Passivation Stress versus Top Metal Profiles by 3D Finite Element Modeling

Xiaokang Huang, Liping Zhu*, Bang Nguyen, Van Tran, Harold Isom

TriQuint Semiconductor, Inc., 500 W Renner Road, Richardson, TX 75080

* TriQuint Semiconductor, Inc., 2300 NE Brookwood Parkway, Hillsboro, OR 97124

Keywords: Passivation Crack, Metal Profile, Finite Element Modeling, Finite Element Analysis

Abstract Figure 2 (a) shows the basic interconnect structure which

Passivation hermiticity is always a concern in contains two metals and two interconnect layers with SiN as
integration circuit fabrication and application. the passivation. Both the top metal (M2) and the underlying
Passivation stress is one of the key factors for the metal (M1) are plated Au connected through vias (VIA1 and
passivation integrity. In this paper, passivation stress is VIA2). The inter-layer dielectric films are
investigated as a function of top metal profiles with a 3- bisbenzocyclobutene (BCB). For simplicity, SiN between
Dimentional Finite Element Modeling (3D FEM). A new BCBs and TiW under each Au are not shown in Figure 2.
structure is proposed for reducing the passivation stress The dimensions of metals and dielectric layers are not scaled
at the corner of a top metal foot to prevent a potential here.
passivation crack. A sloped metal with a curved foot
profile is demonstrated. The pros and corns of the new
structure are discussed at the end.

INTRODUCTION

Passivation hermicity is always a concern for

semiconductor IC in both Si and compound semiconductor
devices. Cracks are one of the passivation defects which is
specifically important for compound semiconductors due to
the materials used for inter metal dielectric layers such as
bisbenzocyclobutene (BCB) in compound semiconductors.
Figure 1 shows an example of a passivation crack at the
corner of the top metal foot. Many investigations have been
conducted by FEM simulation [1], [2] or experiments [3],
[4]. In this paper, the passivation stress is investigated as a
function of the top metal profile and passivation thickness FIGURE 2 DIFFERENT TOP METAL PROFILES PROPOSED FOR SIN STRESS
with a 3-Dimentional Finite Element Modeling (3D FEM). REDUCTION AND PASSIVATION CRACK ELIMINATION. (A) RIGHT ANGLE; (B)
Based on the results of the simulation, a sloped top metal SLOPED ANGLE; (C) RIGHT ANGLE WITH CURVING FOOT; (D) SLOPED
with a curving foot profile is designed and fabricated. The ANGLE WITH CURVING FOOT
pros and corns of the new structure are discussed.
The two red circles are the locations where the stress is
the highest which induces the SiN passivation cracks. There
are different ways to eliminate the cracks, such as
optimization of SiN thickness and stress. However, in
certain cases, there are limitations of what SiN thickness and
stress can be changed. Therefore seeking a different way for
resolving the SiN crack issue is very important.

Since the crack is right at the corner of M2 due to the 10

straight angle of the profile, another way of reducing the b
stress at the corner is to modify the corner profile by
smoothing the straight angle. Figure 2 (b) shows a M2 with
a slope angle. Figure 2 (c) is a M2 with a curving foot at the
FIGURE 1 PASSIVATION CRACK AT THE CORNER OF THE TOP METAL FOOT corner. Figure 2 (d) is a combination of both a slope profile
and a curving foot at the corner.

CS MANTECH Conference, May 13th - 16th, 2013, New Orleans, Louisiana, USA 323
 In the next few sections, we used 3D FEM for the
evaluation of the impact of the three different M2 profiles to TABLE I
MATERIAL PROPERTIES USED IN THE SIMULATION (WWW.METWEB.COM)
the SiN passivation cracks. Since there is a correlation of
SiN crack and thickness established with experiments [4],
we also simulated the correlation of SiN stress and thickness
for both purposes of verifying the model and parameters
used in the simulation and determining the SiN crack
threshold for M2 profile evaluation.

