J O U R N A L O F M A T E R I A L S S C I E N C E 3 9 (2 0 0 4 ) 1 – 49

Review
Adhesively-bonded joints and repairs in metallic
alloys, polymers and composite materials:
Adhesives, adhesion theories
and surface pretreatment
A. BALDAN
Department of Metallurgical and Materials Engineering, University of Mersin,
Ciftlikkoy, Mersin, Turkey
E-mail: [email protected]

In the present paper, the following topics are reviewed in detail: (a) the available adhesives,
as well as their recent advances, (b) thermodynamic factors affecting the surface
pretreatments including adhesion theories, wettability, surface energy, (c) bonding
mechanisms in the adhesive joints, (d) surface pretreatment methods for the adhesively
bonded joints, and as well as their recent advances, and (e) combined effects of surface
pretreatments and environmental conditions on the joint durability and performance.
Surface pretreatment is, perhaps, the most important process step governing the quality of
an adhesively bonded joint. An adhesive is defined as a polymeric substance with
viscoelastic behavior, capable of holding adherends together by surface attachment to
produce a joint with a high shear strength. Adhesive bonding is the most suitable method
of joining both for metallic and non-metallic structures where strength, stiffness and
fatigue life must be maximized at a minimum weight. Polymeric adhesives may be used to
join a large variety of materials combinations including metal-metal, metal-plastic,
metal-composite, composite-composite, plastic-plastic, metal-ceramic systems. Wetting
and adhesion are also studied in some detail in the present paper since the successful
surface pretreatments of the adherends for the short- and long-term durability and
performance of the adhesive joints mostly depend on these factors. Wetting of the
adherends by the adhesive is critical to the formation of secondary bonds in the adsorption
theory. It has been theoretically verified that for complete wetting (i.e., for a contact angle θ
equal to zero), the surface energy of the adhesive must be lower than the surface energy of
the adherend. Therefore, the primary objective of a surface pretreatment is to increase the
surface energy of the adherend as much as possible. The influence of surface pretreatment
and aging conditions on the short- and long-term strength of adhesive bonds should be
taken into account for durability design. Some form of substrate pretreatment is always
necessary to achieve a satisfactory level of long-term bond strength. In order to improve
the performance of adhesive bonds, the adherends surfaces (i.e., metallic or non-metallic)
are generally pretretead using the (a) physical, (b) mechanical, (c) chemical, (d)
photochemical, (e) thermal, or (e) plasma method. Almost all pretreatment methods do
bring some degree of change in surface roughness but mechanical surface pretreatment
such as grit-blasting is usually considered as one of the most effective methods to control
the desired level of surface roughness and joint strength. Moreover, the overall effect of
mechanical surface treatment is not limited to the removal of contamination or to an
increase in surface area. This also relates to changes in the surface chemistry of adherends
and to inherent drawbacks of surface roughness, such as void formations and reduced
wetting. Suitable surface pretreatment increases the bond strength by altering the substrate
surface in a number of ways including (a) increasing surface tension by producing a surface
free from contaminants (i.e., surface contamination may cause insufficient wetting by the
adhesive in the liquid state for the creating of a durable bond) or removal of the weak
cohesion layer or of the pollution present at the surface, (b) increasing surface roughness
on changing surface chemistry and producing of a macro/microscopically rough surface,

0022–2461

C 2004 Kluwer Academic Publishers 1
(c) production of a fresh stable oxide layer, and (d) introducing suitable chemical
composition of the oxide, and (e) introduction of new or an increased number of chemical
functions. All these parameters can contribute to an improvement of the wettability and/or
of the adhesive properties of the surface.
C 2004 Kluwer Academic Publishers

1. Introduction face between the different substrates has the specific

Polymeric materials that fall within the classifications properties.
of thermoplastics, thermosetting resins, elastomeric As a result of theoretical considerations and exten-
compounds, and natural adhesives (animal glue, ca- sive practical testing, the following recommendations
sein, starch, and resin) may serve adhesive func- [1] are made to achieve satisfactory joints: (a) Suitable
tions. Polymeric adhesives may be used to join a surface pretreatment. Surface preparation is, perhaps,
large variety of material combinations: metal-metal, the most important process step governing the qual-
metal-plastic, metal-composite, composite-composite, ity of an adhesive bond [2]; not only is a clean bond
plastic-plastic, metal-ceramic, and so on. The primary surface in the conventional sense required, but fresh
drawback is the service temperature limitation. Organic cleaning to avoid adsorbed gases is often useful. Ac-
polymers maintain their mechanical integrity only at tivated inert gases are sometimes used. (b) Adhesive
relatively low temperatures, and strength decreases choice. The adhesive should wet the adherend and so-
rapidly with increasing temperature. Even though the lidify under production conditions of time, temperature,
inherent strength of the adhesive may be much less than and pressure. Often the desired production conditions
that of the adherend materials, nevertheless, a strong narrow the choice of adhesive. Some thermoplastic and
joint may be produced if the adhesive layer is thin and thermosetting polymers are used as adhesives. Follow-
continuous. If a good joint is formed, the adherend ma- ing forms of adhesives are available: liquids, pastes,
terial may fracture or rupture before the adhesive. and solids. (c) Joint Design. Adhesive joints are gener-
Polymers are attractive materials and usually offer ally more resistant to shearing, compressive, and tensile
the following advantages: (a) weight, (b) cost, (c) mois- stresses than they are to stress systems due to peeling.
ture and chemical resistance, (d) toughness, (e) abra- For example, it is easier to remove adhesive tape from
sive resistance, (f) strength, (g) appearence, (h) insula- a surface by peeling than by any other method of ap-
tion (both thermal and electrical), (i) formability, and plying stress. (d) Service condition. Polymers as adhe-
(j) machinability. Polymeric materials, unlike metals, sives generally have higher thermal expansitivities (i.e.,
do not respond elastically to stress and undergo perme- the coefficient of expansion) than metals and ceramics.
nant deformation under sustained loading. The term Furthermore, their expansion coefficients are not truly
viscoelastic describes the behavior of plastics when constants; that is, the polymers expand markedly in a
subject to stress. After the application of an initial load, nonlinear way with temperature. Epoxy resins have co-
the polymeric structure will move in response to the ap- efficient of linear expansion values between 50 × 10−6
plied stress. The stress-strain curve for plastics change and 100 × 10−6 K−1 while polyester show values be-
with time. When compared with metals, polymeric ma- tween 100 × 10−6 and 200 × 10−6 K−1 [3]. Small
terials lack stiffness and their physical properties are compositional changes can have a marked influence on
temperature-dependent. polymer expansion characteristics. If severe tempera-
Service experience with adhesive bonded structures ture changes are to be encountered, this effect and the
has been extremely varied, with some components pro- required accomodations of the adhesive must be consid-
viding excellent service and others failing or requiring ered. Weathering and solvents that may be encountered
extensive maintenance over a comparatively short ser- in service are also important considerations.
vice life time. The deficient adhesively bonded com-
ponents have performed so badly when compared to
conventionally fastened structures that they have been 1.1. Adhesives
considered representative of the generic lack of relia- Structurely, polymers are giant chainlike molecules
bility of bonded structures, leading to an overall poor (i.e., macromolecules) with covalently bonded carbon
acceptance of the technology. atoms forming the backbone of the chain. Polymers
Adhesively bonded joints offer essential advantages are structurally much more complex than metals or
in comparison with other joining methods, such as ceramic. Prolonged exposure to ultraviolet light and
higher joint stiffness and superior fatigue performance solvents can cause the degradation of the polymer
[1]. For this reason the number of adhesive bonding properties. Due to predominantly covalent bonding,
applications in various industries is steadily growing. polymers are generally poor conductors of heat and
Typical examples of beneficial applications of the adhe- electricity. Polymers, however, are generally more re-
sive bonding technology are in the construction of air- sistant to chemicals than are metals.
craft, rail vehicles and in the automobile industry. Since It is now useful to have an understanding of the ba-
the nature of the bonding is dependent on the atomic sic molecular structure of polymers since the prop-
arrangement and chemical properties of the molecular erties of the adhesives are much dependent on the
conformation, chemical constitution and diffusivity of molecular structure. The molecular structure of most
elements in each constituent, it follows that the inter- polymers are based on hydrocarbons in which carbon

2
and hydrogen combine in the relationship Cn H2n+2 , erally water, forms as a by-product. In contrast to poly-
known as paraffins [4]. Theoretically, these hydrocar- merisation by addition, in which all of the component
bons can be linked together indefinetely to form very atoms appear in the product molecule, condensation
large molecules, the bonds between the atoms being sin- polymerisation occurs as reactive molecules combine
gle pairs of shared (covalent) electrons. Because there with one another in a stepwise fashion, while elimi-
is no provision for additional atoms to be added to the nating small by-product molecules, such as water. The
chain, such molecules are said to be saturated. Bond- structure of condensation polymers is often a three-
ing with the molecule is quite strong, but the attractive dimensional framework in which all atoms are linked
forces between adjacent molecules are much weaker. by strong primary bonds.
Carbon and hydrogen can also form molecules in
which the carbon atoms are held together by double
or triple covalent bonds. Because such molecules do Copolymers are a special category of polymer in
not have the maximum possible number of hydrogen which two or more types of mers are combined into
atoms, they are said to be unsaturated and are impor- the same addition chain. This process (i.e., by addition
tant in polymerisation process, where small molecules mechanism), greatly expands the possibilities of cre-
join to form large ones with the same constituent atoms. ating new types of polymers with improved physical
In polymeric compounds, four electron pairs surround and mechanical properties. The small amounts of Plas-
each carbon atom, and one electron pair is shared with ticizers may be added to change the properties of the
each hydrogen atom. Other atoms or structures can be polymer, thus extending its application; the flexibility,
substituted for carbon and hydrogen, however. Chlorine ductility, and toughness of polymers may be improved
or even a benzene ring can often take the place of hy- with the aid of these additives. Their presence also pro-
drogen. Oxygen, sulfur, or nitrogen can substitute the duces reductions in hardness and stiffness. Plasticizers
place of carbon [4]. Thus, a wide range of polymeric are commonly used in polymers that are intrinsically
compounds can be created. brittle at room temperature, such as polyvinyl chloride
The process of forming large molecules (macro- and some of the acetate copolymers. In effect, the plas-
molecules) from small ones (i.e., monomers) is called ticizers lowers the glass transition temperature, so that
polymerization; polymerization is the process of join- at ambient conditions the polymers may be used in ap-
ing many monomers, the basic building blocks, together plications requiring some degree of pliability and duc-
to form polymers. In other words, the polymerization tility. Plasticizers are usually liquids with high boiling
process is a reaction by which single molecules are points.
linked to form large molecules. There are two impor- Based on their thermal characteristics, polymeric
tant classes (mechanisms) of polymerization: materials are usually divided into two main groups:
(a) thermoplastic or (b) thermosetting, produced by
either condensation or addition polymerization. The
(a) Addition polymerization (or Chain reaction poly- term thermosetting and thermoplastic refer to the mate-
merisation). Addition polymers can be considered rial’s response to elevated temperature. Thermosetting
as long chains of tightly bonded carbon atoms with polymers are those with three-dimensional framework
strongly attached pendants of hydrogen, fluorine, or structure which all atoms are connected by strong cova-
benzene rings. All bonds within the molecules are lent bonds. These materials generally result from con-
strong primary bonds. The attraction between neigh- densation polymerisation, in which elevated tempera-
boring molecules, however, is only by the much weaker ture tends to promote an irreversible reaction, hence
van der Waals forces. In this process monomers join to the term thermosetting. Once these materials are set,
form a polymer without producing any by-product. This additional heatings do not produce softening. Instead,
process is generally carried out in the presence of cata- the materials maintain their mechanical properties up
lysts. Depending upon the properties of each monomer, to the temperature at which they char or burn. Defor-
a range of characteristics may be obtained. For example, mation requires the breaking of primary bonds, so that
the copolymer of vinyl chloride and vinyl acetate to pro- these polymers tend to be strong, but brittle. As a class,
duce PVC is one of the important industrial thermoplas- the thermosetting polymers are significantly stronger
tics. For these materials, the mechanical and physical than the thermoplastics and have a lower ductility, a
properties are largely determined by the intermolecular higher modulus of elasticity, and poorer impact proper-
forces. Because these secondary bonds are weakened ties. Thermosets undergo a curing reaction. Curing can
by the elevated temperature, polymer of this type soften be performed by the use of heat, radiation, or light (pho-
with increasing temperature and become harder and toinitiation), moisture, activators, catalysts, multiple-
stronger when cooled. Activators, like H2 O2 → 2OH, component reactions, or combinations thereof [4]. The
initiate and terminate the chain. Thus, the amount of ac- curing reaction of the thermoset polymers can be initi-
tivator relative to the amount of monomer determines ated by the following methods: (a) appropriate chem-
the average molecular weight or length of the chain. ical agents (b) application of heat and pressure, or (c)
The average number of mers in the polymer is known by exposing the monomer to an electron beam. Ther-
as the degree of polymerisation and ranges from 75 to mosetting plastics are classified according to the resin
750 in most commercial polymers. used. The main classes of thermosetting polymers [5]
(b) Condensation (or Step reaction) polymerization. are: (a) phenol formaldehyde, (b) urea formaldehyde,
In this process a stepwise reaction of molecules occurs (c) melamine formaldehyde, (d) polyesters, (e) epoxies,
and in each step a molecule of a simple compound, gen- and (f) silicones. In general, the most useful adhesive
3
materials for structural stressed applications are the to heavy sections without distortion, (c) joining heat-
thermosets and their alloys. Because the molecules of sensitive alloys, (d) producing bonds with unbroken
the thermosts are densely cross-linked, their resistance surfaces, (e) bonding small adherends. For example, it is
to heat and solvents is good and they show less elastic hard to imagine welding abrasive grains to a paper back-
deformation under load than the thermoplastics. Ther- ing to make sandpaper, or bolting the grains together
moplastics are the polymeric materials that flow under to make a grinding wheel, (f) Adhesive bonds have
the application of heat and pressure, i.e., they soften lower stress concentrations than mechanical joints; in
or become plastic on heating. Thermoplastic materials the adhesive joining of large adherends, resulting in low
harden when cooled to room temperature. As a result of stresses, and holes (necessary for riveting or bolting),
temperature change, their different behavior is due to which invariably act as stress concentrators in the ad-
the following factors: (a) molecular structure and shape, herends, are eliminated, thus lowering the possibility of
(b) molecular size, or mass, and (c) type of bond (cova- adherend failure. (g) In addition to joining, adhesives
lent or van der Waals). Since they become increasingly may also act as seals against the penetration of fluids.
softer with increase in temperature, certain members In the case of corrosive fluids, this, coupled with the
of the thermoplastic family are liable to permenant dis- absence of holes, where corrosion usually gains an ini-
tortion under mechanical strain at relatively low tem- tial foothold, can reduce corrosion problems. (h) Adhe-
perature (i.e., 60◦ C). They may flow to an appreciable sive bonded joints weigh less than mechanically bonded
extent under load at room temperature. Thermoplastics joints; in terms of weight, it does not take much adhe-
have generally higher impact strength and pleasing ap- sive to join much larger adherends. As a result, it is not
pearence, and can be converted into a finished product surprizing that many of the newer high-performance
at lower manufacturing cost. adhesives were originally developed for aerospace ap-
There are five major areas of application for poly- plications. (i) Adhesive bonding has less sensitivity to
mers: (a) plastics, (b) rubbers or elastomers, (c) fibers, cyclic loading. (j) Adhesive bonds permit smooth ex-
(d) surface finishes and protective coatings, and (e) ternal surfaces at the joint. (k) Adhesive joining may
adhesives. Adhesives have been a technologically im- offer economic advantages, often by reducing the hand
portant application of polymers for many years. This labor necessary for other bonding techniques.
class of materials possesses many interesting and use-
ful properties that are completely different from those of
the more tradational engineering materials, such as met- 1.2. Surface pretreatments
als and ceramics, and that can not be explained or han- Surface preparation is, perhaps, the most important pro-
dled in design situations by the tradational approaches. cess step governing the quality of an adhesive bond.
The properties of these materials depend on the molec- Structural adhesive bonding of joints is achieved ei-
ular structure of polymers. This molecular structure is ther by mechanical interlocking of the polymer with
the key to an understading of the science and technolgy the adherend surface or chemical bonding of the poly-
of polymers. An adhesive is defined as a polymeric sub- mer molecules with the metal oxide [2]. To improve
stance with viscoelastic behavior, capable of holding bond strength and durable adhesive joint between dif-
adherends together by surface attachment to produce ferent substrates, surface preparation is a necessary pre-
a joint with a high shear strength. Some thermoplas- treatment prior to adhesive bonding. Certain bonding
tic and thermosetting polymers are used as adhesives. techniques provide adequate static strength, but have
They are tough, strong, and reliable, and can be applied little durability when exposed to hot moisture environ-
almost any comination of materials. ments, while others are susceptible to debonding in the
In thousands of years people have used natural ad- presence of harsh environmental conditions, i.e., fuels,
hesives (glues) made from animal sources. Many of oils and cleaning solvents commonly encountered in
the early natural adhesives are still used. These in- industry such in aircraft application [2]. In addition,
clude starch and protein-based formulations such as the nature of the surface treatment prior to bonding is
hydrolyzed collagen from animal hides, hooves, and found to be a major influence in the control of this effect
bones and casein from milk. However, with advances [12]. Surface pretreatment increases the bond strength
in polymers in recent years, synthetic adhesives with by altering the substrate surface in a number of ways in-
superior characteristics have appeared. As new adhe- cluding [2]: (a) increasing surface tension, (b) increas-
sive formulations based on synthetic polymers (often ing surface roughness or (c) changing surface chem-
the same polymers used in other applications) con- istry. By increasing surface roughness, an increase in
tinue to be developed, the range of applications for surface area occurs which allows the adhesive to flow
adhesives has expanded dramatically (see for example in and around the irregularities on the surface to form a
[6–9]). One important aspects of this rapidly growing mechanical bond [2]. Changing surface chemistry may
field is that thermosets, thermoplastics, and elastomers result in the formation of a chemical bond e.g., between
have all found applications as adhesives, and so-called the polymer molecules in the polymer matrix compos-
alloys have been developed in which more than one ite and the metal oxide layer on the other adherend layer
type of polymer is used. [2, 13–16].
Adhesively bonded joints between various substrates As a result, the nature of the surface will also in-
have the following advantages [5, 10, 11]: (a) the ca- fluence the stability of the joint. When exposed to
pability of joining dissimilar materials without regard hot/humid environmental conditions, a polymeric ad-
to galvanic corrosion, (b) bonding very thin sections hesive/polymer interface is much more stable than the

