Measurement and Characterization of the Moisture-Induced Properties of ACF Package

Transcription

1 Measurement and Characterization of the Moisture-Induced Properties of ACF Package Ji-Young Yoon Ilho Kim Soon-Bok Lee Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Gu-seong, Yu-seong gu, Daejeon , Korea This study is to observe the exact behavior of anisotropic conductive adhesion (ACF) package under humid environments by obtaining the moisture-induced properties such as diffusion coefficient (the rate of moisture movement into the materials), saturated moisture content (the maximum absorbed quantity), and swelling coefficient (length change due to the chemical interaction). So the experiments were performed to get the moistureinduced properties of ACF and FR4 using newly developed method at various temperature and relative humidity conditions. Experimental results showed that the diffusion coefficient of ACF and FR4 follows Arrhenius equation very well, and the saturated moisture content of them follows Henry s law, which means linear relationship between saturated moisture content and relative humidity, but the saturated moisture content of ACF is influenced by temperature as well as relative humidity. And the swelling coefficient of ACF and FR4 increases with temperature. Especially in the case of ACF, it shows the dramatic degradation due to T g (glass transition temperature) at nearby 85 C. Finally, as using these experimental results, the behavior of the ACF package under humid environment is predicted through finite element simulation. When wetness defined by moisture content over saturated moisture content changes from 0 to 0.9, the center of the ACF package is subject to compression and the edge of the ACF package is subject to tension in the case of transient state. After all, because the edge of the ACF package is very weak due to bending moment, the failure is easy to occur under humid environment. DOI: / Keywords: anisotropic conductive adhesion (ACF), diffusion coefficient (D), saturated moisture content (M sat ), swelling coefficient ( ), hygrostress, input and output (I/O), liquid crystal display (LCD), integrated circuit (IC), finite element analysis (FEA) 1 Introduction Anisotropic conductive film ACF has been introduced as a promising flip chip interconnection material, due to its potential in achieving high density input and output I/O interconnection, low processing temperature, and green process lead-free. In particular, it has been extensively used in liquid crystal display LCD packaging area, contactless smart-card module assembly, and bare chip attach on flexible and rigid substrates 1. It consists of an adhesive polymer matrix and fine conductive fillers of metallic or metal-coated polymer particles, and when used for integrated circuit IC assembly provides electrical conduction as well as mechanical interconnections. As schematically shown in Fig. 1, electrical conduction of the ACF package is provided by the deformation of the conductive particles trapped between the chip bumps and substrate pads in the z-axis of the adhesives perpendicular to the plane of the board, while electrical isolation is maintained in the plane of the adhesive layer 2 4. In spite of the increasingly important role of ACF in the assembly of electronic products, there are still concerns about the reliability of any device with ACF in it. After all, when compared with solder interconnect material, ACF is a new material with many unknown properties. Among the many factors that affect the reliability of the ACF package, moisture is one of the most important ones. Because Contributed by the Electrical and Electronic Packaging Division of ASME for publication in the JOURNAL OF ELECTRONIC PACKAGING. Manuscript received January 10, 2007; final manuscript received October 14, 2007; published online April 21, Assoc. Editor: Stephen McKeown. ACF and FR4 substrate of the ACF package component are hydrophilic polymeric material, they absorb moisture when exposed to a humid environment. Since different swelling occurs between the polymeric and nonpolymeric materials as well as among the polymeric materials constituting the electronic packages, it induces swelling mismatch in the package. The swelling mismatches between different materials in the ACF package induce stresses that can be added to the thermal stress in the package leading to package failure in some cases. The stress generated by moisture swelling mismatches is known as hygroscopic stress, and can be calculated similarly to thermomechanical stress. In place of temperature and thermal expansion coefficient, moisture content and swelling coefficient are, respectively, substituted. As using the thermal-hygroanalogy, hygroscopic stresses can be modeled with commercial finite element software 5 8. In order to