Residual stress and adhesion

As a final brief note on thin film growth in general, a technologically important aspect is adhesion of the film to the substrate. Good adhesion normally requires the film to have the lowest possible stress, as any force applied to the interface will encourage decohesion. The most successful adhesives such as epoxy form a stress-free junction. Minimizing stress in a thin film is usually a matter of adjustment of deposition conditions. Two factors contribute to stress in a film, differential thermal expansion and deposition-related processes. 

When the fihn is bombarded during growth with energetic particles, atoms may be knocked into the intergranular voids, filling them even at relatively low temperatures. This knock-on process may drive the film into compressive stress if more atoms are pushed into lower layers of the fihn than it intends to accommodate in its preferred crystal shucture. Thus, the residual stress may be adjusted if a 

The differential thermal expansion contribution to film stress is straightforward to understand. When a film is deposited it typically has only the growth stresses described above. However, if the deposition temperature is significantly different from room temperature (usually higher) then when heating is terminated after growth and the sample cools, the film usually contracts at a different rate than does the substrate. Because the substrate is usually very much stiffer than the film this differential thermal expansion stress is accumulated in the film. 

In addition to residual stresses, adhesion failures result from a lack of chemical reaction between the film and substrate, which leads to a weak interface very 

In production, deposition processes, surface cleaning and careful process control to minimize film stresses and maximize adhesion. 

---

Godoy, C., Souza, E.A., Lima, M.M., and Batista, J.C.A. (2002) Correlation between residual stresses and adhesion of plasma sprayed coatings effects of a postannealing treatment. Thin Solid Films, 420/421, 438 -445. 

W. Herr, B. Matthes, E. Broszeit, M. Meyer and R. Suchentrunk, Influence of substrate material and deposition parameters on the structure, residual stress and adhesion of sputtered Cr N, hard coatings. Surf. Coat. Technol., 1993, 60, 428-433. 

Characterization of the mechanical properties of these thin silica layers, unreinforced or reinforced, is usually conducted by using the nanoindentation technique [33-37] to determine the hardness (H) of the layer and the elastic modulus ( ) using the Oliver-Pharr method [38]. In these tests, a Berkovich indenter is used and low maximum loads are applied (in the range of mN) to avoid the influence of the mechanical response of the substrate. A complete review of how to calculate different key mechanical parameters ( , H, fracture toughness, residual stresses, and adhesion) of thin sol-gel coatings using nanoindentation tests and scratch testing with nanoindenter equipment can be found in the work of Malzbender et al. [39]. 

If a single metal film is deposited on an oxide, the sheet resistance measurement results can by easily interpreted and converted to the thickness. In practice, however, this is not usually the case. For example, in W CVD, the tungsten is not directly deposited on oxide due to high residual stress and unreliable adhesion. A titanium (Ti) layer must be first deposited as a glue layer. In addition, to prevent the fluorine in the CVD-precursor WFg from directly reacting with Ti (a strong catalytic reaction will occur), a barrier layer of titanium nitride (TiN) must be deposited on top of the Ti. As a result, we have a trilayer film of W on TiN on Ti on oxide, as shown schematically in Fig. 21. This poses some problems in accuracy in the four-point probe measurements. Based on the resistivities in Table VI, the... 

Figure 9. Schematic test site for measurement of residual stress, modulus, tensile stress, and adhesion of polyimide.

Measurements of residual stress and strain, Young s modulus and adhesive bonding strength of the hydroxyapatite coating after immersion in HBSS showed... 

Epoxy-amine liquid prepolymers are extensively appHed to metallic substrates and cured to obtain painted materials or adhesively bonded stractures. Overall performances of such systems depend on the interphase created between the organic layer and the substrates. When epoxy-amine Hquid mixtures are appHed to a more or hydrated metaUic oxide layer (such as Al, Ti, Sn, Zn, Fe, Cr, Cu, Ag, Ni, Mg, or E-glass), amine chemical sorption concomitant with metaUic surface dissolution appear, leading to the organometaUic complex or chelate formation [1, 2]. Furthermore, when the solubility product is exceeded, organometaUic complexes may crystaUize. These crystals induce changes of mechanical properties (effective Young s modulus, residual stresses, practical adhesion, durability, etc.). 

Besides delamination, actual cracking of silicon die is a failure mode that can occur from excessive adhesive stresses, voids, and moisture absorption. Residual stresses in adhesive-attached single-chip devices become critical as the size of the device increases and dissimilar die and leadframe materials are used. Several types of fractures that can occur within the die or within the adhesive are shown in Fig. 6.8. A hairline crack in an 1C chip that resulted from adhesive stress is shown in Fig. 6.9. 

Fig. 11 also shows that the adhesive thickness has a significant effect on the F-stress in DCB specimens. As the adhesive thickness decreases, the T-stress decreases, which indicates that directionally unstable cracks are less likely to occur in specimens with thinner adhesive layers. Fig. 13 illustrates representative specimens selected from two groups of symmetric DCB specimens with adhesive C and adhesive thickness of 0.5 and 0.25 mm. The specimens were subjected to mechanical stretching until 1.3% of plastic deformation occurred in the adherends before testing to increase the residual stress and consequently the T-stress. The adhesive thickness was 0.5 mm for specimen a and was 0.25 mm for specimen b. Both specimens have a final residual stress of 38.6 MPa. However, because the adhesive layer in specimen b is thinner, the T-stress in specimen b (26 MPa) is lower that the T-stress in specimen a (35 MPa). The alternating debonding in specimen a represents more directional instability than the predominantly oscillating debond seen in specimen b, as predicted in Fig. 12. 

Dillard, D.A., Park, T.G., 2 ang, H. and Chen, B., Measurement of residual stresses and thermal expansion in adhesive bonds, in preparation. 

The coefficient of thermal expansion (CTE) and the stress-free temperature (SET) are measured to estimate residual stresses in adhesives. 

Davies RE, Fay PA (1993) Int J Adhes Adhes 13 97 Fay PA, Maddison A (1990) Int J Adhes Adhes 10 179 Gledhill RA, Kinloch AJ (1974) J Adhes 6 315 GledhiU RA, Kinloch AJ, Shaw S (1980) J Adhes 1 3 Gurson AL (1977) J Eng Mater Technol 99 2—15 Hua Y, Crocombe AD, Wahab MA, Ashcroft lA (2007) J Adhes Sci Technol 21 179 Jumbo FS (2007) Modelling residual stress and environmental degradation in adhesively bonded joints. PhD thesis, Loughborough University, Loughborough, UK... 

Film Adhesion. The adhesion of an inorganic thin film to a surface depends on the deformation and fracture modes associated with the failure (4). The strength of the adhesion depends on the mechanical properties of the substrate surface, fracture toughness of the interfacial material, and the appHed stress. Adhesion failure can occur owiag to mechanical stressing, corrosion, or diffusion of interfacial species away from the interface. The failure can be exacerbated by residual stresses in the film, a low fracture toughness of the interfacial material, or the chemical and thermal environment or species in the substrate, such as gases, that can diffuse to the interface. 

Good wetting is of course not a sufficient criterion for good contact adhesion because it takes no account of the factors that influence the mechanical loss factor, C, in Eq. 8, nor does it account for residual stress development during cure. But aside from these factors, one might inquire into the validity of the correlation between practical contact adhesion and VEa beyond 0° contact angle , i.e. can any distinction be made based on VEa between different adhesives, all of which perfectly wet the adherend ... 

---