4
equivalent polymeric adhesive/metal interface. A well- some ceramics. They can be applied as drops, beads,
chosen polymeric adhesive/polymer interface is un- pellets, tapes, or coatings (films) and are available in
likely to fail because of the environmental-induced the form of liquids, pastes, gels, and solids.
stress due to the nature of the bond formed [2]. On the In the present review, adhesive systems are classi-
other hand, the durability of a polymeric adhesive/metal fied on the following basis: (a) Organic chemistry, (b)
joint is not as stable. Early studies in the 1960s revealed Intended applications, and (c) high-temperature com-
that these joints did not perform well in hot/wet condi- posites as adherends. With such a range of possibilities,
tions with frequent occurrence of short-term interfacial the selection of the best adhesive for the task at hand
bond failure. can often be quite challenging.
Hart-Smith [17] has studied the interfaces in bonded, Adhesive systems on the basis of organic chemistry
co-bonded, and co-cured composite structures using a are categorized into five different systems that accom-
peel-ply durability test coupon in order to assess dura- plish the objectives [10]: (a) Solvent-based adhesives,
bility of these materials. The composite structures fail (b) Latex adhesives, (c) Pressure-sensitive adhesives,
during the service, even though they appeared to have (d) Hot-melt, and (e) Reactive adhesives. Whereas in-
been manufactured correctly, according to the short- tended applications range [4] from (a) load-bearing
term quality control coupons tested at the time of ini- (structural adhesives) to (b) light-duty holding (non-
tial fabrication. The issue is not one of structural over- structural or fixturing adhesives, to (c) sealing (the
loading. The interface between the adhesive and the forming of liquid or gas-tight joints). Commonly used
composite surfaces simply disbonds without the trace structural (i.e., load-bearing) adhesives include [4] (a)
of either material ever having adherend to the other. epoxies, (b) cyanoacrylates, (c) anaerobics, (d) acrylics,
What appears to be a common element in many of (e) urethanes, (f) silicones, (g) high-temperature adhe-
these problems is contamination introduced by the use sives, and (h) hot melts. For these materials, the bond
of a “released” peel ply without subsequent thorough can be stressed to a high percentage of its maximum
abrasion of the bonding surfaces. Another common but load, for extended periods of time, without failure.
unrelated contributor to premature interfacial failures Because of fatigue considerations, whenever pos-
is some form of pre-bond moisture. Moisture can be sible, it is preferable to bond rather than mechan-
present in an undried laminate cured long before it ically fasten composite structures. For this reason,
was bonded, as condensation on the surface of adhe- the increased usage of high-temperature resin-matrix
sive film not stored or thawed out properly, or absorbed systems for composite materials has necessitated the
within adhesive film which had been left out too long development of compatible and equally heat stable ad-
in the lay-up room. Water absorbed by the nylon fil- hesive systems. Therefore, the adhesives used for bond-
aments before the original laminates are cured with ing the composite materials can be classified as [18]:
the peel ply in place is also identified as a contribut- (a) high-temperature epoxy, (b) epoxy-phenolic, (c)
ing problem. The cause of such weak bonds should be condensation-reaction polyimide, (d) addition-reaction
acknowledged as processing errors. This is that such polyimide, and (e) bismaleimide adhesives [19, 20].
weaknesses are invariably undetected by conventional In the following sections, we will briefly summa-
non-destructive testing (NDT) techniques such as ul- rize the above-mentioned adhesives in relation to the
trasonic inspections, suggesting to some that the parts bonding of various substrates such as metallic alloys,
might not be detective after all. It took almost a decade polymers and composite materials.
to acknowledge the criticality of surface preparation for
metal-bonded structures [17].
In present review, the following subjects have been 2.1. Epoxy adhesives
studied: (a) first of all, the available adhesive mate- Epoxies are a broad family of polymer materials charac-
rials including their recent advances are classified on terized by the presence of epoxy groups in their molec-
the following basis; (i) organic chemistry, (ii) intended ular structure. Although high-molecular-weight linear
applications, and (iii) adhesives for high-temperature epoxies are often used as thermoplastics, they are most
composites as adherends, (b) secondly, the thermody- often used as thermosetting materials that cross-link to
namic factors affecting surface pretreatment methods form a three-dimensional non-melting matrix. The ther-
including the adhesion theory, wettability, and surface mosetting epoxies are the oldest, most common, and
energy of the substrates have been summarized, and (c) most diverse of the adhesive systems, and can be used to
finaly, available surface pretreatment methods as well join most engineering materials, including metal, glass,
as their recent progress for various substrates includ- composite, and ceramic. Just as epoxies are the most
ing polymers, polymer-matrix composites and metallic common matrix for advanced composites, epoxy ad-
alloys have been presented. hesives are also the most common. Epoxy adhesives
can be either one-stage (curing agent already mixed in)
or two-stage where the user mixes in the curing agent
2. Adhesive materials systems: just before use. The form of the one-stage material is
available adhesives most often a sheet, very much like a prepreg without
In adhesive bonding, a nonmetallic material (i.e., adhe- the reinforcement, or in a paste. Both room tempera-
sive) is used to create a joint between two surfaces. The ture and elevated temperature curing systems are used,
actual adhesives span a wide range of material types although many times the room temperature curing ad-
and forms, including (a) thermoplastic resins, (b) ther- hesives require postcuring to develop good mechani-
mosetting resins, (c) artificial elastomers, and (d) even cal properties at elevated temperatures. Cure times can
5
range from a few minutes for simple noncritical parts to significant thermal stresses due to the difference in
to more than 12 h for large, critical-performance parts coefficient of thermal expansion between metal and the
[11]. All adhesives are sensitive to the surface condi- composite. These induced stresses can result in the loss
tions of the materials to be joined. However, epoxies of dimensional stability, disbonds, or delamination in
are more sensitive than other adhesive materials in the these parts. In order to overcome these problems the
case of metal. They have the following advantages: (a) 121◦ C curing film adhesive (i.e., FM
R
300-2) was de-
strong, versatile adhesives that can be designed to offer veloped for providing the performance of 177◦ C curing
high adhesion, (b) good tensile and shear strength, (c) film adhesives (i.e. FM
R
300) in both metal and com-
high rigidity, (d) good chemical resistance, (e) excel- posite bonding applications [21]. This adhesive film
lent bonding, (f) good creep resistance, (g) easy curing designated FM
R
300-2 is a 121◦ C cure version of 177◦ C

R
with little shrinkage, and (h) good tolerance to elevated curing FM 300 adhesive film and provides similar
temperatures. Various epoxies can be used [4] over the stress-strain and mechanical performance to the FM
R

temperature range −51 to 260◦ C. After curing at room 300 system [22, 23]. This modified version of FM R
temperature, shear strengths can be [4] as high as 35 to 300 adhesive (i.e., FM
R
300-2) was developed for var-
70 MPa. ious bonding applications where higher performance is
The heat is used as the curing agent for single- needed but the cure temperaturte can not be higher than
component epoxies. Majority of the epoxies, however, 121◦ C.
are two-component blends involving (a) a resin and (b) The adhesive system, designated as FM
R
300-2, is
a curing agent, plus (c) possible additives such as ac- based on epoxy chemistry and is designed for bonding
celerators, plasticizers, and fillers that serve to enhance metallic and composite structures as well as structures
cure rate, flexibility, peel resistance, impact resistance, fabricated from metallic, Nomex or fiberglass honey-
or other characteristics. Heat may again be used to drive comb. This adhesive system has been formulated to
or accelerate the cure. fully cure within 90 min at 121◦ C and has flow and
Epoxy adhesives based on multifunctional resins [18] handling properties similar to FM
R
300 adhesive.
are available which exhibit excellent strength retention
at temperatures up to about 225◦ C. Where long-term
aging is required, epoxies are generally limited to ap- 2.2. High-temperature adhesives
plications requiring continuous service at temperatures High-temperature adhesives are specified [4] if their
no higher than 175◦ C. The adherends involved are most strengths keep constant at temperatures over 290◦ C. For
commonly aluminum alloys and epoxy-matrix compos- example, epoxy phenolics, modified silicones, pheno-
ite structures. lics, polyamides, and some ceramics can be classified as
However, these adhesives have the following disad- high temperature adhesives. Epoxy-phenolics rival the
vantages: (a) the relatively low peel strength, and flexi- polyamide adhesives in their ability to withstand short-
bility, (b) the bond strength is sensitive to moisture and time exposure to extremely high temperatures. For this
surface contamination, (c) often brittle at low temper- reason, these adhesives are suited for use on missiles,
atures, (d) the rate of curing is comparatively slow, (e) and are generally preferred to polyimides because of
relatively high resin cost. Sufficient strength for struc- their lower cost and ease of processing. Bond pressures
tural applications is generally achieved in 8 to 10 h, of 0.28 MPa and cure temperatures of 150◦ C are usually
with full strength often requiring 2 to 7 days [4]. adequate for most applications [18]. High-temperature
The new epoxy adhesives developed for use on adhesives are primarily used in the aerospace industry.
aerospace industry have excellent strength retention at High-temperature adhesives based on condensation-
temperatures up to 215◦ C. Strength then drops [18] reaction polyimide adhesives (precursors) have been
rather sharply to 6.9 MPa at 260◦ C. After 3000 h aging marketed for over 15 years. Adhesives of this type are
at 215◦ C, the adhesive retains approximately 80% of supplied in both liquid form and as solid films. In addi-
its original lap shear strength [18]. tion, they may be filled or unfilled. Aluminum-powder-
filled adhesives are superior in strength to unfilled ad-
hesives and are generally used for metal-bonding ap-


R
plications [18]. The disadvantages of these adhesives
2.1.1. FM 300 epoxy based film adhesive [4, 18] are (a) high cost, (b) long curing times, (c) high
and its modified version


