Structural integrity of metal–polymer adhesive interfaces in microelectronics
Abstract:
Realization of reliable bonding is the immutable requirement for adhesives. The interfacial adhesive strength is basically determined by the intermolecular interactions at interfaces, such as van der Waals and hydrogen bonds. When coupling agents are utilized for improving interfacial adhesion, it is essential to consider the chemistry of the adhesive molecules and the functional groups of the coupling agents. In addition to the intermolecular interactions, mechanical factors in meso-scale of adhesive layers can also contribute to the bonding strength of adhesive joints. In this chapter, the theoretical background for structural integrity of adhesive joints in microelectronics is discussed.
6.1 Introduction
Adhesion techniques are some of the most important elements in electronics packaging technologies. Advanced adhesives for electronics packaging have recently become diversified, in conjunction with the remarkable progress that is being made in microsystems. The adhesives are required to exhibit functions such as electrical and thermal conductivities in some cases, but realization of reliable bonding is the immutable requirement. In this chapter, the principal factors that determine the bonding strength of adhesive joints are identified and explained in detail, with the aim of establishing appropriate guidelines for developing advanced adhesives from the viewpoint of reliable bonding. Although the chemical composition of adhesive compounds seems complex, their design stems from the basic theory of adhesion and bonding, which is briefly introduced in this chapter.
6.2 Theoretical considerations of work of fracture and bonding strength of adhesive joints
6.2.1 Thermodynamic work of adhesion
Figure 6.1 shows schematically the concept of the thermodynamic work of adhesion;1,2 the surfaces A and B are assumed to be perfectly flat. The interfacial free energy of A–B and the surface free energies of A and B are represented by γAB, γA and γB, respectively. γAB is the interfacial free energy when the atoms on the surfaces of A and B are located at their equilibrium inter-atomic distances. The work of adhesion (Wa) can be expressed by Equation 6.1 (Dupré’s equation):
Wa can be estimated if the interfacial and surface free energies (γAB, γA, γB) are known. However, the surface energy of solids is difficult to measure and is consequently often estimated indirectly using the wettability of the surfaces of solids by liquids. The estimation method for the surface free energy of solids is explained in detail in Section 6.3.3.
When a liquid drop is placed on a solid surface, the equilibrium shape of the drop is determined by the balance between the solid/liquid interfacial tension (interfacial free energy; γSL), and the surface tensions (surface free energies) of the solid and liquid (γS and γL), as shown in Fig. 6.2. Thus, the surface free energy of the solid is represented by Equation 6.2 (Young’s equation):
6.2 The balance between interfacial tension (γSL) and surface tensions (γS and γL) for a liquid drop on a solid surface.
where θ is the contact angle of the liquid on the solid surface. In this situation, we can express Wa between the solid and liquid using Equations 6.1 and 6.2 (Young–Dupré equation):
In addition, it is assumed that γSL can be represented by Fowkes’s equation (Equation 6.4)3’4 and extended Fowkes’s equations (Equations 6.5 and 6.6),5–8 depending on the definition of chemical interaction at the interface:
where γd, γp and γh are the dispersive, polar and hydrogen bonding components, respectively, of the surface free energy. By contrast, γSL is formulated as Equation 6.7 in the acid–base theory:9
where γLW, γ+ and γ− are the Lifshitz–van der Waals, Lewis acid (electron acceptor) and Lewis base (electron donor) components, respectively. The details (premises and limits of application) of these equations are explained in Section 6.3.
We adopt Equation 6.5 to continue discussion of Wa. By using Equation 6.2, 6.3 and 6.5, Wa can be expressed by
In Section 6.2.2, we will estimate ideal adhesive strength between adhesive and substrate using Equation 6.8.
6.2.2 Interfacial interaction energy estimated from ab initio simulations
The interfacial interaction energy caused by the intermolecular interactions between adhesive molecules and substrate surfaces has been recently studied using ab initio simulation techniques. In these simulations, the molecular structure and intermolecular distance between models of adhesive molecules and substrate surfaces are optimized to estimate the total energy of the optimum structure (EAS). The interaction energy (ΔE) is described by
where EA and ES are the total energy of the models of adhesive molecules and surface, respectively. The concept of ΔE is similar to that of the thermodynamic work of adhesion represented by Equation 6.1. Furthermore, ΔE can be decomposed into several interaction components using energy decomposition analyses. For example, Kitaura and Morokuma10 have proposed a methodology of energy decomposition analysis represented by:
where EES, EPL, ECT and EEX are electrostatic, polarization, charge transfer and exchange energies, respectively. The energy decomposition analysis is conceptually similar to Equation 6.8. Minamizaki et al.1,11 have analyzed the interaction energy between several adhesives and substrates using the energy decomposition method based on ab initio simulation.
6.2.3 Ideal adhesive strength and the bonding strength of adhesive joints
The interfacial potential (V(r)) between A and B can be schematically drawn as shown in Fig. 6.3. The ideal adhesive strength is determined from the differential curve, –dV/dr, the maximum value of –dV/dr being the ideal adhesive strength.12 However, the interfacial potential is often difficult to obtain exactly for adhesive joints. Hence in the present study, the ideal adhesive strength is roughly estimated using Wa.13
6.3 Schematic illustration of the interfacial potential and ideal adhesive strength between certain surfaces.
The values of γd and γp for an epoxy-based adhesive and polyimide substrate are assumed as shown in Table 6.1. In this case, Wa between the epoxy and polyimide is estimated as 90 mJ m− 2 using Equation 6.8. Furthermore, we assume that the bonds between epoxy adhesive and polyimide can be broken at a displacement of 0.5–1.0 nm from the equilibrium interatomic distance. The stress to break the chemical bonds is estimated as 90–180 MPa.
Table 6.1
Magnitudes of the dispersive and polar components of the surface free energy of epoxy-based adhesive and polyimide that were assumed to roughly estimate the ideal adhesive strength using the thermodynamic work of adhesion
However, the ideal adhesive strength is always much larger than the bonding strength that can be experimentally measured using real adhesive joints. When we discuss the differences in the ideal adhesive strength and the bonding strength, the following two points should be taken into account. One is the imperfection of interfacial adhesion, including imperfect wetting of adhesives and the effect of internal stress generated by shrinkage of adhesives during the curing process. Imperfect wetting can provide a flaw that will be the origin of a fracture during bonding strength measurement. The internal stress generated in the adhesive layer significantly reduces adhesive strength (see Section 6.4.1). The other important point is the effects of fracture behavior during measurement of the bonding strength that depends on various factors such as geometry of joints, and loading conditions.1 The viscoelastic deformation of adhesives can affect the bonding strength. Furthermore, the bonding strength often depends on the geometry of joints, even if the same adhesive and substrate materials are used for preparing the joints.
6.2.4 Relationship between work of adhesion and work of fracture of adhesive joints
The contribution of the thermodynamic work of adhesion to bonding strength measured by mechanical tests is usually obscured due to the effects of geometrical and loading factors. In the 1970s, Andrews and colleagues proposed a generalized fracture mechanics theory to analyze the fracture energy (Θ) of adhesive joints. According to Andrews and Kinloch,14 Θ can be written:
where Φ is a loss function that depends on the debonding rate, temperature and strain level. Θ0 is the true interfacial fracture energy.
Although Θ0 is proportional to Wa, the experimental data often show that Θ0 is not equal to Wa. Thus, Kamyab and Andrews15 proposed Equation 6.12 to analyze the relationship between Θ0 and Wa:
where Φ′ is a loss function evaluated only over the process zone around the crack, and involves the appropriate parameters for the loss process concerned. Therefore, Θ can be rewritten as:
where Φtotal is the total loss function.
The experimental data showed that Φtotal is usually orders of magnitude greater than Wa. The stress relaxation due to viscoelastic deformation of adhesives during mechanical tests is one of the most effective factors in determining Φtotal. Thus the contribution of the viscoelastic behavior of adhesives is more dominant than the effects of interfacial adhesion during mechanical testing of adhesive joints. However, the effects of interfacial adhesion are still important, because it was established experimentally and theoretically that the total loss function (Φtotal) is proportional to Wa.14 To realize significant stress relaxation due to viscoelastic deformation of adhesives during mechanical tests, sufficient interfacial adhesion between adhesives and substrates that is capable of withstanding stress is essential.
