Upload
imogen-francis
View
218
Download
3
Embed Size (px)
Citation preview
QA Workshop at CERN 3-4 November 2011
2011.11.3
Hamamatsu silicon detectors for high energy physics experiments
Kazuhisa Yamamura* , Shintaro Kamada
QA Workshop at CERN 3-4 November 2011
Outline
1. Development and production history of Hamamatsu SSD ( Silicon Strip Detector )
2. Production flow of SSD
3. Failure analysis
4. New dicing technology
5. Design examination of SSD for HL-LHC
QA Workshop at CERN 3-4 November 2011
Hamamatsu Si detectors for HEP
Direct detector
Silicon Strip Detector(SSD)
Silicon Pixel Detector
Photo detector
Silicon Photo Diode(PD)
Silicon Avalance Diode(APD)
Multi Pixel Photon Counter(MPPC)
QA Workshop at CERN 3-4 November 2011
Development and production history of SSDs at Hamamatsu
1980 1990 2000 2010
1984~ 1987~
SSSD production on 4inch
MarkⅡ , DELPHI , ZEUS , ATLAS , etc
1998~
SSD production on 6inch
GLAST , CMS , etc
1995 ~
N in N on 4inch
ATLAS prot , LHCb prot
1992~
DSSD production on 4inch
DELPHI , CLEOⅡ , Belle , CDF , etc
SSSD R&D
1989~
on 3inch
DSSD R&D DSSD on 6inch
2009~
prot , S-Belle
2inch ⇒ 3inch
QA Workshop at CERN 3-4 November 2011
Review of main SSDs made by Hamamatsu for HEP projectsPROJECT DETECTOR TYPE size QTY. period
MARKⅡ DC-SSSD 3type 3chip/4inch 44 1987
AC-DSSD 3type 1chip/4inch
Pside: punch-through , Nside: poly-Si & DML 2chip/4inch
AC-DSSD 2type
both-side: poly-Si , Nside: DML
DELPHI up grade AC-SSSD , FOXFET 2chip/4inch 330 1994
NOMAD AC-SSSD , FOXFET 2chip/4inch 650 1996~1997
CLEOⅢ DC-DSSD , Pside: DML 2chip/4inch 550 1997~1999
AC-DSSD 3type 1chip/4inch
both-side: poly-Si , Nside: DML 2chip/4inch
AC-DSSD
both-side: poly-Si , Pside: stereo
AC-DSSD
Pside: punch-through , Nside: poly-Si & DML
AC-DSSD
both-side: poly-Si , Nside: DML
ZEUS AC-SSSD 3type , poly-Si 1chip/4inch 950 1999
AGILE AC-SSSD , poly-Si 1chip/6inch 500 2000
PAMELA DC-SSSD 1chip/6inch 300 2000
BELLE up grade AC-DSSD , both-side: poly-Si 2chip/4inch 250 2000~2002
ATLAS AC-SSSD 6type , poly-Si 1chip/4inch 16000 2001~2003
GLAST AC-SSSD , poly-Si 1chip/6inch 11500 2001~2003
CMS AC-SSSD 14type , poly-Si 1chip/6inch 24000 2003~2006
180 1998
CDF-ISL
PAMELA
KEK-B(BELLE) 2chip/4inch
2chip/4inch 60 1997
360
1chip/4inch 550 1998~1999
CLEOⅡ
1997~1999
122 1993~1994
2chip/4inch 130 1993~1994DELPHI
CDF-SVXⅡ
QA Workshop at CERN 3-4 November 2011
Production flow of SSDs
1) Design of Photo masks
2) Wafer process
3) Wafer inspection
4) Dicing
5) Chip proving
6) Visual inspection
7) Packing
8) Test of long time stability
QA Workshop at CERN 3-4 November 2011
Design of Photo masks
CAD soft on PC is mainly used for Si detectors.
Each coordinates of figures can be inputted by macro program
made by EXCEL etc.
Photo-mask design of complicated shape like
any angle or rounded strips pattern are acceptable.
QA Workshop at CERN 3-4 November 2011
Wafer process
About 10,000 of 6 inch wafers are processed per month.
We have kinds of process ( PDs , Bipolar photo IC , CCDs, C-MOS ).
For SSDs
2,000 wafers per month were processed for LHC mass production.
Available type : SSSD or DSSD, AC or DC-coupling ,
Single or Double-metal
Available thickness : 150 to 650um for SSSD , 320um for DSSD
• Oxidation• Photolithography• Ion Implantation• Poly-Si process• Metal Evaporation• Passivation
QA Workshop at CERN 3-4 November 2011
CMS-std Cut-off ( on all CMS wafers) contains
Std-Baby detector , Test structures designed by CMS ,
Mos diode , CMS-Monitor diode ( CMS-MD ) , etc.
