Upload
mwuest
View
221
Download
0
Embed Size (px)
Citation preview
7/28/2019 S03 04 Collier SWTW2010
1/24
ImpactofBondPadCorrosion
June6to9,2010SanDiego,CA USA
Terence Q. CollierCVInc
7/28/2019 S03 04 Collier SWTW2010
2/24
June6to9,2010 IEEESWTestWorkshop 2
Overview
Cleanthewafer nottheprobecard Historyofcontamination Corrosionremoval ExamplesofPreandPostCleans Analysisofpackagedunitsreliability Summary
7/28/2019 S03 04 Collier SWTW2010
3/24
June6to9,2010 IEEESWTestWorkshop 3
Cleanthewafer nottheprobecardRoot cause of high contact resistance (CRES)
is corrosion and contamination, not the probecard. Goal should be to eliminate fluorinecorrosion on bond pads
7/28/2019 S03 04 Collier SWTW2010
4/24
June6to9,2010 IEEESWTestWorkshop 4
WhyCleantheWafer Aprocessthatwillstripthecorrosion/oxidationbutnot
attackthebasemetalsisdesired.Traditionalmethods
includeplasma,phosphoric(PANordilutephosphoric)
acidderivatives
and
even
dilute
TMAH
for
aluminum.
Unfortunatelyphosphoricgoesaftergoodaluminumas
wellandanystrongbasewillhaveadramatic,notagood
one,impactonaluminum.
Improvecostofownership:
Extendhardwarelife. Improveassemblyyield. Reducescrapdieandwafers.
Reducereliabilityriskandregainprocesscontrol.
Damagedpadsleadtocustomerreturns
7/28/2019 S03 04 Collier SWTW2010
5/24
June6to9,2010 IEEESWTestWorkshop 5
Historyofcontamination
7/28/2019 S03 04 Collier SWTW2010
6/24
June6to9,2010 IEEESWTestWorkshop 6
Two items impact post fab yield:Test hardware
Bond pad conditions.
Removing the highly resistive layer can improve yield, extend probe card life, minimizereliability issues and help manage process control constraintsGoal: Remove the resistive layer without damaging the underlying metal layer.
What process has the following capabilities:No etch rate on metalsLow etch rate on passivation layers and Si.Love oxide and corrosion.
TheProblem ContactResistanceDue
to
Corrosion,
Not
Oxidation
7/28/2019 S03 04 Collier SWTW2010
7/24
June6to9,2010 IEEESWTestWorkshop 7
Dry Box Conditions - Fluorine contamination is apparent at time zero asreceived. As time continues the pads become more corroded. Conditions ofthese pads are typical and are the root of poor contact resistance. If the padsare not cleaned then poor yield is likely as well as die and hardware damage.
Zero
Time
One
Week
SixWeeks
ThreeWeeks
AlBondPadAnalysisOverTime
7/28/2019 S03 04 Collier SWTW2010
8/24
June6to9,2010 IEEESWTestWorkshop 8
ProbedPads
Prior
to
Clean
Brown corrosion layer variesfrom pad to pad. Some pads aremore corroded than otherswhich is why a solution isrequired that does not attack
aluminum.
A material that attacksaluminum would likely etch
away the top pad prior toreaching pure aluminum on thebottom pad.
7/28/2019 S03 04 Collier SWTW2010
9/24
June6to9,2010 IEEESWTestWorkshop 9
CorrosionRemovalinFiveMinutes
7/28/2019 S03 04 Collier SWTW2010
10/24
June6to9,2010 IEEESWTestWorkshop 10
CleaningImprovements
AsRecvd
BPS100
5min
7/28/2019 S03 04 Collier SWTW2010
11/24
June6to9,2010 IEEESWTestWorkshop 11
Cleaningand
rinsing
BPS100 can be used to clean thepads but as with any pure aluminumsurface DI water is an enemy.
An oxidized aluminum surface will
be more resilient to DI water than afresh layer.
An example of a cleaned surfaceleft in DI water shows the etching of
the probe mark on the wafer after90 minutes.
7/28/2019 S03 04 Collier SWTW2010
12/24
June6to9,2010 IEEESWTestWorkshop 12
ComparisonofSurfaceTexture
Sharp Al2O3 oxide peaks replacedby spongy corrosion layers.