R
cure temperatures, 260◦ C or even higher, depending on
(i.e., FM 300-2) the expected service temperature, (d) the high volatile
For aerospace bonding applications, FM
R
300 epoxy content (about 12% for aluminum-filled adhesives and
based film adhesive is widely used for bonding metal 30% for unfilled adhesives) of condensation-reaction
to composite structures (i.e., metal-to-composite sub- polyimides.
strate). Some of these applications range from bonding FM34
R
was perhaps the first polyimide adhesive de-
the wing-root assemblies (titanium to graphite epoxy), veloped that gained any significant commercial accep-
to composite sandwich structures on F-18 fighter air- tance. Its performance at temperatures up to 540◦ C
craft, to bonding and surfacing applications on most showed that its strength retention up to 540◦ C did not
commercial and military worldwide [21]. The 177◦ C significantly differ from that of the more easily pro-
curing adhesive is used in both co-cure and secondary cessed epoxy-phenolic adhesive HT
R
424 [18]. The
bonding applications. thermal stability of this adhesive [18] was superior,
In the case of bonding metal to composite parts, the showing no significant drop in strength over 40.000
177◦ C cure temperature for secondary bonding can lead h at 260◦ C.
6
2.3. Hot melts adhesives can be accelerated by heat, and at least one variety cures
Thermoplastic materials (which are solids at room tem- with ultraviolet light. Comparing these with other ad-
perature) are also used as adhesives for composites. hesives, the acrylic-based thermoplastic adhesives offer
These materials are called hot melt adhesives since they strengths comparable to the epoxies, good resistance to
are melted to become fluid so they can be applied to the water and humidity, and the added advantages of room
materials to be joined and then they reharden when temperature curing and a non-mix application system
cooled. Hot melts most often are used when speed in [4]. The disadvantages of these adhesives include (a)
bonding is important. They should be avoided in ap- low high-temperature strength, (b) flammability, (c) an
plications where creep would be a problem. However, unpleasant odor when still uncurred, and (d) compara-
the thermoplastics generally have good peel and en- tively expensive. This materials also have highest opti-
vironmental resistance properties [11]. Although the cal clarity, transmitting over 90% light.
hot-melt adhesives are not generally considered to be
true structural adhesives, they are being used increas-
ingly to transmit loads, especially in composite mate- 2.5. Cyanoacrylates adhesives
rial assemblies. They are thermoplastic resins which These adhesives are liquid monomers that polymerize
are solid at room temperature but melt abruptly when when spread into a thin film between two surfaces [4].
heated into the range of 100 to 150◦ C [4]. The ther- The cyanoacrylates are distinguished by their rapid set-
moplastics often form good adhesives simply by being ting properties at room temperature and by the very
melted to cause flow and then solidifying on cooling wide range of materials which they will bond; trace
after contacting the surfaces under moderate pressure amounts of moisture on the surfaces promote curing at
[10]. Many polymers can be used as hot-melt adhe- very high speeds, often as little as 2 seconds. Thus the
sives on certain adherends, including metals in some cyanoacrylates offer a one-component adhesive system
instances. Polyamides and poly(ethyelene-co-vinyl ac- that cures at room temperature with no external impe-
etate) are used frequently as hot-melt adhesives. Melted tus. Commonly known as superglues, this family of
nylon is capable of wetting many substances and giving adhesives is now available in the form of liquids, gels,
moderate adhesion; similarly the polyacetal “Delrin,” toughened versions designed to overcome brittleness,
and a number of vinyl polymers and polycarbonates. and even nonfrosting varieties. The cyanoacrylates have
The adhesion of all these can generally be improved by the following advantages [4]: (a) the excellent tensile
priming the surface of the adherend with a dilute so- strength, (b) fast curing, (c) good shelf life, and (d)
lution of a phenolic resin-which almost classifies them adhere well to most commercial plastics. The disad-
as two-polymer adhesives. A method of application is vantages of these adhesives, however, are their (a) high
to position the adhesive in the joint prior to operations cost, (b) poor peel strength, (c) brittleness, (d) poor
such as the paint bake process in automobile manu- bond properties at elevated temperatures.
facture. During the baking, the adhesive melts, flows
into seams and crevices, and seals against the entry of
corrosive moisture. The hot melts provide reasonable 2.6. Anaerobic acrylic adhesives
strength within minutes, but do soften and creep when Anaerobic acrylic adhesives are monocomponent sys-
exposed to elevated temperatures and become brittle tems which are able to cure via a redox radical mech-
when cold. anism at room temperature and in the absence of air,
hence their name [25]. Anaerobic adhesives belong
to the large family of thermosetting acrylic polymers.
2.4. Acrylics (-based thermoplastic These polyester acrylics adhesives remain liquid when
adhesives) exposed to air [4]. When confined to small spaces and
Acrylic adhesives permit the tailoring of the flexibil- shut off from oxygen, as in a joint to be bonded, the
ity of the adhesive [11]. Two acrylic molecules (methy polymer becomes unstable. In the presence of iron or
methacrylate and acrylic acid) are copolymerized. Nor- copper, it polymerizes into a bonding-type resin with-
mally these materials are supplied as thin fluids. The out the need for elevated temperature. Additives can
acrylic-based thermoplastic adhesives have the follow- reduce odor, flammability, and toxicity and speed the
ing advantages: (a) one of the most attractive properties curing operation. Slow-curing aneorobics require 6 to
of acrylic polymers is their resistance to aging, which 24 h to attain useful strength [4]. With selected additives
results from the ability of these polymers to absorb ul- and heat, however, curing can be reduced to as little as
traviolet light in the solar spectrum and redissipate it as 5 min. These adhesives can bond almost anything, in-
harmless energy in the infrared wavelength range [24], cluding oily surfaces. The joints resist vibrations and
(b) good strength, toughness, and versatality, and (c) offer good sealing to moisture and other environmental
be able to bond a variety of materials, including plas- influences. Unfortunately, they are rather brittle and are
tics, metals, ceramics, and composites, even through limited to service temperatures below 149◦ C.
oily or dirty surfaces [4]. Most involve application sys- Industrial use (mainly in the fields of automotive,
tems where a catalyst primer (curing agent) is applied aeronautics, electronics etc.) of anaerobics covers five
to one of the surfaces to be joined and the adhesive is applications [26]: (a) fitting (e.g., bearings into hous-
applied to the other. The surface pretreated parts can be ings, (b) locking (e.g., nuts onto bolts), (c) sealing (e.g.,
stored separately for weeks without damage. The com- liquid gaskets), (d) retaining (e.g., shafts into hubs),
ponents react to produce a strong thermoset bond at and (e) bonding (e.g., flanged couplings). As retaining
room temperature when adhesively joined. The curing and bonding media, they can either replace classical
7
systems based on shape (keys, splines, pins) or fric- ibility is required (as in sheet metal parts), or good gas-
tion (interference fits, elastic rings, bolted joints) or, ket or sealing properties are necessary. These adhesives
most profitably, can enhance the overall performance by can be used to join the metals, glass, paper, plastics, and
combining with them. Typical benefits include elimina- rubbers. However, these adhesives have the following
tion of backlash, reduction of fretting wear, improved disadvantages: (a) relatively expensive, (b) curing is
corrosion resistance, airtight sealing action and remark- rather slow.
able increases of stiffness and strength. These adhesives
have the following main advantages [26, 27]: (a) pre-
liminary mixing not required, (b) handling time of ap- 2.9. Pressure-sensitive adhesives
proximatetly 15 min so that they are suited with rhythms As the pressure-sensitive adhesives (PSAs) are viscous
imposed by assembly lines, and (c) final resistance ob- polymer melts at room temperature, these polymers
tained after 24 h, (d) elimination of backlash, (e) reduc- must be used above their glass transition temperatures
tion of fretting wear, (f) improved corrosion resistance, [10, 32]. They are caused to flow and contact the ad-
(g) airtight sealing action, and finally (h) remarkable herends by applied pressure, and when the pressure is
increases of stiffness and strength. However, as most released, the viscosity is high enough to withstand the
adhesives, anaerobic adhesives require a decontamina- stresses produced by the adherends, which obviously
tion of materials to be bonded and a surface treatment can not be very great. The key property for a polymer
which can sometimes modify some characteristics of used in this application is tack, which basically is a vis-
polymerization [28]. cosity low enough to permit good surface contact, yet
Following improvements have been made to ensure high enough to resist separation under stress, something
a higher level of performances for these adhesives on the order of 104 –106 cP, although elasticity probably
[25]: (a) higher toughness by coupling polyurethane plays a role, also [10, 33]. Natural, SBR, and reclaimed
to acrylic monomers [29], (b) increased thermal resis- rubbers are common in this application. The many vari-
tance by use of polyamides with basic monomers [30], eties of pressure-sensitive tape are faced with this type
(c) greater resistance to impact and shearing stress by of adhesive. Contact cements are a variation in which
the incorporation of additives as elastomers [31]. the rubbery polymer is applied to each adherend surface
in the form of a solution, or increasingly, a latex. Evap-
oration of the solvent or water leaves a polymer film
2.7. Urethane adhesives with the tack necessary to grab and hold the adherends
These adhesives are a large and diverse family of poly- when they are pressed together.
mers that are generally applied at temperatures below For good adhesive properties, the glass transition
65◦ C and components that can undergo great elongation temperature for pressure-sensitive adhesives must be
[4]. Urethanes are available as both one-part thermo- between 50 and 70◦ C below the temperature of use
plastic and two-part thermosetting systems. In general, [32, 34, 35]. If the glass transition temperature Tg is too
they cure quickly to handling with 24 h to complete cure far below the temperature of use, the viscoelastic losses
at room temperature. In comparing the other structural in the debonding process will be too low. On the other
adhesives, the urethanes offer good low-temperature hand, if the Tg is too close to the temperature of use,
adhesion coupled with good flexibility and toughness. the compliance will become too low, making it difficult
These adhesives are relatively moisture sensitive, dete- to establish good contact with the adherend. Consider-
riorate in many chemical environments, and can involve ing that most applications for PSAs are for room tem-
toxic components or curing products. perature or slightly above, the most useful polymers
have Tg s between −50 and −30◦ C, with some ex-
ceptions [35]. One common solution for retaining the
2.8. Silicone adhesives useful properties of a polymer while adjusting its Tg in-
Silicones are semiorganic spine molecules with alter- volves compounding with tackifying resins that allow
nating silicon and oxygen atoms.This family of syn- the use of polymers with a lower Tg .
thetic polymers is partly organic and partly inorganic.
Silicones are classified as fluids, elastomers, and resins.
Thus, silicone polymers may be fluid, gel, elastomeric, 2.10. Addition-reaction to polyamide
or rigid in their form. Good physical properties are adhesives
maintained from 232 to 260◦ C. The properties that These adhesives make it possible to obtain bondlines
make silicones attractive to engineers include: (a) low with extremely low void contents. Adhesives of this
surface tension, (b) high lubricity with rubber and plas- type are generally supplied as supported film containing
tic surfaces, (c) excellent water repellency, (d) good sufficient alcohol to impart tack and drape to the adhe-
electrical properties, (e) thermal stability, (f) chem- sive. Unlike the condensation-reaction type, addition-
ical inertness, (h) resistance to oxidation, hot wa- reaction polyimides (PIs) remain thermoplastic after
ter and weather, and (i) retain their flexibility at low imidization and solvent removal have occurred. In con-
temperatures. trast to the condensation-reaction PIs, which can be pro-
The silicone containing thermoset adhesives cure cessed using the postcure concept, addition-reaction PIs
from the moisture in the air or adsorbed moisture from must be held under pressure throughout the cure cycle.
the surface being joined. They form low-strength struc- Therefore, relatively costly high-temperature presses or
tural joints and are usually selected when considerable autoclaves are required to fabricate panels with addition
expansion and contraction is expected in the joint, flex- PI adhesives. Composites made from addition-reaction
8
PIs are generally fabricated using 1.4 MPa pressure to the mixture is compression molded to form grinding
ensure low void content in the laminate. wheels. For example, A B-stage phenolic in a solvent
is used to impregnate tissue paper. The solvent is evap-
orated, and the dry sheets are placed between layers
2.11. Latex adhesives of wood in a heated press, where the resin first melts
Natural rubber originates as a latex, i.e., an aqueous and then cures, bonding the wood to form plywood.
dispersion of minute rubber particles which are kept in Similarly, sheets of paper impregnated with a B-stage
suspension by an adsorbed layer of protein. These ad- melamine-formaldehyde resin are laminated and cured
hesives are based on polymer latexes made by emulsion to form the familiar Formica
R
counter tops.
polymerization; polymer particles are on the order of Unlike the previous examples of reactive adhesives,
0.05 to 0.15 µm stabilized by the soap [10]. The total the phenolics and other formaldehyde condensation
solid particles content of the latex as obtained from the polymers evolve water as they cure. If trapped in the
tree is in the region of 30 percent. Similarly many syn- joint, this can result in serious weakness, which limits
thetic rubbers and other flexible polymers are initially their adhesive applications. Note that all these exam-
produced by a process known as emulsion polymeriza- ples of reactive adhesives are highly polar polymers.
tion in which the polymer is first produced as latex or It is largely this polarity that accounts for their good
emulsion similar in appearence to natural latex but sta- bonding capabilities.
bilized by synthetic stabilizing and dispersing agents
instead of proteins.
They flow easily while the continuous water phase 2.13. Bismaleimide (BMI) adhesives
is present and dry by evaporation of the water, leaving A closely related reactive compound contained within
behind a layer of polymer. In order that the polymer par- additional polyimides (API) is the bismaleimide (BMI)
ticles coalesce to form a continuous joint and be able family of materials [36–38]. The BMI resin system
to flow to contact the adherend surfaces, the polymers can vary from a crystalline diffusional compound to an
used must be above their glass transition temperature oligometric amorphous material with difunctionality. In
at use temperature. These requirements are similar to recent years many BMI resins and modifications have
those for latex paints, so it is not surprizing that some been developed primarily due to good drape and tack
of the same polymers are used in both applications, for and “epoxy-like” cure conditions resulting in resins
example, stryene-butadiene copolymers and polyvinyl with higher temperature characteristics. BMIs possess
acetate. Nitrile and neoprene rubbers are used for in- thermal capabilities between 175–230◦ C (epoxies 120–
creased polarity. A familiar example of a latex adhesive 175◦ C). BMI resins are very rigid system with a cor-
is “white glue,” basically a plasticized polyvinyl acetate responding high degree of crosslinking, hence brittle-
latex. Latex adhesives are displacing solvent-based ad- ness. This characteristic leads to microcracking and
hesives in many applications because of their reduced high shrinkage.
pollution and fire hazards. They are used extensively BMI adhesives fill a niche between high-temperature
for bonding pile to backing in carpets. epoxy and PI adhesives. Unmodified BMI resins are
hard, brittle solids. In order to impart drape to adhesive
films based on these resins, they must be plasticized.
2.12. Reactive adhesives However, unlike the PI resins, plasticization may be
These adhesives are either monomers or low molecu- accomplished without resorting to the use of solvents.
lar weight polymers which solidify by a polymeriza- Volatile-free adhesive films having drape can be for-
tion and/or crosslinking reaction after application [10]. mulated through the use of reactive liquid monomers
They can develop tremendous bond strengths and have as plasticizers. Cure temperature requirements vary de-
good solvent resistance and high-temperature prop- pending on the particular BMI resin employed and the
erties. The most familiar example of reactive adhe- specific monomeric plasticizers selected. In general,
sives are the epoxies generally cured by multifunctional good results can be obtained using a cure of 2 h at
amines. Polyurethanes also make excellent reactive 175◦ C under 0.28 MPa pressure followed by a 2- to 4-h
adhesives. postcure at 200 to 225◦ C [18].
The α-alkyl cyanoacrylate “super glues” are now a BMI adhesives’ strength retention is surprisingly
familiar part of the consumer market. Originally, the good up to about 300◦ C. However, in applications
monomers had extremely low viscosities, and so could where term exposure to high temperatures is involved,
crawl into narrow crevices and wet the adherend sur- BMI adhesives are not expected to be durable beyond
faces rapidly. On the other hand, they would not fill 200 to 225◦ C. In conclusion, almost limitless variations
gaps, and were absorbed into porous adherends, giv- in formulation are possible with BMI adhesives.
ing poor bonds. Newer versions are available with
higher viscosities to overcome these drawbacks. Un-
fortunately, being linear and polar, they have poor re- 2.14. Solvent-based adhesives
sistance to polar solvents (acetone is a good solvent), Here the adhesive polymer is made to flow by dissolv-
and they are subject to hydrolysis, and so have poor ing in an appropriate solvent to form a cement; solutions
environmental stability. based on natural or synthatic rubbers are often called
Phenolic and other formaldehyde condensation poly- “cements”. Good surface-wetting and other application
mers are also important reactive adhesives. Powdered properties can be obtained by dissolving solid adhesive
phenolic resin is mixed with abrasive grains and constituents in organic solvents. The total solids content
9
may vary from 10 to 70 percent depending on the na- fillers can be used to provide thermal conductivity cou-
ture of the solid constituents, the type of solvent, and pled with electrical insulation.
the viscosity and other rheological properties required. Radiation-curing adhesives are are those designed to
The adhesive hardens by evaporation of the solvent. The cure by exposure to radiation, such as visible, infrared,
solvents used are mostly petroleum derivatives. Thus, or ultraviolet light, microwaves, or electron beams.
the polymers used must be linear or branched to allow These adhesives offer rapid conversion from liquid to
solution, and the joints formed will not be resistant to solid at room temperature and a curing mechanism that
solvents of the type used initially to dissolve the poly- occurs throughout, rather than progressing from ex-
mer. To get a good bond, it helps if the solvent attacks posed surfaces. These adhesives are used for applica-
the adherend also. In fact, solvent alone is often used tions as a wide variety of dental amalgams that can fill
to “solvent weld” polymers, dissolving some of the ad- cavities or seal surfaces while matching the color of the
herend to form an adhesive on application. remaining tooth.
One of the drawbacks to solvent-based adhesives
based on rigid polymers is the shrinkage that re-
sults when the solvent evaporates [10]. This can 3. Adhesion, wetting and tack
set up stresses that weakens the joint. An exam- In a simple system, bonding at an interface is due to ad-
ple of this type of adhesive is the familiar model hesion between the adhesive and adherend. One talks
airplane cement, basically a cellulose nitride so- of adhesion when a measurable amount of mechanical
lution, with perhaps some plasticizer. Rubber ce- work is necessary to separate two surfaces of differ-
ments, of course, maintain their flexibility, but can ent chemical composition or shape [35]. In order to get
not support as great a stress. Comercial rubber ce- good adhesion between adherend and adhesive, the first
ments are based on natural, SBR (poly(butadiene- essential is to ensure good molecular contact. Although
co-styrene), nitrile (poly(butadiene-co-acrylonitrile)), the maximum force necessary to break the bond may be
chloroprene(poly(2-chlorobutadiene)), and reclaimed defined as a measure of adhesion, the typical measure
(devulcanized) rubbers [10]. Examples are household of the strength of an adhesive bond across an interface is
rubber cement and Pliobond
R
. Rubber cements may the amount of energy necessary to break it, i.e., to sep-
also incorporate a curing agent to crosslink the poly- arate two surfaces. Such a separation may involve the
mer after application and evaporation of the solvent. breakage of chemical or van der Waals bonds, as well
This greatly increases solvent resistance and strength. as the plastic deformation of one or both of the bulk
materials on either side of the interface. Typically, in
all cases where good adhesion is obtained, the fraction
2.15. Nonstructural adhesives of energy necessary to break the bonds at the interface
The adhesives classified as nonstructural have the lim- is a very small fraction of the total energy necessary for
ited load-bearing capabilities. The nonstructural adhe- the separation of the two surfaces; most of the mechan-
sives are used in many different applications such as ical work is used to deform, under stress, the material
labeling and packaging. The hot-melt adhesives can be adjacent to the interface. Therefore, the measured en-
placed in this category. The followings are the some ergy of adhesion will be dependent on the ability of
examples of the important nonstructural or special ad- the interfacial bonds to sustain stress, as well as on the
hesives [4]: (a) Evaporative adhesives, (b) pressure- amount of plastic deformation caused locally by the
sensitive adhesives (PSAs), (c) Delayed-tack adhe- above-mentioned stress [35, 39].
sives, (d) conductive adhesives, and (e) radiation-curing Tack refers to the adhesion of two surfaces of the
adhesives. same rubbery polymer. When two such surfaces are
Evaporative adhesive use an organic solvent or wa- pressed together and subsquently pulled apart, the max-
ter base, coupled with vinyls, acrylics, phenolics, imum force necessary to break the junction depends on
polyurethanes, or various types of rubbers. Some com- the initial time of contact and the normal force applied,
monly identified evaporative adhesives are rubber ce- as well as the rate separation and the temperature and
ments and floor waxes. Pressure-sensitive adhesives are other variables [40–43]. From the dependence on tem-
usually based on various rubbers, compounded with perature and polymer molecular weight, it can be in-
various adhesives, and bond at room temperature with ferred that the effectiveness of the bond depends partly
a brief applications of pressure. No cure is required, and on the interdiffusion of molecules across the interface
the tacky adhesive-coated surfaces require no activation and hence on molecular motions which are reflected in
by water, solvents, or heat. This group of adhesives are viscoelastic properties [42–44]. However, the effective-
used as peel-and-stick labels, cellophane tape, and Post- ness depends on the ultimate properties of the polymer
it notes. See Section 2.1.9 for more detail description itself.
of these adhesives. If the two materials are the same material, the ad-
Delayed-tack adhesives are similar to the pressure- hesion is called self-adhesion or autohesion. However,
sensitive systems, but are nontacky until activated by in the case of polymers, self-adhesion is usually de-
exposure to heat. They then remain tacky for several pendent on the time of contact and, at infinite time, will
minutes to a few days to permit use or assembly. eventually become the cohesive strength of the material
Although most adhesives are electrical and thermal rather than an adhesive property. In the more general
insulators, conductive adhesives are generally produced case, where the two materials are dissimilar, adhesive
by incorporating selected fillers, such as silver, copper, bonds can be divided according to their strength (struc-
or aluminum flakes or powder. Certain ceramic oxide tural, permenant, and nonpermenant) and to the nature
10
of the adhesive/adherend pair [35]: (a) glassy poly- tle polymers give brittle joints, polymers with high
mer/glassy polymer, (b) glassy polymer/ non-polymer, shear strengths give bonds of high shear strength, heat-
(c) glassy polymer or non-polymer/elastomer or poly- resistant polymers produce bonds with good heat resis-
mer melt, and (d) polymer melt/polymer melt. tance, and so on.
The main applications of polymer/polymer adhesion To form a successful joint, the adhesive must inti-
in glassy polymers are in the development of alloys and mately contact the adherend surface. This requires first
blends made from immiscible polymers, and where the that it wets the surface. The subject of wetting is consid-
interfacial properties between the two phases are crucial ered in detail in treatises on surface chemistry [49, 50].
to the performance of the material [35, 39, 45]. The In general, wetting is promoted by polar secondary
ability of the interface to sustain stress is dependent on forces between adhesive and substrate [10], which is
the number of well-entangled polymer chains that cross another reason why low-polarity polymeric adherends
the interface [35]. are difficult to bond with adhesives. To insure proper
Glassy polymer/non-polymer adhesion occurs in wetting and interfacial bonding, the following require-
structural adhesives, where an adhesive strength equal ments are needed [10]: (a) it is generally necessary to
or superior to the cohesive strength of the base mate- pretreat the surface or “clean” the adherend surfaces
rial is required. Because interdigitation of the polymer carefully before joining [51], and (b) good contact also
chains is not possible, the surface of the solid material requires a viscosity low enough under conditions of ap-
and not equilibrium thermodynamics determines the plication to allow the adhesive to flow over the surface
topology of the interface, so that surface roughness, the and into its nooks and crannies. Once contact has been
chemical interactions between polymer and solid play established, the adhesive must harden to provide the
the role of stress transfer agents and the fracture energy necessary joint strength.
is usually due to the resulting plastic deformation of the It is commonly observed that roughening surfaces
polymer [35]. prior to bonding enhances the strength of adhesive
Polymer melt/polymer melt adhesion is mostly rel- joints, and many manufacturers specify the use of some
evant for the processing of elastomers prior to the form of abrasion as a surface treatment method. This
crosslinking process, and is of particular importance recommendation is based on the perception that the
to the tire industry. This type of adhesion is sometimes abrasive process removes loose contaminated layers
called tack or green strength [46]. and the roughened surface provides some degree of
Finally, polymer melt/solid adhesion is typical of mechanical interlocking or “keying” with the adhesive
non-permenant and relatively weak adhesion and is of- [52]. It is sometimes argued that the increased rough-
ten called pressure-sensitive adhesion, because once a ness also forms a larger effective surface area for the
certain amount of pressure has been applied, a mea- bond. While these mechanisms explain some of the gen-
surable adhesion between the two surfaces is observed. eral characteristics of adhesion to roughened surfaces,
Interdigitation of the polymer chains is impossible, so more detailed analyses suggest that the roughening pro-
chemical interactions, surface roughness, and the rhe- cess may introduce physico-chemical changes, which
ological properties of the adhesive are responsible for affect surface energy and wettability.
the measured energy of adhesion.
Although the detailed study of the adhesion is beyond
the scope of the present work, some important gener- 3.1. Interphase boundaries (or surfaces)
alization can be drawn. Adhesion can be attributed to Surfaces are important to the study of microstructures,
five main mechanisms which can occur at the inter- friction and wear, the joining of all materials by all
face either in isolation or in combination to produce the means, the catalysis of chemical reactions, oxidation,
bond (see Section 4 for more detail). Adhesion results corrosion, the mechanical behavior of small or thin bod-
[10, 47, 48] from (a) the mechanical bonding between ies, the design of electronic devices, and a wide variety
the adhesive and adherend, and (b) chemical forces- of other phenomena. The surfaces of phases always dif-
either primary covalent bonds or polar secondary forces fer in behavior from the bulk of the phases themselves,
between the two. The latter are thought to be the more because of the rapid structural changes which must oc-
important, and this explains in part why inert, non-polar cur at and near phase boundaries. If the forces on a
polymeric substrates such as polyethyelene and poly- molecule in the bulk are compared to the forces on a
tetrafluoroethylene are very difficult to adhesive bond: molecule at the surface, the forces on the bulk molecule
They must first be chemically pretreated to introduce cancel whereas the forces on the surface molecule are
polar sites on the surface. To promote mechanical bond- unbalanced. As a result of this unbalance force, equi-
ing, adherend surfaces are often roughened before join- librium bonding arrangements are disrupted, leading to
ing, but this is sometimes counterproductive. It can trap an excess energy (i.e., surface free energy, γ ), which is
air bubbles at the bottom of crevices which act as stress defined as the energy necessary to form a unit area of
concentrators to promote failure in rigid adhesives. new surface or the energy necessary to move a molecule
With good bonding between adhesive and adherend, from the bulk to the surface.
joint failure occurs cohesively (the adhesive itself or the The excess energy may be minimized by minimizing
substrate fails) [10]. Where the adhesive is weaker than surface area. This tendency is called surface tension if
the substrate, the properties of the adhesive polymer the surfaces are liquid and a vapor, γLV . Surface energy
determine the properties of the adhesive joint; that is, may also be lowered by segregation of the various com-
the bond can be no stronger than the glue line. Brit- ponents to and from the surface; such behavior is called

