3
Motivation

• As the trapping increases the thickness is not so
  important, due to large trapping effects
• Advantages
• the current should be smaller
• the Vfd is smaller, hence the bias voltage needed
• drift is completed sooner less chance for
  trapping
• Drawbacks
• capacitance is larger
• the number of e-h pairs is smaller
• Why epi?
  G. Kramberger et al.,
  NIM A515 (2003) 665.
• Initially to cover the phase space of
  different materials
• Triggered large interest due to positive space
  charge introduction which compensates negative
  space charge during reverse annealing
• smaller damage after neutron irradiation than
  FZ/MCz
• Now, already well studied material!

4
Characteristics of epitaxial wafers
Epitaxial detectors so far successfully
processed by different institutions FBK-Trento,
CiS-Erfurt, CNM-Barcelona, HIP-Helsinki.

• Grown to max. 150 mm by ITME Warsaw (note that
  150 mm is already at the limit)
• So far devices of 25, 50, 75, 150 mm thickness
  were produced.
• Resistivity of the Cz substrate is very low
  (0.015 Wcm) - Sb or B doped
• Epitaxial material is the one with least
  impurities empty DLTS and TSC
• Cz substrate is important, due to
• out-diffusion of O from the Cz wafer during
  processing and growth
• although oxygen concentration is comparable with
  DOFZ, the fraction of O2i in total oxygen
  concentration is much larger ( m(O2i)gtm(Oi) )
• Initial resistivities of epitaxial wafers are
  somewhere between 50 and 500 Wcm depending on
  thickness and type!

5
Characteristics of epitaxial wafers
MCz
D. Eckstein et al., 12th RD50 Workshop,
Ljubljana, June, 2008.
DOFZ
Similar profiles were also measured for 50 and 25
mm thick detectors! Also perfectly homogenous
sheet resistivity profiles!
6
Damage parameters leakage current

• Leakage current damage constant identical with
  all other materials!
• a3.8-4.3 10-17 A/cm for 24 GeV protons
• small deviations in short term annealing (self
  annealing due to larger fluences)
• same also for neutron irradiations

E. Fretwurst et al., RD50 Workshop..
D. Eckstein et al., 12th RD50 Workshop,
Ljubljana, June, 2008.
7
Damage parameters Vfd
The irradiation with 24 GeV p (200 MeV p, 26 MeV
p) introduces positive space charge.
Influence of thickness
SIMS profiling O(25µm) gt O(50µm)
O(75µm)gtO(100µm)gt O(150µm) Stable
Damage Neff(25µm) gt Neff(50µm) geff-0.038
cm-1 gt geff-0.017 cm-1 Neff(75µm) gt Neff(100µm)
gt Neff (150µm) geff-0.015 cm-1 gt geff-0.008
cm-1 gt geff-0.007 cm-1 TSC Defect
Spectroscopy BD(25µm) gt BD(50µm) gtBD(75µm)
CiS - process
G. Lindström et al., NIM A556 (2006) 451.
t08 min_at_80oC
Additional oxydation
Generation of shallow donors BD (Ec-0.23 eV)
strongly related to O Possibly caused by O2i
dimers, outdiffused from Cz with larger diffusion
constant dimers monitored by IO2 complex
geff epi-DOgtgeff epi-ST The oxygenation
increases introduction rate of positive space
charge by some 30 at all thicknesses
J. Lange et al., 13th RD50 Workshop, CERN, 2008.
8
Damage parameters Vfd
Different processing

• 150 mm thick epitaxial detectors produced by
  different
• institutions
• Introduction rate of geff varies by factor of 2
  (more stat.)
• P type sensor undergoes type inversion
• Vfd0
• confirmed also in TCT
• geff0.01 cm-1

K. Kaska et al., presented at 11th RD50
Workshop,2007
Irradiation of p and n type 150 mm thick
epitaxial detectors with protons and neutrons
(FBK and CNM production)! Confirmation smaller
damage after neutrons as for FZ, MCz!
V. Khomenkov et al., presented at IEEE-NSS, 2008
9
Damage parameters Vfd
Around 3-4 times lower damage by neutrons than
for FZ, MCz geff0.02 cm-1

• Proton damage comparable to MCz
• Comparable to DOFZ but different dominant space
  charge

50 Wcm

• Oxygen plays a role (processing, diffusion)
• Resistivity seems to be important, but so does
  for other materials
• donor removal rate
• affect the dominant space charge sign (see CiS 50
  Wcm n irr.)
• Decent reproducibility for 150 mm thick samples.

