Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Humidity Sensitivity of
Large Area Silicon Sensors:
Study and Mitigation
Xavi Fernández-Tejero a, P.P. Allport b, K. Dette c, V. Fadeyev d, C. Fleta a,
D. Gillberg e, L. Gonella b, K. Hara f, C. Helling d, B. Hommels g, J. Keller e,
C. Kleing, T. Koffas e, V. Latonova h,i, M. Mikestikova h, R.S. Orr c, S. Pyatt b,
U. Soldevila j, E. Staats e, J. Thomas b, M. Ullán a, Y. Unno k and S. Wada f
a Centro Nacional de Microelectrónica (CNM, CSIC), Spainb University of Birmingham, United Kingdom
c University of Toronto, Canadad Santa Cruz Institute for Particle Physics (SCIPP), University of California, USA
e Carleton University, Canadaf University of Tsukuba, Japan
g University of Cambridge, United Kingdomh Academy of Sciences of the Czech Republic, Czech Republic
i Charles University, Czech Republicj Instituto de Física Corpuscular (IFIC, CSIC), Spain
k KEK, Japan
December 17th 2019
12th International “Hiroshima” Symposium on the Development and Application
of Semiconductor Tracking Detectors (HSTD12)
Introduction
Observations
o Breakdown Voltage Dependence
o Leakage Current Stability Dependence
o Hotspot Imaging at Breakdown Voltage
o Cleaning and Baking Influence
Incidence Rate
o Statistical Study
o Irradiation Influence
Mechanisms
o Sensor Edge Configuration
o Passivation Thickness
Mitigation Investigation
Summary of Findings and Conclusions
-02-
Outline
J. Fernández-Tejero – HSTD12 – December 2019
Introduction
-03-
• Measurements on Hamamatsu large area prototypes for the Step-2 of the ATLAS ITk Strip
Sensor Market Survey (ATLAS12) show first indications of breakdown voltage
dependence with relative humidity variations, also seen in sensors assembled in
modules (see Poster ID 233 – link).
• Sensitivity of large area prototype batches for the Market Survey Step-3 (ATLAS17) were
extensively tested by many ATLAS institutes.
• This presentation summarizes the findings by the ATLAS ITk strip sensor community, for
prototype batches ATLAS12 and ATLAS17, from Hamamatsu and Infineon, to investigate
the mechanisms and evaluate the impact and implications of the humidity sensitivity
for the ATLAS strip sensor production in particular and for large area silicon sensors in
general.
J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Large area sensors from Hamamatsu
6-inch wafer from Infineon
Observations – Breakdown Voltage Dependence (Large Area Sensors)
-04-
Breakdown voltage (VBD) of ATLAS12EC and ATLAS17LS large area (Main)
sensors sensitive to changes in relative humidity (RH):
VPX22728-W028
ATLAS12EC – Main Sensor
RH↑ ⇒ VBD↓
J. Fernández-Tejero – HSTD12 – December 2019
VPX26244-W040
ATLAS17LS – Main Sensor
0
1
2
3
0 500 1000Le
ak
ag
e C
urr
en
t [µ
A]
Bias [V]
TEST 3 + Measurement 5%
0
1
2
3
0 500 1000Le
ak
ag
e C
urr
en
t [µ
A]
Bias [V]
Storage 20% + Measurement 5%
0
1
2
3
0 500 1000Le
ak
ag
e C
urr
en
t [µ
A]
Bias [V]
TEST 1 + Measurement 60%
0
1
2
3
0 500 1000Le
ak
ag
e C
urr
en
t [µ
A]
Bias [V]
4h Storage 60% + Measurement 60%
After 2h of storage at ~5% RH
TEST 1 TEST 2 TEST 3 TEST 4
ATLAS17LS Main Sensor (VPX26244-W019)
…however, sensor storage at low RH mitigates early breakdown on ATLAS12 and ATLAS17 large
area sensors, recovering the breakdown voltage prior to the high humidity tests:
VBD≈420V VBD≈375V VBD≈310V VBD≈420V
Sensor with initial early
breakdown (see “Cleaning and
Baking Influence” section)
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
-05-
Breakdown voltage of ATLAS17LS Miniature sensors (Mini, MiniSS and MiniLS)
seems less sensitive but also showing clear influence.