11
adsorption [53]. The magnitude of γ may be estimated
for metallic and covalent materials by considering the
number and enegy of the bonds which must be broken to
form the surface. Similarly, calculations of work done
against the Coulomb force lead to approximate values
of γ for ionic materials. In both cases, γ depends on
crystallographic orientation. Direct measurement of γ
is possible by force equilibrium, if the phases are suf-
ficiently mobile.
At the interface between two phases, the crystal struc-
ture, or the state of aggregation, or the composition must
change in a fairly abrupt manner. The atoms in the vicin-
ity of the surface are not in equilibrium states, since they Figure 1 Contact angle θ and surface tension (γ ) for a liquid drop on
are neither one phase nor the other. The excess energy a solid surface: SV, SL and LV stand for solid- vapor, solid-liquid, and
liquid vapor interfaces, respectively. Note that this figure indicates the
due to the perturbed material at the interface is propor- interactions between solid, liquid and vapor materials. The terms γSV ,
tional to the surface area. Thus, a drop of liquid will γLS , and γLV denote the surface energies of solid/vapor, liquid/solid, and
tend to assume a spherical shape in order to minimize liquid/vapor interfaces, respectively. Three different conditions of wet-
its surface area and, thereby, its surface free energy. In ting (i.e., wetting of a substrate by a liquid) [3, 35] are: (a) Partial wetting:
single-phase solids, similar surfaces exist. These are the 0◦ < θ < 180◦ ; γSV = γSL + γLV cos θ , (b) Complete wetting: θ = 0;
γSL + γLV < γSV , and (c) Non-wetting: θ = 180◦ ; γSV + γLV < γSL .
grain boundaries, which exist beteen grains of different
crystallographic orientations.
has a measured γc of 31 mJm−2 , with these resins and
some surface treatment will be required [54].
3.2. Wetting Various mechanisms can assist or impede adhesion.
When two electrically neutral surfaces are brought suf- A key concept in this regard is that of wettability. Wet-
ficiently close together there is a physical attraction tability tells us about the ability of a liquid to spread on
which is best understood by considering the wetting a solid surface [3]. To know the wettability of a surface
of solid surfaces by liquids. Wetting can be expressed is useful to predict its adhesion potential. The contact
in terms of the thermodynamic work of adhesion Wa angle θ on a surface can characterize its wettability by
which represents a physical bonding resulting from the contacting liquid in a given environment.
highly localized intermolecular dispersion forces be- Wettability of a solid by a liquid can be measured
tween different phases. The equation for the thermody- by considering the equilibrium of forces in a system of
namic work of adhesion for the separation of a liquid consisting of a drop of liquid resting on a plane solid
from a solid (both in equilibrium with the vapor phase) surface in the appropriate atmosphere. A liquid spreads
states that [11, 53, 54]: along a solid surface (i.e., wetting) rather than forming
a spherical drop when [53]
Wa = γSV + γLV − γSL (1)
γSL + γLV < γSV (2)
where γSV , γLV and γSL are the specific surface ener-
gies, or surface tensions of the solid-vapor, liquid-vapor that is, when the net free energy is lowered by replacing
and solid-liquid interfaces, respectively. This equation an S-V surface by an S-L and an L-V surface together.
can be related to the physical situation of a liquid drop On the other hand, no wetting at all will occur if
on a solid surface, as shown in Fig. 1a, by using the
Young equation (see Equation 4 below). More specifi- γSV + γLV < γSL (3)
cally, Fig. 1a defines the various phases for this equation
(Equation 4) as well as contact angle, θ, as simply the (Complete) wetting and nonwetting are shown in
solid phase on which the liquid drop rests, and the va- Fig. 1b and c together with the case of partial wet-
por is above the both solid and liquid, while the contact ting (Fig. 1a), which allows quantitative calculation of
angle is the angle between the liquid and the solid at the the surface energies. In the plane of the solid surface
intersection. The surface tension of solids is difficult to force equilibrium must exist between the three surface
measure whereas the surface tension of liquids can be tensions because the liquid droplet is free to move un-
determined relatively easily. It follows that glass and til force equilibrium is established. Hence, when the
graphite with theoretically calculated [55] surface en- forces at a point A are resolved in the horizontal direc-
ergies of 560 and 70 mJm−2 respectively will be readily tion, one can use the following expression [3, 53, 54],
wetted by polyester and epoxy resins with surface en- called Young’s equation,
ergies of 35 and 43 mJm−2 , respectively provided the
viscosity of the resins is not too high. Zisman intro- γSV = γSL + γLV cos θ (4)
duced the concept of critical surface tension of wetting
γc such that only liquids with γLV < γc will sponta- where θ is the contact angle which can be measured
neously spread on the solid [54]. This is a useful param- directly by a goniometer or calculated by using simple
eter in considering the wetting adhesives by adherends. trigonometric relationships involving drop dimensions.
In contrast it will be difficult to wet polyethelene, which A small θ implies good wetting. The extreme cases
12
being θ = 0◦ , corresponding to perfect wetting, and where γLV and γSV are the surface energies at the liq-
θ = 180◦ , corresponding to no wetting. A low contact uid/vapor and solid/vapor interfaces, respectively, θ is
angle, meaning good wettability, is a necessary but not the contact angle and the supercripts d and p represent
sufficient condition for strong bonding. One can have the dispersive and polar components.
excellent wettability and a weak van der Waals-type Wetting of the adherends by the adhesive is critical to
low-energy bond [3]. In practice, it is rarely possible the formation of secondary bonds in the adsorption the-
to obtain a unique equilibrium value of θ . Therefore, ory [57]. It has been theoretically verified that for com-
a range of contact angles exist between the maximum plete wetting (i.e., for a contact angle equal to zero), the
or advancing angle, θa , and the minimum or receding surface energy of the adhesive must be lower than the
angle, θr . This phenomenon is called the contact-angle surface energy of the adherend [58]. Therefore, the pri-
hysteresis, which is generally observed in polymeric mary objective of a surface pretreatment is to increase
systems. This hysteresis can be due to the chemical the surface energy of the adherend as much as possible
attack, dissolution, inhomogeneity of chemical com- [59].
position of solid surface, surface roughness, and local Wetting of liquid on solid surfaces is a topic
adsorption [3]. of fundamental interest with widespread technolog-
In principle the Young’s equation applies only to one- ical implications [60–63]. Examples include adhe-
dimensional spreading and becomes invalid if the sub- sive bonding, coating technology, thin film technol-
strate is not rigid and motion of the contact-line takes ogy, and lubrication. Wetting of solid surfaces is
place in both horizontal and vertical directions. The extremely sensitive to surface geometrical/chemical
force equilibrium ignores the vertical component of the (roughness/contaminants) disorder which manifests it-
surface tension which acts along the line of contact. As self by the contact angle hystheresis phenomenon
the capillary forces are not balanced, external forces [60–62]. In earlier models of geometrical disorder [63],
must be applied to the solid to achieve equilibrium. the case of periodic roughness h(x) = h o sin(2π x/ h)
These forces may produce even deformation in highly was considered. An apparent contact angle, θg ∼ =θ−
deformable solids, such as gels and rubber, destroying (dh/dx), if h o /λ < 1 was defined with respect to the
the co-planarity of interfacial tensions that is assumed average solid surface in the case of a contact line paral-
in Young’s equation and causing ridge formation at the lel to the grooves (allowing local application of Young’s
interfacial region. With the use of Young’s equation, it equation). The maximum advancing or receding appar-
has to be stressed that only a “quasi-equilibrium” exists ent contact angle was found to be θg ∼ = θ ± (2π h o /λ).
within the window of time when observations are made, This model predicts a hysteresis of the contact angle
provided that the solids deformation rate is small. when the contact line is parallel to the grooves [64].
Using Young’s equation (Equation 4) for the contact Palasantzas and De Hosson [64] have also investigated
angle and then substituting in Equation 1 results in the the effects of surface roughness on the wettability of
following equations: the substrate surfaces. In their study, surface rough-
ness was described by an rms amplitude , an in-plane
γSV − γSL roughness correlation length ξ (average distance be-
cos θ = (5)
γLV tween consecutive hills or valleys on the surface), and
a roughness exponent H (0 < H < 1). It was shown
WA = γLV + γLV cos θ (6)
that the apparent contact angle depends critically on the
WA = γLV (1 + cos θ) (7) roughness exponent H and long wavelength ratio /ξ .
For a contact angle θ determined by Young’s equation,
Since the surface tension, γLV , and the contact angle, smaller than certain transition angle θtr , the apparent
θ, can be determined experimentally, the work of ad- contact angle decreases with increasing roughness and
hesion, WA , can be found, as seen in Equation 7. How- vice versa for θ > θtr . The transition angle θtr appears
ever, the work of adhesion, WA , should not be equated to be smaller than 90◦ , and decreases with increas-
to the adhesive bond strength since the bond strength ing roughness exponent H (degree of surface irregu-
also contains the energy of dissipative process such as larity at short length scales, i.e., for the range smaller
viscoelastic deformation, plastic deformation, local mi- than ξ ). Roughness of adherend surface has frequently
crocracking etc. [11]. But under certain conditions, WA been used as a design parameter for adhesive joints. A
approximates the bond strength, Wpeel , The conditions number of researchers have examined its effect on the
where this occurs are low temperatures or high peel rate strength and durability of adhesive joints using various
[11]. adherends and adhesives [65, 66–70]. The relationship
The static contact angle data can be determined using between roughness and adhesion is not very simple.
a Video Contact Angle (VCA) Analyser [52]. The VCA Optimum surface profile varies from one adhesive to
can be used to register angles up to 180◦ on both sides another, and depends upon the type of stress applied
of the drop without moving the substrate. The surface [71]. As mentioned earlier, one of possible positive ef-
energy information is then calculated, in terms of dis- fects of surface roughness [71, 72], increase in surface
persive and polar or non-dispersive components, using area results in increasing intermolecular bond and key-
the geometric mean equation combined with Young’s ing for mechanical adhesion. This in turn can divert
equation (i.e., Equation 4) [56]. the failure path away from the interface into the bulk

 d d 1/2  p p 1/2  of the adhesive [73]. However, the actual microscopic
γLV (1 + cos θ ) = 2 γLV γSV + γLV γSV (8) distribution of stress at the rough interface is complex.

