fine pitch probing and wire bonding and reliability … · june 13, 2000 fine pitch probing and...

June 13, 2000

Fine Pitch Probing and Wire Bonding and Reliability ofAluminum Capped Copper Bond Pads

Fine Pitch Probing and Wire Bonding andFine Pitch Probing and Wire Bonding andReliability ofReliability of

Aluminum Cap Copper Bond PadsAluminum Cap Copper Bond Pads

TuTu Anh Tran Anh TranLois YongLois Yong

Robert Robert RadkeRadke

Electroplate Cu

Cu Seed

Barrier Dielectric

June 13, 2000


Topics to be CoveredTopics to be Covered

• Criteria for Reliable Ball Bonds• Inspection Criteria for Probe Mark Damage• Impact of Probe Marks to Wire Bonding in Al Technology:

– 43µm Ball Bond• Back-end Challenges in Cu Technology• Impact of Probe Marks to Wire Bonding in Cu Technology:

– 70 and 60µm Ball Bonds– Multiple metal layers– Probe conditions

• Conclusions and Recommendations

June 13, 2000


Criteria for Reliable Ball BondsCriteria for Reliable Ball Bonds

Au-Al Intermetallic Coverage of >70%

100% On-bonding Ball Shear Strength/Area = 5.5gf/mil2

No Pad Cratering after KOH

No Pad LiftNo Non-stick

June 13, 2000


Inspection Criteria for Probe Mark DamageInspection Criteria for Probe Mark Damage

Loose criteria for inspecting probe mark damage on pad:– No exposed oxide, no cracked glass– Max 25% probe/pad ratio– Probe/ball bond ratio is NOT specified!!

Large variation in probe mark sizes:– Tends to magnify at finer pitch and smaller ball bonds

0%

20%

40%

60%

80%

100%

120%

140%

160%

180%

65 60 45 35Ball Size (µm)Pad Size (µm) 77 72 59 48Pitch (µm) 91 76 63 52

Pro

be/

Bo

nd

Pro

be/

Pad

1X Heavy - 20x40µm

1X Light - 22x28µm

2X Heavy - 27x49µmData Courtesy of Fuaida Harun

June 13, 2000


Impact of Probe Mark to Wire BondingImpact of Probe Mark to Wire Bonding43µm Ball Bond on Al Technology

Non-stick on Pad(NSOP):– Very significant NSOP rate for Center Probe compared Offset

Probe in 2XHLifted Metal after wire bonding:

– Both Heavy probe marks experience Lifted Metal for Center ofProbe with higher rate for 2XH

– Lifted metal also experienced for the Offset Probe ( no space)

– Large probe marks decrease Au-Al intermetallicscoverage and increase bond non-sticks and pad lifts.

– Must control to < 60% probe/ball bond ratio

1XL

1XH

2XH

NSOP Lifted Metal

Center Probe Offset Probe Center Probe Offset Probe

1 X Light 0% 0% 0% 0%

1 X Heavy 0% 0% 1.17% 0%

2 X Heavy 12% 0.13% 1.95% 0.19%

Data Courtesy of Fuaida Harun

June 13, 2000


Back-end Challenges in Cu TechnologyBack-end Challenges in Cu Technology

Cu Bond Pad á Oxidation

Thin Top Metal á Pad Cupping

Multiple Metal Layers á Pad Sinking

Active Circuitry Under Pads á Circuitry Damage

Fine Pitch Geometry á Probe and Wire Bond

CapabilitiesRequired Production Capability

107

91

7670

6352

9677

7262

59

48Wire Pitch (µm)Pad Size (µm)

1/96 1/97 1/99 1/00 1/01

120

100

80

60

40

20

0Min

imu

m R

equ

ired

Dim

ensi

on

June 13, 2000


Back-end ApproachBack-end Approach

Advantages:• Existing Al fab equipment/process/tooling• Re-metallization is performed in fab, thereby reducing cycle time• Fine pitch capability due to photo lithography• Existing know-how in probe and assembly on Al bond pads

Approach:• Developed 76µm pitch probe and wire bond capability (60 µm Ball ∅)

on thin Cu layer• Al Cap Cu wire bond process window is narrower and shifted to the

right compared to standard Al process.

Cu Bond Pad

Cu

Passivation Opening

Al Cap Re-metallization

Cu

Diffusion/Adhesion Barrier Al

June 13, 2000


Fabrication

Experiment FlowExperiment Flow

Probe Assembly

Variablesq Tip Diameter - 1 milq Overdrive - 60µm (k=1.5gf/mil)q Probe Passes - 6 max

Variablesq Ball Bond Size - 70 and 60µm

Responses/AnalysisStructural Integrity - Thermal Age Study(175ºC; 0, 5, 24, 100, 200, 500, 1000 hrs)q Ball Shear & Failure Modeq Rip Test & Failure ModeReliabilityq MSL 3 + Temp Cycle (-65 to 150ºC)