As mentioned above, the bonding strength of adhesive joints is not determined only by the interfacial adhesive strength, but also by the loss function. When we practically analyze the bonding strength, various factors such as defects at the interface, internal stress generated due to shrinkage of adhesives, geometrical and loading factors of mechanical tests, should be taken into account as well as viscoelastic behavior of adhesives. However, improvement of the interfacial adhesive strength is fundamentally important to increase the bonding strength of adhesive joints.
In the following sections, the influential factors determining the bond strength of adhesive joints are discussed in detail. The discussions will be useful for scientists and engineers who develop novel adhesives and packaging techniques, including nano-technology (nano-packaging).
6.3 Chemical and physical intermolecular interactions at interfaces
6.3.1 General view of the intermolecular interactions
Interfacial adhesive strength is basically determined by the intermolecular interactions at interfaces, although chemical reactions between the surface functional groups of substrates and the molecules of adhesives do not always occur (to form covalent bonds at the interface).
The intermolecular interactions are generally categorized into several types of bonds, such as van der Waals and hydrogen bonds. Although these categories of intermolecular interactions have been widely used, they are not always appropriate for interpreting adhesion phenomena in adhesive joints. The goal is to understand interfacial interactions through quantum chemical (ab initio) and molecular dynamics (MD) simulations. However, these simulation techniques are not yet generally applied for adhesive joints. The concepts of intermolecular interactions that are applied especially for adhesive joints will be introduced in the following sections.
6.3.2 Dispersive and polar components of intermolecular interactions
Van der Waals bonds
Van der Waals forces16 originate from dipole–dipole, dipole–induced dipole, and induced dipole–induced dipole interactions. The intermolecular potential can be generally described as a function of intermolecular distance (r) using:
The first and second terms on the right side of this equation represent the attractive and repulsive interactions. The exponent n in the repulsive term is a constant between 9 and 12. Equation 6.14 represents the Lennard–Jones potential when n = 12.
An attractive interaction is generated between two polar molecules (which have permanent dipoles) depending on their relative orientation. Although the interaction energy between two freely rotating polar molecules is zero, the molecules do not rotate completely freely. The average interaction energy (attractive interaction) of two rotating polar molecules is written as:
where μ, ε0, k, T are dipole moment, permittivity in vacuum, Boltzmann constant and temperature, respectively. Equation 6.15 describes the potential energy of dipole–dipole interaction (Keesom interaction).
The second contribution to van der Waals force is the dipole–induced dipole interaction. A polar molecule (with dipole moment μ1) can induce a dipole in a neighboring polarizable molecule. These two molecules exhibit an attractive interaction between the permanent dipole of the first molecule and the induced dipole of the second molecule. In this case, the average interaction energy (attractive interaction) is written as a function of r:
where α2' is the polarizability volume of molecule 2. The dipole–induced dipole interaction energy is independent of temperature.
An attractive interaction is also generated even between non-polar molecules. Transient dipoles are induced in non-polar molecules as well as polar molecules due to fluctuations in the instantaneous positions of electrons. The intermolecular interactions between the transient dipoles of non-polar molecules are called the dispersive (London) interactions. The dispersive interaction energy (attractive interaction) can be approximated by:
where I1 and I2 are the ionization energies of the two molecules.
Table 6.2 shows the coefficient of the attractive potential for dipole–dipole, dipole–induced dipole and dispersive interactions between several molecules.1,12 Because the dispersive force can be generated in both non-polar and polar molecules, it is considered to be a universal intermolecular interaction. The dispersive interaction tends to increase with increasing molecular (atomic) size. Even in polar molecules, the dispersive interaction makes a significant contribution to the van der Waals force.
Table 6.2
Coefficient of attractive potential for three types of van der Waals interactions between several molecules
By contrast, permanent dipoles in molecules are necessary in order to generate dipole–dipole and dipole–induced dipole interactions. Consequently, these forces should be distinguished from the dispersive force when we analyze intermolecular interactions. Dipole–dipole and dipole–induced dipole interactions can be categorized as the polar component of intermolecular interactions.
Hydrogen bonds
The hydrogen bond is an attractive interaction of the form A–H
B, where A and B are highly electronegative elements such as N, O, F and anionic species (e.g. Cl−).16 In this case, a dominating attractive interaction is supposed to exist because the internuclear distance between formally non-bonded atoms becomes less than their van der Waals contact distance.
The result of an ab initio molecular orbital simulation (using a commercial code, Dmol3) for a primitive model composed two H2O molecules (Fig. 6.4) is presented in order to understand the characteristics of hydrogen bonds. The ab initio simulation was carried out for intermolecular distances (interatomic distance between H(1) and O(2)) in the range of 0.1–0.7 nm. Figure 6.5 shows the interaction energy between the two H2O molecules (at 0 K) obtained by the simulation. The interaction energy exhibits a minimum value at intermolecular distance 0.19 nm. Because the experimentally measured average length of hydrogen bonds between H2O molecules is 0.177 nm, the simulation results are thought to be usable for interpreting the principle of the hydrogen bonding. Figure 6.6 shows the bond orders of H(1)–O(2) and O(1)–O(2) estimated by Mulliken’s method. The population analysis indicates that a bonding orbital is formed between H(1) and O(2) when these atoms approach to within ~ 0.3 nm. Simultaneously, O(2) exhibits an anti-bonding interaction with O(1). The anti-bonding interaction between oxygen atoms increases drastically with decrease of the intermolecular distance below 0.2 nm. The interaction energy between H2O molecules in the present models (shown in Fig. 6.4) is considered to be determined mainly by the balance of these bonding and anti-bonding interactions. Figure 6.7 shows the total electron density between the H2O molecules in the model at an intermolecular distance of 0.19 nm. In this figure, the bond with a weak covalent nature formed between H(1) and O(2) corresponds to the hydrogen bond.
6.4 Primitive model for ab initio molecular orbital simulation for analysis of the hydrogen bond between H2O molecules.
6.5 Interaction energy between H2O molecules at 0 K that was simulated using the model shown in Fig. 6.4.
6.6 Bond orders of H(1)-O(2) and O(1)-O(2) estimated by Mulliken’s method based on ab initio simulation results as a function of intermolecular distance.
6.7 Contour diagram of total electron density between H2O molecules at an intermolecular distance of 0.19 nm.
The mechanisms of the intermolecular interactions between H2O molecules can be briefly summarized as follows. Because the van der Waals contact distance between H and O is 0.272 nm (the van der Waal radii of H and O atoms are 0.120 and 0.152 nm, respectively), H2O molecules are considered to interact mainly by van der Waals forces at an intermolecular distance greater than ~ 0.3 nm. When the molecules approach more closely, hydrogen bonds with weak covalency are formed between neighboring H and O atoms. The bonding energy of hydrogen bonds is strongly limited by the anti-bonding interaction between O atoms of adjacent H2O molecules. The weak covalency is the important characteristic of hydrogen bonds.
Table 6.3 shows typical values of hydrogen bonding energy compared with other chemical and physical bonding energies.13 Hydrogen bonds generally exhibit a higher bonding energy than van der Waals bonds due to the covalency between H and electronegative species.
When hydrogen bonds are formed between the functional groups on a substrate surface and adhesive molecules, the contribution of the hydrogen bonds to adhesive energy is commonly included in the polar component.
By contrast, some researchers separate the effects of weak covalency in hydrogen bonds from the polar component. In this case, the intermolecular interactions are divided into three components including dispersive, polar and hydrogen bonding components.