…additionally, some mini sensors show small RH dependence of baseline current, in contrast to large
area sensors
Increasing RH
50% → 65%
Increasing RH
50% → 65%
J. Fernández-Tejero – HSTD12 – December 2019
RH↑ ⇒ VBD↓
VPX26244 Mini 7VPX26244 Mini 6
50%
5%
50%
5%
50%
5%
Mini Sensor (1x1 cm2) MiniSS Sensor (3x1 cm2) MiniLS Sensor (5x1 cm2)
Mini Sensor (1x1 cm2) Mini Sensor (1x1 cm2)
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Observations – Breakdown Voltage Dependence (Miniature Sensors)
Observations – Leakage Current Stability Dependence
-06-
• Large area sensors showing dependence on current stability only at bias near the breakdown
voltage at high RH
• Quick response of leakage current to RH
ATLAS17LS – Main Sensor
• Long periods biasing sensors in high RH can lead on an irreversible high leakage current
and low VBD
Vbias=-250V Vbias=-300V
J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Uncontrolled RH
Low RH
ATLAS12 (VPX12318 W644)
0.12
0.17
0.22
0.27
0.32
0.37
0.42
0.47
0 2 4 6 8
Leakage C
urr
ent
[µA
]
Time [h]
ATLAS12 (VPX22728 W003)
Low RH
Uncontrolled RH
-600V
-500V
V≈320VRH=~60%
Observations – Hotspot Imaging at Breakdown Voltage
-07-
ATLAS12 and ATLAS17 large area
sensor hotspots, at breakdown voltage,
mainly located at the Bias/Guard/Edge
Rings (edge structure)
ATLAS17LS – Main Sensor
ATLAS17LS – Main Sensor
J. Fernández-Tejero – HSTD12 – December 2019
ATLAS12A – Main Sensor
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Lock-in Infrared Thermography (LIT)[1]
characterization at high RH
[1] M. Vellvehi, et al., Lock-in Infrared Thermography: a tool to locate and analyse failures in power devices, 2017 CDE. doi: 10.1109/CDE.2017.7905234
ISOPROPANOL CLEANING:
• ATLAS17LS MAIN sensor (VPX26244-W019) showing low breakdown voltage
• We decided to clean the sensor with isopropanol and repeat the RH sensitivity test
Results:
Clear increase on initial breakdown voltage (from ~420V to ~925V)
Still similar humidity sensitivity
Observations – Cleaning and Baking Influence
-08-
RH 5%
RH 60%
ISOPROPANOL
RH 60%
RH 5%
J. Fernández-Tejero – HSTD12 – December 2019
(1) 5 min. submerged in isopropanol
applying ultrasounds
(2) 5 min. submerged in deionized
water
(3) 10 min. at 100ºC with low RH
(~5%)
• After the isopropanol cleaning, new
hotspots found on different areas, but
always located at the Edge structure
RH=~60% V≈255V
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Initial early breakdown seems solved after the isopropanol cleaning, but the sensor is still
sensitive to RH
SENSOR BAKING:
Results:
Increased breakdown voltage (from ~925V to >1000V)
Still similar humidity sensitivity
PLASMA CLEANING:
Results:
High breakdown voltage
Still similar humidity sensitivity -09-
RH 5%
RH 60%
BAKING
J. Fernández-Tejero – HSTD12 – December 2019
150ºC 24h with low RH (~5%)
Observations – Cleaning and Baking Influence
PLASMA
Plasma parameters:
• 200 ml/min O2 flow
• 500 W
• 5 min.
