Transcript transparencies - Indico

Comparison of full depletion voltage extracted from
C-V, I-V and Q-V characteristics
for a highly irradiated Epi-detector
E. Fretwurst , N. Hoffmann*, F. Hönniger, G. Lindström
Institute for Experimental Physics, Univ. of Hamburg
*DESY summer student 2005
 Motivation
 C-V measurements, frequency dependence (10kHz – 800 kHz),
I-V, Q-V measurements and annealing behavior,
comparison of extracted Vfd values
 Conclusion
1
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Motivation
 What is the meaning of „full depletion voltage Vfd“ in case of heavily damaged
epitaxial silicon detectors?
 Comparison of Vfd values extracted from different measurements:
C-V characteristics, frequency dependence:
 space charge concentration, shallow and deep defect levels
I-V characteristics:
 concentration of generation centers
Q-V characteristics (charge collection measurements):
 electric field distribution and trapping
 Which Vfd values are relevant for the evaluation of the detector properties under
practical operation in experiments at S-LHC?
2
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Experimental Conditions
 Epitaxial Si on Cz substrate:
epi-layer: 50 µm, n-type 50 cm
Constant P doping profile
[O]: inhomogeneous depth profile, O out-diffusion from Cz
[C]: < 1016 cm-3, near to detection limit
Cz substrate: n-type 0.01 cm, Sb doped
Resistivity [ cm]
102
101
100
10-1
10-2
0
50 m, C-V method
50 m, spreading resistance
10
20
30
40
Depth [m]
50
60
 Irradiation: 24 GeV/c protons, p = 4·1015 p/cm²
 Measurements:
C-V for frequencies between 10 kHz and 800 kHz
all C-values correspond to series mode values (Cs, Rs)
Concentration [cm-3]
1018
Oxygen
I-V measurements: (pad current and guard ring current)
1017
Carbon
1016
1015
0
25
50
Depth [m]
75
100
TCT measurements: generation of charge carriers by a
pulsed 1060 nm laser  simulating mips
Q-V characteristics derived by integrating current
transients with a time window of 30 ns
Annealing measurements: 80 °C, ta = 0min – 240 min
3
E . Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
C-V frequency dependence
9
10 kHz
50 kHz
100 kHz
200 kHz
300 kHz
500 kHz
600 kHz
800 kHz
Capacitance [F]
10
5
Annealing 80oC/0 min
8
Capacitance [10-11F]
Annealing 80oC/0 min
-9
300 kHz
500 kHz
600 kHz
800 kHz
7
6
-10
10
5
5
1
10
Bias voltage [V]
100
1
10
Bias voltage [V]
100
Frequency dependence measured at room temperature:
 strong decrease of C-values with increasing frequency f
 shift of C-V transition to constant value Cg (geometrical value) to lower bias
voltages with increasing f  decrease of Vfd
 specific C-V shape at low f (10 kHz, 50 kHz) indicates non-homogeneous
distribution of electrically active defects  possibly correlated with
non-homogeneous [O] distribution
4
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Annealing Effect
10 kHz
50 kHz
100 kHz
200 kHz
300 kHz
500 kHz
600 kHz
800 kHz
Capacitance [F]
10
5
Annealing 80oC/240 min
-9
10
Capacitance [F]
Annealing 80oC/0 min
-9
5
10-10
10-10
5
5
1
10
Bias voltage [V]
100
10 kHz
50 kHz
100 kHz
200 kHz
300 kHz
500 kHz
600 kHz
800 kHz
1
10
Bias voltage [V]
100
Annealing at 80 °C:
 After 240 minutes the C-V curves are “shifted” to lower bias voltages
 The overall shape in the voltage range 1V-10 V is not influenced but
the strong decay is shifted to lower voltages  shift in Vfd
5
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Vfd frequency dependence
200
Vfd [V]
150
24GeV/c protons, 4.1015 p/cm2
Ta = 80 oC
8 min
0 min
120 min
100
50
0
0
200
400
600
Frequency [kHz]
800
1000
 Vfd decreases with increasing frequency but saturates
 Saturation at about 300 kHz for 0 min and 8 min
 Saturation between 100 kHz and 200 kHz for 120 min
 Relative change Vfd,sat/Vfd,10kHz  50 %
6
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Simulation
one deep acceptor level model
10-9
Parameter:
Simulation
ND = 6·1013 cm-3
5
Capacitance [F]
10 kHz
50 kHz
100 kHz
500 kHz
800 kHz
NA = 0.98·ND
EFn  Et = 0.1 eV
R = 4·105 s-1
10-10
ND - NA
5
X=W-λ
1
10
Bias voltage [V]
100
W.G. Oldham, S.S. Naik; Solid State Electronics 15 (1972) 1085
Deep acceptor level  transition region λ, defined by crossing of the quasi-Fermi level EFn and trap level Et
Capacitance: C = dQ/dV  2 contributions: dQ = dQx + dQw
dQx = dQx() depends on frequency due to the emission rate en of the trap
Low frequency limit: CL  q0·NA·dx/dV+q0·(ND-NA)dw/dV, High frequency limit: CH  q0·(ND-NA)·dw/dV
 C() = CH + (CL  CH)/(1 + (/R)2)
with R  2·en·(1+K), K counts for the coupling of dQx and dQw
7
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Ta = 80 oC
Tm = 21 oC
f = 10 kHz
10-4
0 min
60 min
120 min
240 min
10-5 0
10
Annealing 80oC
10-9
Capacitance [F]
Reverse current [arb. units]
Comparison I-V and C-V
5
10-10
0 min
60 min
120 min
240 min
5
101
102
Bias voltage [V]
100
101
102
Bias voltage [V]
Comparison I-V and C-V curves for same annealing time:
(I-V curves are shifted by an arbitrary value, C-V curves as measured)
 Vfd from I-V (crossing point of fit lines) much smaller compared to values from C-V
 Saturation of I-V not as clean as expected for a 10 µm gap between central pad and
guard ring (possible surface damage effect?)
