Failure of adhesive bonding unveiled by in-situ strain testing in high-resolution scanning transmission electron microscopy

The nano-scale failure behaviors of adhesive interfaces were investigated through in-situ straining testing to observe real-time crack propagations under a scanning transmission electron microscope (STEM). Two different loading modes were applied to thin sections of adhesive interfaces: crack-opening mode applied to pre-cracks made at the interface and shear mode. The failure of aluminum alloy (Al6061) and a second-generation acrylic adhesive (SGA) was examined, enabling observation of the growth of crazing in the adhesive layer, which has a phase-separated structure, preceding the macroscopic failure of the interfaces. Furthermore, the failure of a direct joint of thermoplastic and Al was investigated, with a comparison made to that observed in the adhesive interface. The generation and propagation of cracks near the interface, attributed to the adhesive's phase separation, contribute to the toughness of the adhesive interface. Both the direction of stress acting on the interface and the interface's strength in�uence the initiation and growth of cracks throughout the adhesive layer.


Introduction
Investigating the failure of adhesive interfaces is important to study the bonding mechanisms and evaluate the performance of adhesives and surface treatments.Failure behavior has been generally speculated by inspecting fracture surfaces by optical or scanning electron microscopy [1][2][3].However, such classical fractography limits the capability to understand the complicated bonding mechanism and properties.The direct observation of the failure behavior under high-resolution electron microscopy is expected to provide information on complicated failure processes in adhesive interfaces.Recently, new equipment for performing in-situ experiments to apply tensile force to a joint specimen under highresolution scanning transmission electron microscopy (STEM).We previously reported the real-time observation of the failure processes of the interfaces between a model epoxy/amine mixture and aluminum and the interface in a direct joint of engineering plastic and aluminum [4].STEM allows us to perform high-spatial resolution analysis of interfaces through imaging and local elemental and chemical analysis using a focused electron probe of less than 1 nm diameter [5].The advantages of STEM over conventional transmission electron microscopy (TEM) are that high-quality and high-contrast images can be acquired even with a thicker specimen up to 200 nm.This work attempts in-situ tensile testing under STEM to observe the real-time failure in commercial adhesives with phase-separated morphologies and inorganic llers.

Experimental
Acid treatment of the Al surface was employed to ensure the bonding of Al to the adhesive.3 mm thick Al6051 plates were preliminarily immersed in sodium hydroxide aqueous solution (ph12) at 60°C for 30 or 45 sec, followed by the treatment with 60 wt% nitric acid for 1 min.The acid treatment was employed with two immersion times to control the bonding condition.Then, those two plates were bonded with a commercial structural second-generation acrylic adhesive (SGA) [6-8], HARDLOC C355-20A/20B (DENKA Corp., Tokyo, Japan), or two-component epoxy adhesive [9,10], DENATITE (NAGASE ChemTex Corp., Osaka, Japan).SGA adhesives are two-component room-temperature curing structural acrylic adhesives.After mixing and pasting the adhesives, they were left at approximately 24°C (room temperature, RT) for 24 h and subsequently at 60•C for two hours to cure the SGA adhesive.The epoxy adhesive was cured at 100°C for 30 min.The non-bonded region for introducing a pre-crack into the interface was made by inserting 100 µm thick Kapton lm into one end of the lamination.The Al5052 and polyphenylene sul de (PPS) joint specimens were provided by Taisei Plus Co. (Tokyo, Japan), Ltd., prepared by insert-injection molding of PPS (SGX-120, TOSO Corp., Japan) onto surfacemodi ed Al5052 plate at the melt temperatures of 290-330°C and the mold temperature of 120°C.The Al surface was chemically treated by the method developed by Taisei Plas Co., Ltd., using a hydrazinebased aqueous solution [11], producing nano-sized pores with three-dimensional inter-connected structures within approximately 100 nm thick surface layers.The PPS/Al5052 joint laminate 2 ± 0.1 mm thick was prepared with the non-bonded region at the one side of the joint laminate, which is a pre-crack part.
Figure 1 shows the specimen holder for in-situ tensile testing in a STEM equipped with a device manufactured by Mel-Build Corp. (Fukuoka, Japan).A small device for applying a tensile force to the specimen is built into the tip of the sample holder, as shown in Figs.1a and 1b.The specimen is mounted on the isolated thin metal cartridge attached to the actuator built into the device.Pushing the cartridge by the actuator with 100 nm/sec can open the narrow slit with 20 µm width fabricated in the cartridge, applying tensile load into the specimen xed in the slit.Figures 1c and 1d show an Al/adhesive/Al triple layer sectioned into about 100 nm thick, about 300 µm in size.The thin sections were cut with a diamond knife using an ultramicrotome, and the oating sections on the water in the trough of the diamond knife were collected onto the cartridge [5].Then the sections were xed on the slit as the desired position and direction, as shown in Figs.1c and 1d.As shown in Fig. 1c, the section is xed to apply tensile force to the interface.Here, the interface is arranged parallel to the slit, and both sides are xed with adhesive.On the other hand, for applying a shear force to the interface, the section was positioned with the interface perpendicular to the slit, and the two corners, diagonally opposite each other, were xed with adhesive, as depicted in Fig. 1d.
A focused ion beam (FIB) [5] was used to prepare a thin test specimen for an inorganic ller-containing adhesive.A small section of the aluminum test specimen, along with epoxy adhesive, was xed over a 20 µm wide slit in the metal plate of the specimen holder, and a thin window that included the Al/adhesive interface for electron beam transmission was created.
In-situ STEM experiment was performed using TECNAI Osiris (FEI company, USA) STEM instrument with an accelerating voltage of 200 kV.

