Studies on radiation damage by 900
MeV electrons on standard and
oxygen enriched silicon devices
S. Dittongo1, M. Boscardin2, L. Bosisio1,
G.-F. Dalla Betta2, I. Rachevskaia2
Universita’ di Trieste & INFN - Sezione di Trieste, Italy
2 ITC-IRST, Divisione Microsistemi, Trento, Italy
1
Overview
• Introduction: why high energy electrons?
• Irradiation: experimental setup and
irradiated devices
• Results:
– Full depletion voltage
– Reverse leakage current, damage constant α
– Surface damage effects
• Conclusions
Selenia Dittongo
1st RD50 workshop, October 2002
2
Why high energy electrons?
• In a previous irradiation with 900 MeV electrons, up to
φ~1.5x1014 cm-2, bulk type inversion has been observed to
occur at φ~0.8x1014 cm-2
• New experiment to reach higher fluences
• No experimental results on damage induced on oxygen
enriched silicon by high energy electrons
• Compare radiation hardness of standard and oxygenated
silicon devices
• Help in gaining insight into the microscopic mechanisms of
the oxygen effect
• Possible applications in HEP experiments at linear colliders
Selenia Dittongo
1st RD50 workshop, October 2002
3
Irradiations
• Two irradiation runs
– October 2001: 4 irradiation steps, up to
φ~3.1x1014 cm-2 (D~8.3 Mrad)
– May 2002: 5 irradiation steps, up to
φ~4.5x1014 cm-2 (D~12.1 Mrad)
• Experimental setup
– Devices assembled on a fiberglass support
enclosed in a metal box, mounted on a remotely
controlled x-y stage
– During and after irradiation, device temperature
not controlled. Ambient temperature within the
box monitored with a platinum RTD sensor.
Selenia Dittongo
1st RD50 workshop, October 2002
4
Beam and fluences
• Beam from the Elettra linac (Trieste, Italy)
– 900 MeV electron beam, pulsed at 10 Hz
– 1-6 nC/pulse
• Fluence and dose evaluation
– Electron current pulse measured by a toroidal coil
October 2001
May 2002
step
φ (cm-2)
Dose (Mrad)
step
φ (cm-2)
Dose (Mrad)
1
(4.98±0.03±0.17)x1013
1.33±0.01±0.05
1
(3.87±0.04±0.13)x1013
1.03±0.01±0.04
2
(8.33±0.06±0.29)x1013
2.22±0.02±0.08
2
(9.75±0.04±0.33)x1013
2.60±0.01±0.09
3
(1.655±0.013±0.055)x1014
4.41±0.04±0.15
3
(1.150±0.002±0.040)x1014
3.06±0.05±0.10
4
(3.100±0.080±0.100)x1014
8.3±0.2±0.3
4
(2.07±0.04±0.07)x1014
5.5±0.1±0.2
5
(4.53±0.06±0.16)x1014
12.1±0.2±0.4
– First error: due to the fluctuations of the beam intensity
– Second error: systematic uncertainty on the calibration of the
toroidal coil
Selenia Dittongo
1st RD50 workshop, October 2002
5
Devices
• Test structures (p+nn+ diodes, MOS capacitors)
fabricated by ITC-IRST (Trento, Italy)
– n-type Si wafers, (111) oriented, 300 µm thick,
resistivity ~15kΩcm
• Oxygen enriched silicon devices
October 2001:
10-hour oxidation @ 1150o C +
16-hour diffusion @ 1150o C
May 2002:
12-hour oxidation @ 1150o C +
36-hour diffusion @1150o C
Selenia Dittongo
1st RD50 workshop, October 2002
6
Full depletion voltage
• Diode C-2-V characteristics measured with LCR meter
(f=10kHz) at room temperature a couple of days after
irradiation
Selenia Dittongo
φ
(cm-2)
Vdepl
(OXY)
Vdepl
(STD)
φ
(cm-2)
Vdepl
(OXY)
Vdepl
(STD)
0
22.6 V
20.3 V
0
20.3 V
26.3 V
4.98x10 13
10.8 V
11.4 V
3.87x10 13
10.0 V
14.9 V
8.33x10 13
7.3 V
4.9 V
9.75x10 13
4.1 V
6.5 V
1.65x10 14
5.0 V
6.9 V
1.15x10 14
4.6 V
6.2 V
3.10x10 14
18.8 V
22.9 V
2.07x10 14
16.3 V
15.1 V
4.53x10 14
38.7 V
42.5 V
1st RD50 workshop, October 2002
7
Effective dopant concentration
• Bulk type inversion reached after φ~1x10 14 cm-2
Both before and after bulk
type inversion, NO significant
difference in Neff (Vdepl)
between oxygenated and
standard devices, up to
φ~4.5x1014 cm-2
Selenia Dittongo
1st RD50 workshop, October 2002
8
Annealing of full depletion voltage
•
Monitored at room temperature for ~4 months after exposure, for two
couples of diodes irradiated in May to 5.5 Mrad and 12.1 Mrad
•