SIMULATION MODEL

The software used in the simulation is Ansys. The

Finite Element Analysis (FEA) model was used to conduct SIMULATION RESULTS
sensitivity analysis of design form factor effects, such as
metal thickness, metal side wall angle and corner filet on First, we simulated the SiN stress at the M2 corner with a
PSN stress which is a metric used to assess risk of PSN right angle slope as a function of SiN thickness shown in
crack and provide solutions for optimal process. A slice of Figure 4. The highest SiN stress is at the corner of M2,
FEA model was constructed by using hex brick elements and where the crack observed in our experiment. The highest
constrained in normal surfaces of front and back, bottom, stress in SiN passivation decreases with the thickness of
and two sides with the top free. The selection of proper mesh SiN. This is consistent with the experimental observations
size is very important for achieving valid simulation results. in Ref. [4]. At a certain point where the maximum stress is
By comparing the simulation results among different mesh lower than the strength of SiN, the crack will not happen.
sizes, one fourth (1/4) of standard SiN thickness was used By comparing with the results in Ref. [4], the stress
for the mesh size for the most of the structure in the threshold of the SiN crack shown in Figure 4 (d) is
simulation as shown in Figure 3 (a). At the area near the determined as a reference for M2 profile evaluation in this
surface of SiN passivation, a fine mesh is used as shown in paper.
Figure 3 (b).

FIGURE 3 MESH STRUCTURES (A) IN A STRAIGHT ANGLE METAL PROFILE

WITH NORMAL MESH AND (B) IN A CURVING FOOT METAL PROFILE WITH FINE
MESH NEAR SIN SURFACE

The thermal load from 320C to 23C was imposed to

simulate deposition and cool down process. Material
properties used in the simulation are given in the Table 1.
is the material density. is the Poissons ratio. E is the
elastic modulus. CTE is the coefficient of temperature
expansion. The data are from www.MetWeb.com. The
FIGURE 4 SIN HIGHEST STRESS AS A FUNCTION OF SIN THICKNESS
strength of SiN has a value between 800 to 2100MPa, which
depends on the film deposition.
Second, we investigated the SiN highest stress as a
function of different sloped angles with a fixed SiN
According to the First Principal Stress Theory, we
thickness. Figure 5 shows the impact of sloped angle
compared the maximum stress value in the SiN film to the
profiles to the SiN stress at the M2 corner. The smaller the
strength of SiN. Once the maximum stress value was larger
M2 sloped angle is, the lower the SiN stress is at the M2
than the strength of SiN, then cracks would occur. The
corner. However, by comparing the highest stress at 85 to
strength of SiN or the crack threshold in our case is
the stress threshold of crack occurring, the highest SiN stress
determined by comparing the simulation results to the
at 85 is still higher than the crack threshold. This means
experiment data [4].

324 CS MANTECH Conference, May 13th - 16th, 2013, New Orleans, Louisiana, USA
that even with 85 sloped M2 profile, SiN will still crack at from the right angle profile. However, by comparing to the
the corner. There are two reasons for choosing 85 here. crack threshold, these two profiles are still not good enough
One is that a smaller angle gives a larger difference between to fully eliminate the cracks. The only one that has the
the sizes of the top and bottom of M2, which can cause maximum stress lower than the critical crack threshold is the
enclosure concerns in both layout design and wafer process. combination of both the sloped angle and the curving foot
The other one is the limitation of what we can achieve profiles.
experimentally.

FIGURE 7 STRESS COMPARISON OF 4 DIFFERENT TOP METAL PROFILES

FIGURE 5 SIN HIGHEST STRESS AS A FUNCTION OF M2 SLOPED ANGLE RELATIVE TO CRACK THRESHOLD

Finally, we simulated four different M2 profiles as EXPERIMENT AND DISCUSSION

shown in Figure 6. These four profiles are (a) right angle,
(b) sloped angle, (c) right angle with curving foot, and (d) Experimentally, we fabricated a plated metal with a
sloped angle with curving foot. As expected, the highest sloped angle profile and curving foot shown in Figure 8.
stresses for different cases are all at the corner of the M2 The slope of the metal profile comes from the optimization
foot corner. of the focus depth, exposure, and development during the
metal lithography. The curving foot is realized by
controlling the plating process.