6.3.3 Analysis of interfacial free energy using the intermolecular interaction concept
The discussion on the components of intermolecular interactions suggests that the interfacial free energy can be described as a function of these components.17 To analyze the interfacial free energy, Fowkes3,4 assumed that the surface free energy of substrate i is represented by Equation 6.18 using the dispersive component γid and the other components γix:
When material j that has only the dispersive component of surface free energy (γjd) forms an interface with substrate i, the interaction energy between i and j is assumed to be described using the Berthelot relation based on the regular solution approximation. Under these conditions, Equation 6.19 can be used to describe the interfacial free energy between i and j (γij):
Equations 6.4 and 6.19 are well known as the Fowkes’s equation. However, the Fowkes’s equation is applicable to interfacial interactions based solely on dispersive interaction.
Owen et al.5,6 and Kaelble et al.7 have taken the polar component into account as well as the dispersive component in their analyses of the interfacial free energy. According to their assumption, the interfacial free energy is written as Equation 6.5, using the additional term for the polar component. Furthermore, Kitazaki and Hata8 separated the hydrogen bonding component from the polar component, and expressed the interfacial free energy by Equation 6.6, using a supplementary term for the effect of hydrogen bonds. The Berthelot geometric average may not strictly represent the hydrogen bonding component because of the covalency in hydrogen bonds. However, they introduced supplementary terms that deal with the Berthelot relation for the hydrogen bonding component, to establish Equation 6.6.1 The components of the surface free energy for substrates and adhesives provide a useful guideline for the surface modification of substrates and the development of adhesives.
By contrast, intermolecular interactions at interfaces are expressed by another concept in the case of the acid–base theory,9 as expressed by Equation 6.7. The Lifshitz–van der Waals component (γLW) includes the dispersive, dipole–dipole and dipole–induced dipole interactions. The Lewis acid (γ+) and base (γ−) components represent the interaction ability of surfaces as electron acceptor and electron donor, respectively. In the acid–base theory, the intermolecular interactions are classified from the viewpoint of charge transfer between the adhesive molecules and functional groups on substrate surfaces. In this classification, the hydrogen bonding component can be apparently distinguished from the polar component including dipole–dipole and dipole–induced dipole interactions. Fowkes9 claimed that the acid–base interaction theory is adequate for describing the polarity or hydrophilicity of surfaces rather than the polar component introduced in the extended Fowkes’ s equations (6.5 and 6.6).
From the concept of interfacial interaction based on surface free energy, the maximum interfacial adhesive strength is considered to be ideally obtained when γSL = 0. In this ideal condition, the components of surface free energy are perfectly matched between adhesives and substrates (extended Fowkes’s theories: γSd = γLd, γSp = γLp, γSh = γLh; acid–base theory: γSLW = γLLW, γS+ = γL−, γS− = γL+). Thus, mismatch in the intermolecular components of surface free energy between adhesives and substrates should be decreased in order to improve the interfacial adhesive strength.
The experimental technique for estimation of the surface free energy of solid surfaces is introduced next. When the surface free energy is assumed to be composed of dispersive and polar components, the surface free energy of solids can be estimated using Equations 6.2 and 6.5, which lead to:
When we measure the contact angle for the solid surface using two different liquids that have known values for γL, γLd and γLp, the values for γSd and γSp can be estimated using Equation 6.20 (liquid drop method). The total value for the surface free energy (γS) is obtained by:
To estimate the surface free energy of solids composed of three components (dispersive, polar and hydrogen bonding components) using Equation 6.6, three different liquids are required for the contact angle measurement. The surface free energy can be similarly analyzed based on the acid–base theory (using Lifshitz–van der Waals, Lewis acid and Lewis base components) using Equation 6.7.
The results of surface free energy analysis by the liquid drop method for several polymer films are shown in Table 6.4. The surface free energy was separated into dispersive and polar components using H2O and CH2I2 as the probe liquids, by means of Equations 6.20 and 6.21. PTFE film exhibited a small surface free energy due to extremely small dispersive and polar components. The surface free energy of polyimide film was increased remarkably by soaking in 5 m NaOH aqueous solution for 30 s due to a significant increase in the polar component. Polar functional groups were effectively introduced on the surface of the polyimide film by the treatment with strongly alkaline solution (formation of carboxylic groups by breaking imide rings).18 The analysis of surface free energy possibly provides useful information on the chemical properties of substrate surfaces.
Table 6.4
Surface free energy of several polymer films estimated by the liquid drop method using H2O and CH2I2
The surface energies of metals and metalized surfaces can also be estimated by the liquid drop method. The experimentally estimated values of the surface energy of metals are strongly influenced by surface contamination, which can be categorized as inorganic or organic, depending on the nature of the adsorbate. O2 and H2O molecules are typical inorganic adsorbates. O2 molecules adsorbed on metal surfaces (physisorption) are often decomposed to atomic oxygen (O) (chemisorption), and in the case of metals with high oxidation potential, an oxide layer is formed spontaneously. H2O molecules can also chemisorb on metal surfaces to form hydroxyl groups on the surfaces. Physisorption of organic molecules on the surface of metals results in significant decrease of the polar component of the surface free energy.
Table 6.5 shows the variations in surface free energy of a Cu foil (used for conductive patterns on flexible substrates) depending on the surface finishing. The surface free energy was estimated by the liquid drop method using H2O and CH2I2. The Cu foil surface was plated with Ni (5 μm thickness), then Au flash plated (0.1 μm thickness). Although the Cu surface exhibited a low value of the polar component, the magnitude of the polar component increased significantly due to the Ni plating. The Au flash plating increased the dispersive component and decreased the polar component. The variation in surface free energy is considered to be related to characteristics of the adsorbates on the surfaces. In general, the surface free energy of noble metals (such as Au, Ag, Cu, Pt and Pd) is easily affected by the adsorption of organic molecules, and the surface free energies of the original and the Ni/Au plated Cu foil are likely to have been influenced by surface contamination with organic molecules.19,20 By contrast, the effect of organic adsorbates on the surface free energy of transition metals such as Ni is relatively small.19,20 Hence, the Ni-plated specimen exhibited a relatively higher value for the polar component.
Table 6.5
Variation of surface free energy of a Cu foil with surface finishing. The surface free energy was estimated by the liquid drop method using H2O and CH2I2
Metals are considered to have intrinsically high surface free energies due to the large polar component. Table 6.6 shows the surface free energy of mechanically polished Ag, Cu and Ni sheets before and after exposure to air at ambient temperature for 48 h. The surface free energy of these sheets significantly decreased during exposure to air, due to decrease in the polar component. Hence, surface cleaning techniques using physical and chemical methods are effective in improving the adhesion strength of adhesives on metal surfaces. The ultimate bonding process utilizing surfaces cleaned in vacuum chambers is the surface activated bonding (SAB) process developed by Suga and colleagues.21,22 This process enables direct metal/metal, metal/ inorganic and metal/organic substances bonding without adhesives at low temperatures.
6.3.4 Solubility parameters for analyzing intermolecular interactions
When novel adhesives are developed, the solubility parameters23 are often referred to in compounding the adhesives. Although solubility parameters do not directly relate to adhesive strength, they are useful for analyzing intermolecular interactions at interfaces as well as within adhesive compounds.
The cohesive energy density (c) of molecules (i) is defined by:
where ΔEiv and Vio are the energy of vaporization and molar volume of the molecule. Hildebrand proposed that the solubility parameter (δ) of a molecule can be written based on the regular solution approximation using the cohesive energy density, as:
The energy for mixing components 1 and 2 (ΔEmix) is described by Eq. 6.24:
The components exhibit good solubility when the magnitude of δ1 − δ2 is small, because ΔEmix consequently becomes small.
The solubility parameter (cohesive energy density) is related to intermolecular interactions in the same way as the surface free energy. Hansen23 assumed that the solubility parameter is composed of dispersive (δd), polar (δp) and hydrogen bonding (δh) components, as represented by:
In addition, some researchers have formalized the relationships between the surface free energy and the solubility parameter. Consequently, the solubility parameter is useful for designing adhesive compounds from the viewpoints of solubility of the components in adhesives, and the adhesive ability of the compounds on substrate surfaces.