- 0, 500, 1000 cyclesq MSL 3 + Autoclave - 0, 96, 144 hrs

Variablesq 1-Metalq 3-Metal

1-metal device6.7 x 6.7mm230 pads256 17x17 BGA

3-metal device4.2 x 4.7mm174 pads196 15x15 BGA

Evaluate Effects of:• Ball Sizes - 70 and 60 µm• Multiple Layers - 1 and 3• Probe Conditions - up to 6 touches

June 13, 2000


Au-Al Intermetallics Coverage %Au-Al Intermetallics Coverage %

Un-probed Pad90%

Pad Probed 4-6x 75%

Pad Probed 8x 25%

Photo Courtesy of Fuaida Harun

June 13, 2000


Effects of Smaller Bond Sizes -Effects of Smaller Bond Sizes - Ball Shear Results Ball Shear Results

Effect of smaller bond sizes:– Different bond sizes yielded similar ball shear/area with optimized wire bond

processes.

Sample Size/cell :16 balls/unit x 5 units = 80 balls

14.0

22.0

30.0

38.0

46.0

0 100 200 300 400 500

Time at 175οC (hrs)

Bal

l Sh

ear

Rea

din

g(g

f)

Min Spec Limit: 15gf

1-metal Device, Probed 6x, Ball Bonds 70 and 60µm

5.0

6.0

7.0

0 100 200 300 400 500


Bal

l Sh

ear/

Are

a (g

f/m

il2)

1-metal / 60um 1-metal / 70um

61.61 µm Ball Size

71 µm Ball Size

Min Spec Limit: 5.5gf/mil2

June 13, 2000


Ball Shear Failure ModesBall Shear Failure Modes

Mode 1Less than 25% gold lefton pad

Mode 2More than 25% gold lefton pad

Mode 3APad lift below barrierexposed Cu or oxide

Mode 3BPad lift at barrier

June 13, 2000


Wire Rip Failure ModesWire Rip Failure Modes

Cu

Mode 1Gold wire breaks atneck

Mode 2Ball lifts pad remainsintact

Mode 3APad lift below barrierexposed Cu or oxide

Mode 3BPad lift at barrier

June 13, 2000


Failure Mode ResultsFailure Mode Results

Effect of Smaller Ball Bond:– Increasing mode 2s (Break through Au-Al intermetallics) with increasing thermal age

indicates weakening bonds.

– Smaller bonds weakened earlier than larger bonds.

1-metal Device, Probed 6x, Ball Bonds 70 and 60µm

Ball ShearBall Shear Wire RipWire Rip

0%

20%

40%

60%

80%

100%

0 24 150 504 0 24 150 504

1 2 3A 3BTime at 175οC (hrs)

1-metal / 60um Ball1-metal / 70um Ball

0%

20%

40%

60%

80%

100%

0 24 150 504 0 24 150 504


1-metal / 60um 1-metal / 70um Ball

Sample Size/cell :16 balls/unit x 5 units = 80 balls

173 wires/unit x 2 units = 346 wires

June 13, 2000


20.0

24.0

28.0

32.0

36.0

0 100 200 300 400 500


Bal

l Sh

ear

Rea

din

g(g

f)Effects of Multi-layer Structure -Effects of Multi-layer Structure -

Ball Shear Results Ball Shear Results

Min Spec Limit: 15gf

1-metal and 3-metal Devices, Probed 6x, Ball Bonds 60µm

2

3

4

5

6

7

8

0 100 200 300 400 500


Bal

l Sh

ear/

Are

a (g

f/m

il2 )

1-metal / 60um 3-metal / 60um

60.35 µm Ball Size61.61 µm Ball Size

Min Spec Limit: 5.5gf/mil2

Effect of Multi-layer Structure:– Slight degradation in ball shear strength for 3-metal device compared to 1-metal

device.

June 13, 2000


Failure Mode ResultsFailure Mode Results

Effect of Multi-layer Structure:– More mode 3s was observed on 3-metal device.

– When evaluating failure mode with ball shear strength, mode 3s arenot indicative of a problem.

• High shear strength indicates that a robust Au-Al intermetallic formationis stronger than the Al cap interfacial strength.

– Bonding onto 3-metal device weakened faster than on 1-metal device.

1-metal and 3-metal Devices, Probed 6x, Ball Bonds 60µm


0%

20%

40%

60%

80%

100%

0 24 150 504 0 24 150 504


1-metal 3-metal

0%

20%

40%

60%

80%

100%

0 24 150 504 0 24 150 504


1-metal 3-metal

Mode 3A

June 13, 2000


Effect of probe conditions:– Beyond t(504 hrs) P3 ball shear strength deteriorates more rapidly than the P1 ball

shear strength.