observe the tendency of moisture-induced properties such as diffusion coefficient, saturated moisture content, and swelling coefficient of ACF and FR4 on temperature and relative humidity, this paper obtains these properties with a new experimental method at various temperature and relative humidity conditions. Diffusion coefficient D represents the rate with which moisture diffuses through a material. Saturated moisture content M sat represents the maximum content of moisture that can be contained per unit dry weight. Swelling coefficient is defined as hygrostrain induced by moisture per unit moisture content. Finally, the behavior of the ACF package under moisture environment is predicted through finite element FE simulation using these moisture-induced properties of ACF and FR4. Journal of Electronic Packaging Copyright 2009 by ASME JUNE 2009, Vol. 131 /

2 Fig. 1 Schematics of the ACF flip chip joint 2 The Evaluation Method of Moisture-Induced Properties Fig. 3 The microscopic structure of the polymer 2.1 Diffusion Coefficient and Saturated Moisture Content (D and M sat ). Moisture sorption can be considered to occur in two steps. At first, the moisture dissolves at the surface of the polymer, establishing a chemical potential gradient within the material. Diffusion then follows under the influence of this gradient. A complete understanding of the moisture sorption process thus requires knowledge of both the diffusion coefficient and saturated moisture content of the moisture within the polymer. Typically, diffusion is described by Fick s law of diffusion, which states that the flux of a diffusing species is proportional to the moisture content gradient measured normal to the area. If diffusion coefficient is constant and dominant in only one direction like Fig. 2 here chosen as x, such as in a material, then M = D 2 M t x 2 1 where D cm 2 /s is the diffusion coefficient given by Eq. 3 and M is the moisture content given as wet weight dry weight M % = dry weight q q D = D o exp = D kt K o exp 3 k T C where k J/mol K is Boltzmann s constant, q J/mol is active energy, which may be thought of as energy required to produce the diffusive motion of 1 mole of atoms, T K and T C are, respectively, absolute temperature and Celsius temperature, and D 0 is constant. The saturated moisture content is defined as substituting saturate weight for wet weight from Eq. 2. This obeys Henry s law related only to relative humidity and not to temperature in the linear regime as follows: M sat = K RH 4 where K is constant and RH is relative humidity. Deviation from this correlation is common, especially at high moisture content and/or in heterogeneous systems containing fillers 9. We introduce the following new procedure to be easy to obtain diffusion coefficient 10. First, the separation method of variable is given as M = X x T t where X x is the function of position and T t is the function of time. The moisture content of Eq. 5 is substituted for Eq. 1. And moisture content according to time can be found as initial condition:t =0, M = M sat, 0 x h boundary condition:t 0, M =0, x =0,h 6 where h is the thickness of material. Finally, the model of diffusion coefficient is given as sin Dt M x,t 4M sat 2 x exp h h 2 Here, the dominant initial term is approximately chosen. And when the thickness is very thin, it is difficult to find moisture content distribution according to position. So, the average moisture content must be considered for experiment. When considering unit area, the volume is h. 5 7 M t = 1 h 0 h M x,t dx = 8M sat 2 Dt 2 exp h 2 8 Here, M t means average moisture content. Finally, Eq. 8 is very simply expressed into ln M t M sat =ln D h 2 t 9 Fig. 2 Diagram of the model to calculate the diffusion coefficient 2.2 Swelling Coefficient. As shown in Fig. 3, it has been shown through spectroscopic methods that water molecules are present in polymeric materials in two states: free volume composed of void and/or microcrack and occupied volume composed of polymer chain and polymer molecules 11,12. Many studies have postulated that swelling attributes to the occupied volume. As polar water molecules slide into the polymer chains and form hydrogen bond with the polymer molecules, as / Vol. 131, JUNE 2009 Transactions of the ASME

