Interface Between SiO 2 and Commercial Epoxy

The extent of adhesion strength degradation at epoxy interfaces during hygrothermal aging has been correlated to the hydrophobicity of the interface materials. It was previously reported that the die shear strengths of hydrophilic SiO2/epoxy and Si3N4/epoxy interfaces

decreased to almost zero after 24 h of exposure to a 121 °C, 100% relative humidity (RH) environment.8 However, relatively more hydrophobic interfaces such as benzocyclobutene/epoxy and polyimide/epoxy interfaces maintained about 32% and 21%, respectively, of their original adhesion strength after the same treatment.8 The loss of adhesion strength at the hydrophilic interfaces was attributed to an increased number of interfacial water molecules that broke the original adhesive hydrogen bonds, dipole–dipole interactions, and covalent bonds between the epoxy underfill and the substrate; however, no molecular-level evidence was reported. Adhesion strength loss at the more hydrophobic interfaces was likely due to the moisture-induced degradation of the epoxy elastic modulus. Absorbed water can reduce the elastic modulus of epoxy by acting as a plasticizer and by hydrolyzing bonds.48 Plasticization and hydrolysis reactions in epoxy can both result in molecular structural changes of the epoxy polymer chains, which in turn can alter the molecular structure of adhesive interfaces, so the interfacial and bulk effects of epoxy moisture uptake are often related.

In this study, we probed buried epoxy molecular structures at interfaces with varying degrees of hydrophobicity. We first collected SFG spectra from a hydrophilic SiO2/commercial

SFG spectrum of the untreated SiO2/epoxy interface is shown in Figure 4.3a. One feature was

observed near 2950 cm–1 and can be assigned to the Fermi resonance of the BADGE methyl group. The detection of the SFG signal originating from a methyl vibrational mode suggests that methyl groups were initially ordered at the SiO2/epoxy interface. The ATR-FTIR spectrum of the

untreated SiO2/epoxy interfacial region is shown in Figure 4.3b. One broad feature was observed

in the 2800–3000 cm–1 region with peaks near 2870, 2930, and 2965 cm–1 that can be assigned to methyl symmetric stretching, methylene asymmetric stretching, and methyl asymmetric stretching modes, respectively, in the epoxy bulk.49

The SFG spectrum collected from the SiO2/epoxy interface after 24 h of hygrothermal

aging is shown in Figure 4.3a. Two features near 2875 and 2935 cm–1 and one broad feature in the 3000–3400 cm–1 region were observed and can be assigned to the epoxy methyl symmetric stretch, epoxy methyl Fermi resonance, and water, respectively. New features in the spectrum suggest that molecular structural changes took place because of hygrothermal aging. After 24 h of hygrothermal aging, the lap shear adhesion strength of the SiO2/epoxy interface decreased to

nearly zero. The adhesion strength of untreated SiO2/epoxy could not be quantitatively measured

using the lap shear method because the adhesive strength was stronger than the cohesive strength of fused silica. However, the dramatic decrease in adhesion strength observed as a result of hygrothermal aging is consistent with previous reports that studied the interface between epoxy underfill and silicon dioxide passivation.8 After 48 h of hygrothermal aging, the SFG spectrum again contained features near 2875 and 2935 cm–1 and a broad peak in the 3000–3500 cm–1 region with the center near 3175 cm–1.

Interfacial SFG water signals with peak centers near 3200 cm–1 are contributed by strongly hydrogen bonded water or icelike water, as suggested in previous research.39 The

presence of an SFG feature centered near 3200 cm–1 suggests that water diffused to hydrogen bonding sites at the SiO2/epoxy interface. Although the presence of interfacial water has been

correlated to low friction at buried sapphire/poly(dimethylsiloxane) (PDMS)39 and quartz/poly(vinyl alcohol) (PVA)40 interfaces in previous SFG studies, the effect of interfacial ordered water on the adhesion strength has not, to our knowledge, been reported. It is interesting that the ATR-FTIR spectra collected from the SiO2/epoxy interface after 24 h of aging contained

a broad peak from 2970 to 3700 cm–1 that is centered at 3400 cm–1, as dominated by contributions from relatively weakly hydrogen bonded water. Such infrared features in the epoxy bulk have been attributed to water located in small pores within the epoxy bulk.50 The broad feature of the IR signal indicates that some strongly hydrogen bonded water molecules were also present, which has been attributed to strong hydrogen bonding interactions between water and hydrophilic epoxy sites, leading to a red shift in the water OH oscillator vibrational frequency.50

The SFG signal is sensitive to the degree of ordering of functional groups. The SFG water signal collected from a buried interface thus suggests that ordered water is present at the interface. One of the driving forces for water ordering at the buried SiO2/epoxy interface is the

formation of hydrogen bonds. Ordered water detected at the buried interface likely replaced original adhesive hydrogen bonding sites between the terminal hydroxyl groups on the SiO2 and

polar functional groups on the epoxy that reduced the interfacial adhesive strength. The presence of strongly hydrogen bonded water at the SiO2/epoxy interface suggests that hydrogen bonds are

one of the primary contributors to adhesion because their disruption was directly correlated to an adhesion strength decrease to nearly zero. Furthermore, although both strongly hydrogen bonded water and relatively weakly hydrogen bonded water were detected in the epoxy bulk, only strongly hydrogen bonded water was detected at the buried interface.

Figure 4.3. SFG and (b) ATR-FTIR spectra collected from the SiO2/epoxy interface after

hygrothermal aging time periods.