368 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 11, NO. 3, SEPTEMBER 2011
Failure Probability Estimation of Anisotropic
Conductive Film Packages With Asymmetric
Upper-to-Lower Pad Size and Misalignment Offsets
Chao-Ming Lin, Tzu-Chao Lin, Te-Hua Fang, and Yen-Chun Liu
Abstract—Anisotropic conductive films (ACFs) are widely used
in the packaging of flat panel displays and liquid crystal displays and for attaching bare chips to both flexible and rigid substrates. This paper utilizes the V-shaped curve method to analyze the fail-ure probability of ACF packages with an asymmetric upper/lower pad size and misalignment offsets. In the proposed method, the probability of opening failures is modeled using a Poisson func-tion, modified to take into account the effects of the pad-width difference and misalignment offset on the effective conductive area between opposing pads. Meanwhile, the probability of bridging failures is evaluated using an enhanced bridging model based on the distance between the neighboring pad pairs in the array. The failure probability of the pad array is evaluated as a function of both the difference in width of the upper and lower pads and the degree of misalignment between the opposing pads in the array. The results show that the V-shaped curve method provides the means to predict the ACF volume fraction which minimizes the failure probability of the ACF assembly given a knowledge of the pad-width difference and the misalignment offset. In addition, it is shown that when the misalignment offset is greater than the pad-width difference, the minimum failure probability reduces as the pad-width difference increases due to the correspond-ing increase in the effective conductive area between opposcorrespond-ing pads. Conversely, when the misalignment offset is less than the pad-width difference, the minimum failure probability increases with an increasing pad-width difference due to the corresponding reduction in the effective conductive area between the pads.
Index Terms—Anisotropic Conductive Film (ACF), asymmetric
pad size, failure probability, misalignment, V-shaped curve. I. INTRODUCTION
A
NISOTROPIC CONDUCTIVE FILMS (ACFs), consist-ing of an insulatconsist-ing adhesive polymer matrix with a fine dispersion of conductive metallic particles or metal-coated polymer balls, are used in a wide variety of fine-pitch in-terconnection methods, e.g., outer lead bonding, flex to PCB Manuscript received September 26, 2010; revised January 27, 2011, March 1, 2011, and March 2, 2011; accepted March 2, 2011. Date of publication March 14, 2011; date of current version September 2, 2011. This work was supported by the National Science Council, R.O.C., under Grants NSC 98-2221-E-274-005- and 99-2221-E-274-003-.C.-M. Lin and Y.-C. Liu are with the Graduate School of Opto-Mechatronics and Materials, WuFeng University, Chia-Yi 621, Taiwan (e-mail: cmlin@ wfu.edu.tw; chaoming.lin@gmail.com; blliu@wfu.edu.tw).
T.-C. Lin is with the Department of Computer Science and Information Engi-neering, WuFeng University, Chia-Yi 621, Taiwan (e-mail: tclin@wfu.edu.tw). T.-H. Fang is with the Department of Mechanical Engineering, National Kaohsiung, University of Applied Sciences, Kaohsiung 807, Taiwan (e-mail: fang.tehua@msa.hinet.net).
Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TDMR.2011.2126576
Fig. 1. Schematic illustration showing the use of ACF in COG and COF packaging.
bonding (PCB), chip-on-glass (COG), and chip-on-film (COF) (see Fig. 1). ACFs have a number of key advantages for packaging applications, including the following: 1) fine pitch; 2) low processing temperature; 3) low cost; 4) good flexibility; 5) the avoidance of lead or other toxic metals; and 6) good compatibility with a wide range of surfaces. As a result, ACF-based packaging techniques are an ideal alternative to tradi-tional soldering methods for the packaging of flat panel displays (FPDs) and liquid crystal displays (LCDs) [1]–[4].
LCDs have many applications nowadays, including note-books, computer monitors, TVs, 3C products, and so on. The ACF used in the packaging of these devices plays a critical role in ensuring their electrical reliability by providing unidirec-tional conductivity in the vertical, i.e., z-axis direction. Fig. 2 presents a schematic illustration of a typical ACF processing. In the IC/substrate assembly failure, the bridging probability (i.e., an undesirable conductivity in the horizontal direction) increases as the volume fraction increases. Conversely, the opening probability (i.e., a loss of electrical conductivity in the vertical direction) increases as the volume fraction de-creases. Hence, in specifying an appropriate volume fraction, a compromise must be obtained which minimizes the risk of bridging failure while simultaneously ensuring full electrical conductivity between the IC and the substrate.
In designing an ACF assembly, the volume fraction of the conductive particles within the ACF resin is assigned a low value of approximately 5%–10%. In practice, the volume frac-tion is specified in such a way as to provide conductivity in the vertical direction and electrical insulation in the pitch direction. An inappropriate specification of the package geometry or ACF composition may result in opening failures in the vertical direction or bridging failures in the pitch direction. In recent studies, the current group presented a method designated as the 1530-4388/$26.00 © 2011 IEEE
LIN et al.: FAILURE PROBABILITY ESTIMATION OF ACF PACKAGES 369
Fig. 2. Schematic illustration of typical ACF processing.
Fig. 3. Illustrative V-shaped curve for the estimation of ACF failure probability.
V-shaped curve method for predicting the failure probability of ACF-packaged IC/substrate assemblies [5]–[8].
As shown in Fig. 3, the V-shaped curve plots the failure probability against the volume fraction of conductive particles in the ACF for a given set of geometry parameters. To establish the V-shaped curve, the probability of opening failures is calcu-lated using a Poisson function [9], [10], while that of bridging failures is calculated using a box model [11]. The overall failure probability of the package is then estimated by combining the opening and bridging probabilities using a probability theory. In the resulting V-shaped curve, the portion to the left of the min-imum point corresponds to the opening probability, while that to the right of the minimum point corresponds to the bridging
Fig. 4. Schematic illustrations of (a) symmetric and (b) asymmetric upper/ lower pads in ACF assembly.
probability. Furthermore, the volume fraction corresponding to the minimum point of the V-shaped curve is the optimal volume fraction for the package.
The current group proposed four different failure prediction models for fine-pitch connections using ACF [10]. Comparing the four methods, it was found that the model utilizing a Poisson function to evaluate the probability of fewer than six particles existing per pad yielded the most accurate predictions of the opening failure probability [12], [13]. In a recent study, the current group proposed an enhanced box model for predicting the bridging failure probability in ACF packages based upon a consideration of all the possible conductive and linear paths between adjacent pad pairs [11]. The results showed that the enhanced bridging model yielded more accurate estimates of the minimum failure probability and optimal volume fraction than existing bridging models. In another recent study, the current group examined the effect of the upper-to-lower pad-height ratio on the failure probability of ACF packages and showed that the failure probability can be reduced by reducing the total height of the two pad arrays or by utilizing pad arrays with an equal height [14]. It should be mentioned that the whole failure models in this paper are based on the geometry effects, and the resistivity of the particles is not considered.
For analytical convenience, the pads on the IC and substrate of an ACF package are generally assumed to be of the same size [see Fig. 4(a)]. However, in practice, the two sets of pads often have different dimensions [see Fig. 4(b)]. Thus, in analyzing the true failure probability of ACF assemblies, it is necessary to consider an asymmetric IC/substrate pad size. In [15], the current group examined the effects of alignment errors on the reliability of ACF packages with a symmetric pad size. The results showed that a misalignment of the upper and lower pads affects both the overall failure probability of the package and the optimal volume fraction of conductive particles. In this paper, the V-shaped curve method is used to analyze the failure probability of ACF packages with both an asymmetric IC/substrate pad size and misalignment offsets.