13
3.3. Thermodynamics of adhesion 3.4. Factors affecting the adhesion
In the past many researchers have studied the molecular at the interface between the adhesive
interactions between two surfaces. Prediction of bond and substrate systems
strength between adhesive and adherend, or between Adhesion is the condition by which two surfaces are
the interphase and the adhesive, can be obtained by ap- held together in one of the following ways [74]: (a)
plying the general principles of surface chemistry [11]. physically, that is, the components are held together
Creton [35] has recently theoretically investigated the by interlocking due to physical penetration; (b) chem-
molecular interactions between two surfaces assuming ically, that is, the components are bound together by
that the contact between the adhesive and substrate is valence forces of the same type which give rise to
perfact, i.e., that there is no surface roughness and that cohesion. To control the interface which can provide
the bond forms instantaneously upon contact. This as- the material with improved mechanical performance
sumption is clearly not true and has significant impli- and structural integrity, it is essential to understand the
cations for the measurement of quick tack. However, mechanisms of adhesion which are specific to each ma-
it is easier to concentrate on the perfect contact case terial system and the mechanics of load transfer at the
first, which is a reasonable assumption when interpret- interface.
ing peel test results. The energy measured during the separation of the
In the simplest case of the contact of an adhesive A adhesive/substrate assembly is a complex function of
and a rigid substrate B, which do not react chemically the adhesion energy and also of the dissipative proper-
nor have specific interactions, the process of forming ties of both adhesive and substrate [75]. For example,
A-B contact and then separating the surfaces involves when a stress is applied to a polymer, a part of the
the formation and then destruction of van der Waals energy is dissipated by the movements of chains (vis-
bonds. The energy necessary to break these bonds per cous flow). These viscous losses are time (or speed
unit area of surface is given [35] by or frequency) and temperature dependent. The ef-
fects of time and temperature are related by the well
known time/temperature superposition principle. Dis-
Wadh = γa + γs − γas (9) sipative phenomena occuring during an assembly sep-
aration are reflected through the dependence of the
measured energy with the separation rate. The depen-
where γa and γs are the specific surface energies of dence of the adherence energy G with the separation
the adhesive and of the substrate, respectively, and γas speed V has been first determined by Gent and Shultz
is the interfacial free energy. If the forces are purely [76]:
dispersive, the interfacial energy is approximately given
by
G = Wa φ(aT V ) (13)

γas = γa + γs − 2(γa γs )1/2 (10) where Wa is the thermodynamic energy of ad-

hesion, φ is the dissipation function and aT the
William-Landel-Ferry shift factor for time-temperature
so that the thermodynamic energy of adhesion is given
equivalence.
by
A derived expression (Equation 14) which separates
G into two terms, has been proposed by Maugis and
Wadh = 2(γa γs )1/2 (11) Barquins [77]:

G − W = W φ(aT V ) (14)
On the other hand, the cohesive energy of the adhesive
will be given by
This equation indicates that the adherence energy G
is equal to the sum of the adhesion energy W plus
WcohA = 2γa (12) the dissipation energy, which also depends on the ad-
hesion energy Wa . As a consequence, a higher ad-
hesion will induce more energy dissipation, even if
Two main observations should be made from these re- the intrinsic dissipation ability of the polymer remains
sults: (a) first, the adhesion energy is predicted to in- identical [76, 78, 79]. Different energy dissipation
crease with the surface energy of the substrate (see mechanisms have been proposed in the past [80–82].
Equation 11) and, (b) second, such an approach would Dissipation losses can be induced by chains pull-off,
predict cohesive failure in the adhesive if the surface en- at a molecular scale, and can also affect a greater
ergy of the substrate is higher than that of the adhesive volume [83, 84] due to high entanglements density.
(i.e., γs > γa ). None of these simple predictions have Guillement et al. [75] have investigated the adherence
actually been quantitatively confirmed experimentally level of steel/polymer/steel assemblies (i.e., metal-to-
yet. Furthermore, it is well known that the measured ad- metal joint) by using a wedge test, performed at differ-
hesion energy W is usually several orders of magnitude ent wedge introduction rates. The adherence energy G
higher than what is predicted by the thermodynamic values measured for various wedge introduction speeds
equations. V are presented in Table I. The following expression
14
T A B L E I Adherence energy value G as a function of the wedge in-
troduction rate for steel/polymer/steel assemblies [75]

Speed (mm/min) G (J/m2 )

0.8 30263
7.5 29408
1.5 30825
26.5 32524
37.5 35607
52.5 34991
75 36864

[75] has been used for the calculation of G [85, 86]:

Figure 2 Effect [35] of the weight average molecular weight Mw on the
3Eh3 δ 2 1 180◦ peel fracture Energy W at a peel rate of 10−3 m/s (squares) and
G= 
16L 4 1 + λo

h 4
(15) ◦
8.5 × 10−3 m/s (circles) for polyisobutylene
. Cohesive failure and
λ 0.64 L • : Adhesive failure. Note that these data were taken from the reference
[88b].
where E is Young’s modulus of the steel substrate
(2.1 × 105 MPa), h the substrate thickness (0.4 mm),
δ the wedge height (0.4 mm), L the crack length, E a
the modulus of the adhesive (3.03 × 10102 MPa), h a
the thickness of the adhesive (4.5 × 10−5 m) and b
the width of the assembly, λ = (3 k/Ebh)1/4 with
k = bEa / h a and λo = (3 ko /Ebh3 )1/4 = 61/4 / h with
ko = 2Eb/ h. The adherence energy globally increases
when the wedge introduction speed increases, and a
transition was observed. A similar transition was de-
tected on the curve of Young’s modulus of the polymer
versus tensile speed [75], attributed to the glass relax-
ation. The failure is pseudo-adhesive, i.e., partly co-
hesive, inside the adhesive layer, but very close to the
steel/polymer interface. Strong interfacial bonds and
great energy dissipation in the adhesive layer could ex-
plain the great adherence values.
Creton [35] has studied the effect of molecular Figure 3 Peel fracture enery {log10 [W ](J/m 2 ) ∗ 296/T }-peel rate
weight on the adhesion properties (i.e., peel resis- {log10 [RaT ](m/s)} relation for SBR adhering to Mylar [35]. The raw
material () has a high molecular weight which reduced by the milling
tance, quick tack, and shear resistance) of the adhe-
operation (◦).
sives (i.e., polymer-to-polymer substrates). The molec-
ular weight has an important effect on the adhesion
properties through its influence on the intermediate and
long relaxation times. A high molecular weight will glass transition. The main difference between the two
favor adhesive separation, by slowing the process of polymers is the terminal relaxation time which must
chain disentanglement, which is responsible for fibril be longer for the raw polymer. However, the transition
creep and eventually cohesive fracture [87]. However, in fracture energy observed for the milled sample at
the molecular weight can also influence the microme- low rates could be due microscopically to the transition
chanical deformation mechanisms within the adhesive from high dissipation fibrillar fracture to low dissipa-
layer which are eventually responsible for the measured tion adhesive fracture.
energy of adhesion [35]. A high molecular weight Mn Quick tack is more negatively affected by an increase
will negatively affect the formation of fibrils during the in molecular weight because, in addition to the reason
debonding process. Also, as shown in Fig. 2, the peel already mentioned for the adhesion properties, a high
resistance (180◦ peel fracture energy) decreases slowly M will also increase the modulus at intermediate and
with molecular weight for a series of polyisobutylenes long times, and can negatively affect the ability of the
[88a]. polymer to form a bond. Shear resistance, however, is
Gent and Petrich [89a] performed peel fracture tests positively affected by molecular weight [35]. The sim-
against the reduced peel rate on an SBR polymer on plest approximation is to correlate creep resistance with
Mylar before and after a milling operation (see Fig. 3). viscosity, and it is now well established that the viscos-
As shown in this figure, the measurements of W (peel ity follows the empirical relation η ∝ M 3.4 . Therefore,
fracture energy) versus the reduced peel rate showed a small change in molecular weight can greatly affect
two transitions for the milled sample, while only one the shear resistance.
transition was observed for the raw sample. The high In conclusion, the general rule is that within the range
peel rate transition was attributed to the onset of the of molecular weight that gives the adhesive separation,
15
 the expected properties rather than brings any enhance-
ment in itself. Therefore, when no transcrystalline zone
is present, a weak boundary layer is formed, which de-
creases considerably the expected quality of adhesion
[93]. Transcrystallinity prevents such a detrimental ef-
fect by extending the bulk properties of the matrix to
the surface region. If the transcrystalline zone is dam-
aged, this protective effect is lost. It was found [92]
that a similar result can be achieved by crosslinking the
surface layer through activated species. They used an
anology between the behavior of a transcrystalline layer
and that of a crosslinked layer to explain the increased
Figure 4 Effect of peel rate on the peel adhesion energy and mode of
cohesive strength. Because of extensive surface nucle-
failure for: vinyl acetate-2-ethyl hexyl acrylate copolymer [35]. The ad-
hesive films were on the polyester backing and peeled off clean glass at a ation, there is a strong compatition between the chains
◦
90◦ angle. : Cohesive fibrillar fracture, : Adhesive fibrillar fracture, at the surface.
•
and : Adhesive nonfibrillar fracture. The vertical arrows show the high Creton [35] has also studied the adhesion behavior
and low values of the fracture energy in the domains of stick slip fracture at the adhesive-substrate interface. Although the bulk
where the force oscillates [Data taken from [89b]].
viscoelastic properties are of primary importance for
an adhesive, a necessary condition for good adhesion is
the tack and peel strengths decrease slowly with molec- that the interaction at the interface is sufficiently strong
ular weight while the creep resistance increases [35]. to sustain the stress necessary to cause extensive plas-
In practice, however, most polymers used industri- tic deformation in the bulk adhesive [35]. In order to
ally have a broad polydispersity, and an experimental understand the influence of the adherend on the peel
study on polyisobutylenes showed that a wide molec- energy and, in particular, the relation between the ther-
ular weight distributed can have a beneficial effect on modynamic surface energy and the measured adhesion,
all three properties [88]. some works were undertaken in the past [76, 94, 95].
Aubrey et al. [87] has studied the molecular effects In the absence of primary bonds at the interface, the
of different polymer melts on adhesion using differ- measured peel strength is proportional to the thermo-
ent approaches. The peel strength (the peel adhesion dynamic energy of adhesion Wadh and the proportion-
energy) of a polymer is shown in Fig. 4 against the ality factor is dependent on the bulk viscoelastic prop-
reduced rate of peel. As shown in this figure three dif- erties of the adhesive [35]. These observations have
ferent microscopic deformation mechanisms exist [87]: been confirmed for crosslinked adhesive that did not
(a) At low peel rate, the separation is fibrillar and cohe- show any fibrillar fracture. Toyama et al. [96] stud-
sive for the polymer. (b) However, at higher rates, the ied the probe tack and peel strength of a poly(n-butyl
vinyl acetate-2-ethylhexyl acrylate (VA-EA) copoly- acrylate) on a variety of substrates and analyzed the
mer shows a transition in fracture mechanism to fibrillar dependence of their data on the surface energy of the
adhesive failure. It is important to emphasize that this adherends. They found a maximum in the peel strength
transition is not associated with a discontinuity in the when the surface tensions of the adhesive, γa , and of
fracture energy versus peel rate. (c) When the peel rate the adherends, γs , were closely matched and pointed
is raised further, the polymer goes through a slip-stick out that in this situation the interfacial energy is mini-
regime, which is related to the mechanics of the peel mum. Other workers have, however, found an increase
test, and then undergo a transition to adhesive nonfib- in the peel strength at low surface energies of the ad-
rillar fracture. herend, followed by a plateau when γs > γa [34], or a
Schonhorn and co-workers [90–92] have studied the monotonic increase in the tack with γs [97]. As men-
adhesion of polyethelene to various adherends. They tioned by Creton [35], two general conclusions can
found that the surface layer morphology strongly in- be drawn from these results: (a) the surface energy
fluences the adhesive joint strength. In a simple peel- of the adherend should not be below that of the ad-
ing test of bonded polyethylene-aluminum joints (i.e., hesive, this being particularly true for tack. (b) If the
polymer-metal substrate) [90], a relatively thick layer surface energy of the adherend is comparable or higher
of polyethelene was found to adhere to the substrate sur- than that of the adhesive, the influence of surface forces
face. The observation by optical microscopy revealed on the measured adhesion is relatively small compared
that this layer consisted of transcrystalline material. to the effect of the viscoelastic losses in the adhesive
The thickness of the transcrystalline layer increased layer.
with the bond-preparation time at the selected temper- As Creton pointed out [35] the effect of surface
ature. Later, Schonhorn and Ryan [92] studied the ad- roughness is also important for the adhesion in the case
hesion in epoxy adhesive-polyethelene film-epoxy ad- of tack. Fuller and Tabor [98] showed that the measured
hesive pieces (i.e., polymer-polymer substrate). It was adhesion of a crosslinked adhesive is dependent on its
shown that the presence of a large transcrystalline sur- surface roughness down to fractions of a micrometer,
face layer in the polyethelene film considerably en- and that a surface roughness of the order of a microm-
hances the adhesive-joint strength. As Ishida and Bussi eter could decrease the measured adhesion by a factor
[93] concluded from these works, it appears that the ma- of 10. The topology of the surface is never, however,
jor interest of a transcrystalline layer is that it preserves considered in tack studies and may be responsible for