FIGURE 8 PLATED METAL WITH A SLOPED PROFILE AND CURVING FOOT

FIGURE 6 FOUR DIFFERENT TOP METAL PROFILES SIMULATED WITH 3D-
FEM. (A) RIGHT ANGLE; (B) SLOPED ANGLE; (C) RIGHT ANGLE WITH Clearly, the sloped angle M2 with curving foot changes
CURVING FOOT; (D) SLOPED ANGLE WITH CURVING FOOT the original dimension of the metal. The bottom size of the
metal becomes larger than the top. Therefore, there should
10
The summary and the comparison of the passivation be no concern for the enclosure of M2 to the VIA below
stress of these four different profiles are shown in Figure 7.
b
practically. Since the M2 is the top metal in the case here
The slope angle is 85 and the radius of the curving foot is and normally the density of the top metal is low comparing
0.3m. For easy comparison, the maximum stress values of to metals underneath, the increasing of the bottom size of the
SiN film for the different M2 profiles are normalized to the metal should not impact the overall layout and die size. Pad
value of the basic structure, the right angle profile, which enclosure to the top metal should be considered in the real
gives the highest stress within the four different profiles we case. Depending on layout design rules, it could be an issue
interested. The maximum stresses of both the right angle and the pad size may have to be modified accordingly.
with curving foot and the sloped angle profiles are improved

CS MANTECH Conference, May 13th - 16th, 2013, New Orleans, Louisiana, USA 325
 In fact, there should be still a limitation of how small of
the slope angle can go and how large the radius of the
curving foot can use. The limitation can come from either
the fabrication capability or the layout design rules. That is
why a slope of 85 and radius of 0.3m are mostly used in
our simulation. The impact of this profile change to the RF
performance requires electrical simulation, which is beyond
the discussion here, but it should be verified.

CONCLUSIONS

Based on our 3D FEM simulation, neither a sloped

profile nor a curving foot profile of the top metal is enough
to eliminate the passivation cracks by itself. Only the
combination of both sloped and curving foot profiles of the
top metal is promising as a solution. Experimentally, we
fabricated a sloped metal profile with a curving foot on a
monitor wafer. Further effort is needed to transfer it to the
product chip, which is still a challenge.

ACKNOWLEDGEMENTS

The authors would like to thank Chi-hing Choi, Peter

Moon, Jinhong Yang, Jeremy Middleton, Duofeng Yue, and
Eric McCormick for valuable discussions. Special thanks to
Gaurav Gupta and David Gonzalez for passivation crack
detection.

REFERENCES

[1] J. Ackaert, On the Impact of the Edge Profile of the Interconnects on

the Occurrence of Passivation Cracks of Plastic Encapsulated electronic
Power Devices, IC Design & Technology (ICICDT), 2011 IEEE
International Conference, 2-4 May 2011

[2] L. Zhu, et al, Stress Analysis for PSN under Wafer Process, TriQuint
Internal Report, Sept. 2009

[3] W. J. Roesch, et al., Assessing Circuit Hermeticity By Electrolysis,

2000 GaAs MANTECH, 2000

[4] C. Choi, et al, Experiment Investigation of PSN stress and Thickness

for Passivation Crack, TriQuint Internal Report, 2010

ACRONYMS

FEA: Finite Element Analysis

FEM: Finite Element Modeling
BCB: bisbenzocyclobutene
TiW: Titanium Tungsten
SiN: Silicon Nitride

326 CS MANTECH Conference, May 13th - 16th, 2013, New Orleans, Louisiana, USA