II. FAILUREPROBABILITYANALYSIS OFACF PACKAGES WITHMISALIGNMENTOFFSETS
A. Opening Failure Analysis Using Poisson Function
As described in the previous section, ACF assemblies are liable to two different types of failure, namely, opening failures
370 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 11, NO. 3, SEPTEMBER 2011
and bridging failures. An opening event occurs when the gap between opposing pads in the IC/substrate assembly contains an insufficient number of conductive particles to form an electrical contact between them (see Figs. 2 and 4). In a previous study by the current group, it was shown that a density of more than five particles per pad is required to ensure a reliable electrical contact between the IC and the substrate. In addition, it was shown that the probability of there being n particles between the pads could be described by a Poisson distribution [10]. Hence, for a given assembly geometry, an effective contact between the upper and lower pad arrays can be ensured by specifying the volume fraction of conductive particles in the ACF compound such that the following condition is satisfied:
PO(n≤ 5, μo(f )) = 5 n=0 μnoe−μo n! = 5 n=0 6l∗2f π n e−6l∗ 2 f π n! (1) where Pois the opening failure probability between opposing
pads, μois the average number of particles between opposing
pads, f is the volume fraction of particles within the ACF compound, l is the side length of the IC/substrate pads, r is the radius of the conductive particles, and l∗is a dimensionless integer defined as l∗= [l/2r] (where [ ] is a Gaussian operator).
B. Opening Failure Analysis Subject to Misalignment Effects
To analyze the effects of misalignment offsets in the x- and
y-directions on the opening failure probability, (1) should be
reformulated as follows [15]: P(Ex,Ey) O (n≤ 5, μo(f )) = 5 n=0 μn oe−μo n! = 5 n=0 6(1−Ex)(1−Ey)·l∗2·f π n e−6(1−Ex)(1−Ey)·l∗2·fπ n! (2) where P(Ex,Ey)
O is the opening failure probability between
opposing pads given a misalignment of the pads in the
x- and y-directions; Δx and Δy are the misalignment offsets
in the x- and y-directions, respectively; Δx∗ and Δy∗ are the dimensionless offsets in the x- and y-directions, respectively; and Ex(= (Δx/l)· 100% = (Δx∗/l∗)· 100%) and Ey(=
(Δy/l)· 100% = (Δy∗/l∗)· 100%) are the offset percentages in the x- and y-directions, respectively.
C. Bridging Analysis Using Box Model
A bridging failure occurs when a continuous chain of con-ductive particles is formed between neighboring pads (see Figs. 2 and 4). Mannan et al. [9] proposed a simple box model for estimating the bridging probability between neighboring pads. Lin et al. [6] modified this model to take into account the possibility of bridging in multiple directions. The bridging model was further enhanced by Lin et al. [11]. According
to this enhanced model, the bridging failure probability is given by PB(f ) = 1− i1,i2:1→l∗ j1,j2:1→h∗ 1− μ √ (i1−i2)2+(j1−j2)2+d∗2 b (3)
where i1, j1, i2, and j2 are coordinate indices referencing the
individual cells on the side walls of the pads; μb = 6f /π is
de-scribed in the box model; h∗is a dimensionless integer defined as h∗= [h/2r]; and d∗ is a dimensionless integer defined as
d∗= [d/2r].
D. Bridging Analysis Subject to Misalignment Effects
To evaluate the effects of misalignment on the probability of bridging failures in the x-direction, (3) should be refor-mulated as PB,x(f, l∗, h∗, d∗, Ex) = 1− i1,i2:1→l∗ j1,j2:1→h∗ 1−μ √ (i1−i2)2+(j1−j2)2+(d∗±l∗·Ex)2 b (4)
where PB,xis the bridging failure probability in the x-direction.
Similarly, to evaluate the effects of misalignment on the probability of bridging failures in the y-direction, (3) should be reformulated as PB,y(f, l∗, h∗, d∗, Ey) = 1− i1,i2:1→l∗ j1,j2:1→h∗ 1− μ √ (i1−i2)2+(j1−j2)2+(d∗±l∗·Ey)2 b (5)
where PB,yis the bridging failure probability in the y-direction.