The values of solubility parameter for many substances have been listed in the literatures.23–25 If values of the solubility parameter for particular substances are not available, they can be estimated using the group contribution methods23–25 that have been proposed by Hoy, van Kreveren and Hoftyzer, and others. The solubility parameters of the components of adhesives (main resin, diluents, curing agents and modifiers) should be determined using these methods before compounding the adhesives. The solubility parameter of a mixture is calculated by volume-wise summation of the solubility parameters of the individual components of the mixture.
6.3.5 Improvement of interfacial interactions using coupling agents
Adhesives usually adhere to substrate surfaces via intermolecular interactions such as dispersive, polar and hydrogen bonding interactions, as described previously. To improve the interfacial adhesion of adhesives to substrates and fillers, organometallic compounds containing Si, Ti, Al and Zr may be utilized as coupling agents. Typical molecular structures of coupling agents are illustrated schematically in Fig. 6.8. The molecules have functional groups both for hydrolytic reactions and for interactions with adhesive molecules. The groups for hydrolytic reactions transform to hydroxyl groups that can condense with hydroxyl groups on the surface of fillers and substrates. Consequently, the coupling agents are strongly fixed on the surfaces of fillers and substrates due to the condensation reactions.
There are two mechanisms for the interaction of adhesive molecules with substrate and filler surfaces modified by coupling agents. If the modified surface contains no functional groups that are reactive to the adhesive molecules, intermolecular interactions including dispersive, polar and hydrogen bonding interactions are induced between the molecules and modified surfaces. In this case, the interfacial interactions can be analyzed based on the concepts of the extended Fowkes’s equations for interfacial free energy, and the Hansen’s solubility parameters. Matching of the components between the adhesive molecules and the functional groups on the modified surfaces is the key to improving interfacial adhesion.
When the modified surfaces contain some functional groups that are reactive toward the adhesive molecules, chemical reactions can be promoted to form covalent bonds between the molecules and modified surfaces.2 For the formation of covalent bonds, the extended Fowkes’s equations are no longer valid. Here, the surface free energies of the modified surfaces and adhesive molecules are assumed to be given by Equation 6.21. The interfacial free energy with covalent bonding cannot be written as Equation 6.5.
The formation of covalent bonds completely deviates from the regular solution approximation that is the premise of the concept of solubility parameters, and consideration of interfacial interactions based on the solubility parameters is meaningless in this case. In fact, the bonding strength between adhesives and surfaces that are treated with silane coupling agents containing reactive functional groups is usually determined without regard to solubility parameters.
When coupling agents are utilized for improving interfacial adhesion, it is essential to consider the chemistry of the adhesive molecules and the functional groups of the coupling agents. If they do not exhibit a significant reactivity, the concepts of components (dispersive, polar and hydrogen bonding) of surface free energy and solubility parameter provide an effective guideline for improvement of interfacial adhesion. However, the reactivity between the adhesive molecules and the functional groups of coupling agents is the most important issue for selection of coupling agents when the interfacial adhesion is designed based on the formation of covalent bonds.
There are two methods of utilizing coupling agents. Fillers and substrates can be directly treated by dry and wet processes before mixing and bonding with adhesives, or the coupling agents can be pre-mixed into the adhesives (integral blend method). The integral-blended coupling agent molecules can diffuse in the adhesive paste to modify the surfaces of fillers and substrates. In this case, the coupling agents also contribute to modifying several properties of the adhesives. The integral blend method is often used for adhesives for electronics packaging.
6.3.6 Self-assembled monolayers (SAMs) for surface and interfacial modifications
In 1946, Zisman and colleagues26 reported monolayer adsorption of a surfactant on a metal surface, although the possibility of self-assembly was not recognized. Since Nuzzo and Allara27 demonstrated formation of SAMs of alkanethiolates on gold using dilute solutions (in 1983), various self- assembly systems have been investigated as shown in Table 6.7.
Table 6.7
| Surfactants | Substrates |
| R-SH, RS-SR’ | Au, Ag, Cu, Pt, Pd |
| R-SCN | Fe, GaAs, InP |
| R-CN | Pt, Pd, Au, Ag |
| R-COOH | Al2O3, AgO, CuO |
| R-SiH3 | Au |
| R-SiCl3 | SiO2, SnO2, TiO2 |
| R-SiOCH3, R-SiOC2H5 | GeO2, ZrO2, Al2O3 |
Figure 6.9 schematically illustrates the formation mechanism of SAMs on a substrate surface.28,29 When surfactant molecules adsorb on the surface by chemical interactions (reactions) of surface-active functional groups, ordered molecular assemblies are realized if intermolecular interactions adequately occur between alkyl groups of the molecules. The function of a surface covered with SAMs is determined by the chemical properties of the terminal group of the surfactant molecules. To improve interfacial bonding, several groups (such as carboxyl, amino, phosphoric, sulfonic, thiol) may be used as the terminal group of surfactants.
6.9 Schematic illustration of formation mechanism of self-assembled monolayers (SAMs) on a solid surface.
Some siliane coupling agents (alkylalkoxysilanes, alkylaminosilanes and alkylchlorosilanes) can form SAMs on metal and oxide surfaces under appropriate conditions.28,30,31 In the case of agents that have three active groups for the surface, siloxane bonds are formed between adjacent molecules as well as between molecules and substrate surface, as shown in Fig. 6.10, SAMs of such organosilicon compounds exhibit excellent mechanical, chemical and thermal stabilities due to the siloxane network formed on the surface.31
6.10 Schematic illustration of molecular structure of SAMs formed using organosilanes (R–SiX3; X = Cl, OCH3, OC2H5).
Recently, SAMs have attracted great interest for improving interconnect (mechanical and electrical) properties of adhesive joints. For example, Wong and colleagues32 have investigated the electrical properties of anisotropic conductive adhesive (ACA) joints when SAM compounds are introduced into the interface between the metal filler and the substrate bond pad. The ACA joints with SAM-treated conductive fillers and bond pads exhibited superior electrical properties to the non-treated joints.
6.4 Other influential factors determining bond strength of real adhesive joints
6.4.1 Internal stress generated in adhesives
Because adhesives always shrink during curing, residual stress (internal stress) is induced in the adhesive layer of joints.1 Figure 6.11 shows schematically the generation of internal stress in a simple joint. Substrates 1 and 2 are assumed to be not fixed when the adhesive layer shrinks. Although the adhesive layer shrinks isotropically, compressive stress is not directly induced in the vertical direction by the shrinkage because the substrates can move simultaneously with shrinkage of the adhesive. However, since the adhesive layer adheres to the substrates, shear stress is generated in the horizontal direction, to warp the joint as shown in Fig. 6.11. Consequently, compressive stress in the vertical direction is induced in the adhesive layer. The internal stress induced by the shrinkage of adhesives always decreases the bond strength of the joints.
The internal stress is considered to be generated through two steps during the adhesive curing process. In the case of thermosetting adhesives, adhesives shrink during the curing reaction (Fig. 6.12). In most cases, the curing process is performed at a higher temperature than the glass transition temperature (Tg) of fully cured adhesives. Thus, most of the internal stress generated in this step should be relaxed by micro-Brownian motion of polymer chains in the adhesive. The residual stresses after relaxation are accumulated as the internal stress in this step.
After curing the adhesive, a joint is cooled from the curing temperature to ambient temperature. During the cooling step, the adhesive layer of the joint undergoes significant shrinkage (cooling shrinkage), as shown in Fig. 6.12. In the temperature range below Tg of adhesives, there are no significant relaxation mechanisms to relax the stresses because the micro-Brownian motion of the polymer chains is frozen. Consequently, a large internal stress is induced during the cooling process from Tg to ambient temperature. The internal stress induced in this step is expressed by Equation 6.26:
where Ea and αa are the elastic modulus and linear expansion coefficient of the adhesive, respectively, and αs is the linear expansion coefficient of the substrate.
Equation 6.26 suggests a guideline for reducing the internal stress generated in the adhesive layer of the joints. Because the magnitude of internal stress is proportional to Ea and Δα = αa – αs, the stress can be reduced by decreasing at least one of these parameters. The magnitude of the internal stress is also reduced by decreasing Tg, but adhesives that have very low Tg are useless for high-temperature applications, for example, because a decrease in Tg results in decreased heat resistance of the adhesive (see in Section 6.5.1). Thus, to reduce internal stress, Ea and αa should be controlled without significant decrease in Tg of the adhesive.