Results and Discussions
Page 4/14 3.1 Failure of the adhesive interfaces between SGA and Al in crack opening mode bright domains and a continuous dark phase.There are two types of domains that are dispersed, with signi cantly different sizes.Many small domains are dispersed in the narrow gap between the large spherical domains.Moreover, it was found that the bright phase preferentially covers the entire Al surface.It is indicated that the acrylic monomers in the adhesive are separated into two phases during the polymerization process [12], and the minor component tends to form the matrix phase.It means that this adhesive exhibits phase-inversion during the increase in molecular weights of the components.At the same time, the major component prefers to lay on the Al surface.
Figure 3 shows the STEM-HAADF images captured in the in-situ tensile experiment of the specimen with a pre-crack at the interface.When tensile force was applied in the lateral direction of the specimen, the pre-crack widened, and the crack tip became round-shaped before the crack propagated.It was con rmed that numerous branched ne cracks were generated from the crack tip on that round face (Fig. 3a).The tensile load continued to be applied to the specimen in the lateral direction.Then, the crack started propagating into the adhesive layer, producing the branched ne cracks in the wide region ahead of the crack (Fig. 3b).The crack, thus, escaped from the interface between Al and the adhesive, preventing the interfacial failure of the adhesive interface.Figure 3c is a magni ed image of the ne cracks, showing that the ne cracks are running through the gap between the spherical dispersed domains.Figure 3d shows that microvoids and small brils are included in the ne cracks around the crack tip, which are elongated in the direction perpendicular to the crack growth direction.These features observed in the ne cracks indicate crazing occurs upon the failure of the adhesive bond when the tensile force is applied to the interfacial pre-crack.The ne brils elongate and break, causing the microvoids to grow and coalesce, then cracks forming.The real-time observation of the failure of adhesive bonding suggests that the adhesive forms the phase separation with continuous soft phase and the hard dispersed domains.

Failure of the adhesive interfaces between SGA and Al under shear force
The failure behavior of the Al/adhesive/Al triple layer under shear force was examined through in-situ tensile testing with STEM observation.Figure 4 shows the crazing in the adhesive layer of the Al plates under two different acid treatment conditions.Figure 4a presents the development of crazes and interface delamination with the Al treated in a short immersion time with a sodium hydroxide (soft treatment), while Fig. 4b illustrates the failure process by the long immersion time (hard treatment).With the soft acid treatment, the failure of the interface of the upper Al layer occurred at the beginning of the shear force loading (left panel of Fig. 4a), and crazing started to develop in the middle part of the adhesive.Meanwhile, the voids were produced after the growth of the crazing (middle panel of Fig. 4a), and then the adhesive layer was entirely raptured with the delamination of the adhesive from the Al substrate (right panel of Fig. 4a).The bonding by the hard acid treatment, the interfacial delamination could be avoided in the earlier stage of the shear loading, as shown in the left panel of Fig. 4b.The crazing was observed to be produced simultaneously in the entire part of the adhesive layer, and then the adhesive layer was raptured entirely in the adhesive after the crazing grew into cracks.
The ndings suggest that the crazing developed in the soft component, acting as the matrix phase as the narrow gap, effectively resists crack propagation, thereby enhancing interface toughness.Furthermore, the selective segregation of the hard component at the Al/adhesive interface can contribute to a tough interface, directing crazing towards the adhesive layer rather than the Al interface.The bonding performance of the SGA adhesive used in this study has been well investigated, and the mode I fracture energy of the bonding to steel was reported to be 1.7-2.0kJ/m 2 [6-8].Such high toughness could be achieved due to its phase-separated structure, which promotes crazing and effectively absorbs energy to cause failure, and also due to the selective segregation of the hard component on the Al surface.

Failure of the adhesive interfaces between the SGA adhesive and Al under shear force
To understand deeply the role of the phase-separated structure of the adhesive on interfacial toughness, the failure behavior of the interface in the direct bonding of thermoplastic and Al via injection molding was investigated.The details of this bonding technology are shown in the literature [11][12][13].As shown in the left panel of Fig. 6a, the polyphenylene sul de (PPS) contains elastomer as the dispersed domains.
When applying the tensile force to open the pre-crack, cavities were produced in the elastomer domains close to the interface.The cavities could promote local plastic deformation around the domains, and then crazes were produced to connect the cavities (middle panel of Fig. 6a).Finally, the crazes coalesced, separating the PPS from the Al with a small amount of elongated PPS remaining on the Al surface (right panel in Fig. 6a).As compared to the SGA adhesive, crazing occurred in the limited region around the crack in the PPS/Al direct joint, indicating that the phase separation structure obtained in the SGA adhesive is effective in introducing crazing into the adhesive.
Figure 6b depicts the STEM-HAADF images captured during shear loading.In the PPS/Al double layer, crazing was initiated in the middle part of the PPS layer (left in Fig. 6b) and propagated towards the Al surface (middle in Fig. 6b).It was found that the shear loading induced extremely widespread crazing in the PPS layer as compared with the tensile loading (right panel in Fig. 6b).      Figure 7

Figure 2
Figure 2 is a STEM image in high-angle annular dark eld (HAADF) mode showing the phase separation in the SGA adhesive layer in the interfacial region.It indicates that the adhesive contains two phases:

Figures Figure 1
Figures