3 Fig. 4 The composition of the occupied volume shown in Fig. 4, the space between the polymer chain will increase and the polymer will swell 13. So the change in dimension induced by moisture absorption can be related to swelling coefficient as follows: h = M 10 where h is the hygrostrain and is the swelling coefficient. It is similar to thermal expansion in that both are physical phenomenon involving molecules equilibrating at a larger distance apart as a result of temperature increment and moisture absorption, respectively. 3 The Experimental Procedure and Results 3.1 Specimen. The experimental samples are used in ACF and FR4 of the ACF package components. The dimensions of the specimen for each experiment are shown in Table ACF A Diffusion Coefficient and Saturated Moisture Content (D and M sat ). The technique to measure the diffusion coefficient and saturated moisture content of ACF at 85 C/ 90% involves the following. 1 Prepare the ACF sample cured at 170 C for 30 min in a thermal chamber. 2 Saturate the ACF sample at 85 C/90% in a humidity chamber. 3 Measure the weight decrease in the ACF sample using thermal gravitational analysis TGA, the equipment in which the weight change in sample under constant load is measured as a function of temperature or time, at 85 C/0% until it reaches to the stable equilibrium. Here, the inside TGA is assumed as 0% because nitrogen gas filled in the TGA chamber has the crystal structure that is not to react with water molecule. 4 Obtain diffusion coefficient using the slope of the graph shown in Fig. 5 through Eq Obtain saturated moisture content using dry weight and saturated weight from Eq. 2. Fig. 6 The weight reduction in ACF according to time at various conditions This procedure was also performed at 25 C/ 90%, 60 C/ 90%, 70 C/90%, 85 C/60%, and 85 C/70% to find influence of the temperature and relative humidity on the diffusion coefficient and saturated moisture content. The weight reduction in ACF according to time at various conditions is shown in Fig. 6. And the results of diffusion coefficient according to temperature and saturated moisture content according to relative humidity are shown in Figs. 7 and 8, respectively. Here, it takes twice or longer time to finish measuring weight change in ACF as shown by Fig. 6 and diffusion action is very weak at 25 C. Therefore, I exclude the data point at 25 C/90% to obtain Arrhenius equation and modified Henry s law of ACF due to environmental factors and resolution problem of TGA. When we find two unknown parameters D o and q using the slope and y-intercept values in the Fig. 9 curve Arrhenius equation can be completed. The analytic results expressed as line correspond very well with the experimental results as shown in Fig. 7. And it was found that the diffusion coefficient of ACF is not influenced by relative humidity. Table 1 The sample dimensions for each experiment D and M sat ACF mm mm 3 FR mm mm 3 Fig. 7 The D of ACF versus temperature Fig. 5 Variation of moisture according to time Fig. 8 The M sat of ACF versus relative humidity Journal of Electronic Packaging JUNE 2009, Vol. 131 /

4 Fig. 9 In D versus 1/kT K curve of ACF Fig. 11 The method to measure of ACF The saturated moisture content of ACF follows Henry s law, M sat =0.02*RH % obtained by least squares method at constant temperature from Eq. 4. However, Fig. 8 depicts that it is exceedingly influenced by temperature, which is because the rate of volume change in ACF is susceptible for temperature. To estimate the saturated moisture content of ACF at various conditions, a modified Henry s law is suggested to include the function of temperature as follows: E M sat = A exp RH 11 kt K where A is a constant and E is an active energy. When the slope and y-intercept values from the Fig. 10 curve are found, the modified Henry s law is completed with the two unknown parameters A and E as the case for Arrhenius equation. Finally, the modified model of the saturated moisture content of ACF is given as M sat = 0.2 exp kt K 7652 RH 12 While Arrhenius equation and modified Henry s law of ACF are known, diffusion coefficient and saturated moisture content can be found at all temperatures and relative humidity conditions Swelling Coefficient. The technique in Fig. 11 is developed on two standard thermal analysis machines, namely, thermomechanical analysis TMA, the equipment in which the deformation of sample under constant load is measured as a function of temperature or time, and TGA machines to measure the swelling coefficient of ACF at 85 C/ 90%. The procedure involves the following. 1 Prepare two ACF samples cured at 170 C for 30 min in a thermal chamber. 2 Saturate ACF at 85 C/90% in a humidity chamber. 3 Desorp moisture from the samples at 85 C in TMA and TGA, respectively. 4 Extract dimensional change and weight decrease in the samples at the same desorption duration. 