The magnitude of Ea can be controlled by the adhesive chemistry, and αa can be decreased by mixing fillers with adhesives. However, introduction of fillers with high elastic modulus in adhesives simultaneously results in increased Ea, so that incorporation of fillers in adhesives is not always effective in decreasing internal stress. Furthermore, the interfacial adhesive strength significantly decreases when the adhesive contains an excessive amount of filler. Optimum material design is required for adhesives to control internal stress.
6.4.2 Mechanical behavior of adhesives and substrates
The fracture modes of adhesive joints are categorized into three modes, as shown schematically in Fig. 6.13. In the cohesive fracture mode, the bond strength of adhesive joints is governed by the fracture strength of the adhesives or the substrates.
When thermosetting adhesives (such as epoxy-based adhesives) are used for preparing joints, the degree of cure of the adhesives is a key factor for obtaining sufficient bonding strength.33 The degree of cure is usually defined as the extent of conversion of the curing reaction of the adhesive. of the methods for estimating the degree of conversion of adhesives, Fourier transform infrared (FTIR) spectroscopy and differential scanning calorimetry (DSC) are often utilized. Because the reactive functional groups in adhesive pastes can be measured quantitatively by FTIR spectroscopy, the degree of conversion is estimated by comparing the concentration of the reactive groups before and after curing. The heat of reaction that is released during curing is often detectable using DSC, and the degree of conversion can be estimated from the heat of reaction as an indicator for the curing reaction. Furthermore, kinetic analysis of the curing reaction of adhesives can be performed using DSC measurements to predict optimum curing conditions.
The values of Tg for adhesives often show a one-to-one relationship with the degree of conversion.34,35 Figure 6.14 gives experimental Tg and degree of conversion data for an epoxy-based isotropic conductive adhesive (ICA), as an example.36 In this case, the values of Tg are determined by the degree of conversion, regardless of curing temperature, hence the degree of conversion can be estimated by measurement of Tg.
6.14 Relationship between degree of conversion (α) and glass transition temperature (Tg) in an epoxy-based isotropic conductive adhesive (ICA) cured at several temperatures.
An experimental result for the relationship between peel strength of adhesive joints and degree of conversion of adhesives37 is shown next. Figure 6.15 shows the 90° peel strength of model joints composed of ITO glass substrate/ epoxy-based anisotropic conductive film (ACF)/polyimide flex bonded at 180 °C. The peel test was performed at a test speed of 8.33 μm s− 1. The peel strength of the ACF joints increased significantly with increasing degree of conversion, and a drastic increase in peel strength observed at 40–60% conversion coincided with a significant increase in Tg. Hence, the drastic increase in peel strength is considered to be caused by the formation of a cross-linked polymer structure in the adhesive binder. The peel strength of the ACF joints increased further above 80% conversion.
6.15 The 90° peel strength of model joints composed of ITO glass substrates and polyimide-based flex with an interdigitated Cu pattern, bonded using an epoxy-based anisotropic conductive film (ACF) at 180 °C for several bonding times. The peel test was conducted in the direction parallel to the Cu pattern at a test speed of 8.33 μm s− 1.
The mechanical behavior of adhesives usually gives a characteristic test speed dependence of the bonding strength of adhesive joints. As Andrews and colleagues14,15 showed, adhesives exhibit significant viscoelastic deformation during the peel test of adhesive joints. The viscoelasticity of adhesives can affect the bonding strength of adhesive joints. Because the viscoelastic deformation of adhesives is a time-dependent phenomenon, adhesive joints usually exhibit higher bonding strength with increasing test speed of peel tests. The bonding strength is decreased by the stress relaxation due to viscoelastic deformation of adhesives when the peel test is conducted at slow test speed. Figure 6.16 shows the test speed dependence of a 90° peel strength of the model (epoxy-based) ACF joints between polyimide-based flex and ITO glass substrate bonded at 180 °C for 15 s (82.7% degree of conversion of the adhesive binder), as an example.
6.16 Test speed dependence of 90° peel strength of model ACF joints composed of ITO glass substrate and polyimide-based flex bonded at 180 °C for 15 s.
In addition to the effect of test speed on bonding strength of adhesives, there is also an effect of stressing-mode,1 which is discussed in Section 6.5.1.
6.4.3 Relationship of physical factors at the interface to bonding strength
Physical effects (anchoring effects) that contribute to improving the bonding strength of adhesive joints can be realized when substrates have microscopic and macroscopic morphologies (anchor patterns) on their surface. The magnitude of the anchoring effects is known to depend on the geometrical factor of the anchor patterns.
Experimental peel test data for an epoxy-based ACF (anisotropic conductive film) joint37 have been shown to explain the contribution of anchoring effects to the bonding strength. In these experiments, a polyimide-based double layer flex with an 18 μm thick interdigitated Cu pattern was used for preparing model joints with ITO glass substrates (Fig. 6.17a). A large number of dimples with dimensions of ~ 1 μm were also formed on the polyimide surface of the flex (Fig. 6.17b).
6.17 Polyimide-based flex that was used for preparing model ACF joints. (a) Interdigitated Cu pattern, (b) dimples formed on the surface of the polyimide-based film.
It was established by cross-sectional SEM observation of the model joints that the adhesive binder had completely infiltrated into the micro-dimples on the polyimide surface (see Fig. 6.18). As a consequence, the micro-dimples were likely to contribute to increased bonding strength of the joints due to an anchoring effect.
6.18 Cross-sectional SEM micrograph of the bonding interface between ACF and polyimide in the model joint.
In addition, the macroscopic Cu circuit pattern appeared to exhibit an anchoring effect. Figure 6.19a shows peel strength profiles (measured at a test speed of 8.33 μm s− 1) of joints bonded at 180 °C for 15 s, in directions parallel and perpendicular to the interdigitated Cu pattern. The peel strength in the direction parallel to the Cu pattern was significantly higher than that in the perpendicular direction. During the peel test in the parallel direction, periodic oscillation of the peel strength was clearly observed (Fig. 6.19b), and was correlated with the interdigitated Cu pattern, since the period of the oscillation coincided with the Cu pattern. The peel strength profiles indicate that an anchoring effect due to the Cu pattern was effectively obtained in the parallel direction. Thus, anchoring effects due to the surface morphology of the flex can be induced cooperatively by the microscopic dimples on polyimide and the macroscopic Cu pattern.
6.19 (a) The 90° peel strength profiles of the model ACF joints (bonded at 180 °C for 15 s) in the directions parallel and perpendicular to the interdigitated Cu pattern measured at test speed 8.33 μm s− 1. (b) Periodic oscillation of the peel strength observed during the peel test in the direction parallel to the interdigitated Cu pattern.
Figure 6.20 shows peel strength profiles (measured at a test speed of 8.33 μm s− 1) in the parallel direction for ACF joints bonded at 140, 180 and 200 °C for 15 s. Because the degree of cure of the adhesive binder increased with increasing bonding temperature (9.3, 82.7 and 92.5% at 140, 180 and 200 °C, respectively), the joints bonded at higher temperatures provided higher peel strengths. The periodic oscillation due to the Cu pattern was observed even in the joints bonded at 140 °C, but the magnitude of the oscillations of peel strength was quite small in those joints. Hence, adhesives are required to have sufficient strength in order to obtain significant anchoring effects.
6.20 The 90° peel strength profiles of the model ACF joints bonded at 140, 180, and 200 °C for 15 s, in a direction parallel to the interdigitated Cu pattern. The peel test was conducted at a test speed of 8.33 μm s− 1.
The surface morphology of the substrates is a necessary factor for realizing anchoring effects.38 However, good wetting of adhesives at the interface and sufficient mechanical strength of adhesives and substrates are also needed to obtain significant magnitude of anchoring effects.