16
Figure 5 Effect of surface roughness on the surface energy of an alu-
minum alloy [52].

Figure 6 Effect of grit-blasting media on the surface energy of an alu-

some quantitative discrepancies between experimental minum alloy [52]. : Dispersive energy and : Polar energy.
results [35].
Harris and Beevers [52] have investigated the effects cos θrough = r cos θsmooth . This correlation increased the
of surface texture on surface energy, which might also contact angles and reduced the derived value of surface
be expected to affect adhesion using a mechanical sur- energy. Carre and Shultz [105] used a reverse concept to
face pretreatment methods (i.e., grit blasting treatment). Wenzel’s analysis and proposed that a roughness factor
As shown in Fig. 5, in general, smoother grit-blasted could be determined from the contact angles measured
surfaces display higher surface energy values for an alu- on smooth and rough surfaces. Shuttleworth and Bailey
minum alloy adherend. Hitchcock et al. [99] and Tamai [106] derived an alternative technique for correlation of
and Aratanic [100] also observed that, within certain the contact angles in which the adjustment was simply
limits, roughening a substrate surface usually causes based on the mean slope of surface roughness obtained
its wettability to decrease. Several workers have inves- from two-dimensional profile measurements.
tigated this aspect of substrate surface behavior and In recent years, many researchers in the adhesion
a number of concepts have been proposed to explain field [107–114] have employed the JKR technique to
the differences in wetting characteristics of rough and analyze the fundamental forces of adhesion between
smooth surfaces. Harris and Beevers [52] have propesed two materials. An important advantage of the contact
a simple model that the peaks, ridges and asperities mechanics experiment is the very low cracks growth
form barriers which restrict the spreading of the droplet. speeds which are accessible with this technique with
Huh and Mason [101] and Yost et al. [102] have pointed higher sensitivity as compared with other adhesion tests
out that with acute contact angles the three-phase line such as the peel test [107–114]. Woerdeman et al. [112]
is reluctant to flow over ridges and peaks. However, have reviewed this JKR contact mechanics approach.
they also reported that the droplet “seek out” areas of The JKR theory, developed by Johnson et al. [115]
the surface where it can spread more easily, particularly is an extension of the Hertz analysis [116], which re-
through roughs and and valleys, perhaps as a result of lies on a fracture mechanics-energy balance approach
capillary channelling. Bikerman [103] suggested that to describe the mechanics of interaction in the pres-
wetting was independent of roughness on grooved and ence of adhesive forces. A most significant feature of
ridged surfaces and implied that as a drop front moves the JKR contact mechanics approach is that it allows
over a ridge that the true contact angle (i.e., the angle re- one to make a direct estimate of the work of adhe-
lated to the actual surface of the solid and not the mean sion, W , between the two solids of interest, unlike the
plane) remains constant. Harris and Beevers [52] sug- conventional contact angle experiment, which requires
gested that the changes in the surface energy might be the user to construct a calibration curve. Chaudury and
attributable to changes in surface composition. Their Whitesides [113] led the renewed interest in the JKR
proposal was supported by the data in Fig. 6, which contact mechanics technique, having developed an ex-
indicate considerable shift in the polar component of periment based on the use of soft, hemispherical lenses
surface energy between the different treatments. Also on rigid, flat plates. In this technique, a material of low
it was shown [52] that after grit blasting of the mild elastic modulus, in the shape of a hemispherical lens
steel and aluminum alloy adherends, small amounts of is brought into contact with a solid surface of interest,
Na and/or Mg appear on the treated surfaces and these and the resultant contact area is monitored as a func-
correlate with the grit impurities. tion of applied loading and lens penetrating depth. The
Using a different approach, Wenzel [104] proposed contact area varies with applied loading according to
that wetting and spreading properties are influenced the JKR equation (Equation 16), which describes the
more by the amount of effective surface area that can system behavior as a function of its geometry, material
actually interact with the liquid, than the surface texture properties, and adhesive forces:
characteristics. He drived the ratio of the true surface
R
area divided by the projected geometric area (Wenzel’s a3 = {P + 3π WR + (6πWRP + (3πWR)2 )1/2 }
roughness factor, r ) and used this to provide a “cor- K
rection” for the contact angle θ from the expression (16)

17
where a is the contact radius, R is the effective radius
of the system, K is the effective bulk modulus of the
material, P is the contact load, and W is the work of
adhesion.
The JKR predictions give rise to a second relation-
ship, which describes the lens penetration depth, δ, as
a function of P, a, R, and K

a3 2P
δ= + (17)
3R 3a K

Researchers have previouly demonstrated

[110, 114, 117] that use of these two equations
simultaneously provides an internal check of the
validity of the small deformation approximation used
in the JKR analysis. In particular, JKR theory assumes
small deformations so that the depth of the strain
field at the elastomer-rigid support interface is negli-
gible compared with the depth of penetration [110].
Therefore, when the JKR contact mechanics technique
is used to measure the work of adhesion between
two solids in contact, it is now an adopted custom
to analyze the plot of δ versus a to assess whether
or not the experimental data deviate from the JKR
prediction at large contact areas [108, 110, 114, 117].
Deruelle et al. [110] have demostrated that one can
circumvent this problem by intercalating a thick
ribbon of the same elastomer between the lens and
its support. Alternatively, the experimentalist can
make a special point of not over-compressing the
elastomeric hemisphere during the loading experiment
[108, 117], or in other words, can avoid deforming the
hemisphere laterally, which has been known to occur
when measurements are made at exceedingly large
contact areas [117].
Fig. 7 shows the schematic description of the JKR
experiment, whereas Fig. 8 indicates the sample data
obtained during a JKR compression experiment [108].
In all cases, the symbols represent the data points and Figure 8 Typical curves obtained from a single JKR loading experiment
[108]. In all cases, the symbols represent the data points and the solid lines
the solid lines trace the fits to the data for an epoxy lens trace the fits to the data for an epoxy lens against a γ -aminopropylane-
against a γ -amino-propyl-triethoxysilane-modified sil- modified silicon wafer: (a) Contact radius cubed versus applied load, (b)
icon substrate. Fig. 8a depicts the contact radius cubed displacement versus contact radius, and (c) displacement versus applied
versus applied load (a 3 versus P), while Fig. 8b and c load.
illustrate the relationships between δ and a, and δ and
P, respectively. In this particular case, the suitability
of the JKR model was good, which suggests that the 4. Bonding mechanisms (or types)
experiment had been performed within the constraints in adhesively bonded joints
of the theory. It is important to control the interfaces between the
adhesive and substrates. To do so, it is necessary to un-
derstand all the different possible bonding mechanisms,
one or more of which may be acting at any given in-
stant. Despite the wide use of adhesives, a good deal
of conroversy surrounds the nature of the bond. There
are following bonding mechanisms or types existing in
the literature, which are particularly useful in explain-
ing certain phenomena associated with adhesive bond-
ing [1, 3, 54, 118]: (a) physical bonding, (b) chemical
bonding, (c) diffusion or interdiffusion theory, and (d)
mechanical bonding or mechanical interlock theory. It
Figure 7 Schematic illustration of the JKR contact mechanics experi- is beneficial to summarize these bonding mechanisms
ment, where P is the load, a is the lens contact radius, δ is the penetration because they indicate procedures commonly followed
depth of the lens, and R is the radius of curvature of the lens. for optimal bonding.
18
4.1. Physical bonding at the interface may contribute to the force of attraction
Physical bonding involves weak, secondary or van der bonding. The strength of the interface will depend on
Waals forces, dipolar interactions, hydrogen bonding the charge density [54]. This attraction is unlikely to
and other low energy forces [3, 118]. The bond energy make a major contribution to the final bond strength
in such physical bonding is approximately 8–16 kJ/mol of the interface [118]. The bonding of this type will
[3]. Physical bonding contains the following bonding explain why silane finishes are especially effective for
types: (a) the absorption theory, and (b) the electrostatic certain acidic or neutral reinforcements like glass, silica
attraction theory. and alumina, but are less effective with alkaline surfaces
like magnesium, asbestos and calcium carbonate [118].

4.1.1. The absorption theory

This theory states that to be successful, an adhesive 4.2. Chemical bonding
must wet the surface to be bonded (called the adherend). A chemical bond is formed between a chemical group-
This theory has led to the development of materials ing on the adhesive surface and a compatible chem-
with lower surface tension than that of the adherend ical group in the adherend (Fig. 9b). The strength of
[1]. Supporting this theory is the fact that epoxy wets the chemical bond depends on the number and type of
steel and provides a good bond, whereas it does not wet bonds and interface failure must involve bond break-
the olefins PE, PP, and PTFE and does not bond them. age. The processes of bond formation and breakage
are in some form of thermally activated dynamic equi-
librium [54, 118]. The chemical bonding theory is the
4.1.2. The electrostatic attraction theory oldest and best known of all bonding theories. The na-
This theory postulates that as a result of the interaction ture of the chemical bonding is the key to the physical
of the adhesive and the adherend, an electrostatically and chemical behavior of matter. Atomic or molecular
charged double layer of ions develops at the interface. transport, by diffusional processes, is involved in chem-
In another words, forces of attraction occur between ical bonding. Solid solution and compound formation
two surfaces when one surface carries a net positive may occur at the interface, resulting a reaction zone
charge and the other surface a net negative charge as with a certain thickness. This encompasses all types of
in the case of acid-base interactions and ionic bonding covalent, ionic, and metallic bonding. Chemical bond-
[54] (see Fig. 9a). The fact that electrical discharges are ing involves primary forces and the bond energy in the
observed when an adhesive is peeled from a substrate range of approximately 40–400 kJ/mol [3]. For exam-
is cited as evidence of these attractive forces [1]. A ple, a chemical reaction at the interface is of particular
difference in electrostatic charge between constituents interest for polymer matrix composites because it offers
the major explanation for the use of coupling agents on
glass fibers and porabably the surface oxidative treat-
ments on carbon fibers for application with most ther-
moset and amorphous thermoplastic matrices.
Surface treatments often involve chemicals which
produce surfaces with different chemical composi-
(a) tions and oxide stoichiometries. These morphological
changes influence the nature of the chemical bonds.
Subsequently, a relationship exists between chemical
composition of the surface and the bond durability
[57, 59, 119].

(b)

4.3. Diffusion or interdiffusion theory

This bonding mechanism is applicable to cases in which
the adhesive contains a solvent for the adherend [1]. A
type of bonding similar to diffusion bonding in metals
develops, and molecules pass across the interface. This
(c) diffusion can obliterate the mechanical plane of the in-
terface and its weakness. It is possible to form a bond
between two polymer surfaces by the diffusion of the
polymer molecules on one surface into the molecular
network of the other surface, as illustrated schemati-
cally in Fig. 9c. The bond strength will depend on the
(d) amount of molecular entanglement and the number of
molecules involved. The phenomena of interdiffusion
Figure 9 Types or mechanisms of bonding [54]: (a) Bond formed by
has been called autohesion in relation to adhesives.
electrostatic attraction, (b) Chemical bond formed between groups A
on one surface and group B on the other surface, (c) bond formed by A bond between two surfaces may be formed by in-
molecular entanglement following interdiffusion, and (d) mechanical terdiffusion of atoms or molecules across the surface.
bond formed when a liquid polymer wets a rough solid surface. As an example, the bond strength in polymer matrix
19
composites will depend on the amount of molecular en- unlikely to be very high in transverse tension, but the
tanglement, the number of molecules involved and the strength in shear may be significant dependent on the
strength of the bonding between molecules [118]. Inter- degree of roughness [118]. However, mechanical bond-
diffusion may be promoted by the presence of solvents ing is efficient in load transfer when the applied force
and the amount of diffusion will depend on the molec- is parallel to the interface [3]. In addition to the sim-
ular conformation, constituents involved and the ease ple geometrical aspects of mechanical bonding, there
of molecular motion. For example, bonding between are many internal or residual stresses in composite ma-
glass fibers and polymer resins through silane coupling terials which develop during fabrication process due
agents by other than chemical bonding can be explained to matrix shrinkage and differential thermal expansion
by inter-diffusion and interpenetrating network (IPN) between fiber and matrix. Among these stresses, the
formation at the interfae region [60, 120–122]. The in- residual clamping stress on the fiber provides a major
terface region thus formed has a substantial thickness, bonding at the interface of many ceramic matrix com-
and its chemical, physical and mechanical properties posites which plays a decisive role in controlling the
are different from those of either the bulk fiber and fracture resistance of these materials.
matrix [123–126]. In metal matrix composites, inter- Roughness or an increase in the surface are results
diffusion is also necessary for proper reaction between in increased mechanical interlocking of the adhesive
elements of each constituent to take place. However, in- to the adherend. It has been shown that as a result of
terdiffusion here may not always be beneficial because the high stability of the fresh oxide layer to moisture
undesirable compounds are often formed, particularly degradation, good durable bonds can be achieved [38].
when the oxide films present on the fibers are com-
pletely disrupted under extremely high temperature and
pressure in a solid state process [127]. To prevent or re- 5. Surface pretreatments for polymers,
duce the interaction it is necessary to apply an effective polymer composites, and metallic
diffusion barrier in the form of a coating on the fiber. materials substrates
The selection of an appropriate diffusion barrier relies Good surface pretreatment is achieved by one or a com-
on a detailed knowledge of the nature of the interaction bination of the following [58]: (a) the production of
which is specific to each other. a surface free from contaminants, (b) the production
It is generally agreed that the highest strength is of a macro/microscopically rough surface, (c) the pro-
achieved when, upon stressing, the fracture occurs in duction of a fresh stable oxide layer, and the chemical
the body of the adherend or within the adhesive, not composition of the oxide.
at the interface. The weak boundary layer theory holds Surface treatments enable the nature of the chemical
that for an adhesive to perform satisfactorily, the weak groups present at the surface to be modified and they
boundary layer should be eliminated. For example, in may be used to modify the topography. Depending on
case of metals with a scaly oxide layer, failure has a the treatment used, the modification may be carried out
coherent oxide layer. Similarly, in the case of polyethy- without changing the bulk properties of the polymer.
lene, a weak, low-molecular-weight additive is present The alterations to the polymer may come under many
througouht the structure, and this leads to a weak inter- different forms, as stated by Lennon et al. [128]: (a)
face. In both cases the potentially weak layers can be removal of the weak cohesion layer, or of the pollution
removed by surface treatments [1]. present at the surface, (b) introduction of new or an in-
creased number of chemical functions, and (c) increase
the roughness of the surface. All these parameters can
4.4. Mechanical bonding or mechanical contribute to an improvement of the wettability and/or
interlock theory of the adhesive properties of the surface. However, the
Some bonding may occur purely by the mechanical in- efficiency of a surface treatment depends on the nature
terlocking of two surfaces as illustrated in Fig. 9d .This of the substrate and on the depth of treatment. There is
theory points out that surfaces on a micro scale are often a compromise between the functionalisation and
very rough [1]. Therefore, when a liquid adhesive is the degradation of the surface.
placed between two surfaces, it penetrates the crevices There are many methods available to pretreat poly-
and pores and then solidifies. Thus a cement interlocks mers and metallic alloys substrates. These include both
with the surface layers on both sides and provides a physical and chemical surface pretreatments. With the
mechanical bond. The fact that fresh, roughened sur- former, material with low cohesive strength is removed
faces provide the best bond support this theory. Simple and the topography may be changed. With chemical
mechanical keying or interlocking effects between two methods a number of changes may occur [129], namely,
surfaces can lead to a considerable degree of bonding. (a) removal of weak material, (b) roughening, and (c)
In general, mechanical bonding is a low-energy bond the introduction of functional groups into the polymer.
vis a vis a chemical bond, i.e., the strength of a mechan- The methods used to pretreat a polymer or metallic
ical bonding is lower than that of a chemical bond [3]. material surface are highly varied. More specifically,
Pure mechanical bonding alone is not enough in most they can be [128, 129]: (a) chemical or electrochem-
cases. However, mechanical interlock bonding could ical, (b) mechanical, (c) thermal, (d) photochemical,
add, in the presence of reaction bonding, to the overall or (e) plasma. Over the last 25 years, the possibility
bonding. of using electrochemical methods to pretreat polymers
Mechanical bonds involve solely mechanical inter- has been examined in particular by Brewis, Dahm and
locking at the surface. The strength of this interface is co-workers (see for example [129]). Many chemical
20
T A B L E I I Examples of surface pretreatments for plastics or elas- latter generally contain large quantities of cohesively
tomers substrates [129] weak additives. These are often present, in substan-
Plastics Elastomers tial quantities, on the surface and if they can not be
absorbed by the adhesive used, poor bond strengths
Abrasion Abrasion will result. For those elastomers which contain a sub-
Solvent cleaning Solvent cleaning stantial quantity of functional groups, e.g., nitrile rub-
Coronaa Conc. sulphuric acida
Flamea Trichloroisocyanuric acida
ber, a physical method which removes any weak layers
Plasmaa Halogen gasesa may be sufficient to provide good adhesion. However,
Active gasesa with elastomers which possess few functional groups,
Specific reagentsa it may be necessary to remove any weak layers and
a Expected chemically modify the elastomers. It may not be nec-
to cause chemical modifications to substrate surfaces.
essary to remove the weak layer if this is chemically
modified to make it compatible with the adhesive used
so that the layer can be absorbed. Selection of surface
treatments involve oxidative methods but reagents usu- pretreatments for polymers depends on several factors
ally used with fully fluorinated polymers e.g., sodium [129] including (a) cost, (b) safety, (c) environmen-
in liquid ammonia, are powerful reducing agents. It is tal impact, (d) effectiveness, and (e) stability of the
possible to provide both powerful reducing or oxidising treated surface in service. In some cases it will be nec-
conditions using electrochemical methods. essary to pretreat 3-dimensional objects while in other
In their review article, Brewis and Dahm [129] dis- cases it may be required that one side of a sheet is
cussed the pretreatments of polymers. In general, the treated.
methods that have been shown to be effective with plas- A variety of surface pretreatments have been used
tics are considerably different from those that are ef- with various degrees of success to increase surface ten-
fective with elastomers (Table II). Typical composite sion, increase surface roughness, change surface chem-
surface treatments include tradational abrasion/solvent istry, and thereby increase bond strength and durability
cleaning techniques for thermoset composites, while of polymer composite adhesive joints and are shown [2]
thermoplastic composites require surface chemistry in Table III. Wetting of the adherends by the adhesive
and surface topographical changes to ensure strong and is critical for the formation of secondary bonds in the
durable bond strengths. However, To increase strength absorption theory [57]. It has been theoretically veri-
and durability a more intimate bond is necessary; to fied that for complete wetting (i.e., for a contact angle
achieve this the following conditions should be met θ equal to zero) the surface energy of the adhesive must
[129]: (a) increasing surface tension, (b) increasing sur- be lower than the surface energy of the adherend [58].
face roughness, and (c) changing the surface chemistry. For thermoplastic composites the primary aim of the
Clearly, not all the methods listed for plastics, or elas- surface treatment is to increase the surface energy of
tomers, are effective with all plastics or all elastomers. the adherend as much as possible. Surface treatments
For example, physical methods such as abrasion are [2] (a) decrease water contact angle, (b) increase sur-
not effective with polyethyelene (PE). It is normally face tension, and (c) as a result increase bond strength.
necassary to chemically modify PE to achieve good The researchers [65, 139–146] have used many dif-
adhesion, for example with a corona discharge. A key ferent titanium alloys as substrates in the past, however
difference between plastics and elastomers is that the Ti-6V-4Al is the most widely used one in the aerospace