5 Plot hygrostrain versus moisture content. This procedure was also performed at 25 C/ 90%, 50 C/ 90%, 60 C/90%, 70 C/90%, 85 C/60%, and 85 C/70% to find the influence of temperature and relative humidity on the swelling coefficient of ACF. The results are shown in Fig. 12 and the slope of these curves means swelling coefficient defined by Eq. 10. The swelling coefficient of ACF increases with temperature as shown in Fig. 13, because the interaction between water molecule and polymer molecule is gradually weak. But the decline is shown at nearby 85 C. This sudden decline can be explained as the effect of T g glass transition temperature. T g can be defined by dramatic change in the thermal expansion coefficient. After the dimensional change in dry ACF is measured using TMA, it was found that T g has the region from 110 C to 140 C as shown in Fig. 14. But in the case of wet ACF, as decreasing from 30 C to 40 C, T g has the region from 90 C to 120 C as shown in Fig. 15. If we consider the experimental error of 5 C, 85 C can belong to the T g region of wet ACF. The free volume of ACF increases in the T g region. This is why the decline of swelling coefficient occurs at 85 C. Fig. 10 ln M sat versus 1/kT curve of ACF Fig. 12 Hygrostrain results from TGA and TMA for ACF / Vol. 131, JUNE 2009 Transactions of the ASME

5 Fig. 16 The weight change in FR4 versus time at various conditions Fig. 13 The swelling coefficient of ACF versus temperature 3.3 FR A Diffusion Coefficient and Saturated Moisture Content (D and M sat ). The technique to measure the diffusion coefficient and saturated moisture content of FR4 at 85 C/90% is a little different from that of ACF. This procedure involves the following. 1 Prepare the FR4 sample dry for 24 h in a thermal chamber at 180 C. 2 Measure the dry weight of FR4 through electronic scale. 3 Put the FR4 sample in a humidity chamber at 85 C/90%. 4 Measure the weight increase in the FR4 sample in situ with an electronic scale of high sensitivity with 0.1 mg until the saturated state. 5 Obtain the diffusion coefficient using the slope of the graph shown in Fig. 5 through Eq Obtain the saturated moisture content using the dry weight and saturated weight from Eq. 2. This procedure was also performed at 25 C/ 60%, 50 C/ 60%, 70 C/60%, 85 C/50%, 85 C/60%, and 85 C/70% to obtain the influence of the temperature and relative humidity on the diffusion coefficient and saturated moisture content. The weight change in FR4 versus time at various conditions is shown in Fig. 16. And the diffusion coefficient of FR4 versus temperature is depicted in Fig. 17. Here, the line and dot represent Arrhenius equation and experimental results, respectively. Arrhenius equation of FR4 can be completed through the similar method as ACF. Experimental results follow Arrhenius equation of Eq. 3 similar to the case of ACF, but the influence of relative humidity is very small on D of FR4. The saturated moisture content of FR4 versus relative humidity is shown in Fig. 18. This follows Henry s law fairly well related to Eq. 4. Fig. 14 The measurement for T g of dry ACF Fig. 15 The measurement for T g of wet ACF Fig. 17 The diffusion coefficient of FR4 versus temperature at various conditions Journal of Electronic Packaging JUNE 2009, Vol. 131 /

6 Fig. 19 The fringe result of FR4 at 85 C/90% This procedure was also repeated at 25 C/ 60%, 70 C/ 60%, 85 C/60%, and 85 C/70% to find the effect of temperature and relative humidity. The swelling coefficient of FR4 increases with temperature as shown in Fig. 20, and the influence of relative humidity is similar to the case of ACF. Fig. 18 The saturated moisture content of FR4 versus relative humidity at various conditions Swelling Coefficient. To measure the hygrostrain of FR4, the two-beam moiré interferometry is used. Moiré interferometry is a whole-field optical interference technique with high resolution and high sensitivity for measuring the strain fields 14. The widely used moiré interferometer in electronic packages analysis is the portable engineering moiré interferometer from photomechanics PEMI. In moiré interferometry, deformed specimen gratings interfere with the reference grating to produce the moiré fringe pattern. The resulting fringe patterns generate contour maps of U displacement fields, which are defined as in-plane displacement. The displacement then can be determined from fringe orders by the following relationship: U = 1 f N x 13 where f is the frequency of the virtual reference grating, and N x is fringe orders in the U field moiré patterns. A virtual reference grating with a frequency f of 2400 lines/mm is used, which provides a sensitivity of m per fringe order. When the strain is required, they can be decided from the displacement fields by the relationship for small engineering strain. x = U x = 1 N x 14 f x The procedure to obtain the swelling coefficient of FR4 at 85 C/ 90% involves the following. 1 Prepare the FR4 sample dry for 24 h in a thermal chamber at 180 C. 2 Attach the reference grating to the surface of FR4 at 85 C. 3 Saturate this sample in a humidity chamber at 85 C/90%. 4 Measure the fringe induced by desorption using PEMI and thermal chamber for keeping 85 C in real-time until no fringe changes. 5 Calculate the hygrodisplacement and strain by Eqs. 13 and 14 with respect to saturated moisture content, respectively. 6 Obtain the swelling coefficient of FR4 at 85 C/90% through Eq. 10. This fringe result is shown in Fig. 19, where a means the initial state after just drawing it from relative humidity and b indicates the final state after 3 h, which represent at no fringe change condition. The hygrostrain is obtained from fringe order change between a and b. 4 The Reliability Evaluation of ACF Package Under Humidity Environment 4.1 Moisture Diffusion and Hygromechanical Modeling. Most commercial FE software is not equipped with moisture diffusion coupled stress analysis modeling capability. So with appropriate thermal-moisture analogy, moisture diffusion can be modeled using the thermal diffusion function of the software 8. The finite element analysis FEA implementation scheme is presented in Table 2, where wetness is defined as W=M /M sat and M sat is saturated moisture content. The dimensions of chip, ACF, and FR4 are mm 3, mm 3, and mm 3, respectively. The analysis model used a quarter of total model as shown in Fig Results of Finite Element Analysis. The material properties of the ACF package are presented in Table 3 to analyze the moisture diffusion and hygromechanical behavior at 70 C/ 90%. The initial and boundary conditions on the outside surfaces are given as initial condition:w =0, t =0 boundary condition:w = % wet, t 0 15 Fig. 20 The swelling coefficient of FR4 versus relative humidity at various conditions Table 2 FEA thermal-moisture analogy for hygromechanical modeling of the ACF package Property Thermal Moisture Field variable Temperature T Wetness W Density 1 Conductivity k DM sat Specific capacity C M sat Coefficient of expansion M sat / Vol. 131, JUNE 2009 Transactions of the ASME

7 Fig. 23 The von Mises stress distribution in ACF after 1 year Fig. 21 Finite element model The result of FEA moisture diffusion modeling of the ACF package is illustrated in Fig. 22. With the knowledge of wetness distribution in the package as well as the hygroscopic swelling characteristic of materials, hygroscopic stress shown by Fig. 23 von Mises stress and Fig. 24 normal stress can be readily computed through hygromechanical modeling. To observe the stress more exactly, the finite elements of ACF are chosen from center to end as shown in Fig. 25 from Fig. 23 and Fig. 26 from Fig. 24. In Fig. 25, the von Mises stress has very different values between the center element point and the end element point until 31 h. However, it is gradually saturated. Finally, the ACF package will be opening mode at static state. The normal stress in the ACF package is the most important failure criterion, because the performance of the ACF package is determined by the contact of conductive particles between the pad and bump 15. Therefore, the normal stress is chosen as the essential hygroscopic stress. During transient state, the large tensile normal stress occurs at the edge, but the compression normal stress occurs at the center. Because of the bending moment depicted in Fig. 27, it is most Table 3 The material properties of the ACF package at 70 C/90% Property Chip ACF FR4 E GPa Density Conductivity % mm 2 /s Specific capacity Coefficient of expansion Fig. 24 The normal stress distribution in ACF after 1 year Fig. 22 The wetness distribution in the ACF package after 1 year Fig. 25 The change in von Mises stress according to time Journal of Electronic Packaging JUNE 2009, Vol. 131 /

8 Fig. 26 The change in normal stress according to time 5 Summary and Discussion The moisture-induced properties of the ACF package materials ACF and FR4 at various temperatures and relative humidities are measured and their characterizations are presented. The diffusion coefficient of ACF and FR4 increases with temperature exponentially, according to Arrhenius equation. Arrhenius equation is completed by finding D o and q from Eq. 3 through experimental results at different temperatures and same relative humidity. Finally, the diffusion coefficient of ACF and FR4 can be predicted at all temperatures and relative humidities. The saturated moisture content of ACF and FR4 follows Henry s law very well, which is linear relationship between saturated moisture content and relative humidity at same temperature. The model expressed by Eq. 9 to predict the saturated moisture content of ACF is completed, by assuming K value as a function of temperature to include the effect of temperature. And the model to predict the saturated moisture content of FR4 is also completed, by assuming K value to constant from Eq. 4. The proposed models estimate the saturated moisture content of ACF and FR4 at various temperatures and relative humidities. The swelling coefficient of ACF and FR4 increases with temperature regardless of relative humidity, which can be explained by the interaction between water molecule and