6.5 Effect of environmental factors
6.5.1 Temperature dependence of shear and peel strengths of adhesive joints
The bonding strength of adhesive joints is significantly different for different stressing modes. Figure 6.21 shows a schematic illustration of the relationship between elastic modulus of adhesives and bonding strengths (shear and peel strengths) of adhesive joints.1 The shear strength of the joints tends to increase with increasing elastic modulus of the adhesives. By contrast, the peel strength of the joints exhibited a maximum value at a low elastic modulus. To obtain high peel strength of the joints, the stress relaxation ability of adhesives is the most important factor. The relationship between elastic modulus and bonding strength illustrated in Fig. 6.21 should be taken into account in the design of adhesive joints.
6.21 Schematic illustration of variations in shear and peel strength of adhesive joints with elastic modulus of adhesives.
To interpret the temperature dependence of the bonding strength, the elastic modulus dependence shown in Fig. 6.21 provides useful information. Shear strength of adhesive joints significantly decreases with decreasing elastic modulus (storage modulus) of adhesives in the temperature range around Tg of the adhesives, as illustrated in Fig. 6.22a. By contrast, the peel strength of joints and the loss modulus of adhesives exhibit a maximum value around Tg of adhesives since the stress relaxation effect of adhesives is significantly induced as shown in Fig. 6.22b.
6.22 Schematic illustration of temperature dependence of (a) shear strength and (b) peel strength of adhesive joints.
If high shear strength is required to be realized together with high peel strength in the same adhesive joint, highly-elastic polymer-based composites containing rubber (elastomeric) fillers are good candidates. Fortunately, the rubber fillers also contribute to decreasing the internal stress generated during curing and cooling.
6.5.2 Effects of moisture absorption
General remarks on moisture absorption of adhesive joints
Moisture absorption in the adhesive layer is a serious problem in forming reliable adhesive joints. The amount of moisture absorption generally increases with increasing concentration of polar functional groups in adhesives. The polymer structure of adhesives contains free volume, as shown in Fig. 6.23a. The free volume spaces are considered to be divided into two categories namely micro-voids (which are different from large voids or pores with sizes greater than a few hundred nanometers) and more narrow spaces due to the density variation of polymer chains (Fig. 6.23b).39 H2O molecules can diffuse into the free volume spaces of adhesives from the surrounding environments. Since a large number of micro-voids are considered to exist at the interface between adhesive and substrate rather than within the adhesive layer, the H2O molecules predominantly diffuse along the interface.
Figure 6.24 shows the near-infrared (NIR, 7500–6000 cm− 1) absorption spectrum of an epoxy-based ACF exposed to an 85 °C/85%RH environment for 1000 h. From this spectrum, several types of H2O molecules are suggested to exist in the free volume spaces. The molecules can exist both as free molecules (S0, ~ 7075 cm− 1) and as molecules hydrogen-bonded to polymer chains in the adhesive.40,41 The free molecules are likely to exist in the micro-voids. The bound molecules can be divided into S1 (with one bond, ~ 6820 cm− 1) and S2 (with two bonds, ~ 6535 cm− 1) species depending on the number of hydrogen bonds.
6.24 Near-infrared absorption spectrum of an epoxy-based ACF exposed to an 85 °C/85%RH environment for 1000 h.
Various properties, including mechanical and electrical properties, of adhesive joints are influenced by the H2O molecules that diffuse into the adhesives. For example, the bound H2O molecules can affect mechanical properties of the joints since the intermolecular interactions are weakened at the interfaces as well as within the adhesive layer. In addition, the free H2O molecules in micro-voids are considered to form a continuous channel within adhesives. Because chemical species such as ions and oxygen molecules can diffuse through the channel, chemical and electrochemical phenomena are induced in the adhesive joints due to exposure to humid environments. This section focuses on the effects of moisture absorption on the mechanical properties of adhesives.
Plasticization42 of adhesives is always induced by moisture absorption via diffusion of H2O molecules into the narrow free volume spaces as well as the micro-voids. The elastic modulus and Tg of adhesives are decreased, depending on the amount of moisture absorption. By contrast, the ability of adhesives to undergo stress relaxation is considered to be improved somewhat. The plasticization effect of adhesives usually disappears reversibly by drying the adhesives.
H 2O molecules that diffuse along the interfaces43 can break the intermolecular interactions between adhesive molecules and functional groups on substrate surfaces, if the H2O molecules create hydrogen bonds with the groups on the substrate surface. However, the adhesive molecules reversibly recover the intermolecular interactions with substrate surfaces by drying the adhesive joints, thereby removing H2O molecules.2 When covalent bonds are introduced at the interfaces by using silane coupling agents, the adhesive joints are often reported to exhibit high bonding strength even during and after exposure to humid conditions.
Moisture absorption sometimes promotes irreversible chemical and electrochemical reactions at interfaces. For example, metal surfaces such as Cu and Ni can be oxidized and hydroxylated during exposure to humid environments. If such irreversible reactions occur at a substrate surface, the bonding strength is not recovered, even after removing H2O molecules.
When variations in the bonding strength of adhesive joints are discussed, the stressing modes should be borne in mind. The shear strength of joints is significantly decreased by plasticization and interfacial weakening due to moisture absorption. By contrast, the peel strength of joints is sometimes increased by enhancement of the ability of adhesives to undergo stress relaxation, accompanied by plasticization due to the moisture absorption.
Variation of mechanical properties of an Anisotropic conductive film (ACF) joint due to moisture absorption
Variation in the bonding strength of adhesive joints due to moisture absorption may be quite complex, depending on the situation. Experimental data for variation of the mechanical properties of ACF joints composed of ITO glass substrate and polyimide-based flex, that were referred to in Sections 6.4.2 and 6.4.3, due to exposure to an 85 °C/85%RH environment are shown as an example.
First, the moisture absorption behavior of the ACF specimens is explained. Figure 6.25 shows the increase in weight due to moisture absorption for non-cured and fully-cured ACF during exposure to the 85 °C/85%RH environment. The weight increment was almost saturated in the first few hours of exposure. Although the curing reaction of the adhesive binder was promoted in the non-cured specimen during exposure, the magnitude of moisture absorption increased with decreasing degree of conversion before exposure.
6.25 Variation in weight of epoxy-based ACF specimens during the exposure to the 85 °C/85%RH environment. The ACF specimens were uncured and fully cured before exposure.
Variations in the values of Tg for the ACF specimens bonded at 140 and 180 °C for 15 s during the exposure to the 85 °C and 85 °C/85%RH environments for 1000 h are shown in Fig. 6.26. The specimens were cured during the bonding process at 140 and 180 °C to 9.3 and 82.7% conversion, respectively. These specimens were subsequently post-cured for the first 100 h of exposure to these environments. It is apparent that the curing reaction (increase in Tg) was similarly promoted during exposure to these environments, although moisture absorption occurred significantly in the 85 °C/85%RH environment. The post-curing reaction of the adhesive binder occurred independently of moisture absorption during the early stages of exposure. The implication is that H2O molecules predominantly diffused into the micro-voids during this period. Plasticization (decrease in Tg) of the specimens became significant in the later stages (more than ~ 200 h) of exposure.
6.26 Variation of the glass transition temperature of epoxy-based ACF specimens bonded at 140 and 180 °C for 15 s, during exposure to an 85 °C/85%RH environment. The degree of conversion of the specimens bonded at 140 and 180 °C was 9.3 and 82.7%, respectively, before the exposure.
Next, the effects of moisture absorption on the peel strength of the ACF joints are discussed. Figure 6.27 shows the 90° peel strength of ACF joints composed of ITO glass substrate and polyimide-based flex covered with Cu foil (with no interdigitated pattern), that were bonded at 180 °C for 15 s, as a function of exposure time to the 85 °C/85%RH environment. In the early stages (~ 100 h) of exposure, the peel strength of the ACF joints decreased although post-curing was promoted. By contrast, the peel strength increased during exposure for more than 200 h. The increase in the peel strength is thought to be caused by the increase in stress relaxation ability of the adhesive binder due to plasticization, because the peel strength significantly decreased after drying at ambient temperature in a vacuum chamber, as shown in Fig. 6.27.