T A B L E I I I Showing the effects of various surface pretreatments methods on the surface tension, surface roughness, surface chemistry, bond
strength and durability of the polymer composites [2]

Nature of Surface Surface Surface

Treatment type Material treatment tension roughness chemistry Bond strength Durability Sources

(1) Abrasion and Thermoset and Remove mold Y Increased found Good for [130, 131]
solvent wipe thermoplastic release for thermosets thermosets
(2) Grit blasting Thermoset and Remove mold Y Increased found Good for [131–133]
thermoplastic release for thermosets thermosets
(3) Acid etch Thermoset and Etcha Y Y Slight increase Poor [131, 134, 135]
thermoplastic
(4) Peel-ply Thermoset Remove mold Y Increase Good [131]
release
(5) Tear-ply Thermoset Remove mold Increase Good [136]
release
(6) Corona discharge Thermoplastic Oxidising Y Y Double Good (90 days) [133]
(7) Plasma treatment Thermoplastic Ablation and/or Y Y Y Increase Good (90 days) [130, 135–137]
oxidationa
(8) Flame treatment Thermoplastic Oxidasinga Y Increase [131, 138]
(9) Laser treatment Thermoset and Ablation and or Y Y Increase More research [131]
thermoplastic oxidation is necessary

a Depends on polymer matrix material.

21
industry [2]. Durability studies of T-6Al-4V reveal that The various surface pretreatment methods used for
surface preperations that produce no roughness (macro polymeric materials and titanium alloys, and the quali-
or micro) yield the poorest bond durability. Those that tative descriptions of their effects on the surface tension,
produce significant macro-roughness but little micro- surface roughness, bonding strengths and durability of
roughness yield moderate to good durability. Finally the these materials are presented [2] in Tables III and
those that produce significant microroughness yield the IV. Both thermoset and thermoplastic composite ma-
best durability (Table IVA). Table IVB gives the six terials have been bonded to titanium with varying de-
surface pretreatment procedures and their effects for grees of success [154]. However, it is noted that thermo-
titanium alloys. plastic materials are inherently more difficult to bond
The surface pretreatments used for polymeric and [155]. Carbon-fiber-reinforced polymer composite ma-
metallic materials are highly varied (see for exam- terial has been widely used as the adherend in the past
ple Tables III and IV), and can be conveniently clas- [156]. Although glass-fiber-reinforced composite ma-
sified (see for example [2, 52, 59, 128, 129, 151– terial has a much lower modulus than carbon fiber, bond
153]) into: (a) Mechanical, (b) chemical, (c) elec- integrity tends to be dominated by the characteristics of
trochemical, (d) thermal, (e) photochemical, and (f) the matrix, rather than the fiber and for this reason the
plasma. The followings are surface pretreatment meth- research conducted on fiber composite is also relevant
ods used for polymeric materials: (a) Abrasion/solvent to studies involving glass-fiber [2].
cleaning, (b) grit blasting, (c) peel-ply (or tear films), Thermoset polyester or epoxy composites can have
(d) tear-ply, (e) acid etching, (f) corona discharge a resin-rich surface layer, which can cause particular
treatment, (g) plasma treatment, (h) flame treatment, problems in bonding [2]. The reason for this is that
(i) laser treatment, and (j) silver electrolytic pre- most of them contains a gel coat at the surface, or
treatment process. Whereas for metallic materials such if the latter has not been deliberately created the sur-
as titanium alloys the following methods are used to face layer usually contains a higher proportion of resin
surface preatreat them: (i) Abrasion and grit blast- as opposed to the interior. Both the gel coat and the
ing, (ii) VAST (Vought Abrasive Surface Treatment), resin-rich surface layer are very brittle and can sub-
(iii) Acid etchants, (iv) Alkaline peroxide etch, (v) sequently fail catastrophically when overload. Highly
Phosphate fluoride process, (vi) Modified phospho- compliant adhesives are a particularly good choice,
tate fluoride process, (vii) TURCO 5578, (viii) DAP- as they spread the applied load over a large area and
COtreat, (ix) Pasajell 107 Treatment, (x) Electrochem- hence reduce the stress borne by the surface of the
ical reactions, (xi) Chromic acid anodization, (xii) composite.
Sodium hydroxide anodisation, (xiii) Catodically de- These surface pretreatment methods mentioned
posited aluminum oxide, (xiv) Plasma Treatment, (xv) above are summarized in the following sections (see
Sol/Gel Methods, (xvi) Primers, (xvii) γ -APS, (xviii) for example recent works by Molitor et al. [2, 59] and
Laser. Green et al. [153].

T A B L E I V A Showing the effects of various surface pretreatment methods on the surface roughness, oxide layer, bond strength, and durability of
titanium alloys [2]

Nature of Surface Oxide layer

Treatment type Alloy treatment roughness (nm) Bond strength Durability Sources

Abrasion and Ti-6Al-4V Remove mold Macro Poor Poor [145]

solvent wipe release
Grit blasting Ti-6Al-4V Remove mold Macro Increase Adequate [131, 145]
release
VAST Ti-6Al-4V Remove mold Macro Good Poor [131, 145]
release
Acid etch Ti-6Al-4V Etch Micro Adequate poor Poor [145, 147]
Alkaline etch Ti-6Al-4V Etch Micro 60–200 Good Good [145, 147, 148]
Phosphate-fluoride T-6Al-4V Etch None 20 Adequate Poor [145, 147]
Modified phosphate- Ti-6Al-4V Etchant and None 8 Adequate Better than [145, 147]
fluoride oxidation phosphate-fluoride
TURCO Ti-6Al-4V Oxidising Macro 17.5 Adequate Adequate [145, 147]
DAPCOtreat Ti-6Al-4V Macro 6 Increase Good [147]
Pasajell Ti-6Al-4V Oxidising Macro 10–20 Adequate Adequate [145, 147]
Chromic acid Ti-6Al-4V Oxidising Micro 40–140 High Excellent [147]
anodisation
NaOH anodisation Ti-6Al-4V Oxidising Micro 80–90 High Excellent [147]
Cathodically deposited Oxidising Adequate Adequate [145, 147]
Al2 O3
Plasma spray Ti-6Al-4V Ablation and Micro 130 High Excellent [145, 147]
oxidation
Sol gel Ti-6Al-4V Coupling and High Good [140]
oxidation
Laser treatment Ti-6Al-4V Ablation and Macro High Poor [146]
oxidation

22
T A B L E I V B Showing the procedures applied and their surface modification effects of selected surface pretreatment methods for titanium alloys
substrates [59]

Surface treatment Procedure Surface effect

Surface modification Produced using a wavelength of 308 nm wit XeCl. The focal Riblets or wavelike structures were produced
by excimer laser length was 100 mm and the fluence produced was 77 J/cm2 . on the surface which were approximately
40 µm deep with 100 µm period
Surface cleaning by Produced using a wavelength of 308 nm with XeCl. The focal A homogenised area of ∼4 mm2 was produced.
excimer laser lenght was 150 mm and the fluence produced was 2.8 J/cm2 . Each irradiated zone was overlapped in both x
and y directions.
Plasma spray The titanium specimens were heated to 100◦ C in an oven to A TiO2 coating of thickness 50 µm was
remove moisture, grit blasted with 180/220 Al2 O3 grit, rinsed producded.
with a cleaner and subsequently dried. Plasma spraying was
applied by spraying Hamdry 6506. TiO2 powder.
Pasa Jell 107 Pasa Jell 107 was applied to grit blasted samples. It was applied Etched the existing oxide film
for 12–15 min. After which it was rinsided with deionised water
at room temperature. The specimens were subsequently dried
and sprayed with a thin layer of the primer Redux 101 using a
De Vilbiss suction feed cup spray gun, type JGA.
Sodium hydroxide Performed as described by Ingram and Ramani [149]. The samples Cleaned and created a fresh oxide film
anodisation were degreased in methylthylketone (MEK) for 10 min, pickled
by in 15% volume of 70% nitric acid, 3% by volume of 50%
hydrofluoric acid at room temperature for 30 s. They were
subsequently rinsed in 2 baths of agitated deionised water for 5 min.
Anodisation was performed in a 5 M NaOH electrolyte at 10 V for
20–30 min. Upon removal from the solution, the samples were rinsed
for 5 min in running water. The samples were allowed to dry and
sprayed with a thin layer of the primer Redux 101 using a spray gun.
Chromic acid Performed as described by Arnold et al. [150]. The samples were Cleaned and created a fresh oxide
anodisation degreased in MEK for 10 min, pickled in 15% by volume of 70% nitric
acid, 3% by volume of 50% hydrofluoric acid at RT for 30 s. They
were subsequently rinsed in 2 baths of agitated deionised water for
5 min. Anodisation was performed in an electrolyte containing 5%
CrO3 and 0.1% NH4 HF2 at RT. The voltage was applied after the
specimens were immersed in the solution and increased from 2 to 10 V
at a rate of 2 V/min. The voltage held constant for 20 min. Upon removal
from the solution, the samples were rinsed in the agitated deionised water
and subsequently rinsed for for 5 min in running water. The samples
were allowed to dry and sprayed with a thin layer of the primer
Redux 101 using a spray gun.

5.1. Mechanical surface pretreatments these composites have very smooth low surface energy
5.1.1. Abrasion, grit blasting surfaces [133, 135, 137].
and solvent cleaning Solvent degreasing must be performed in a con-
A mechanical treatment is used primarily to produce trolled atmosphere to prevent surface re-contamination
a clean macroscopically rough surface and to remove [132]. Surface contamination should be removed as
some of the existing oxide layer [2] (see Tables III the first step in surface preparation as degreasing af-
and IV). The combination of a clean adherend surface ter abrasion or chemical treatment will result in a layer
with significant macro-roughness improves the initial of partially solvent dissolved contamination remaining
dry strength [145]. However, abrasion and grit blast- on the prepared surface which may inhibit adhesion
ing techniques are not aduquate methods of surface [139].
treatment, but when combined with chemical or elec- The type of solvent is also important. Care is re-
trochemical treatment, durable bond strengths can be quired to ensure that the solvent does not degrade the
obtained [157]. structure being degreased. For example, many chlori-
Abrasion/solvent cleaning may be employed to de- nated hydrocarbon solvents can cause stress corrosion
grease the surface and remove mold release agents from cracking [58]. Solvents with a rapid evaporation rate are
the adherend [133, 135, 136]. For example, the com- best for hand degreasing, because any residual pools
posite sheets can be lightly abraded using 180/220- of solvent will spread partially dissolved contamina-
mesh alumina, then wiped clean with a solvent such tion as the material evaporates; the faster the solvent
as methylethylketone (MEK) and allowed to dry. Pre- evaporates, the less surface is contaminated. Degreas-
vious researchers have found a significant increase in ing must be performed using a solvent rather than a
surface roughness and bond strength for thermosets [2]. detergent. Detergents function by wetting the surface
However, the same treatment for thermoplastic polymer better than the contamination, thus displacing it from
composites did not reveal any significant increase in the surface. This means that the detergent will become a
bond strength or durability, due to the fact that some of contaminant unless it can be removed from the surface

23
[132]. Note that water solubility of a detergent is not a abrasion by either high pressures or slow pass rates will
guarantee of removal. cause folding of the surface [159] leading to trapped
As important as selection of an appropriate solvent moisture which evaporates during heat cure of the ad-
is the selection of an appropriate wipe cloth which will hesive causing voiding in the adhesive layer. The grit
not contaminate the surface. Shop rags, cheescloth and blast should be light enough to avoid folding, but suffi-
even some proprietary wipe cloths may contain contam- ciently heavy to remove the surface oxides. This can be
inants. Others contain lanolin (which is a food release easily controlled by abrading until the surface sheet is
agent) to reduce skin abrasion on the hands of the user. removed, but not so hard that the surface of aluminum
These must be avoided. alloys turn a dull, dark grey. With limited training,
Alternatively, an alumina grit blast (of particle size most operators easily master the appropriate level of
45 µm) with three passes at a distance of 15–20 cm abrasion.
may be employed, followed by solvent rinse and dried
in nitrogen [2]. MEK may not be compatible with lo- 5.1.1.1. Effect of grit blasting on surface properties for
cal government regulations for industrial applications adhesion. Harris and Beevers [52] have investigated
and as a result may have to be phased out and be re- the surface energy and adhesion characterics of a mild
placed by a less hazardous solvent. The DLR (German steel and an aluminum alloy using the grit blasting
Aerospace Research Establishment) has investigated with diffrent alumina grits. The coarser grit produced
three different types of cleaning agents to be used with rougher surfaces and in general the rougher surfaces
mechanical methods of surface treatment [132]. These exhibited lower surface energies. Harris and Beevers
included acetone, frigen and ethylacetate. Of these, the have explained this observation using the effects of ge-
latter gave the highest lap shear values. Again, as with ometric features of the surfaces on the spreading of
the abrasion and solvent wipe, grit blasting gave strong the droplet and the measurement of the contact angle.
and durable bond strength for thermosets but revealed However, it was also observed that the grit-blasting pro-
very little degree of success with thermoplastic materi- cess introduces chemical changes on the surface of the
als [2]. substrate and these in turn affect the surface energy
In summary, grit blasting is far more effective in pro- [52].
duction of an active surface, mainly because it is a non- The surface energy and adhesion characteristics of a
contact process with a clear visible measure of effec- surface are dependent on a complex interaction between
tiveness [152]. One requirement for grit blasting is that surface texture and chemical components [52]. The fol-
the abrasive must cut the surface, not punch the surface. lowing factors can also influence these interactions: (a)
Thus abrasives such as glass beads, walnut husks, sand the substrate material, (b) the grit-blasting media purity
and steel shot are not suited. Aluminume oxide powder and contamination, and (c) the grit-blasting media par-
and even zirconia have been found to be effective [158]. ticle size. Although it is difficult to derive a generalised
It is important that the abrasive powder is not recycled, hypothesis for the relationship between surface texture
to prevent re-contamination of exposed surfaces. and surface energy and the resulting adhesion charac-
The grit should be delivered in a dry inert gas stream. teristics, an interpretation of these complex interactions
Compressed air is not suited, due to oil and water con- may be drawn from the summary of changes resulting
tent. Current practice in defence applications is to use from the surface pretreatment of surfaces using differ-
dry nitrogen gas which readily available at service bases ent sized grit-blasting media as depicted [52] in Table V.
because of its use for inflation of tyres and shock struts. It should be emphasized that this table does not show
Grit particulates must be contained within the worksite, actual properties for a particular treatment but indicates
and extracted using a vacuum system which must ex- changes which occur with increasing grit size and the
haust away from the aircraft. Particles may break into associated greater surface roughness. From this table it
fine needles, pass through fuel filters and cause fuel may be concluded that initial strengths were relatively
system malfunctions. Failure to extract fine dust from insensitive to grit type or size. However, durability char-
fuel tanks has led to in-flight engine failures. acteristics were found to give different responses and
One factor which influences the adhesive void con- thus Table V attempts to correlate these changes with
tent is the level of abrasion by grit blasting. Excessive other effects.