RH 5%
RH 60%
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
J. Fernández-Tejero – HSTD12 – December 2019
Incidence Rate – Statistical Study (Market Survey Step-3 Batches)
-10-
HAMAMATSU Batch VPX26244
Device Edge Wafer (Device ID) RH Sensitivity
Main
Slim-edge
W002 High
W019 High
W035 High
W037 High
W038 Low
W040 High
MiniLS
(5x1 cm2)
W003 (1) High
W004 (1) High
W005 (1) No
MiniSS
(3x1 cm2)
W004 (2) High
W006 (2) High
Mini
(1x1 cm2)
W002 (2) No
W002 (3) Low
W002 (4) No
W004 (7) High
W005 (?) High
W006 (?) High
W017 (8) No
W049 (8) No
Diode
(0.8x0.8 cm2)
W015 (P1) High
W038 (P1) No
W038 (P2) High
W038 (P3) No
Diode
(0.4x0.4 mm2)
W014 (P1) No
W014 (P4) No
W014 (P7) No
INFINEON Batch VC820647Device Edge Wafer (Device ID) RH Sensitivity
MainStandard-edge
W001 No
W004 Low
Slim-edge
W004 High
MiniSS
(3x1 cm2)W006 (1) High
Mini
(1x1 cm2)
W004 (8) High
W006 (1) High
Diode
(0.8x0.8 cm2)Standard-edge
W001 (E) No
W001 (W) No
W005 (W) No
W006 (W) No
W007 (W) No
Diode
(0.2x0.2 mm2)
Slim-edge W007 (TestEdge) No
(Slim-edge) – 60 µm W007 (TestEdge) High
Humidity Sensitivity tests performed in batches from
ATLAS ITk Strip Sensor Market Survey Step-3:
Hamamatsu Batch VPX26244: 26 devices
Infineon Batch VC820647: 13 devices
CRITERIA: Devices tested at high humidity showing a
decrease in breakdown voltage…
lower than 200 V → No sensitivity
between 200 and 300 V → Low sensitivity
higher than 300 V → High sensitivity
• Both batches are highly sensitive to RH variations
• Devices with higher dimensions and/or with slim-
edge seem to be more sensitive
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
HAMAMATSU Batch VPX29549 (Special Passivation)
Device Edge Wafer (Device ID) RH Sensitivity
Main Slim-edgeW071 No
W072 Low
MiniSS
(3x1 cm2) Standard-edgeW068 (2) No
Mini
(1x1 cm2)
W068 (8) No
Slim-edge
W068 (4) No
W068 (6) No
W071 (3) No
W071 (4) Low
W071 (7) No
W071 (8) High
Diode
(0.4x0.4 mm2)
W064 (?) No
W068 (?) No
W068 (?) No
W068 (?) No
W071 (P2) No
W071 (P7) No
W071 (P11) No
HAMAMATSU Batch VPX30816
Device Edge Wafer (Device ID) RH Sensitivity
Main
Slim-edge
W208 Low
MiniSS
(3x1 cm2)
W172 (2) No
W179 (2) No
W182 (2) No
Mini
(1x1 cm2)
W172 (6) No
W172 (7) No
W177 (6) No
W177 (7) Low
W179 (6) No
W179 (7) No
W182 (6) No
W182 (7) No
W185 (6) No
W185 (7) No
• After the Market Survey, Hamamatsu fabricated a “Special Passivation” batch (VPX29549) to
investigate the humidity sensitivity of the Market Survey sensors, and a final prototype batch
(VPX30816)
• “Special Passivation” batch shows clear improvement, but the
passivation layer is only for investigation purposes, not for
sensor production.
• Last batch VPX30816 shows the best response to humidity,
only with 2 out of 14 devices sensitive.
Humidity Sensitivity tests also performed in:
HPK “Special Passiv” Batch VPX29549: 17 devices
HPK Final Batch VPX30816: 13 devices
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Incidence Rate – Statistical Study (Batches After Market Survey)
-11-
Incidence Rate – Irradiation Influence• Sensors showing sensitivity to RH variations were irradiated and tested to evaluate the evolution
of the dependence
• ATLAS12 and ATLAS17 sensors drastically reduce their humidity sensitivity after (0.5-1)e15 neq/cm2
proton irradiation (measuring at -20ºC)
• Post-irradiated sensors not sensitive to changes in relative humidity
0 200 400 600 800 1000
voltage[-V]
9-10
8-10
7-10
6-10
5-10
4-10
3-10
leak[-
A]
30%
50%
70%
irrad
ATLAS12EC - MAINATLAS17LS - Mini
PROTON 5e14 neq/cm2
NEUTRON 5e14 neq/cm2
-12-J. Fernández-Tejero – HSTD12 – December 2019
PROTON 1e14 neq/cm2
ATLAS17LS - Mini ATLAS17LS - Mini
PROTON 5e14 neq/cm2
ATLAS17LS - Mini
PROTON 2e15 neq/cm2
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
Mechanisms – Sensor Edge Configuration
-13-J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
• The sensor edge configuration in the new large area sensors is one of the new features of the
future HL-LHC strip sensors.
• Devices with “Slim-edge” seems to be more sensitive to humidity changes than devices with
“Standard-edge” (Statistical Study in slides 10 and 11).