polymer molecule, and becomes gradually weak as temperature goes up. In the case of ACF, after the swelling coefficient increases up to 70 C, a dramatic decrease was observed at nearby 85 C. The swelling phenomenon only occurs in the occupied volume of polymer composed of polymer molecular and polymer chain, but does not relate to the free volume composed of voids existing among polymer segments. After the T g region of wet ACF is measured using TMA, it was found that 85 C belongs to the T g region. Because the free volume of ACF increases in the T g region, the decline of the swelling coefficient that occurred at nearby 85 C can be explained by this T g effect. The finite element simulation was performed for the reliability of the ACF package with the experimentally obtained moistureinduced properties of ACF and FR4 and moisture diffusion model through appropriate thermal-moisture analogy. The large tensile normal stress occurs at the edge, but the compression normal stress occurs at the center during transient state, which produces the bending moment in humid environment. Since it is very easy to lose the contact of conductive particle between the bump and pad at the edge of the ACF package, it can be predicted that the edge of the ACF package is weak. When it reaches stable state, a small tensile stress occurs at the ACF. The prediction of deformation behavior of the ACF package under humid environment is provided as an effective method for evaluating reliability of ACF packages. Acknowledgment This research was supported by a grant from the Center for Nanoscale Mechatronics & Manufacturing, one of the 21st Century Frontier Research Programs, which are supported by the Ministry of Education, Science and Technology, Korea. Fig. 27 The deformation of the ACF package by bending moment under moisture environment likely to lose the contact of conductive particle between the bump and pad at the edge of the ACF package. Finally, the edge of the ACF package can be predicted to be weak. References 1 Chan, Y. C., Hung, K. C., Tang, C. W., and Wu, C. M. L., 2000, Degradation Mechanism of Anisotropic Conductive Adhesive Joints for Flip Chip on Flex Application, Proceedings of the Fourth International Conference on Adhesive Joining and Coating Technology in Electronics Manufacturing, pp Yim, M. J., Paik, K. W., Kim, Y. K., and Hwang, H. N., 1997, A Study on the Electrical Conduction Mechanism of Anisotropically Conductive Film for LCD Packaging Application, Advances in Electronic Packaging, Vol. 19 1, EEP, New York, pp Fu, Y., Wang, Y., Wang, X., Liu, J., Lai, Z., Chen, G., and Willander, M., 2000, Experimental Characterization and Theoretical Characterization of Electrical Contact in Anisotropically Conductive Adhesives, IEEE Trans. Adv. Packag., 23 1, pp Lai, Z., and Liu, J., 1990, Anisotropically Conductive Adhesive Flip-Chip Bonding on Rigid and Flexible Printed Circuit Substrates, IEEE Trans. Adv. Packag., 19 3, pp Tanaka, N., Kitano, M., Kumazawa, T., and Nishimura, A., 1997, Evaluation of Interface Delamination in IC Packages by Considering Swelling of the Molding Compound Due to Moisture Absorption, 47th ECTC, pp Wong, E. H., Chan, K. C., Tee, T. Y., and Rajoo, R., 2002, Comprehensive Treatment of Moisture Induced-Recent Advances, IEEE Trans. Electron. Packag. Manuf., 25 3, pp Xu, T., and Farris, R. J., 2006, Stresses Associated With Diffusion in Polyimide and Polyacrylic Films, J. Appl. Polym. Sci., 99 5, pp Wong, E. H., Rajoo, R., Koh, S. W., and Lim, T. B., 2002, The Mechanics and Impact of Hygroscopic Swelling of Polymeric Materials in Electronic Packaging, ASME J. Electron. Packag., 124 2, pp Springer, G. S., 1981, Environment Effect on Composite Materials, Technomic, Lancaster. 10 Park, C.-O., 2003, The Book: Diffusion in Solids, BRAINKOREA, Seoul, South Korea. 11 Vanderhart, D. L., Schen, M. A., and Davis, G. T., 1999, Partitioning of Water Between Voids and the Polymer Matrix in a Polymer Compound by Proton NMR: The Role of Larger Voids in the Phenomena of Popcorning and Delamination, Int. J. Microcircuits Electron. Packag., 22 4, pp Toprak, C., Agar, J. N., and Falk, M., 1979, State of Water in Cellulose Acetate Membranes, J. Chem. Soc., Faraday Trans. 1, 75, pp Adamson, M. J., 1980, Thermal Expansion and Swelling of Cured Epoxy Resin Used in Graphite/Epoxy Composite Materials, J. Mater. Sci., 15 7, pp Post, D., Han, B., and Ifju, P., 1994, High Sensitivity Moiré, Springer-Verlag, New York. 15 Yin, C., Lu, H., Bailey, C., and Chan, Y.-C., 2005, Moisture Effects on the Reliability of Anisotropic Conductive Film Interconnection for Flip Chip on Flex A Applications, IPACK2005, Proceeding of the ASME/Pacific Rim Technical Conference and Exhibition on Integration and Packaging of MEMS, NEMS and Electronic Systems, pp / Vol. 131, JUNE 2009 Transactions of the ASME