Some reflections on contemporary views of theories of adhesion
Post on 21-Jun-2016
Some reflections on contemporary views of theories of adhesion
(The City University, UK)
This paper reviews the literature on mechanical interlocking and adsorption interactions, the two main mechanisms postulated to account for adhesive bonding. It is shown that interpenetration of adhesive and adherend is important in obtaining optimum adhesion in a wide range of applications. While for fibrous materials this interpenetration is fairly obvious on a macro scale, for metal/metal oxide surfaces it is on a micro scale and revealed by sophisticated surface examination techniques. Wetting and the thermodynamics of solid/liquid interfaces are reviewed and the development of theories relating to secondary interaction forces is discussed. Recent evidence for the existence and significance of primary chemical bonding in adhesion is then outlined.
Key words: adhesion; mechanical interlocking; adsorption interactions; review
For a very considerable period when reviewing theories or postulates of the fundamental basis of adhesive bonding it has been usual to distinguish four distinct mechanisms: mechanical interlocking, adsorption interactions, electrostatic forces and diffusion mechanisms 1. Of these, the last two have always been recognized as being of significance in only a few, rather restricted instances; although important within these limitations. For the main areas of industrial application, where adhesive bonding is required to resist significant stress and to play a serious part in the integrity of structures, then it is the first two which are relevant. It is the deeper understanding of these which has underpinned the confidence that has enabled the recent steady growth of the use of structural adhesives in engineering. Accordingly it is these two -- mechanical interlocking and adsorption interactions -- which will be considered in this present paper.
Mechanical interlocking In essence, the idea of mechanical interlocking of the adhesive within irregularities of the adherend surface is the oldest and simplest explanation of the whole phenomenon of adhesive bonding. It has an intuitive appeal and immediately led to the idea of roughening surfaces to increase adhesion. While the main uses of adhesives were in joining papyrus, paper, leather or
wood, this was largely true. This was because all these substrates were fibrous materials and successful bonding generally did involve penetration of the adhesive between the fibres and thus embedding in the cured adhesive. However, there were oddities and apparent exceptions as illustrated in Table 1, where the bond strength between wood specimens decreases with increasing roughness.
This idea was justified by the classical work of Borroff and Wake 3 on the adhesion of rubber to textiles in the production of motor tyres. The early tyre cords were made from spun staple of natural origin, mainly cotton. This work demonstrated conclusively that the bond strength depended upon the penetration of the fibre ends from the spun yarn into the rubber (and not significantly upon the penetration of the rubber into the weave). If the spun staple natural fibre was replace by a synthetic monofilament (e.g. nylon), which might give enhanced mechanical properties, the bond strength to the rubber was very considerably reduced. A variety of special treatments and finishes had to be devised and used to provide specific interaction between the surface of these monofilaments and the rubber to achieve satisfactory bonding when they were used.
Another instance in which a mechanical interlocking was clearly demonstrated to be important was the electroless plating of certain plastics with
0143-7496/93/020067-06 1993 Butterworth-Heinemann Ltd
INT,J.ADHESION AND ADHESIVES VOL. 13 NO, 2 APRIL 1993 67
Table 1. Maple wood samples, bonded side grain with urea-formaldehyde resin under 5 Ibf in -2 pressure, tested in shear 2
Surface Shear strength (Ibf in -2)
Planed 3120 Sanded 2360 Sawn 2690 Combed 2400
1 Ibf in -2 = 6 .895 kPa
metal. The plastics involved, either ABS (acrylonitrile/ butadiene/styrene) or high impact polystyrene, consist of a matrix of glassy polymer with the elastomer dispersed within it. The first step in the process is etching with an oxidizing agent; this oxidizes and removes the rubbery material, leaving a highly porous and reticulate surface upon which the metal is then deposited. Electron micrographs of sections 4" 5 show quite clearly the penetration and interlocking of the metal within the plastic to a depth of up to 10pm. A very careful study by Perrins and Pettett 6 demonstrated that there were two mechanisms involved and reinforcing each other -- mechanical interlocking and a specific interaction depending upon the surface chemistry of the plastic. It has been authoritatively suggested that the synergy of these two mechanisms should be represented by a multiplicative rather than an additive mathematical function.