6.27 Variation in 90° peel strength of the ACF joints composed of ITO glass substrate and polyimide-based flex covered with a Cu foil (without any patterns) bonded at 180 °C for 15 s, during exposure to an 85 °C/85%RH environment. The peel strength of the joints after drying in a vacuum chamber is also shown.
Figure 6.28 shows the 90° peel strength of the ACF joints (bonded at 140 and 180 °C for 15 s) composed of ITO glass substrate and polyimide- based flex with a Cu interdigitated pattern in the direction parallel to the Cu pattern, as a function of exposure time to the 85 °C/85%RH environment. In the early stages of exposure, the peel strength of the joints significantly increased due to the post-curing of the adhesive binder. Subsequently, the peel strength exhibited an almost constant value regardless of bonding temperature and exposure environment. In this case, the anchoring effects due to the Cu pattern dominated the peel strength of the ACF joints, as discussed in Section 6.5.3. Consequently, the effects of the moisture absorption were not reflected clearly in the peel strength. The geometrical factor of adhesion interfaces is also important, in consideration of the mechanical reliability of adhesive joints.
6.6 Interconnections using electrically conductive adhesives
In the case of adhesives for interconnections, electrical conductivity is required to be ensured as well as mechanical bonding. This section discusses the relationship between mechanical bonding and electrical conductivity in adhesives for making interconnections.
The mechanism for electrical conductivity in electrically-conductive adhesives (ECAs) including ICAs and ACAs is discussed first. Figure 6.29 shows a schematic illustration of the cross-sectional microstructure of ICAs, which usually contain 40–50 vol% of metallic fillers (such as Ag) in order to form a percolation network that provides a conduction path for electrons. The electrical resistance of ICAs is divided into two components, namely intra-particle and interfacial (filler/filler and filler/electrode) resistance.44,45 The interfacial resistance is further categorized into constriction and tunneling resistance.
During the curing process, the adhesive binder of the ICA adheres to the substrate surfaces. In addition, the curing and cooling shrinkages of the adhesive binder result in generation of internal stress. The mechanical bonding strength of the ICA joints is determined by these processes, and the conductive percolation network composed of metallic fillers is formed simultaneously.46,47 Figure 6.30 shows the result of in situ monitoring of electrical conduction in an epoxy-based ICA specimen during curing (at 130 °C for 3000 s) and cooling (air cooling). Prior to curing, the electrical current was difficult to measure using this apparatus because the ICA had high electrical resistance. The specimen showed a significant decrease in electrical resistance (increase in electrical current in Fig. 6.30) during heating at the curing temperature. The onset of decrease in electrical resistance (for curing time ~ 300 s) corresponds to 70–80% conversion from the viewpoint of the curing reaction of the adhesive binder. It is considered that the internal stress caused by the curing shrinkage began to be accumulated in the adhesive layer during this conversion range, because the stress-relaxation ability of the adhesive binder decreased due to the development of a cross-linked polymer structure in the binder. The internal stress (compression stress) can contribute to a decrease in the interfacial resistance, such as tunneling resistance, to generate electrical conductivity in the ICA specimen. Furthermore, decrease in the electrical resistance is also observed during the cooling process. The interfacial resistance of the ICA specimen is considered to be decreased by the internal stress generated by the cooling shrinkage of the adhesive.
6.30 Result of in situ monitoring of electrical conduction in an epoxy- based ICA specimen during curing (at 130 °C for 3000 s) and cooling (air cooling) processes.
Next, the generation of electrical conductivity during the bonding process of ACF is explained. Figure 6.31 shows a schematic illustration of the cross- sectional microstructure of ACF. The conductive particles (such as metal-coated polymer balls) are sandwiched between electrodes during the bonding process. Figure 6.32 shows the result of in situ monitoring of electrical conduction in a model (epoxy-based) ACF joint composed of ITO glass substrate and polyimide-based flex during the bonding (180 °C, 1.5 MPa, 15 s) and cooling (air cooling) processes. Significant plastic flow of adhesive binder and capture of conductive particles between the electrodes occurred concurrently just after starting the bonding process (~ 2 s). The conductive particles captured between the electrodes were significantly deformed to make conductive contacts with the electrodes during this period. Subsequently, the adhesive binder was cured to fix the conductive contacts between the particles and electrodes. After removing the tool from the specimen, the electrical current gradually increased during the cooling process. This increase in electrical current indicates that the interfacial resistance decreased, which is attributed to generation of internal stress during the cooling shrinkage of the adhesive binder.
6.32 Result of in situ monitoring of electrical conduction in an epoxy- based ACF joint composed of ITO glass substrate and polyimide- based flex during the bonding (at 180 °C for 15 s with1.5 MPa applied pressure) and cooling (air cooling) processes.
Based on the above discussions, the relationship between electrical conductivity and bonding strength of ICA and ACA joints is now considered. The electrical conductivity of ICAs is significantly influenced by the internal stress induced in the adhesive binder during the curing and cooling processes. In the case of ACAs, the elastic recovery stress of conductive particles captured between the electrodes also contributes to decreasing the contact resistance between the particles and electrodes, as well as the internal stress induced in the adhesive binder. Although these stresses are essential to obtain electrical interconnections, at the same time they have a negative effect on the bonding strength of the joints. The balance between these stresses and the adhesion force is the key to realizing high (mechanical and electrical) reliability of the adhesive joints, as shown in Figures 6.29 and 6.31.
6.7 Conclusions
The theoretical basis and influential factors for adhesion and bonding strength of adhesive joints have been explained in detail with the aim of formulating guidelines for development of advanced adhesives for electronics packaging applications.
The bonding strength of joints is determined not only by the thermodynamic work of adhesion, but also by various factors such as the viscoelastic deformation of adhesives. However, the role of interfacial adhesion strength is still important for realization of high bonding strength, as Andrews and colleagues established.
Interfacial adhesion usually occurs by intermolecular interactions, including van der Waals forces and hydrogen bonding. In such cases, the concepts of surface free energy and solubility parameter based on dispersive, polar and hydrogen bonding components are effectively applied for designing the chemical composition of adhesives. By contrast, these concepts are not valid when covalent bonds are introduced at interfaces using coupling agents. Since the common theories are not established for the introduction of covalent bonds at interfaces, individual considerations are required for the selection of reactive coupling agents from the viewpoint of reactivity of their functional groups. In any case, the chemistry of adhesive compounds is one of the most important subjects in developing advanced adhesives. The concepts of interfacial chemistry are also applicable to molecular adhesion in nanotechnology, such as with self-assembled monolayers (SAMs).
In addition to the chemistry of interfacial adhesion, modifying the geometry of interfaces to significantly induce the anchoring effect plays an important role in determining the bonding strength of adhesive joints.
Because the viscoelastic deformation of adhesives and the anchoring effect are effectively obtained when the adhesives are sufficiently cured, the curing process should be controlled in order to achieve higher bonding strength. Kinetic analysis for the curing reaction provides useful information for control of the process.
By contrast, internal stress due to the shrinkage of adhesives generated during the curing and cooling processes, has a negative effect on bonding strength, although the stress is essential to secure electrical conductivity in ECAs. Hence, a balance between the interfacial adhesive force and the internal stress should be sought to obtain highly reliable joints.
Microsystems will continue to be further diversified in the future. Consequently, the importance of advanced adhesives will increase in electronics packaging technology, and the design concepts for adhesive compounds based on the scientific fundamentals that are discussed in this chapter will be increasingly important for engineers and scientists in this field.
6.8 References
1. Japan Adhesion Society. Setchaku Handbook, 4th ed. Tokyo: Nikkan Kougyo Shinbun-sha, 2007. [ibid. 2nd ed. (1980)].
2. Adams, R.D. Adhesive Bonding: Science Technology and Applications. Cambridge, UK: Woodhead Publishing Ltd; 2005.
3. Fowkes, F.M. Determination of interfacial tensions, contact angles, and dispersion forces in surfaces by assuming additivity of intermolecular interactions in surfaces. J. Phys. Chem. 1962; 66:382.