 C-V shape after 60 min annealing shows a double shoulder vanishing after 120 min
8
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Vfd Annealing Curves
200
Ta=80 oC
10 kHz
50 kHz
100 kHz
Vfd [V]
150
200 kHz
300 kHz
500 kHz
100
600 kHz
800 kHz
50
0 0
10
I-V
10 1
10 2
Annealing time [min]
10 3
 Nearly identical time dependence for all frequencies
 Shift due to frequency dependent charging and discharging of deep defects
 Slightly different annealing behavior of Vfd from I-V, values are comparable
with values from C-V at high frequencies
9
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
TCT-Measurements
Pulsed laser
1060 nm
Bias
Epi-layer: 50 µm Cz-substrate: 300 µm
Signal current [arb. units]
0.025
Ta = 80 oC
t = 30 min
0.02
230 V
200 V
185 V
0.015
149 V
83 V
40 V
0.01
20 V
0.005
0
0
10
20
30
t [ns]
40
50
60
 Signal shape dominated by laser pulse and R-C time constant
(diode capacitance, 50  input resistance of the amplifier, charge
collection time in the order of 500 ps at 150 V)
 Collected charge: integration of the current pulse with a time
window of 30 ns
10
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Collected charge [arb. units]
Q-V characteristics and annealing
101
240 min
120 min
80 min
30 min
8 min
10-1
Measurements
performed at 20 °C
0 min
100
101
102
Bias voltage [V]
Q-V curves shifted
by an arbitrary value
103
 Extraction of Vfd indicated for the Q-V curves taken at 0 min and 240 min annealing
 In the log-log presentation the slope of the increasing part of Q-V increases from about
0.8 at 0 min to 1.1 at 240 min
 Charge trapping is clearly seen above “full depletion” and annealing time  120 min
11
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Comparison of Vfd annealing from Q-V and C-V
200
Ta=80 oC
Vfd [V]
150
100
C-V, 10 kHz
C-V, 50 kHz
50
TCT, 1060 nm
0 -1
10
10 0
10 1
10 2
Annealing time [min]
10 3
 Vfd values from Q-V are in between the values from C-V taken at 10 kHz and 50 kHz
at 0 min and for ta > 30 min the Q-V values coincide almost with those at 10 kHz
 The time dependence is nearly identical
 This indicates that “full charge collection” is achieved when the transition region λ of
the space charge region (low field region) becomes zero as approximately deduced
from C-V measurements at low frequencies
12
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005
Conclusion
 The meaning of “full depletion voltage” Vfd has to be taken with care:
Vfd depends strongly on type of measurement (C-V, I-V, Q-V) and method of extraction
 Systematic studies on a highly irradiated Epi-device show:
C-V: Vfd decreases with increasing frequency, but saturates
I-V: Vfd comparable with those from C-V at high frequencies (> 500 kHz)
Q-V: Vfd comparable with those from C-V at low frequencies ( 10 kHz)
 Which Vfd value is relevant for detector operation?
The values extracted from Q-V which coincide with values from C-V at low frequencies,
but keep in mind:
Vfd from Q-V depends on integration time and trapping:
For 50 µm thick layers the integration time is less important than for 300 µm (matter of
collection time)
 Proposed interpretation of relevant Vfd for non-inverted Epi-detectors:
The “voltage for full depletion” is achieved when the low field region of the transition
region λ vanishes or the crossing point x of the quasi-Fermi level with the deep acceptor
level approaches the rear contact (x=d, d=detector thickness)
13
E. Fretwurst, Univ. Hamburg, RD50 workshop, CERN, November 2005