advanced vertical technologies for low damage probing of ...scrub dia. ratio (%) scrub dia (um)...
Post on 28-Jan-2021
3 Views
Preview:
TRANSCRIPT
-
Advanced Vertical Technologies for Low Damage Probing of Bumps, Pillars, and Pads
Gwen GerardJean-Pierre Gibaux
Texas Test CorporationHsinchu, Taiwan
Jerry Broz, Ph.D.Darren Aaberge, Shota HetsugiMicronics Japan Corp. (MJC)
Tokyo, Japan
Texas Test Corporation
-
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
Agenda
2End User / TTC / MJC
-
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
Agenda
3End User / TTC / MJC
-
LOCATIONHeadquarters: Dallas, Texas, U.S.A.Product and Test Engineering: Sophia Antipolis, FranceManufacturing Operations: Hsinchu, Taiwan and Shanghai, China
WHO WE ARETeam of former Texas Instruments Engineers.
WHAT WE OFFERThe full range of product and test engineering services.
Texas Test Corporation (TTC)
4End User / TTC / MJC
-
• Bumped devices are moving from 150um pitches to 80um (and smaller) with high numbers of smaller bumps– Bump pitch range from 130 to 100um today moving into 80um (or less)– Bump diameters shrinking from 70um sub-30um for certain structures– Reducing Cost of SoC Test is driven by large arrays for x256 multi-sites and pin-counts as high as 40,000 probes
• Pitch reductions and minimum allowable damage for assembly and die stacks– Mobile Devices 80um– Automotive Devices 65um– HBM Memory Devices 50um– Wide I/O and Wide I/O 2 Devices 40um
• Electrical performance – Low and stable CRES for functional test– Increased CCC for steady state and pulsed current– High speed performance testing
• Tri-temperature characterization– Demanding automotive standards– Multiprobe -55C to 25C to 200C– Same test cell and probecard
Next Gen Requirements Create Probe Challenges
Source: B. Mair SW-Test Archives
5End User / TTC / MJC
-
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
Agenda
6End User / TTC / MJC
-
Flip Chip Type DeviceMulti-die test of devices withArea Array (Bump / Cu Pillar)
MJC MEMS-SP (MSP)
MJC MEMS Spring Probe (MSP) Technology• MEMS fabricated, non-oxidizing barrel & spring • Proprietary fabrication process for probe architecture• Spring force is controlled and defined by barrel geometries• Preload at space transformer for stable contact• On-site needle replacement capability
MJC MEMS Spring Probe for Solder Bumps, Cu Pillars, and Pads
7End User / TTC / MJC
Flat type Crown typePoint type
Wafer Side ST Side
for Al Pad for Solder Bump for Solder Bump
Wafer Side Space Transformer Side
-
Flip Chip Type DeviceMulti-die test of devices withArea Array (Bump / Cu Pillar)
MJC MEMS-SP (MSP)
MJC MEMS Spring Probe (MSP) Technology• MEMS fabricated, non-oxidizing barrel & spring • Proprietary fabrication process for probe architecture• Spring force is controlled and defined by barrel geometries• Preload at space transformer for stable contact• On-site needle replacement capability
MJC MEMS Spring Probe for Solder Bumps, Cu Pillars, and Pads
6
5
4
3
2
1
00 50 100 150 200 250 300
Allowable Overdrive
Forc
e [g
f]
O.D [um]
Spring Force can be designed to be anywhere within this range.
8End User / TTC / MJC
-
• Plunger material and tip shape can be selected depending on the application.
MJC MEMS Spring Probe Tip Plunger Geometry
Solder Bump
Flat PlungerCrown Plunger
Cu Pillar Bump
Flat Plunger Pointed Plunger
Cu Pillar / TSV Al Pads
9End User / TTC / MJC
-
Stroke = Preload + Overtravel (micron)
Solder Bump Probe with MJC MSP Crown Tip Plunger
Solder Bump Deformation vs. Applied Stroke (um)
4gf
2gf
6gf
8gf
Crown type
10End User / TTC / MJC
-
Stroke = Preload + Overtravel (micron)
Solder Bump(6gf Probe)
● Initial● After-1M
Durability Evaluation「Needle position」Durability Evaluation [ Needle position ]
Initial After -1MInitial
Forc
e (g
ram
)
Durability Evaluation [ p-Force ]
Solder Bump Probe with MJC MSP Crown Tip Plunger
Solder Bump Deformation vs. Applied Stroke (um)
4gf
2gf
6gf
8gf
11End User / TTC / MJC
-
Stroke = Preload + Overtravel (micron)
Probe Mark
dia. 14um
d/D: 27%
3gf
Probe Mark
dia. 15um
d/D: 30%
4gf
Probe Mark
dia. 12um
d/D: 24%
2gf
Room
Tem
p.