• Additionally, the metal separation between guard ring and edge ring can be critical if the
passivation thickness is not thick and conformal enough to cover the metal steps:
Guard ring Edge ring
D(HPK) = 125 µm
D(IFX) = 105 µm
h
passivation
HPK: 180 µm
IFX: 180 µm
Si e
dg
e
h
Guard ring
Edge ring
• Metal separation (D) between guard ring and edge ring: 125 µm (HPK) and 105 µm (IFX)
• Dielectric strength:
o Air dielectric strength: SAir=3 V/µm
o SiO2 dielectric strength: SSiO2=103 V/µm (Passivation)
Vmax = D·Sair + 2·h·SSiO2 ⇒ h(Vmax=1kV) ≥ 0.3-0.4 µm
• If the passivation thickness (h) is close to 0.3-0.4 µm, we can be close to the dielectric
breakdown between the guard ring metal and edge metal
• High sensitivity of Infineon batch compatible with their lower metal separation (D)
• Dielectric strength (S) can be reduced in high humidity, facilitating the breakdown
• Passivation thickness and conformity may vary between batches
RH↑ ⇒ S↓ [2]
[2] C.M. Osburn, et al., Dielectric Breakdown in SiO2 Films, J. Electrochem. Soc., Vol. 119, No. 5, 1972
Mechanisms – Passivation Thickness
-14-J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
• The passivation thickness was measured in miniature sensors from the 4 prototype batches
using different techniques in different ATLAS institutes.
Guard ring Edge ring
D(HPK) = 125 µm
D(IFX) = 105 µm
h1
passivation
HPK: 180 µm
IFX: 180 µm
Si e
dge
h1
Sensor from HPK batch VPX26244 showing less conformal metal step coverage (h1<h2) in
guard and edge rings.
Sensor from HPK batch VPX30816 showing conformal passivation (h1=h2) and no humidity
sensitivity.
Sensor from HPK “special passivation” batch VPX29549 showing less conformal metal step
coverage, but no sensitivity probably due to the different passivation material.
Sensor from IFX batch with thicker conformal passivation, but showing moderate humidity
sensitivity probably due to the lower separation between metals.
h2h2
• The high sensitivity of batch VPX26244 could be partially associated to a
thinner and/or less conformal passivation.
Mitigation Investigation
-15-J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
• Sensitive miniature sensor selected from the highly sensitive Hamamatsu batch
(VPX26244), and glued + wire-bonded to a PCB board:
50%
5%
• Manual sensor passivation using a Silicone Conformal Coating Aerosol (RS ref: 494-714):
Mitigation Investigation
-16-J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Mechanisms | Mitigation | Conclusion
MANUAL PASSIVATION WITH COATING AEROSOL:
• Extensive humidity sensitivity test after manual passivation, showing clear improvement
even at humidity conditions near the dew point:
50%
5%
COATING
• No variation of baseline current
• No influence of long exposition (1 week) to ambient
humidity (55%)
• Highly stable leakage current even for long periods
(5h) biasing the sensor at high humidity (55%):
Vbias = 600 V
RH 55%
• Encouraging results – Total coverage, not proving anything yet
• Further experiments: Humidity sensitivity test coating only the sensor edge area to
try and demonstrate passivation mechanism AND radiation tolerance of coating
• Still, in a first approach, possible mitigation method demonstrated
Summary of Findings and Conclusions
-17-
Observations:
• Humidity sensitivity of breakdown voltage observed for ATLAS12 and ATLAS17 MAIN sensors.
• After measurements at high humidity, fast recovery of breakdown voltage placing the sensor in low
humidity.
• Humidity sensitivity less common on miniature sensors but showing clear influence.
• Long periods biasing a sensor in high humidity can lead on an irreversible high leakage current and low
breakdown voltage.
• Hotspots at breakdown voltage mainly located at the edge structure of the sensor.
• Cleaning and Baking clearly improve breakdown voltage at low humidity, but still sensitive to humidity.
Incidence Rate:
• HPK batch VPX26244 and IFX batch highly affected by humidity changes.
• Humidity sensitivity less common on devices from subsequent HPK batches.
• Less sensitivity observed as sensors get irradiated.
Mechanisms:
• Influence of Guard/Edge ring metals separation and passivation thickness and conformity of step
coverage.
• Dielectric strength also can be reduced at high humidity, facilitating the breakdown.
Mitigation Investigation:
• Manual passivation with Silicone Conformal Coating Aerosol totally mitigates the humidity sensitivity,
even in highly sensitive batches and humidity conditions near the dew point.
• Further experiments to be performed coating only the edge region and testing radiation tolerance.
J. Fernández-Tejero – HSTD12 – December 2019
Intro | Observations | Incidence | Possible Causes | Mitigation | Conclusion
• Procedures established to keep low humidity conditions for reception and storage.
• Sensors should be handle in clean conditions.