ALL cut-offs are kept in envelop and delivered with main detector
HPK test structure
contains
test structures of poly-Si resistance ,
implant resistance , coupling capacitance , etc.
W6A main detector
W6A-baby detector ( the same design of W6A main detector )
HPK-Monitor diode ( HPK-MD )
CMS-W6A chip & Cut-offs from 6inch wafer
QA Workshop at CERN 3-4 November 2011
Wafer proving
1) Lot check using TEG
by manual or auto prober
・ implant resistance
・ poly-Si resistance
・ Flat-band voltage
・ capacitance of Cc
・ IV & CV of Monitor PD
2) IV curve of main chips
Chose good wafers
3) Put chip serial number Binary notation at
Scratch PADs on chip
< Auto prober system >
QA Workshop at CERN 3-4 November 2011
Vfd and Leakage current for CMS-thick type SSSDs VR
minmaxavespec
CMS- thick Vfd
0 0 1 4
193
698
998
1676
1932
1407
383
18 0 00
500
1000
1500
2000
2500
60
~
80
~ 100
~
120
~
140
~
160
~
180
~
200
~
220
~
240
~
260
~
280
~
300
~ 300
~
Vfd [V]
Number of Detectors
total 7310min 90Vmax 280Vave 203Vσ 29V
CMS- thick ID VR=600V)(
0 0 0 0 7
305
1298
1895
1578
1002
495
270
151766741262315 6
55
0
500
1000
1500
2000
50
~ 150
~
250
~
350
~
450
~
550
~
650
~
750
~
850
~
950
~
1000
~
ID [nA]
Number of Detectors
total 7310min 238nAmax 2367nAave 430nAσ 134nA
Vfd distribution is due to
the resistance variation of wafer
Relatively small leakage current
though big and 500um thickness
QA Workshop at CERN 3-4 November 2011
Bad channel rate for CMS-SSDs
CMS- 768ch type Bad ch.
6360
23680 28 8 4 3 2
0
1000
2000
3000
4000
5000
6000
7000
0 1 2 3 4 5 6 7
Number of bad channel
Number of Detectors
total 6721min 0max 7ave 0.084 0.011%
CMS- 512ch type Bad ch.
189 40 24 3 4
6471
0
1000
2000
3000
4000
5000
6000
7000
0 1 2 3 4 5
Number of bad channel
Number of Detectors
total 6731min 0max 5ave 0.055 0.010%
Breakdown of 937 Bad channels of all13452 detectors
42.2%
19.0%
6.7%
0.0%
4.2%
28.0%
0% 10% 20% 30% 40% 50%
Cc short
ACALshort
ACAL open
Poly- Siopen
badisolation
Implantopen
DC
AC
More than 95% of detector have no bad channels.
The average of bad channel rate is around 0.01%.
Short of Cc, Open of strip implant, short of AC AL are main factors of bad channel.
QA Workshop at CERN 3-4 November 2011
Leakage current and bad channel rate for Belle-L5-DSSDs
Vfd distribution is due to
the resistance variation of wafer
Relatively small leakage current
though DSSD
Id(nA Vr=100V)
0
5
53
38
27
23
96
24 3 2
0 0 1 0 0 0 0 00
10
20
30
40
50
60
-100
-300
-500
-700
-900
-1100
-1300
-1500
-1700
-1900
total 173min 150max 1481ave 434σ 218
number of bad ch
0
6
12
17
26
18
10
15
989
11
2 2
4
1
3 32 2
1212 2
32
0 0 0 00
5
10
15
20
25
30
0 2 4 6 810
1214
1618
202224
2628
30
total 173min 1max 26ave 8.06 0.52%
The average of bad channel rate is 0.52% and larger than SSSD
QA Workshop at CERN 3-4 November 2011
Long time stability of ATLAS proto-type SSSD
・ Long time stability of dark current under operating at 600V.
・ For 13 days biasing, no abnormalities have been identified.
< 4 probers for checking long time stability>
QA Workshop at CERN 3-4 November 2011
Failure Analysis
・ A picture of delivered sensor, whose voltage tolerancecame to be bad during the evaluation at the customer. ・ Hot electron emission was observed at the edge of guard ring. ( reported by customer )
・ We will check whether there are some weak points at the structure of guard ring.
Hamamatsu PHEMOS200
QA Workshop at CERN 3-4 November 2011
New Dicing TechnologyWe also have Stealth dicer in addition to the traditional blade dicer.
Cut as narrow as 1μm width without chipping.
Baby detectors, Test structures, Mos diodes , Monitor diodes can be cut by stealth dicer in one set and one program.
[ from Hamamatsu homepage]
QA Workshop at CERN 3-4 November 2011
Comparison between cut edge and leakage current
IV curve after dicing
1.E-09
1.E-08
1.E-07
1.E-06
1.E-05
1.E-04
0.01 0.1 1 10 100VR(V)
ID(A
)
28-128-229-129-230-130-230-333-133-2
Leakage current of stealth diced sample
are smaller than that of blade diced one.