T A B L E V Summary of responses of surface properties to changes (including the surface energy, surface chemistry and durability) for mild steel
and aluminum alloy substrates in grit size from 180/220 to 60 grit [52]

Response of surface properties to increase in grit size

Substrate material Grit type Surface energy Surface chemistry Durability

Mild steel Brown No change Mg ↓ Al ↓

Pink Polar ↓ Na ↓ Al ↓ Lap no change
Butt ↓
Aluminum alloy White Polar ↓ Na ↓ Al ↓
Brown No change No change
Pink Polar ↓ Na ↓ Lap ↓ Butt ↓
White Polar ↓ Na ↓

24
5.1.2. VAST (vought abrasive surface For thermosetting plastic matrix composite surfaces,
treatment) practice is to use tear films, a sacricial layer placed on
Using this pretreatment (Table IVA), the titanium the surface during the laminating process. The tear film
is blasted with a slurry of fine abrasive alumina remains in place until the adhesive is to be applied. It
(220 mesh) containing 2% fluorosilic acid under high is then removed, taking surface contaminants with it.
pressure [2]. A post treatment rinse in 5% nitric acid is The concept is flawed and is based again on the “clean
required for Ti-6Al-4V titanium alloys to remove grey surface” theory.
smut after blasting. An oxide film is produced which
is crystalline in nature. This results in joint durability,
which is reported to be slightly lower than TURCO 5.1.4. Tear-ply
5578 and slightly higher than the phosphate fluoride Tear-ply is mainly used for thermoset composites
process [65, 145]. Most chemical treatments alter the (Table III) as thermosets are reactive upon heating and
titanium substrate by etching the existing oxide film. If hence do not require a chemical surface treatment [2]. It
the solvents are reducing agents, then a fresh thin oxide is fundamentally different from peel-ply and consists of
film is produced after the metal is removed from the a fabric, which is completely impregnated by the lami-
solution, while in an oxidising solvent a thicker oxide nate matrix resin, and as the name implies subsequently
film is produced [145]. peeled off the molding. Superficially, the surfaces of the
laminates treated with either peel-ply or tear-ply appear
very similar as both exhibit the pattern of the ply fabric.
Tear-ply exploits a laminate property, which is gener-
5.1.3. Peel-ply (or tear films) ally regarded as being a serious disadvantage: the lack
Peel-ply (also known as tear films) (Table III), an im- of strength in the z direction (normal to the surface).
pregnated ply is removed immediately prior to bond- Immediately prior to a bonding operation, a corner of
ing [2]. Previous research has shown that successful the fabric is raised with a knife and the tear-ply is then
joints were obtained by increasing surface roughness, torn off the molding, thus fracturing the resin interface
degreasing and removing of mold release agents [157]. between the tear-ply and the bulk molding.
The peel-ply is released from the surface because of The tear-ply must be carefully selected to enable it to
the non-stick nature of the substrate within which it is be removed without any difficulty following the mold-
impregnated. These are frequently fluorine silicon or ing process. Polyester resin has fracture characteristics,
fluorine compounds, which contaminate the laminate which permit this, and have little adhesion to the nylon
surface. Consequently, this implies that further treat- fibers in the fabric. However, as pointed out by Molitor
ment of the surface, e.g., by abrasion, is required if the et al. [2], some resins are not compatible with this tech-
strength of the subsequent bond is not to be seriously nique and hence peel-ply must be employed despite the
impaired [148]. If the ply contains release agent, the difficulties arising from contamination.
adherend must be subsequently cleaned with a solvent
and dried in a stream of nitrogen.
Typically, three forms of tear films are used: (a) The 5.2. Etching
first type are a thin fiber-glass layer which is difficult For any adhesive bonding process, a fresh chemically
to remove, and will cause delamination damage in thin active surface is essential [152]. This may be generated
laminates. There is a real danger of removing the first chemically by etching, but may also be produced by
few layers of laminate with the tear film; (b) The sec- abrasion. Production facilities usually use tank etching
ond type are usually a nylon product. Unfortunately, methods for metallic surfaces, and field level facilities
nylon forms good chemical bonds with epoxy resins usually rely on abrasion of the surface. The function of
so some form of release agent is applied to the fibers abrasion is often misuderstood to be enable the adhesive
to enable the operator to remove the tear film without to key into the surface. Removal of sanding debris is
causing excessive damage to the laminate. Tests [160] difficult, leading many specifications to require solvent
have shown that this release agent is transferred to the cleaning after abrasion. Because bonding relies on the
bonding surface and reduces the effectiveness of any chemical activity of the surface and the absence of con-
adhesive bond formed with that now contaminated sur- tamination, solvent cleaning after abrasion only suc-
face. Most reputable manufacturers clearly state that ceeds in partially dissolving contaminants and spread-
all coated fabrics have the potential to transfer release ing them over the surface, reducing the effectiveness
agent to the surface exposed by removal of the tear of the active surface. Sanding debris is best removed
film; (c) The third form of tear film uses heat setting or using a dry wipe process until all debris is removed.
corona discharge treatment to glaze the surface of the
fibers. This is the usual process for polyester tear films.
The glazed surface is so slick that it releases from the 5.2.1. Acid etching
resin when required. Unfortunately, it leaves a replicate Acid etching has produced similar results to abrasion
of that glazed surface on the laminate surface again re- and grit blasting, in that an increase in bond strength is
sulting in a reduced bond effectiveness on that surface. recorded for thermoset polymer composites, whereas
While the surface remaining is clean, it is not chemi- little or no effect was recorded for thermoplastics
cally active, apart from a few locations where the weave [130, 136, 148].
bundles cross over each other and the resin is fractured A number of acid mixtures have been used to treat the
[160]. surface of titanium, including nitric-hydrofluoric acid,
25
hot sulphuric acid, and sodium dichromate-chromic 5.2.4. TURCO 5578
acid [2]. Most of these treatments give adequate dry TURCO 5578 is an alkaline-based etchant, which con-
strengths but durability is reported to be very poor tains caustic soda, sodium metasilicate and pyrophos-
[161]. Both stressed and unstressed peel and lap shear phate [2]. This pretreatment process produces a large
joints have shown to perform equal to or better than amount of macroroughness with little or no micror-
phosphate fluoride or modified phosphate fluoride pro- oughness, which can be used to treat the surface of ti-
cesses [65]. An additional problem associated with tanium alloys (Table IVA). The oxide film produced is
these treatments is hydrogen pick up. However, this ≈17.5 nm thick with a macroroughness of 3.4 µm peak
problem can be avoided with the use of alkaline solu- to valley [163]. The durability of this treatment is much
tions as opposed to acid etchants. better than that of the phosphate fluoride treatments. In
durability trials, TURCO 5578 is only out-performed
by chromic acid anodisation [65]. However, the added
5.2.2. Alkaline peroxide etching advantage of the TURCO 5578 treatment is that there
As shown in Table IVA, Molitor et al. [2] have used the is no hydrogen embrittlement as observed in the acid
alkaline peroxide etching intially at room temperature etching process.
for the Ti-6Al-4V alloy, but this process, which lasted Another surface pretreatment, DAPCOtreat, which is
up to 36 h, was too lengthy for industrial applications. similar to TURCO 5578, is also applied to titanium al-
The immersion time can be reduced to 20 min by in- loys, in that it results in an increase in macroroughness
creasing the temperature to 50–70◦ C [145]. Depending with little or no microroughness [2]. DAPCOtreat pro-
upon the concentration of sodium hydroxide and hy- duces a thin 6 nm oxide layer on the titanium substrate
drogen peroxide, the metal is either etched or oxidised. [65].
Those concentrations, which produce grey oxides, have
been found to produce adhesive wettable surfaces. An
oxide layer up to 2 µm thick, which is stable up to 5.2.5. Pasajell 107 treatment
200◦ C and capable of forming high bond strengths at This pretreatment is recommended (Table IVA) as a pre-
elevated temperatures and high-humidity environments bond treatment for titanium [65, 144, 145, 164, 165].
were produced using a sodium hydrogen peroxide etch The chemical composition of pasajell 107 is 40% ni-
[148]. Alkaline peroxide etch is reported to be satisfac- tric acid, 10% combined fluorides, 10% chromic acid,
tory but has long heat up times and has high hydrogen 1% couplers and the balance is water. It is available as a
peroxide consumption. Oxide formation is related to thixotropic paste for brush applications or as an immer-
the rate of hydrogen peroxide decomposition. The rate sion solution for tank treatment. A recommended treat-
of decomposition is increased by adding 5–10% heavy ment time is 10–15 min. Pasajell 107 is applied in com-
metal ions to the hydrogen peroxide solution. bination with a pre-treatment of MEK wipe, blast with
320 grit non-silicone sand, followed by non-chlorinated
solvent wipe [136]. After applying the pasajell for 10–
5.2.3. Phosphate fluoride process 15 min, the coated area is rinsed with deionised water.
This process has been used to pretreat the titanium al- This process produces [2] an amorphous looking ox-
loys (see Table IVA), which is based on combinations ide, which has anatase structure, which is stable up to
of trisodium phosphate, disodium tetraborate, potasium 175◦ C and converts to rutile at 350◦ C. Comparing this
fluoride and hydrofluoric acid [162]. A process of pre- treatment to the alkaline peroxide etch, the bond dura-
etching in 3% hydrofluoric acid and 15% nitric acid fol- bility of the pasajell treatment is reported to greatly
lowed by a 2 min dip in the 5% trisodium phosphate, 2% exceed that of the alkaline peroxide etch at elevated
potassium fluoride and 2.6% hydrofluoric acid mixture temperatures and humid environments [144].
has become known as the phosphate fluoride process
(US Patent 2 864 732). The rinsing procedure after this
treatment is critical to remove excess chemicals and a 5.3. Corona discharge treatment
15 min soak in di-ionised water at 60◦ C has been rem- Corona discharge, namely exposing the substrate sur-
monded [145]. face to excited atoms, ions and free radicals at
Modified phosphate fluoride process has also been atmospheric pressure has been widely used to pretreat
applied to treat the surface of titanium alloys polymeric materials surfaces (Table III) for adhesive
(Table IVA), which has been shown that on exposure bonding [135, 166, 167]. The success of corona dis-
to warm moist environments, the anatase oxide layer charge in treating carbon-fiber-epoxide, carbon-fiber-
produced by the phosphate, fluoride process slowly re- PEEK and glass-fiber-polypropylene composites for
verts to rutile [2, 145]. As a result, there is a decrease bonding is reported, as is its use for treating polyolefin
in volume of about 8%, which results in the develop- films to make them receptive to printing inks. Results
ment of stresses at the adhesive/oxide layer interface. [133] reveal that corona treatment increases surface ten-
By stabilising the antase structure the joint durability is sion and in some cases alters the surface chemistry by
enhanced. This is achieved by adding 0.75% anhydrous oxidizing the polymer matrix, which results in the in-
sodium sulphate into the etch solution [65, 145]. How- crease in bond strengths.
ever, both of these treatments have been out-performed Corona discharge pretreatment method and tech-
by treatments such as TURCO 5578, sodium hydroxide nique was described in detail by Tohra Vehara [168]
anodisation and chromic acid anodisation [145]. and Sutherland et al. [169] using a modified continuous
26
treatment type machine allowing small number of sam- lar technique to plasma spray called silicon sputtering,
ples on a flat electrode plate to be treated, with the arc where a 200 Å thick silicon layer was sputtered onto ti-
struck between it and the moving bar electrode above tanium. An increase in bond strength and durability was
it. Optimization of the treatment of the maximum sur- recorded and attributed to an increase in surface ten-
face energy, using standard wettability inks and was sion and surface roughness, which allowed the molten
found to be 58 dyn · cm−1 . For example, the process pa- polymer adhesive to flow in and around the increased
rameters for homopolymer polypropylene, HF 133 M surface area and interact with the silicon and oxides.
were as follows [153]: number of passes = 1, treatment Use of microwave plasma in post discharge has the
width = 0.18 m, speed = 15 m min−1 , electrode-sample advantage of a less severe effect on the surface proper-
gap = 1 mm and power = 33 W (power density = ties than treatments where the polymer is placed di-
120 Wm−2 min−1 ). rectly within discharge [173]. In post-discharge, the
polymer substrate is not exposed to high-energy en-
tities such as electrons or high-energy ions which can
5.4. Plasma treatment have a highly destructive impact. The principal effects
As presented in Table III, the use of plasma treatment of a microwave post-discharge treatment are known to
to pretreat polymers has been known for more than be the cleaning of the surface (removing pollution or
20 years [2]. The plasma treatment involves a low- weak boundary layer), the crosslinking of the surface
pressure plasma gas, which is electrically conductive and the chemical modification by introducing of func-
and consists of excited atoms, ions and free radicals tional groups [128]. The modification of the roughness
[170]. This allows polymer surfaces to be cleaned, of the surface is generally less marked for this type of
etched or chemically modified [136, 170]. The plasma treatment but has been shown to occur in certain cases
particles react not only with each other but also with the especially for drastic treatment conditions (high power
surfaces, which are exposed to the gas, giving rise to the and long treatment times) [174].
following effects [136]: (a) surface cleaning, (b) degra- The use of nitrogen/oxygen mixtures for the plasma
dation of the polymer chains, (c) removal of material treatment of polymers is well-known and has been stud-
from the surface, (d) formation of radicals on the sur- ied [175–178]. They are used to incorporate oxygen
face, and (e) change of tacticity of the polymer chains. functions at the polymer surface, and leading to an in-
The combined effect of these processes results in crease in wettability. Several studies [179, 180] indicate
an improvement of the adhesion properties of the sur- that the use of a mixture of the two gases is more ef-
face [2]. Plasma treatments involving various gases ficient than the use of oxygen alone as they lead to a
have been found to enhance the surface tension, oxide higher concentration of atomic oxygen, the main active
the polymer matrix and increase the bond strength of species in the oxygen plasma.
PEEK composite [137]. Results obtained by Blackman Low-pressure O/N vacuum plasma pre-treatment
[130] and Barron [155, 171] echo the results obtained can be used for the Polyolefins components, using a
by corona treatment [135], in that plasma treating large-scale evacuation chamber. The processing and
the PEEK composite produced an increase in surface equipment of large-scale low-pressure plasma cham-
roughness, surface tension and bond strength. bers was properly described by Wertheimer [168] and
Both glow discharges and plasma sprays have been covers the equipment and processing technique used.
examined as methods of surface treatment for titanium For example, the process parameters for homopolymer
alloys [2]. Aronsson et al. [172] investigated the use polypropylene, HF 133 M were as follows [153]: pres-
of glow discharge. Results obtained revealed that this sure = 2.6 × 10−2 mbar, voltage = 2500 V, current =
method of plasma treatment produced clean surfaces 2.3 A, pump-down time = 5 min, plasma reaction time
and reproducible results, depending upon the degree of = 35 s, hold time following plasma = 60 s and re-
treatment. