You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/4336133

Wire bonding capillary profile and bonding process parameter optimization


simulation

Conference Paper · May 2008


DOI: 10.1109/ESIME.2008.4525066 · Source: IEEE Xplore

CITATIONS READS

9 2,123

4 authors, including:

Scott John Irving


Petrus Technology LLC
53 PUBLICATIONS   547 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermal stress simulation in the Metal-insulator-metal (MIM) wafer fabrication process View project

All content following this page was uploaded by Scott John Irving on 15 October 2015.

The user has requested enhancement of the downloaded file.


Wire Bonding Capillary Profile and Bonding Process Parameter Optimization Simulation
Qiuxiao Qian,Yong liu, Timwah Luk and Scott Irving
Fairchild Semiconductor Corp., 82 Running Hill Road, South Portland, ME 04106, USA
Email: yliu@fairchildsemi.com; Phone: (207) 761-3155; Fax: (207) 761-6339

device are investigated. Modeling studies the stress


Abstract
impact with different FAB ball diameters, different ball
diameters, different bonding temperatures, different
In this paper, a methodology for wire bonding bonding capillary profiles and different bond wire
parameter modeling is developed, which considers the material properties. A comparison of the results with and
capillary, FAB and device on silicon. The impact of without the modeling optimization will be presented.
capillary profile and bonding process parameters which
include ball diameter, bonding temperature and bond wire
material properties are studied to optimize the wire 2. Wire bonding Process and Finite Element Models
bonding assembly process. Finally, the comparison of the description
results with and without the modeling optimization shows
that the probability of bonding failure is reduced after the The conceptual 2D model is shown in Fig. 1, which is
wire bonding process is optimized. cut from a die with 3 layer metallization and 3 dielectric
(ILD) layers above the silicon. The capillary moves down
a certain height to press the FAB with a amplitude and
1. Introduction frequency. The impact on wire bonding ultrasonic
amplitude, ultrasonic frequency and friction coefficients
Wire bonding is a critical manufacturing step in the between bond pad and FAB have been throughly studied
assembly process for connection between semiconductor in [1]. This paper will study the impact of capillary
chip and external world. An improper wire bonding profile, FAB diameter, bonding temperature and FAB
process will lead to metal lift failure during wire pull tests, material properites.
which will induce fatal open circuit in a later processing,
such as reflow. A lot of wire bonding parameters, such as
capillary profile, ball diameter, bonding temperature and
bond wire material properties, can affect wire bonding
quality. The wire bonding process without parameter
optimization may induce potential risk for metal lifting Passivatio
Met 3

and silicon cratering. ILD W

The goal of this paper is to reduce metal lifting by


Met

simulation and test. In this paper the impact from the Met
ILD
S

capillary profile and bonding process parameters which


include ball diameter, bonding temperature and bond wire
(a) Bond Pad Structure (b) bonding system
material properties are studied to optimize the wire
Fig. 1: A Conceptual wire bonding system
bonding assembly process and to reduce stress on the
interface between bond pad and FAB.
A non-linear wire bonding modeling methodology is
developed. The wire bonding capillary is considered as a
rigid body due to its high hardness. The gold wire and
FAB is considered as non-linear material at the bonding
temperature. Rigid and flexible contact pair is defined at
the interface between capillary and FAB. The aluminum
metal bond pad is considered as an elastic plastic material.
Thus, the contact pair at the interface between FAB and
bond pad are non-linear contact pair. The bottom of
silicon is defined to be fixed during wire bonding process.
The bonding force is applied on the rigid capillary by a
numerical pilot point. The capillary deforms the FAB and IC angle
the bonding force is transferred to the metal pad and
Fig. 2: Standard capillary profile and inside chamfer
device on silicon. A large number of simulation with
angle ( IC angle ) dimension
different capillary profile and wire bonding parameters
(FAB diameter, bonding force, bonding temperature and
material) are carried out. Both stresses on bond pad and
Fig. 2 gives standard capillary profile and inside Table 2: Material properties ( nonlinear ) for numerical
chamfer angle (IC Angle) description. The parameter of model
IC angles selected for simulation are 70 degree, 90 degree Yield stress
and 120 degree. ( MPa )
The parameter of FAB diameters selected for
simulation are 2.8 mils, 2.9 mils and 3.0 mils separately. 200 @ 25C
Fig. 3 shows a typical free air ball (FAB) picture. Fig. 4 Al 50 @ 450C
gives actual ball height in the squashed ball figure. Au 32

The wire bonding material properties are listed in Table


1 and Table 2. Silicon, ILD and Tungsten are considered
to be linear elastic materials. FAB, bond pad and metal
layer are considered as nonlinear (bi-linear) materials.
The FAB initial yield stress 0 is 32 MPa.
FAB diameter
3. Effect of capillary profile

The results of impact of capillary inside chamfer angle


Fig. 3: A free air ball (FAB) are shown in Fig. 5 and Table 3-Table 4. These results are
obtained under a fixed FAB diameter 2.8 mil and a fixed
bonding temperature 205 C.

ball height

Sy: 332.7 Mpa Shear stress: 144.8 Mpa


Fig. 4: ball height in the deformed ball picture Capillary IC Angle: 70 degree

The yield stress function of FAB may be expressed as


the rate dependent Peirce Model[1]:

• pl
σ s = 1+ ε σ0 (1)
γ Sy: 322.8 Mpa Shear stress: 145.9 Mpa
Capillary IC Angle: 90 degree

Where m=1 and γ =561.4 (1/s).

Table 1: Material properties for numerical model


Density Young’s Poisson’s CTE
Modulus ratio
Unit Kg/mm^3 MPa --- ppm
Silicon 2.3 E-6 169000 0.23 3 Sy: 317 Mpa Shear stress: 147.5 Mpa
ILD 2.2 E-6 70000 0.25 0.55 Capillary IC Angle: 120 degree
Metal 2.71 E-6 70300 0.345 20
Tungsten 19.3 E-6 409600 0.28 4.5 Fig. 5: Stress distribution in silicon, tungsten, metal and
Gold 19.3 E-6 60000 0.44 14.2 ILD (Capillary IC Angle: 70 degree, 90 degree and 120
degree)
Fig. 5 shows that the max Y component stress
decreases from 332.7 Mpa to 317 Mpa as capillary IC
Angle increase from 70 degree to 120 degree. However,
the max shear stress increases from 144.8 Mpa to 147.5
Mpa as capillary IC angle increase from 70 degree to 120
degree.
Table 3 shows that ILD layer’s max first principle
stress decreases as the capillary IC angle increases. Sy: 408.8 Mpa Shear stress: 158.2 Mpa
However, the max third principle stress increases as the FAB diameter: 2.9 mil
capillary IC angle increases. Table 4 shows that the max
Y component stress decreases as the capillary IC angle
increases. But the max shear stress increases as the
capillary IC angle increases.

Table 3: Max stress on ILD layers


IC Angle S1 of ILD S3 of ILD
( degree ) ( Mpa ) ( Mpa )
70 63 282.3 Sy: 579.7 Mpa Shear stress: 265.5 Mpa
90 61.4 283.5 FAB diameter: 3.0 mil
120 61 305.9
Fig. 6: Stress distribution in silicon, tungsten, metal and
Table 4: Max stress of the ball-pad interface ILD (FAB diameter: 2.9 mil vs 3.0 mil, same deformed
Sy of ball-pad Shear stress of ball- ball height)
IC Angle interface pad interface
( degree ) ( Mpa ) ( Mpa ) Table 5: Max stress on ILD layers
70 312.8 88.6 FAB diameter S1 of ILD S3 of ILD
( mil ) ( Mpa ) ( Mpa )
90 307.9 89.1
120 304.2 91.5 2.8 61.4 283.5
2.9 68.1 320.2
3.0 141.6 403
4. Effect of FAB dimension

As part of this work we examined the effect of FAB


diameter. Comparison were made both with the same Table 6: Max stress of the ball-pad interface
squashed ball height and with the same bonding force. FAB diameter Sy of ball-pad Shear stress of ball-
The capillary inside chamfer angle is 90 degree and the ( mil ) interface pad interface
bonding temperature is 205 degree. ( Mpa ) ( Mpa )
2.8 307.9 89.1
(1) Impact of FAB diameter with the same squashed 2.9 371.9 100.1
ball height 3.0 390.3 124.7
Fig. 6 and Table 5-Table 6 give the stress comparison
of bond pad and below structure with the same squashed (2) Impact of FAB diameter with the same bonding
ball height. This indicates that both of the max Y force
component stress and the max shear stress increase as Fig.7 and Table 7-Table 8 give the stress comparison of
FAB diameter increases. bondpad and below structure with the same bonding force.
Table 5 shows that ILD layer’s max first principle The results show that the Y component stress decreases
stress and max third principle stress increase as FAB as the FAB diameter increases, but the max shear stress
diameter increases. Table 6 shows that the max Y has a little increment as the FAB diameter increase.
component stress and the max shear stress at the ball-pad Table 7 shows that the ILD layer’s max first principle
interface increases also as FAB diameter increases. The stress increases a little as the FAB diameter increase to
above results show that the FAB diameter has significant 2.9 mil then decreases as the FAB diameter increase to
impact on bond pad and device’s stress. 3.0 mil. The max third principle stress decreases as the
FAB diameter increases. Table 8 shows that both the max
Y component stress and the max shear stress on ball-pad
interface decreases as the FAB diameter increases. The
above results show that the FAB diameter’s impact on
bond pad and the device below is not very significant and Fig. 8 shows that both of the max Y component stress
when the bonding force is set to be the same as compared and the max shear stress decreases as bonding
to the case with the same squashed ball height. temperature increase.

Sy: 320.1 Mpa Shear stress: 146.7 Mpa Sy: 324 Mpa Shear stress: 150.5 Mpa
FAB diameter: 2.9 mil Bonding temperature: 190 C

Sy: 313 Mpa Shear stress: 148.1 Mpa Sy: 321.4 Mpa Shear stress: 143.8 Mpa
FAB diameter: 3.0 mil Bonding temperature: 220 C

Fig. 7: Stress distribution in silicon, tungsten, metal and Fig. 8: Stress distribution in silicon, tungsten, metal and
ILD (FAB diameter: 2.9 mil vs 3.0 mil, same bonding ILD at different bonding temperature.
force )
Table 9 shows that ILD layer’s max first principle
Table 7: Max stress on ILD layers stress increases as bonding temperature increases. While
FAB diameter S1 of ILD S3 of ILD the max third principle stress does not change much when
( mil ) ( Mpa ) ( Mpa ) bonding temperature changes. This may be induced by the
2.8 61.4 283.5 metal layer’s yield stress decreasing as the bonding
temperature increases. A higher temperature allows a
2.9 62.7 276.8
larger metal deformation which further induces a higher
3.0 58.3 274.7 first principle stress (tensive/bend) in the ILD layer. The
bonding force applied is the same for all three
. temperatures in table 9, thus, the third principle stress
Table 8: Max stress of the ball-pad interface (compressive) of ILD is not significantly affected by
FAB diameter Sy of ball-pad Shear stress of ball- bonding temperature.
( mil ) interface pad interface
( Mpa ) ( Mpa ) Table 9: Max stress on ILD layers at bonding
2.8 307.9 89.1 temperature
2.9 298.2 86.9 Bonding S1 of ILD S3 of ILD
temperature ( Mpa ) ( Mpa )
3.0 295.7 81
( degree )
190 57.2 277
5. Effect of bonding temperature
205 61.4 283.5
220 68 276.3
Here we discuss bonging substrate temperature. Both
the stress at bonding temperature and stress at room
temperature are compared. The bonding force is set to the
same value in both cases. The capillary IC Angle is set to Table 10 shows stress comparison of ball-pad interface.
be 90 degrees. The max Y component stress has a little increment when
the bonding temperature increases to 205C but then
(1) Stress at bonding temperature decreases when bonding temperature increases to 220 C.
The max shear stress does not have much change when
Fig. 8 shows the Y component stress and the shear
bonding temperature increases.
stress comparison at different bonding temperature. Fig. 5
Table 10: Max stress of the ball-pad interface at and the max shear stress at the ball-pad interface decrease
bonding temperature as temperature decrease to room temperature.
Bonding Sy of ball-pad Shear stress of ball-
temperature interface pad interface Table 11: Max stress on ILD layers at 25C
( degree ) ( Mpa ) ( Mpa ) Bonding S1 of ILD S3 of ILD
190 303.7 88.9 temperature ( Mpa ) ( Mpa )
205 307.9 89.1 ( degree )
220 287.6 89.3 190 225.6 146.9
205 224.1 156.5
220 221.5 166.6
(2) stress at room temperature

Table 12: Max stress of the ball-pad interface at 25C


Bonding Sy of ball-pad Shear stress of ball-
temperature interface pad interface
( degree ) ( Mpa ) ( Mpa )
190 109.3 57.1
205 91.9 51.9
220 82.1 48.1
Sy: 87.1 Mpa Shear stress: 110 Mpa
Temperature: 190 C 25 C
6. Effect of FAB material properties

The FAB’s material properties is now defined as bi-


linear plastic material with an initial yield stress of
32MPa and three different tangent modulus. The results
of impact of the FAB’s tangent modulus are shown in Fig.
10 and Table 13-Table 14 which are obtained with a FAB
diameter of 2.8 mil and a bonding temperature of 205 C.

Sy: 89.2 Mpa Shear stress: 110.8 Mpa


Temperature: 205 C 25 C

Sy: 271 Mpa Shear stress: 138.4 Mpa


Tangent modulus: 250 Mpa

Sy: 91.1 Mpa Shear stress: 122.6 Mpa


Temperature: 220 C 25 C

Fig. 9: Stress distribution in silicon, tungsten, metal and


ILD at room temperature.