• Epi sensors ideal for controlling the space
  charge by compensation
• acceptors from reverse annealing
• mixed radiation fields

10
Annealing studies
short term annealing
Stable damage NC NC0(1-exp(-cFeq) gCFeq
250 min _at_ 80oC 3 y at 20oC
Long term (reverse) annealingTwo components ?
NY,1(?,t(T)), first order process introduction
rate gY1.7 10-2 cm-1 (neu.), 3.110-2 cm-2
(pro.), tY1 few 100 min and scales with Ea1.3
eV (same process as in FZ?)? NY,2(?,t(T)),
second order process tY2 few 1000 min, at
times not so interested for SLHC, but second
order process - similarly observed also lately
in MCz Long term annealing amplitude seems to
saturate for DO epi sensors largely open to
studies.
The dominant SCSI at Vfdgt0 for higher fluences
double junction field profile The minimum Vfd
increases with fluence!
11
Annealing at 20oC for 3 years

• In order to check long term annealing in
  realistic conditions annealing at 20oC was
  initiated
• Roughly the scaling works so annealing at RT
  can be reasonably well approximated with
  annealing at elevated temperatures!

I. Dolenc, PhD thesis, Uni. Ljubljana, 2008
I. Dolenc, PhD thesis, Uni. Ljubljana, 2008
Similar long term annealing for neutron and
proton irradiated samples irradiated to the
similar fluences!

• The minimum in Vfd
• after 300 days at 20oC
• after 80 min at 80oC

Compatible with acceleration of 6500 for Ea1.3eV
12
Annealing of 150 mm epi- sensors

• Annealing of proton irradiated detectors is
  different for n and p type (only qualitatively
  similar)!
• Neutron irradiated sensors behave similarly
  regardless of the type!
• Assuming that 3 years at 20oC corresponds to 250
  min at 80oC the reverse annealing is not too
  harmful!

13
Warm SLHC scenario
G. Lindström et al., NIM A556 (2006) 451.
Stable donor generation at high F would lead to
larger Vfd, but acceptor generation during RT
anneal could compensate this. Proposed Benefit
Storage of EPI-detectors during beam off periods
at RT (in contrast to required cold storage for
FZ) Check by dedicated experiment

• Simulationreproducing the experimental scenario
• with damage parameters from analysis

• Experimental parameter
• Irradiationfluence steps ? 2.2?1015 cm-2
  irradiation temperature ? 25C
• After each irradiation stepannealing at 80C for
  50 min,corresponding 265 days at 20C

Excellent agreement between experimental data
and simulated results ? Simulation parameters
reliable!
14
Trapping parameters
J. Lange, Diploma thesis, Uni. Hamburg,2008.

• 150 mm thick epitaxial detectors studied with TCT
  (combination of thickness and fast TCT make it
  possible to use CCM for studies)
• Effective trapping times of electrons measured up
  to 41015 cm-2 are in accordance with previous
  measurements (FZ, MCz)

15
Trapping parameters
J. Lange, Diploma thesis, Uni. Hamburg,2008.
100, 150 mm thick epitaxial detectors studied
with alpha TCT (absolute measurement)
possibility to compare CCE of different detectors
(penetration depth around 25 mm)

• Simulation of the CCE fails to describe the
  measured charge at fluences of gt1015 cm-2 (even
  if voltage dependent trapping is taken into
  account)
• Charge at very high fluences and voltages rises
  rapidly can not be explained by the present
  device modeling assumptions - true also for
  strip detectors with n readout
• Thin detectors (100 mm) show even larger
  discrepancy. The simplest explanation would be
  avalanche effects
• Thin detectors larger peak fields at the same
  voltage applied
• Large fluences larger Neff hence higher peak
  fields
• But why dont we see CCEgt1 detectors break down
  at the point of CCEgt1

16
Modeling of the damage
n-type Epi (DO) 75 mm 2.331014 cm-2
I. Pintilie et al., Wodean workshop, 2008, HH

• Using the defects identified from TSC (position,
  concentration, cross-sections)
• Assuming the only free parameter is the remaining
  initial donor concentration
• Short term annealing added as there are no TSC
  (unlike C-V) measurements before 20 min