4. Fowkes, F.M. Attractive forces at interfaces. J. Ind. Eng. Chem. 1964; 56:40–52.
5. Owens, D.K., Wendt, R.C. Estimation of the surface free energy of polymers. J. Appl. Polm. Sci. 1969; 13:1741–1747.
6. Owens, D.K. Some thermodynamic aspects of polymer adhesion. J. Appl. Polym. Sci. 1970; 14:1725–1730.
7. Kaelble, D.H., Uy, K.C. Reinterpretation of organic liquid-polytetrafluoroethylene surface interactions. J. Adhesion. 1970; 2:50–60.
8. Kitazaki, Y., Hata, T. Extension of Fowkes’ equation and estimation of surface tension of polymer solids. J. Adhesion Soc. Jpn. 1972; 8:131–142.
9. Fowkes, F.M.Mittal K.L., Anderson H.R., Jr., eds. Acid–Base Interactions. 1991:93–115. [VSP, Utrecht].
10. Kitaura, K., Morokuma, K. A new energy decomposition scheme for molecular interactions within the Hartree–Fock approximation. Int. J. Quantum Chem. 1976; 10:325–340.
11. Minamizaki, Y., Tanaka, Y., Kobayashi, K. Molecular orbital calculations of interfacial energy in adhesion. J. Adhesion Soc. Jpn. 2004; 40:282–288.
12. Ikada, Y., Matsunaga, T. Surface energy of polymers. J. Adhesion Soc. Jpn. 1978; 14:427–434.
13. Takemoto, K., Mitou, M. Setchaku no Kagaku. Tokyo: Kodansha; 1997.
14. Andrews, E.H., Kinloch, A.J. Mechanics of adhesive failure. I, Proc. R. Soc. Lond. A. 1973; 332:385–399.
15. Kamyab, I., Andrews, E.H. Interfacial and bulk contributions to peeling energy. J. Adhesion. 1996; 56:121–134.
16. Atkins, P., du Paula, J. Atkins’ Physical Chemistry, 8th ed. Oxford: Oxford University Press; 2006.
17. Kloubek, J. Development of methods for surface free energy determination using contact angles of liquids on solids. Advances in Colloid and Interface Science. 1992; 38:99–142.
18. Akamatsu, K., Ikeda, S., Nawafune, H. Site-selective direct silver metallization on surface-modified polyimide layers. Langmuir. 2003; 19:10366–10371.
19. Kawai, A., Kumagai, T., Takata, M. Dispersion and polar components of surface free energy of transition metal films measured by contact angle method in air. J. Adhesion Soc. Jpn. 1995; 31:307–311.
20. Kawai, A. J., Surface free energy variation of sputtering-formed Au and Pt thin films after exposure to air. Adhesion Soc. Jpn. 1996; 32:39–43.
21. Suga, T., Takahashi, Y., Takagi, H., Gibbesch, B., Elssner, G. Structure of Al-Al and Al-Si3N4 interfaces bonded at room temperature by means of the surface activation method. Acta Metall. Mater. 40, 1992. [Suppl., S133–S137].
22. Takagi, H., Kikuchi, K., Maeda, R., Chung, T.R., Suga, T. Surface activated bonding of silicon wafers at room temperature. Appl. Phys. Lett. 1996; 68:2222–2224.
23. Hansen, C.M. Hansen Solubility Parameters, A User’s Handbook, 2nd ed. Boca Raton: CRC Press; 2007.
24. Van Kreveken, D.W. Properties of Polymers. Amsterdam: Elsevier, 1990; 189–225.
25. Barton, A.F.M. CRC Handbook of Solubility Parameters and Other Cohesion Parameters, 2nd ed. Boca Raton: CRC Press; 1991.
26. Bigelow, W.C., Glass, E., Zisman, W.A. Oleophobic monolayers; temperature effects and energy of adsorption. J. Colloid Interface Sci. 1947; 2:563–591.
27. Nuzzo, R.G., Allara, D.L. Adsorption of bifunctional organic disulfides on gold surfaces. J. Am. Chem. Soc. 1983; 105:4481–4483.
28. Ulman, A. Formation and structure of self-assembled monolayers. Chem. Rev. 1996; 96:1533–1554.
29. Smith, R.K., Lewis, P.A., Weiss, P.S. Patterning self-assembled monolayers. Prog. Surf. Sci. 2004; 75:1–68.
30. Wirth, M.J., Fairbank, R.W.P., Fatunmbi, H.O. Mixed self-assembled monolayers in chemical separations. Science. 1997; 275:44–47.
31. Hayashi, K., Sugimura, H., Takai, O. Frictional properties of organosilane self- assembled monolayer in vacuum. Jpn. J. Appl. Phys. 2001; 40:4344–4348.
32. Li, Y., Moon, K.-S., Wong, C.P. Adherence of self-assembled monolayers on gold and their effects for high-performance anisotropic conductive adhesives. J. Electron. Mater. 2005; 34:266–271.
33. Uddin, M.A., Alam, M.O., Chan, Y.C., Chan, H.P. Adhesion strength and contact resistance of flip chip on flex packages – Effect of curing degree of anisotropic conductive film. Microelectron. Reliab. 2004; 44:505–514.
34. Hale, A., Macosko, C.W., Bair, H.E. Glass transition temperature as a function of conversion in thermosetting polymers. Macromole. 1991; 24:2610–2621.
35. Pascault, J.P., Williams, R.J.J. Glass transition temperature versus conversion relationships for thermosetting polymers. J. Polym. Sci.: Part B. 1990; 28:85–95.
36. Inoue, M., Suganuma, K. The dependence on thermal history of the electrical properties of an epoxy-based isotropic conductive adhesive. J. Electron. Mater. 2007; 36:669–675.
37. Inoue, M., Suganuma, K. Influential factors in determining the adhesive strength of ACF joints. J. Mater. Sci.: Mater. Electron. 2009; 20:1247–1254.
38. Arrowsmith, D.J. Adhesion of electroformed copper and nickel to plastic laminates. Trans. Inst. Metal Finishing. 48(88–92), 1970.
39. Tsujita, Y. Gas sorption and permeation of glassy polymers with microvoids. Progress in Polym. Sci. 2003; 28:1377–1401.
40. Musto, P., Ragosta, G., Scarinzi, G., Mascia, L. Probing the molecular interactions in the diffusion of water through epoxy and epoxy–bismaleimide networks. J. Polym. Sci.: Part B. 2002; 40:922–938.
41. Musto, P., Ragosta, G., Mascia, L. Vibrational spectroscopy evidence for the dual nature of water sorbed into epoxy resins. Chem. Mater. 2000; 12:1331–1341.
42. Nogueira, P., Ramirez, C., Torres, A., Abad, M.J., Cano, J., Lopez-Bueno, J., Barral, L. Effect of water sorption on the structure and mechanical properties of an epoxy resin system. J. Appl. Polym. Sci. 2001; 80:71–80.
43. Zanni-Deffarges, M.P., Shanahan, M.E.R. Diffusion of water into an epoxy adhesive: Comparison between bulk behaviour and adhesive joints. Int. J. Adhesion and Adhesives. 1995; 15:137–142.
44. Li, L., Lizzul, C., Kim, H., Sacolick, I., Morris, J.E. Electrical, structural and processing properties of electrically conductive adhesives. IEEE Trans. Components, Hybrids, and Manufacturing Technol. 1993; 16:843–851.
45. Morris, J.E.Liu J., ed. Conductive Adhesives for Electronics Packaging. Electrochemical Publications: Port Erin, 1999; 36.
46. Lu, D., Wong, C.P. Effects of shrinkage on conductivity of isotropic conductive adhesives. Int. J. Adhesion and Adhesives. 2000; 20:189–193.
47. Inoue, M., Suganuma, K. Effect of curing conditions on the electrical properties of isotropic conductive adhesives composed of an epoxy-based binder. Soldering and Surface Mount Technology. 2006; 18:40–46.