However, these, which all involve roughness on a relatively macro scale in the micrometre range, are largely irrelevant to the mainstream of structural bonding. In more recent times, particularly with more sophisticated techniques available, attention has focused on roughness on a much smaller scale, at the AngstrOm level rather than the micrometre level.
Probably the first clear example was due to Packham 7-9, who was investigating the adhesion of molten polyethylene to metals and especially aluminium. It had already been established that for good adhesion the polyethylene needed to be oxidized and that if this was prevented then only poor adhesion could be achieved. It was also well established that anodic oxidation of aluminium in acid electrolytes gave a barrier layer of oxide immediately on the metal and beyond this a structure of hexagonal oxide cells each with a pore, approximately circular in cross- section, at its centre. The dimensions of the cells was governed by the anodizing voltage but the diameter of the pores was not. However, the nature and concentration of the electrolyte controlled the diameter of the pores.
Packham found that, while for aluminium oxide surfaces without pores it was necessary for polyethylene to be oxidized if good bonds were to be produced, for oxide surfaces with pores goods bonds were produced with polyethylene which was carefully protected (in various ways) from oxidation. Indeed, the bond strength was directly related to the depth of the pores and the overall porosity of the surface. Electron micrographs were taken of surfaces of polyethylene which had been sintered on to aluminium anodized in 4% phosphoric acid and on to steel surfaces and from
which the metal had been dissolved away. The polyethylene which had been in contact with steel showed tne surface roughness of the metal but no other features. The polyethylene which had been in contact with the aluminium, at a magnification of 15 000, showed regular projections normal to the surface which were fibrous in appearance. The tufts were of the order of I pm (1 X 10",~) in diameter. The best explanation, and the one which certainly corresponds with the general appearance, is that each projection is a tuft of a number of individual fibres, each one of which was originally in a single pore. Estimates of the diameter of the fibres gives a value of approximately 550A, which is not far removed from the diameter of the pores of 330,h, given by Keller et al. Io
Clearly here the mechanical interlocking of the polyethylene (regarded as the adhesive) with the surface is an important factor in achieving good adhesion.
A little later, in studies of the bonding of titanium alloys II with epoxide adhesives, a significant effect of surface roughness was noted. A variety of chemical treatments all gave a surface layer of the rutile form of titanium oxide. The one which resulted in the greatest bond strength also gave the roughest surface, with needles of ruffle strongly adhering to the substrate and with the a-phase of the alloy preferentially removed.
More recently Venables and colleagues 12 13, using a variety of surface analytical techniques but especially scanning transmission electron microscopy (STEM) at 30A resolution, studied the surfaces of aluminium and titanium which had been subjected to the standard treatment used in preparation for adhesive bonding in aerospace applications. The chromic acid etching of aluminium, usually known as the Forest Products Laboratories (FPL) process, produces a shallow hexagonal cell structure with a high concentration of protruding oxide 'whiskers' (~50A diameter and ~400,~ high). If this etching is followed by anodizing in phosphoric acid (PAA), the whole oxide structure is considerably deeper with much more pronounced cellular formation and longer (~ 1000A) oxide whiskers. Further evidence leads to the conclusion that, in the absence of any degradation due to the environment and especially water, the bond only fails when the polymer itself fails by viscoelastic deformation. When the metal oxide surface does not have this particular pore/whisker structure, then failure occurs at considerably lower stress levels.
Closely similar results were produced for titanium alloys.
Venables et al. also demonstrated that moisture caused a considerable change in the surface morphology quite quickly. The oxide is converted to hydroxide which has a totally different structure. The pores and whiskers are replaced by a lamellar 'cornflake' structure which adheres only weakly to the underlying material, and the bond strength is considerably reduced. There is, however, some controversy about whether this hydration degradation occurs before the bond is parted or afterwards.
Nevertheless, all this very elegant work undoubtedly demonstrates that mechanical interlocking on a micro scale is an important factor (even if not the dominant one) in achieving high adhesive bond strength.