For the 1st time the microscopic picture in
hadron irradiated detector reproduces the
measurements
17
CCE bias voltage dependence
Epitaxial 150 mm thick (p-type) detectors
irradiated with protons 75 mm (n-type) irradiated
by neutrons! Measurements of most probable signal
of 90Sr electrons (25 ns shaping, -10oC) after
completed beneficial annealing
Epi 75, CiS
CNM
neutrons

• Vfd and onset of saturation coincide well for p
  irradiated an less for neutron irradiated (Vfd
  denoted by vertical bars)
• This correlation was shown for all combinations
  of material type, irradiation particle and
  producer
• Charge is larger than predicted by simulations,
  although no avalanche effects seen (voltages too
  low?)!
• Nice reduction of the charge with fluence can be
  observed. Junction model still works.

18
CCE fluence dependence
Averaging over the points above the full
depletion voltage!
Charge

• Maximum bias voltage applied (500 V for 150 mm,
  300 V for 75 mm and 200 V for 50 mm)
• The thickness becomes less important as the
  fluence grows as expected.
• The CCE at high fluences is significantly larger
  than predicted!
• Signal for given thickness is independent on
  material type of producer (not listed, but CNM,
  FBK, CiS and HIP)
• p-type epi shows better performance after neutron
  irradiation than the rest

after beneficial annealing
Equivalent fluence
G. Kramberger et al., NIM 552 (2005). K. Kaska et
al., presented at 11th RD50 Workshop,2007. V.
Khomenkov et al., presented at IEEE-NSS, 2008
19
CCE epi strips (neutron irradiated)
ATLAS 80 mm pitch detectors, SCT128A, 25 ns,
T-25oC
CNM
CNM
Vfd
T. Affolder et al., presented at 13th RD50
Workshop, 2008.
T. Affolder et al., presented at 13th RD50
Workshop, 2008.

• Signals for 150 mm Epi-p and Epi-n pad detectors
  agree with strip data (taken 10 in absolute
  charge calibration and difference in
  segmentation)!
• At Feq81015 cm-2 and 500 V bias voltage 6.7ke
  are collected (on the limit to be enough for
  SLHC)!
• Vfd is in agreement with kink in Q-V
• Also p-n detectors perform well
• smaller thickness (smaller difference between
  holes and electrons)
• saturating drift velocities (over-depleted
  detectors)

20
CCE strips compared (neutrons)
VFZ800 V
Vepi500 V
T. Affolder et al., presented at 13th RD50
Workshop, 2008.
T. Affolder et al., presented at 13th RD50
Workshop, 2008.

• Epi detectors biased to 400-500 V perform as good
  or better as other detectors from 1015 cm-2 on at
  higher voltages (confirmation of lower damage by
  neutrons than FZ)
• Same behavior seen also by Pisa group (A.
  Messineo et al., presented at 11th RD50 Workshop,
  2007)

• To be investigated
• it will be interesting to see performance of
  strip detectors after proton irradiations
• could lower neutron damage in epi-si detectors be
  a problem for innermost layers (no compensation
  of damage)?
• If bias voltage up to 1000 V can be used are
  Epi-p detectors at 500 V better than FZ at 1000
  V!
• Currents were shown to be comparable, but this
  lacks a solid explanation ?
• Do thinner detector help the physics material
  budget ?
• What is the availability on relatively larger
  scale?

21
Conclusions

• Epitaxial detectors are successfully produced by
  different institutions.
• They show a smaller stable damage after neutron
  irradiation than any other material.
• The reverse annealing is such that a warm SLHC
  scenario is not only possible but desired for
  innermost layers.
• CCE data show good agreement with Vfd up to the
  highest fluences (junction model). The
  degradation of CCE is more severe for thick
  detectors
• CCE on strips prove that if bias is limited to
  500-800 V the epitaxial detectors are the best
  in CCE at very high fluences and are a viable
  solution for SLHC tracking.
• For the first time prediction based on a
  microscopic measurement of hadron irradiated
  detector described the macroscopic measurements
  with reasonable accuracy.
• All in all epitaxial detector perform in
  accordance with expectation up to the highest
  SLHC fluences
• To be done
• Proton irradiated epi-p type strips sensors as
  well as mixed irradiated sensors need to be
  tested
• A warm CERN scenario tests need to be performed
  with 150 mm thick p-type epi-Si sensors
  irradiated with protons.