Initial
Dia: 50um
Dia_initial = 50um diameter Cu Pillar Bumps with Sn/Ag lead-free solder cap
Flat type
Cu Pillar
D
d
12End User / TTC / MJC
Cu-Pillar Bump Probe (RT) with MJC MSP Flat Tip Plunger
-
Stroke = Preload + Overtravel (micron)
Cu Pillar Bump(3gf Probe)Durability Evaluation [ Needle position ]
● Initial● After-1M
After -1MInitial
Forc
e (g
ram
)
Durability Evaluation [ p-Force ]
Probe Mark
dia. 14um
d/D: 27%
3gf
Probe Mark
dia. 15um
d/D: 30%
4gf
Probe Mark
dia. 12um
d/D: 24%
2gf
Room
Tem
p.
Initial
Dia: 50um
Dia_initial = 50um diameter Cu Pillar Bumps with Sn/Ag lead-free solder cap
Cu Pillar
D
d
13End User / TTC / MJC
Cu-Pillar Bump Probe (RT) with MJC MSP Flat Tip Plunger
-
Stroke = Preload + Overtravel (micron)
Probe MarkDia.18umd/D:20%a/A: 4%
Probe MarkDia.19umd/D:22%a/A: 5%
4gf
Probe MarkDia.23umd/D:26%a/A: 7%
Probe MarkDia.27umd/D:30%a/A: 9%
6gf
Probe MarkDia.25umd/D:28%a/A: 8%
Probe MarkDia.31umd/D:35%a/A:12%
8gf
RT
HT9
0C
Solder Bump Deformation vs. Applied Stroke (um)
Dia_initial = 90um diameter bumps (lead-free solder alloy, Sn/Ag3/Cu0.5) *SMIC_M705
Flat type
Cu-Pillar Bump Probe (HT) with MJC MSP Flat Tip Plunger
14End User / TTC / MJC
-
• Multisite cards with > 28K pin counts have been successfully installed.
MJC MEMS SP-Probe … In the Field !
15End User / TTC / MJC
-
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
Agenda
16End User / TTC / MJC
-
End User Objectives
• Improve Sort Performance over Cobra Style Vertical Card– Probe force consistency and “tune-ability” for uniform probe marks– Low damage on small bumps– Probe-to-bump alignment (PTBA) – Stable electrical performance – High 1st Pass Yields w/ reduced recovery test
• Reduced Cost of Ownership over Cobra Style Vertical Card– Reduced maintain and increased lifetime performance (MTBR and End of Life)– Minimize test cell down due to contact related – Easy repair and simple single pin replacement capabilities
17End User / TTC / MJC
-
• Equipment– Tester = VLCT Platform– Prober = Accretech 300mm Prober– Probe Type = MEMS SP w/ flat tip plunger– Probe Card Test Vehicle = 1 x 4 DUT w/ 2140 pins
• Test Conditions– Test temp = 30C– OD = Variable as defined during testing
• Cleaning conditions for all testing– Octagonal movement– Cleaning Overtravel = 100um– Cleaning Frequency = 1 clean per 200 wafer TD
Test Cell Overviewwww.accretech.com
Wegleitner, SWTW 2005
18End User / TTC / MJC
-
Property Specification
Target Material Cu Pillar Bump
Minimum pitch 95um pitch with full array
Tip shape Flat Plunger
Contact force 3gf / OD150um
C.C.C.(ISMI 20% force drop) 800 mA
Alignment +15um
Planarity Less than 50um
Temperature -40C to 90C
1 x 4 DUT layout w/ 2140 pins
Probe Card Test Vehicle and MSP Probe
MEMS-Spring
Flat Plunger
Cu Pillar Bump
19End User / TTC / MJC
-
Qualification Test Plan• Contact Resistance (CRES) Assessment
– Determine OD to attain stable CRES– CRES vs. Repeated Touchdowns
• Damage Assessment– Bump Damage vs. Overdrive– Bump Damage vs. Repeated Touchdowns
• Bin-to-bin Reproducibility• Stable Correlation Wafer Yield Results• CRES Determination and Trending
– MJC MSP Probe vs. Cobra Style Vertical
20End User / TTC / MJC
-
10
12
14
16
18
20
22
24
26
28
30
25 50 75 100 125 150 175 200
Cont
act R
esis
tanc
e (oh
ms)
Overtravel (micron)
CRES vs. Overtravel
Stable resistance attained @ Overtravel > 50um
Overtravel to Attain Stable CRES
Stable CRES
Overtravel Zones
• Overtravel Applied to Wafer– Overtravel Range = 25 to 200um– Overtravel Increment = 25um– Single Touchdown
21End User / TTC / MJC
200u
m17
5um
150u
m12
5um
100u
m75
um50
um25
umOD
FailPass
-
0
10
20
30
40
50
60
70
80
90
100
25 50 75 100 125 150 175
Corr
elat
ion
Waf
er Y
ield
(%)
Overtravel (micron)
Correlation Wafer Yield vs. Overtravel
PassFailOpen
Stable Correlation Wafer Yield @ Overtravel > 75um
Overtravel to Attain Stable Correlation Yield
Stable Yield (~93%)
• Overtravel Applied to Wafer– Overtravel Range = 25 to 200um– Overtravel Increment = 25um– Single Touchdown
22End User / TTC / MJC
Overtravel Zones
200u
m17
5um
150u
m12
5um
100u
m75
um50
um25
umOD
FailPass
-
Cu Pillar
D
d
Scrub Dia. Ratio (%) =d/D
D
d0
10
20
30
40
50
60
0 25 50 75 100 125 150 175 200 225
Bum
p Da
mag
e
Overtravel (micron)
Cu-Pillar Bump Damage vs. Overtravel
Scrub Dia. Ratio (%)
Scrub Dia (um)
Overtravel vs. Bump Deformation and Damage• At higher overtravel, the bump are more deformed and tend to higher %-damage.• At OD = 200um, the %-Damage is significantly less than the specification limit (% Damage < 50%).