In conclusion, it is now clear that. in a wide range
68 INT.J.ADHESION AND ADHESIVES APRIL 1993
of applications, the interpenetration of adhesive and adherend is important in securing the optimum adhesion. For fibrous materials this will be on a fairly obvious and macro scale, while for metal/metal oxide surfaces it is on a micro scale and only demonstrable by surface examination of a high degree of sophistication.
Adsorption interactions The cohesive strength of any solid depends upon the various forces of attraction that exist between the ultimate particles of which it is composed. These forces are of various types depending upon the particular nature of the solid but, in general, will include those regarded as of chemical origin -- covalent, ionic and metallic bonding, hydrogen bonding -- as well as those considered as of physical origin -- the secondary or van der Waals' forces including dipole interactions and the dispersion forces. In any particular instance, which of these are present and their relative significance will depend upon the chemical nature of the material, but the dispersion forces will always be present and effective to a greater or less extent. A quantitative consideration of these forces will enable an ideal strength of the material to be calculated. However, this strength is never normally achieved. This is because of irregularities, flaws and defects both throughout the structure and more particularly in the surface layer, as was demonstrated originally by Griffith t4.
This is evident from simple calculations 15, which indicated a theoretical strength for steel of about 3 104 MPa compared with the experimental values of 4 X 102 MPa for ordinary commercial material and 3 l03 MPa for the strongest steel wires. Moreover, steel is rather exceptional in achieving a real strength which is only one order of magnitude less than the theoretical value. For most materials the difference is two or three orders of magnitude.
It is only under very special conditions such as when 'whiskers' are grown or very fine fibres are drawn that these ideal strengths can be approached.
It is these same forces of cohesion within a single material that are responsible for adhesion when it occurs across the interface between two different materials. Ultimately there are no other forces available. One important feature which is common to all these forces is that they are only significant over very short distances. Most of them are quite negligible beyond a few AngstrOms, and those which have the greatest range are only effective over distances less than l0 times as great. This means that if they are to have any effect to hold two surfaces together, these must be in very close contact indeed.
It is for this reason that all normal adhesives are, at one stage in their use, mobile liquids which can wet and flow across the adherend, penetrating all the irregularities and roughness of the surface so that the very closest and most intimate contact is achieved. It is from this requirement that all the concern with, and study of, the phenomenon of wetting and the thermodynamics of liquid/solid interfaces arises.
The first quantitative ideas about the wetting of a solid surface by a liquid are due to Thomas Young 16, who conceived the relationship between surface tensions and contact angle that is now very familiar:
~'SV = VSL + YLV cOs0 (Young equation)
The theoretical consideration of the situations involved in adhesion begins from the expression for the thermodynamic work of adhesion first enunciated by Dupr617:
WA = 71 + 72 - 712 (Dupr6 equation)
That is, the thermodynamic work of adhesion of a liquid to a solid is equal to the sum of the surface free energies of the two separate phases minus the interfacial free energy. It is important to recognize immediately that the two surface free energies relate to clean surfaces, each in equilibrium with its own vapour. They are not the values which may commonly be measured or considered, where the surface of the solid is in equilibrium with the vapour resulting from the liquid. If this equation is to be combined with that due to Young (which involves equilibrium with the vapour from the liquid), then the term n, the spreading pressure, must be introduced in order to relate }'s and }'sv to give
]IS = }'SL + 7LV cOsO+ n (Young-Dupr6 equation)
where n, the equilibrium spreading pressure, is given by Is
~ o tr = Ys - }'sv = RT Fd(lnp) and F is the surface concentration of adsorbed vapour, p is the vapour pressure and P0 is the equilibrium vapour pressure.
But going back to the Dupr6 equation, the }'12 term and how it might be expressed in useful terms remain important questions. Two separate and distinct paths of development have emerged.
Good and Girifalco I,' 20 gave their attention to this interfacial free energy and how this might be expressed in terms of the properties of the two separated surfaces. They began with Berthelot's relation 21 for the attractive constants between like molecules Aaa,...