The peculiar time dependence of Vdepl observed for the second couple of
diodes has to be ascribed to measurement problems (not totally
understood); it is not a physical effect
Reverse annealing of Vdepl started ~1 month after the irradiation
•
Selenia Dittongo
1st RD50 workshop, October 2002
9
Reverse annealing of full depletion voltage
•
Studied by means of accelerated isothermal annealing above room
temperature: 8 cumulative annealing steps at 80oC up to 86 hours
•
•
Vdepl keeps increasing up to the last annealing step
Difference between curves of standard and oxygenated devices increases
with the annealing time
V (std )
V (oxy)
= 0.94 → 1.10
depl
depl
•
5 .5 Mrad
V (std )
V (oxy)
= 1.18 → 1.34
depl
depl
12.1 Mrad
Reverse annealing seems to be reduced on oxygenated silicon substrates
Selenia Dittongo
1st RD50 workshop, October 2002
10
Reverse leakage current
•
•
Measured @ Vbias=50V (>Vdepl) and normalized to 20oC
For the couple of diodes irradiated in May to 12.1 Mrad, Ileak
monitored for ~68 hours
– After 1 hour, Ileak ~75% of initial value
– After 2 days, Ileak ~60% of initial value
– No obvious functional dependence versus time
Selenia Dittongo
1st RD50 workshop, October 2002
11
Damage constant α
•
•
Normalized value at 20oC, measured at different times after the
end of exposure, for the May run
time
α (OXY) (A/cm)
α (STD) (A/cm)
10 min
4.88x10 -18
4.46x10 -18
15 min
4.75x10 -18
4.38x10 -18
20 min
4.53x10 -18
4.13x10 -18
60 min
4.11x10 -18
4.56x10 -18
> 3 weeks
2.44x10 -18
2.50x10 -18
For the October run, ~8 months after the exposure we get
α(STD)= 1.46x10 -18 A/cm
•
α(OXY)=1.37x10-18 A/cm
No significant difference between devices made on standard and
oxygenated silicon substrates
Selenia Dittongo
1st RD50 workshop, October 2002
12
α: comparison with other particles
• Assume the validity of NIEL scaling for electrons
• Use asymptotic value of displacement cross section for 200
MeV electrons [G.P.Summers et al., IEEE Trans.Nucl.Sci.
40(6) (1993), 1372-1378] (no higher-energy values available)
D(E)/95 MeV mb = 8.106x10-2
particles
α(A/cm)
measured
1 MeV n equiv.
1.8 MeV e-
4.5x10-20
1.9x10-18
900 MeV e-
4.88x10-18
6.02x10-17
1 MeV n
8.0x10-17
8.0x10-17
• At equal NIEL, high energy electrons are:
– ~30 times more effective than low energy electrons
– ~equivalent to 1 MeV neutrons
in damaging the bulk silicon
Selenia Dittongo
1st RD50 workshop, October 2002
13
Surface damage
•
•
•
•
MOS capacitor C-V characteritics measured with LCR meter (f=10kHz) and
corrected to account for the series resistance of the undepleted bulk
region
φ (cm-2)
VFB (OXY)
VFB(STD)
0
-12.74 V
-12.98 V
3.87x1013
Not avail.
-38.5 V
9.75x1013
-5.95 V
-8.89 V
1.15x1014
-4.39 V
-5.16 V
2.07x1014
-8.00 V
-7.06 V
4.53x1014
-2.47 V
-4.00 V
Flat-band voltage VFB starts decreasing after 1st irradiation step, when the
substrate is not yet inverted => no simple relation between VFB changes and
bulk type inversion
The stretch-out of the C-V curves indicates a build-up of interface trap
states
The C-V curves appear representative of n-type substrate even after bulk
inversion
Selenia Dittongo
1st RD50 workshop, October 2002
14
Conclusions
• High energy electrons produce important bulk damage:
– bulk type inversion observed at φ~1x1014 cm-2
– damage constant α, expressed in terms of 1 MeV neutron
equivalent fluence, comparable to heavy hadrons
• No clear beneficial effect of oxygen enrichment of Si
substrate on the radiation hardness has been observed for
high energy electrons up to φ~4.5x1014 cm-2
– only the reverse annealing of V depl appears to benefit
somewhat from the oxygenation of silicon substrates
• Further irradiation tests are planned to investigate possible
improvements in the radiation hardness of oxygenated
devices at higher fluences
Selenia Dittongo
1st RD50 workshop, October 2002
15