E. Analysis of Local Failure Probability
Since an ACF package may fail as a result of either open-ing or bridgopen-ing, the local ACF failure probability, i.e., the probability of failure at a single pad pair within the array, is given by
PO∪B(μo(f ), μb(f )) = PO(μo(f )) + PB(μb(f ))
− PO∩B(μo(f ), μb(f )) (6)
where PO∪B(f ) is the probability of an opening event or a bridging event and PO∩B(f ) is the probability of both an
opening event and a bridging event. Assuming that the opening and bridging events are statistically independent, the overall probability of a local failure of the ACF package can be expressed as
PO∩B(μo(f ), μb(f )) = PO(μo(f ))· PB(μb(f )) . (7)
Substituting (7) into (6) and defining the failure probability as PO∪B(f ), the local failure probability of the ACF assembly
can be formulated as
PO∪B(μo(f ), μb(f )) = PO(μo(f )) + PB(μb(f ))
376 IEEE TRANSACTIONS ON DEVICE AND MATERIALS RELIABILITY, VOL. 11, NO. 3, SEPTEMBER 2011
Fig. 17. Variation of minimum failure probability with Dp as function of (Ex, Ey).
misalignment error is still kept in the Dplimitation, so the tip
failure probabilities of the V-shaped curves are increasing with the Dp increasing, and the major factor is the bridging paths
(see Paths I–III in Fig. 7) shortened as the Dp is increasing
under the condition Ex= Ey< Dpeven though the effective
conductive area is unchanged.
V. CONCLUSION
This paper has used the V-shaped curve method to analyze the failure probability of ACF packages in which the upper and lower pads have a different side length and are subject to alignment errors. In formulating the V-shaped curve, the opening probability has been modeled using a Poisson function, suitably modified to take into account the effects of the pad-width difference and the misalignment offset on the effective conductive area between opposing pads. Meanwhile, the bridg-ing probability has been modeled usbridg-ing an enhanced bridgbridg-ing model, modified to take into account the effects of asymmetry and misalignment on the lengths of the bridging paths between neighboring pad pairs. The major findings of this paper can be summarized as follows.
1) The V-shaped curve method provides the means to ana-lyze the failure probability of asymmetric pad-width ACF packages with misalignment effects.
2) The V-shaped curve method provides the means to es-timate the optimal ACF volume fraction for any given value of the pad-width difference or misalignment offset provided that the geometry parameters of the ACF assem-bly are known.
3) When the misalignment offset is greater than the pad-width difference, the failure probability can be reduced by increasing the asymmetry of the upper and lower pads. Conversely, when the pad-width difference is greater than the misalignment offset, the failure probability can be reduced by reducing the asymmetry of the upper and lower pads.
REFERENCES
[1] J. Lau, Low Cost Flip Chip Technologies. New York: McGraw-Hill, 1996, pp. 1–17.
[2] J. Liu, Conductive Adhesives for Electrics Packaging. Isle of Man, U.K.: Electrochemical Publications Ltd., Jun. 1999.
[3] H. Date, Y. Hozumi, H. Tokuhira, M. Usui, and E. Horikoshi, “Anisotropic conductive adhesive for fine pitch interconnections,” in Proc. Int. Symp.
Microelectron., 1994, pp. 570–575.
[4] A. Torri, M. Takizawa, and K. Sasahara, “Development of flip chip bonding technology using anisotropic conductive film,” in Proc. 9th Int.
Microelectron. Conf., 1996, pp. 324–327.
[5] C. M. Lin, M. H. Su, and W. J. Chang, “The prediction of failure prob-ability in anisotropic conductive adhesive (ACA),” IEEE Trans. Device
Mater. Rel., vol. 5, no. 2, pp. 255–261, Jun. 2005.
[6] C. M. Lin, W. J. Chang, and T. H. Fang, “Analysis of new anisotropic conductive film (ACF),” IEEE Trans. Device Mater. Rel., vol. 5, no. 4, pp. 694–700, Dec. 2005.
[7] C. M. Lin, Y. L. Chen, and H. M. Chu, “Modeling of real-time reliabil-ity predicting system on anisotropic conductive film (ACF) processing,”
Mater. Sci. Eng. A, vol. 432, no. 1/2, pp. 239–244, Sep. 2006.
[8] C. M. Lin, “Effects of systematic and stochastic errors on estimated failure probability of anisotropic conductive film,” IEEE Trans. Device Mater.
Rel., vol. 7, no. 3, pp. 387–398, Sep. 2007.