Fig. 9 shows the max Y component stress and the max


shear stress decreases as temperature decrease to room Sy: 381.3 Mpa Shear stress: 160.6 Mpa
temperature. The max stress is at the interface between Tangent modulus: 400 Mpa
ILD layer and metal layer. This is induced by the CTE
mismatch between ILD and metal. Fig. 10: Stress distribution in silicon, tungsten, metal and
Table 11 shows that the max first principle stress on ILD ( impact on ball hardness )
ILD layer significantly increases when cooled to room
temperature. However, the max third principle stress
Fig. 10 shows the max Y component stress comparison
decreases as temperature goes down to room temperature.
and the max shear stress comparison on bond pad and
Table 12 shows that both of the max Y component stress
device. Table 13 gives the max first principle stress and
the max third principle stress comparison on ILD layer.
Table 14 gives the max Y component stress comparison
and the max shear stress comparison at ball-pad interface.
All above results show that all stresses increase as the
FAB’s tangent modulus increases. The FAB’s tangent
modulus has significant impact to the stress on bond pad
and device.

Table 13: Max stress on ILD layers


Tangent S1 of ILD S3 of ILD
modulus ( Mpa ) ( Mpa )
( Mpa )
250 31.8 219.4
320 61.4 283.5
400 85.3 334.2

Table 14: Max stress of the ball-pad interface Fig. 11: Die with 12 bonded wires.
Tangent Sy of ball-pad Shear stress of ball-
Modulus interface pad interface
( Mpa ) ( Mpa ) ( Mpa )
250 254.9 72.3
320 307.9 89.1
400 359 104.6

7. Experimental data

Wire pull test results are shown in Table 15 and Fig.


11- Fig. 12.
Table 15 gives comparison of bonding force and the Pad 4 Pad 3
FAB’s tangent modulus. The parameter of bonding force
failure unit good unit
selected for test are 0.2 N and 0.4 N. Two types of wire
are selected for comparison. Wire A’s tangent modulus is
higher than Wire B’s tangent modulus. A DOE Fig. 12: Failing unit and good unit after wire pull test.
experiment was made for these two parameters. There is
no failing unit in the group with low bonding force and 8. Conclusion
low wire’s tangent modulus. The other 3 groups have 1
failing unit each. Table 15 also shows that a lower FAB Wire bonding is a complicated multi-physics problem.
tangent modulus results in a lower average pull force. This paper develops a parameter modeling methodology
to study wire bonding assembly process. Through the
Table 15: wire pull test results modeling we understand:
Bonding Wire Test failed Average
force (N) units units pull force 1. Increasing the FAB diameter and keeping the
(N) same ball height will require increasing the
0.2 A 24 1 0.1912 bonding force. All of the stress in bond pad and
0.4 A 24 1 0.1893 device will increase at this case. Increasing the
0.2 B 24 0 0.1733 FAB diameter and keeping the same bonding
0.4 B 24 1 0.1727 force may decrease the Y component stress but
increase shear stress in bond pad and device.
Fig. 11 shows the device with 12 bonded wires. Fig. 12 2. Simulations show that the capillary IC Angle and
shows pad 3 and pad 4 of the device after wire pull test. temperature does not have a very significant
The wire bonded to pad 3 is a good unit which shows impact on bond pad and the device’s stress.
wire break at neck of the ball to 5 mils above the ball. The 3. There is a stress trade off with different bonding
wire bonded to pad 4 is a failing unit which shows metal temperature. When the bonding temperature
lifting. increases, the ILD first principal stress (tensile
stress) will increase while its pull stress Sy at ball
and bondpad will decrease. When cooling down
to room temperature, the ILD first principal stress
will significantly incease.
4. Both modeling and experimental data shows a
significant improvement in the wire pull test by
reducing FAB’s tangent modulus. This means
harder FAB will induce more failures of bond
wire lift.

Acknowledgements

The authors wish to thank the support from Automation


Development department in Maine and assembly Penang,
Fairchild Semiconductor Corp.

References
1. Yong Liu, Scott Irving and Timwah Luk,
“Thermosonic Wire Bonding Process Simulation and
Bond Pad Over Active Stress Analysis”, Electronic
Components and Technology Conf , June 1-4, 2004.
pp 383-391
2. Joseph Alison King, “Material Handbook for Hybrid
Microelectronics”
3. Yong Liu. Don Desbiens, Scott Irving and Timwah
Luk, “Probe Test Failure Analysis of Bond Pad Over
Active Structure by Modeling and Experiment”, Proc
55st Electronic Components and Technology Conf, ,
May31 - June 3, 2005. pp 861-866

View publication stats

You might also like