23End User / TTC / MJC
11.5
11.5
13.8
14.5
16.1
16.8
17.8
18.4
Scrub Diam.
(micron)
23.0
23.0
27.6
28.9
32.2
33.6
35.5
36.8
Scrub Ratio
(%)
ScrubMark
25um
50um
75um
100um
125um
150um
175um
200um
OD(micron)
-
Multi-Touchdowns vs. CRES• Touchdowns Applied to Wafer
– Overtravel = 150um– Number of TDs = 1 to 6
• Function Test @ 1TD• CRES Test @ 2 to 6TD
• Electrical testing performed across 10-critical pins.
Touchdowns Electrical Tests Performed1 24422 20133 15834 11505 7146 281
0
1
2
3
4
5
6
7
8
9
10
1 2 3 4 5 6
Cont
act R
esis
tanc
e (oh
ms)
Number of Touchdowns
CRES vs. Touchdowns
Stable Contact Resistance @ TD 1 to 6
Stable CRES
24End User / TTC / MJC
-
Multi-Touchdowns vs. Correlation Wafer Yield• Touchdowns Applied to Wafer at OD = 150um
– Number of Touchdowns = 1 to 6 (Function Test @ 1TD w/ retest check; CRES Test @ 2 to 6TD)
• 1st Pass Correlation Wafer Yield = 91.5%• Re-Test Recovery 13 chips were recovered• No further recovery for TD = 2 thru 6 • Pass / Fail / Open of each touchdown
sequence were well correlated.
1st Pass Yield 91.5%1st Pass Fail 8.5%
1st Pass Open 0.7%Retest Recovery 13 ChipContaminations
(Immediate after cleaning)
■Pass■Fail※Open
Contaminations(Immediate after cleaning)
1 Touchdown 2 Touchdown 3 Touchdown
4 Touchdown 5 Touchdown 6 Touchdown
25End User / TTC / MJC
-
0
10
20
30
40
50
60
0 1 2 3 4 5 6 7
Bum
p Da
mag
e
Overtravel (micron)
Cu-Pillar Bump Damage vs. Multi-Touch @ OD = 150um
Scrub Dia. Ratio (%)
Scrub Dia (um)
Multi-TD vs. Bump Deformation and Damage• After multiple touchdowns at OD = 150um, the bump are more deformed and have higher damage.• After 2 x TDs, the %-Damage is remains constant and less than the specification limit (% Damage < 50%)
26End User / TTC / MJC
Cu Pillar
D
d
Scrub Dia. Ratio (%) =d/D
D
d
16.1
20.1
19.7
19.7
23.0
19.1
Scrub Diam.
(micron)
32.2
40.1
39.5
39.5
46.1
38.2
Scrub Ratio
(%)
ScrubMark
1
2
3
4
5
6
TouchDown
-
• Test Conditions– Wafer Overtravel = 150um– Octagonal movement– Cleaning Overtravel = 100um– Cleaning Frequency = 1 clean per 200 wafer TD
Correlation Determination > 99%
FailPass
Highly significant positive correlation confirmed
27End User / TTC / MJC
-
歩留まり確認結果
歩留まり[%]
60
65
70
75
80
85
90
95
K115
G01
K115
G04
K115
G07
K115
G10
K115
G12
K115
G15
K115
G17
K115
G20
K115
G23
K116
301
K116
304
K116
306
K116
309
K116
311
K116
314
K116
317
K116
320
K116
323
K116
325
K117
703
K117
705
K117
708
K117
711
K117
714
K117
717
K117
719
K117
722
K117
724
K118
C02
K118
C05
K118
C08
K118
C11
K118
C13
K118
C16
K118
C18
K118
C21
K118
C24
K14F
A02
Wafers (4 lots of 25-wafers per LOT)
% Yield Result
Sani
tized
Yie
ld (%
)
4 x Wafer Lot % Yield Results• Total Yield Average for 4-LOT
– First Pass Yield Target was attained and confirmed
– MSP card had significantly better first pass yields than a comparable Cobra card.