Clockwise from left, Al+oxide, Al atambient for 3 days, Al + ambient +CF4 plasma.
7/28/2019 S03 04 Collier SWTW2010
13/24
June6to9,2010 IEEESWTestWorkshop 13
Voltage(mV
)
Touchdowns
BeforeBPS
AfterBPS
Lower CRES
Less Noise
Fewer false fails
Less CRES increaseover touchdowns
DramaticResultsAfterBPSonAl
Real resistancebetween two pinson the pad at10um overtravel
7/28/2019 S03 04 Collier SWTW2010
14/24
June6to9,2010 IEEESWTestWorkshop 14
BaselineSample;
No
Clean
Fuzzy, oxidized aluminum pads.
7/28/2019 S03 04 Collier SWTW2010
15/24
June6to9,2010 IEEESWTestWorkshop 15
BPS100(Sample
100
5)
Well defined grain boundaries..
7/28/2019 S03 04 Collier SWTW2010
16/24
June6to9,2010 IEEESWTestWorkshop 16
SideBy
Side
Comparison
of
the
BondPads
As Received, without BPS100.With BPS100 for 5 minutes.
7/28/2019 S03 04 Collier SWTW2010
17/24
June6to9,2010 IEEESWTestWorkshop 17
AugerDepthProfiles 3DaysC09ZG15711.pro: Sample 0 EAG
2009 Oct 19 10.0 keV 0 FRR 1.5904e+005 max
Al2/Full
0 20 40 60 80 100 120 140 160 180 2000
10
20
30
40
50
60
70
80
90
100C09ZG15711.pro
Sputter Depth ( )
AtomicConc
entration(%)
O
F
Al(O)
Si Ti
Al
O1.ls1C1.ls1F1.ls1
Al4.ls1Al2.ls2Si4.ls1Ti1.ls1
C09ZG15721.pro: 100-5 EAG2009 Oct 19 10.0 keV 0 FRR 9.8344e+001 max
Al2/Full
0 20 40 60 80 100 120 140 160 180 2000
10
20
30
40
50
60
70
80
90
100C09ZG15721.pro
Sputter Depth ( )
AtomicConce
ntration(%)
Al
O
C
F
Al2O1.ls1C1.ls1F1.ls1
Sample 0
100-5
70
20
F present until 40A
Si3N4 at 25A
Ti unusually present
AlO up to 80A; typicalfor Al
O at +120A
90% Al at 200A
F removed at 10A
Si3N4 eliminated
Ti eliminated
AlO eliminated
O in some form up to40A
90% Al at 40A
As Received
BPS100
7/28/2019 S03 04 Collier SWTW2010
18/24
June6to9,2010 IEEESWTestWorkshop 18
Analysisofpackagedunitsreliability
7/28/2019 S03 04 Collier SWTW2010
19/24
June6to9,2010 IEEESWTestWorkshop 19
WireBond
Evaluation
7/28/2019 S03 04 Collier SWTW2010
20/24
June6to9,2010 IEEESWTestWorkshop 20
CrossSections
7/28/2019 S03 04 Collier SWTW2010
21/24
June6to9,2010 IEEESWTestWorkshop 21
WireBondTestatMSL3
7/28/2019 S03 04 Collier SWTW2010
22/24
June6to9,2010 IEEESWTestWorkshop 22
CrossSectionsatMSL3
7/28/2019 S03 04 Collier SWTW2010
23/24
June6to9,2010 IEEESWTestWorkshop 23
Low etch rate on baseline aluminum pad.
Five minutes in BPS will remove CRES layer for up to 3-4 days.
Pads can be brought to t=0 from nitride etch.
Reduces bottlenecks and hardware costs.Improves yield and cycle time.
Decreases reliability risks with no detrimental impact to assembly
Clean the wafer, not the probes!
Summary
7/28/2019 S03 04 Collier SWTW2010
24/24
June6to9,2010 IEEESWTestWorkshop 24
ThankYou!Thankyoufortakingtimetolistingtothispresentation.TheauthorwouldalsoliketothankthefollowingfolksatAirProductsfortheirhelpandsupport:
DavidRennieRajRamamurthi
Questions?