• Sensors will work at dry environment therefore no expected problems of humidity during experiment.
Actions Taken for Sensor Production
Backup Slides
-18-J. Fernández-Tejero – HSTD12 – December 2019
-19-
BACKUP: Breakdown Voltage Dependence – ATLAS07
No humidity sensitivity observed for ATLAS07 prototype sensors (only for high RH ~70%):
J. Fernández-Tejero – HSTD12 – December 2019
BACKUP: Breakdown Voltage Dependence
-20-
• The variation range of VBD, for
different RH, is reduced when the
temperature decreases
Low temperature seems to reduce
the humidity sensitivity dependence
of breakdown voltage20ºC
-11ºC
ATLAS17LS - Mini
ATLAS17LS - Mini
0ºC
ATLAS17LS - Mini
J. Fernández-Tejero – HSTD12 – December 2019
-21-
• Hamamatsu initially doesn’t observe these
dependence, then started repeat tests on time of the
humidity sensitivity of ATLAS17LS MAIN sensors
• Long period of storage at 25ºC and ambient humidity
(50%), and tested at ambient humidity (50%)
• Different levels of breakdown voltage degradation
observed
• The observed time dependence explains why we are
seeing the humidity sensitivity at RH levels
comparable to the conditions of HPK QC tests
J. Fernández-Tejero – HSTD12 – December 2019
BACKUP: Breakdown Voltage Dependence
POST-ISOPROPANOL
-22-
BACKUP: Microscope Inspection of Hotspots after Cleaning/Baking
• Prior to cleaning/baking, several stains found on hotspot areas
After the cleaning/baking, the sensor looks pretty clean
Initial early breakdown voltage solved
But humidity sensitivity seems not related with sensor cleanness
POST-ISOPROPANOL POST-BAKE
POST-BAKE
POST-BAKE
POST-PLASMA
POST-PLASMA
POST-PLASMA
INITIAL INSPECTION
POST-ISOPROPANOL
J. Fernández-Tejero – HSTD12 – December 2019
-23-
BACKUP: Repairing Treatments – Bake-out
stable unstable
humidity,
time
bake-out
pote
ntial
During the Hamamatsu-ATLAS meeting in September 2017, a model for sensor stability was
presented, based on findings on an ATLAS12EC sensor:
• The sensor can reside in a “stable” state, and an “unstable” state
• A potential barrier exists between the stable and unstable states, favoring the latter
• Boltzmann statistics / Arrhenius equation applies to transition
• High humidity encourages the transition into the unstable state over time
• Bake-out (16h at 160ºC and dry environment) is supposed to bring the sensor back in the
stable state
• Procedure can be repeated and leads to longer lifetime in stable state
J. Fernández-Tejero – HSTD12 – December 2019
-24-
BACKUP: Repairing Treatments – Bake-out
Bake-out should be done in dry conditions to avoid possible sensor degradation:
Uncontrolled RH
Low RH
J. Fernández-Tejero – HSTD12 – December 2019
BACKUP: Sensors with Special Passivation
0
1
2
3
0 200 400 600 800 1000
Le
akag
e C
urr
en
t [µ
A]
Bias [V]
VPX29549-W071 (special passivation)
RH 5% (Storage RH 20%)
RH 45% (Storage RH 20%)
RH 45% (3h Storage RH 50%)
RH 55% (3h Storage RH 50%)
RH 65% (3h Storage RH 50%)
RH 75% (3h Storage RH 50%)
0
1
2
3
0 200 400 600 800 1000
Le
akag
e C
urr
en
t [µ
A]
Bias [V]
VPX29549-W072 (special passivation)
RH 5% (Storage RH 20%)
RH 45% (Storage RH 20%)
RH 45% (3h Storage RH 50%)
RH 65% (3h Storage RH 50%)
RH 75% (3h Storage RH 50%)
• ATLAS17LS batch with special passivation for
humidity sensitivity studies, NOT considered to be
included on production
• Preliminary results for three MAIN sensors:
- W064: Decrease of VBD for RH > 65%
- W071: Decrease of VBD for RH > 65%
- W072: No decrease of VBD
• Relative humidity of 65% at 20ºC can lead on
water condensation on sensor surface (dew point
14ºC)
Two independent groups confirmed that no
humidity sensitivity is observed for safe dew point
conditions
• First MAIN sensors found without sensitivity
• Passivation seems to play a key role on humidity
sensitivity
-25-Plots overlap
J. Fernández-Tejero – HSTD12 – December 2019