Blade 30um
Stealth 30um
Stealth 50um
QA Workshop at CERN 3-4 November 2011
Requirement of Silicon detector for HL-LHC ■ HL-LHC radiation environment SSDs (R=30cm)> 1E15 【 1MeVn-eq/cm2 】 PIXELs (R=10cm) > 4E15 【 1MeVn-eq/cm2 】
■ Radiation damages ・ Increase in leakage current ・ Change of the full depletion voltage(Vdep) N type Si : decrease Vdep invert ⇒to P type ⇒increase Vdep P type Si : increase Vdep
・ Decrease in charge collection efficiency■ Higher bias tolerance to use in HL-LHC condition for long time.
■ Small dead space
We are now doing experiment for edge design of slim edge and GR on 320um thickness P type Si material ( initial Vfd is around 200V)
QA Workshop at CERN 3-4 November 201119
N+ N+ P+
GR
Field width
Distance to edgeP-sub
Sample No Field width Distance to edge
Q 160 280
R 200 320
S 240 360
T 240 400
U 240 450
V 240 600
W 240 750
X 240 1011
■ Sample No. Q to S N+GR width & P+ width are the same Distance to edge are changed ■ Sample No. T to X Field width are the same Distance to edge are changed
Experiment for Slim Edge design
QA Workshop at CERN 3-4 November 201120
Experiment for Slim Edge design
Onset Voltage vs Edge Distance (70MeV Proton irradiation)
0
200
400
600
800
1000
1200
1400
1600
1800
0 200 400 600 800 1000Edge distance (um)
Ons
et V
olta
ge(V
)
1.0E+141.0E+155.0E+15
Slim Edge I-V (non irradiation)
1.E- 11
1.E- 10
1.E- 09
1.E- 08
1.E- 07
1.E- 06
0 200 400 600 800 1000Vr (V)
I (A)
Q280R320S360T400U450V600W750X1011
[ data from Mitsui et al 2011]
(※ annealed for 80min@60℃)
・ Before irradiation, the sensor of edge distance lager than 450um hold over 1000V tolerance.
・ After irradiation, the voltage tolerance tends to rise.
・ So, if the sensor meets high voltage tolerance before irradiation, It also meets high voltage tolerance after irradiation.
QA Workshop at CERN 3-4 November 201121
1GR-Narrow
3GR-Wide2GR-Wide2GR-Middle
1GR-Wide1GR-Middle
Experiment for Guard Ring design
■ Distance to edge are the same, field width are the same
■ Number of GR and total width of GR are changed. 1GR ( Narrow , Middle , Wide ) 2GR ( Middle , Wide ) 3GR ( Wide )
QA Workshop at CERN 3-4 November 201122
Experiment for Guard Ring ( GR ) design
Various GuardRing Onset Volatge(70MeV Proton irraditaion)
800900
100011001200130014001500160017001800
1.0E+13 1.0E+14 1.0E+15 1.0E+16Fluence [neq/ cm2]
Ons
et V
olta
ge [V]
1GR- N 1GR- M1GR- W 2GR- M2GR- W 3GR- W
Various GuardRing I-V(non irradiation)
1.E-11
1.E-10
1.E-09
1.E-08
1.E-07
1.E-06
0 500 1000 1500 2000Vr (V)
I (A)
1GR- N 1GR- M
1GR- W 2GR- M
2GR- W 3GR- W
1GR 2GR
3GR
(※ annealed for 80min@60℃)
・ Before irradiation, multi GR ( 2GR,3GR ) holds higher voltage tolerance than single GR.
・ Width of single GR doesn't influence I-V characteristic.
・ After irradiation, the difference between single GR and multi GR tends to disappear.
・ As mentioned before, GR problem (degradation of voltage tolerance) is reported. So, we must consider from viewpoint of both slim edge design and reliability.
[ data from Mitsui et al 2011]
QA Workshop at CERN 3-4 November 2011
Summary
1. The history of Hamamatsu SSD is more than 25 years, and have been used for many HEP experiment.
2. We have enough capability to design, process, inspection and check reliability for Si detector for HEP.
3. We have a stealth dicer and will be useful for Si detectors.
4. Now we are studying the experiment to design Si detector for the next HEP experiment. And we got following results.
・ In design of edge distance of over 450um, the detector can hold over 1000V tolerance, both before and after irradiation.
・ Before irradiation, multi GR ( 2GR,3GR ) holds higher voltage tolerance than single GR
but after irradiation, the difference between single GR and multi GR tends to disappear.
・ As for GR design, we must consider from viewpoint of both slim edge design and reliability.
QA Workshop at CERN 3-4 November 2011
We Hamamatsu are proud of the fact that our Si-detectors are being used in many physics experiments.
We are continuously making efforts to provide a better Si detectors.
Thank you for your attention !