[9] S. H. Mannan, D. J. Williams, D. C. Whalley, and A. O. Ogunjimi, “Con-ductive adhesives for electronics packaging,” in Models to Determine
Guidelines for the Anisotropic Conducting Adhesives Joining Process,
J. Liu, Ed. Isle of Man, U.K.: Electrochemical Publications Ltd, Jun. 1999, ch. 4.
[10] C. M. Lin, W. J. Chang, and T. H. Fang, “Reliability analysis of the fine pitch connection using anisotropic conductive film (ACF),”
Micro-electron. J., vol. 37, no. 7, pp. 565–568, Jul. 2006.
[11] C. M. Lin, C. M. Tan, Y. C. Liu, and C. Y. Chu, “An enhanced bridging model for evaluating the failure probability of anisotropic conductive film packages,” Optoelectron. Adv. Mater.—Rapid Commun., vol. 3, no. 5, pp. 493–500, 2009.
[12] H. Ando, Y. Yamada, and Y. Suga, “Anisotropic conductive film,”
Electron. Mater., pp. 72–76, Jan. 1995.
[13] Y. Yamada, “ACF for fine pitch connection,” in Proc. Fine Process
Technol., Tokyo, Japan, 1995, p. R12.
[14] C. M. Lin, T. C. Lin, T. H. Fang, and K. S. Chao, “Failure analysis of pad-height effects in the fine-pitch interconnection of the anisotropic conductive films,” Microelectron. Reliab., vol. 48, no. 7, pp. 1087–1092, Jul. 2008.
[15] C. M. Lin, C. M. Tan, T. H. Tasi, and Y. C. Liu, “Failure analysis of anisotropic conductive film packages with misalignment offsets,” IEEE
Trans. Device Mater. Rel., vol. 10, no. 1, pp. 9–17, Mar. 2010.
[16] S. Ghahramani, Fundamentals of Probability., 2nd ed. Englewood Cliffs, NJ: Prentice-Hall, 2000.
Chao-Ming Lin received the M.S. and Ph.D. degrees
in mechanical engineering from the National Cheng Kung University, Tainan, Taiwan, in 1993 and 1999, respectively.
He is currently with the Graduate School of Opto-Mechatronics and Materials, WuFeng University, Chia-Yi, Taiwan, as a Professor with research inter-ests in IC packaging, injection molding packaging, electrically conductive adhesive/films, MEMS and polymer packaging composites, etc. He has some publications in the areas of electronic packaging, microfluidics, and polymer processing.
Tzu-Chao Lin received the M.S. degree from
the Department of Applied Mathematics, National Chung Hsing University, Taichung, Taiwan, in 1992, and the Ph.D. degree from the Department of Com-puter Science and Information Engineering, National Chung Cheng University, Chia-Yi, Taiwan, in 2004.
He has been with the Institute of Computer Sci-ence and Information Engineering, Wufeng Univer-sity, Chia-Yi, where he is currently an Associate Professor. His current research interests include ex-pert system, artificial neural network, fuzzy system, watermarking, and image filter design.
LIN et al.: FAILURE PROBABILITY ESTIMATION OF ACF PACKAGES 377
Te-Hua Fang received the B.S. and M.S.
de-grees in mechanical engineering from the National Taiwan University of Science and Technology, Taipei, Taiwan, in 1993 and 1995, respectively, and the Ph.D. degree in mechanical engineering from National Cheng Kung University, Tainan, Taiwan, in 2000.
He is currently a Full Professor with the Depart-ment of Mechanical Engineering, National Kaoh-siung University of Applied Sciences, KaohKaoh-siung, Taiwan. His main research interests include molecu-lar dynamics, scanning probe microscopy, nanomaterials, and nanotechnology applications. He currently serves as a member of the Editorial Advisory Board of The Open Surface Science and Current Nanoscience.
Yen-Chun Liu received the Ph.D. degree in
chemical engineering from the National Tsing Hua University, Hsinchu, Taiwan, in 1995.
He is currently with the Graduate School of Opto-Mechatronics and Materials, WuFeng Univer-sity, Chia-Yi, Taiwan, as an Associate Professor with research interests in IC packaging, nanoscience and nanotechnology, fine ceramics/sol-gel process-ing to thermal barrier coatprocess-ings, chemical engineer-ing, etc. He has some publications in the areas of nanoscience and nanotechnology, electronic packag-ing, and chemical engineering.