• MSP card had low recovery rate vs. a high recovery rate observed with the Cobra card.– Recovery Rate for MJC MSP PC ≤ 0.13%– Recovery Rate for Cobra PC ≥ 0.50 %
Recovery Rate for MSP PC ≤ 0.13%
YieldTarget
28End User / TTC / MJC
-
MJC MSP vs. Cobra Probe CRES Trending
29End User / TTC / MJC
Resis
tanc
e (o
hms)
Cobra Style Probe Card
Resis
tanc
e (o
hms)
MJC MSP Probe Card
Resis
tanc
e (o
hms)
MSP Probe Cobra Probe
• Test Conditions– Wafer Overtravel = 150um– Octagonal movement– Cleaning Overtravel = 100um– Cleaning Frequency = 1 clean per 200 wafer TD
-
Summary of QualificationProperty Requirement Result Assessment
CRES vs. OD OD = 75 to 200umTarget OD = 150umStable CRES at
OD ≥ 50um
Correlation Yield vs. OD
OD = 75 to 200umTargety OD = 150um
Stable Yield atOD ≥ 75um
Bump Damage vs. OD
% Damage ≤ 50%OD = 75 to 200um
% Damage ≤ 40% Max. OD = 200um
CRES vs. TD TD = 1 to 6Stable CRES at
Max. Allowable TD = 6
Correlation Yield vs. TD TD = 1 to 6
No additional recoveryafter 1st retest
Bump Damage vs. TD
% Damage ≤ 50 %at OD = 150um
% Damage ≤ 40% Max. Allowable TD = 6
Test ReproducibilityStatistical Correlation
better than 99%All bins within 0.3%
(R2 = 0.999)
Wafer Lot ResultHigh 1st Pass YieldLow Recovery Rate
Recovery Rate ≤ 0.13%
ResistanceVariance
Improved Stability over Cobra Type
Statistically Reduced Variance
30End User / TTC / MJC
-
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
Agenda
31End User / TTC / MJC
-
• Production Qualification Overview– More than 5000 production device wafers were split across multiple test cells – Split Lot Ratio MJC MSP (15%) and Cobra Style (85%)
• Production Metrics comparison for MJC MSP vs. Cobra Style – First Pass Yield Improvement 0.37% increase in FPY over Cobra Style– Significant Reprobe Rate Reduction across 5000 wafers:
– Avg. 2.69% wafer test time reduction was realized with MJC MSP Probe
High Volume Manufacturing Validation
32End User / TTC / MJC
Card Type Average StdDev
COBRA 0.34% 0.83%
MJC MSP 0.09% 0.08%
-
Agenda
33End User / TTC / MJC
• Background• MJC MEMS Spring Probe (MSP) for Solder Bumps, Cu Pillars, and Pads• Qualification of MJC MSP Probe on Cu Pillar Devices at End User
– End User Objectives– Qualification Test Plan– Results Summary
• High Volume Manufacturing Validation• Summary / Conclusions
-
• Bumps, Pillars, and Pads of next-gen devices continue to shrink in size for higher pin counts with tighter pitches requiring precise and low force contact under various test conditions.
• MJC MEMS SP has “tune-ability” of probe force, plunger shape, and contactor metallurgy.– Engineered low probe forces facilitate high pin counts with reduced %-damage– Tip geometries designed for CRES stability with low %-damage after multiple touchdowns– Contactor shapes and metallurgies optimized for high CCC, low CRES, and %-Yield stability
• Under HVM conditions at End User, MEMS SP had superior performance over Cobra Style probe technologies.– A 0.37% improvement in First Pass Yield– Significantly reduced reprobe and recovery rates– Overall 2.69% reduction in average wafer test times
• Future work– Production probe card performance and lifetime characterization– Elevated temperature (125C) test conditions
Summary / Conclusions
End User / TTC / MJC 34
-
• End User Probe Process Team (Special Thanks !)• Ardentec Probe Process Team• Philippe Cavalier (Texas Test Corporation)• Tom Stewart (Texas Test Corporation)• T.T. Kanenari (Micronics Japan Co., LTD)• Hideo Kuroyanagi (Micronics Japan Co., LTD)• Keita Kudo (Micronics Japan, Co., LTD)
Acknowledgements
35End User / TTC / MJC
top related