10.00

15.00

20.00

25.00

30.00

35.00

0 200 400 600 800 1000


Bal

l Sh

ear

Rea

din

g (

gf)

P1P3

3

4

5

6

7

8

0 200 400 600 800 1000

Time at 175οC

Bal

l Sh

ear/

Are

a (g

f/m

il2 )

P1P3

60.12 µm Ball Size

60.31 µm Ball Size

Min Spec Limit: 15gf Min Spec Limit: 5.5gf/mil2

Effects of Probe ConditionsEffects of Probe Conditions - - Ball Shear Results Ball Shear Results

3-metal Device, Probed 1x and 3x, Ball Bonds 60µm

June 13, 2000


Ball Shear Failure Mode ResultsBall Shear Failure Mode Results 3-metal Device, Probed 1x and 3x, Ball Bonds 60µm

Ball shear failure modes evaluated in conjunction with relatively stableshear strength readings indicate:

– Mode 3s are not indicative of a problem. Shear strength of a robust Au-Al intermetallicformation is stronger than the Al cap interfacial strength.

– Decrease in mode 3s and increase in mode 2 at t(1008 hrs) points to a weakening bond.– P1 bonds are more robust than P3 bonds (more mode 3s).

Wire rip: P1 bonds are more robust than P3 bonds (less mode 2s).

0%

20%

40%

60%

80%

100%

0 24 150P1

504 1008 0 24 150P3

504 1008

3B

3A

2

10%

20%

40%

60%

80%

100%

0 150P1

1008 0 150P3

1008

3B3A21


June 13, 2000


High Condition:1 mil tip (k=1.5gf/mil),

60um OD,3 Double-touch Passes

Size: 18 x 36 µmProbe/Bond: 28% O hrs 24 hrs 200 hrs

Worst Condition:1 mil tip (k=1.5gf/mil),

80um OD,4 Double-touch Passes

Size: 22 x 72 µmProbe/Bond: 70% O hrs 24 hrs 200 hrs

Probe Effects on Wire BondabilityProbe Effects on Wire Bondability

NOT RECOMMENDED

June 13, 2000


Cross-sectionsCross-sectionsT

(0

ho

ur)

T (

200

ho

urs

)

Voids

Al richAu rich

Voids

June 13, 2000


3-metal Reliability Results3-metal Reliability Results

Dice probed 6 times passed reliability testing.

MSL3 Pre-conditioning

0 / 180

Air-to-Air Temperature Cycling -65C to 150C

Autoclave, 121C / 100%RH / 15PSIG

500 Cycles 1000 Cycles 96 hours 144 hours0/90 0/90 0/90 1/90 (1)

(1) 1 failure is acceptable. Wafers probed at1mil tip/50µm OD/6 touches

June 13, 2000


Conclusions and RecommendationsConclusions and Recommendations

• 3-Cu-layer pad structure is more “sensitive” than 1-layer structure.– Narrower wire bond process window than Al technology’s– Slight ball shear degradation– More prone to failure mode 3s– 3-metal devices probed 6 times passed package reliability evaluations.

• Probe affects wire bonding.– Large probe marks create greater voids that:

• Reduce Au-Al intermetallic formation• Increase bond non-sticks and pad lifts• Degrade ball shear strength faster at t(1000 hours)

– Probe mark variation is large and its impact is magnified at finer pitch.– 1 mil probe tip (k=1.5gf/mil2) / 60µm OD / 6x max is recommended for 76µm pitch and

72µm pad opening.– Heavier overdrive and excessive probing is strongly discouraged.

• Better probe characterization, control and in-process inspection isrequired.

– Use wire bondability as one probe response

áProbe and assembly must work together to ensure highest system yield.

June 13, 2000


Looking out to Finer Pitch ...Looking out to Finer Pitch ...

Break-away probe technology with less force contact andminimal pad damage.

– Maintain max 60% probe/ball bond ratio– At 45µm ball bond, probe size is 20x40µm– At 35µm ball bond, probe size is 15x30µm

0%

20%

40%

60%

80%

100%

120%

140%

160%

180%

65 60 45 35Ball Size (µm)Pad Size (µm) 77 72 59 48Pitch (µm) 91 76 63 52

2X Heavy

1X HeavyUnacc

eptab

le

Break-away Break-away ProbeProbe

June 13, 2000


Thank You!Thank You!Probe

Tony AngeloKelvin HolubTom ScuderiJoan SibbittBill Williams

Packaging & Assembly

Audi ChenScott Chen

Ewa Orlowski Janusz Orlowski

Pete HarperRobert Radke

Matthew RustonTu Anh TranLois Yong

Gloria Estrada

FabGail Benjaminson

Greg BraeckelmannWayne Clark

Nedu DuraiswamiDave Farber

Thom Kobayashi Franklin Nkansah

Scott PozderTab Stephens

Product EngineeringR&QA

Jeff Bosworth Van Ho

Julie KernDave Wontor

Rosanna Yang

Die Design Substrate Design

Mark AbercrombieRodney Klabunde

Mark Peterson Fujio Takeda

Reliability & Analysis Roy Arldt

Frank ByersGary Clark

Steve HeinekeKeven Hussey Chongnan Kim

Thomas Koschmieder Andrew Mawer

Chuck MillerTricia Slovacek

Packaging &Assembly

(KLM)Mohd FaizairiFuaida Harun

K.Y. LeeK.W. MuiK.H. TanL.C. Tan

C.C. Yong

fine pitch probing and wire bonding and reliability … · june 13, 2000 fine pitch probing and...

Documents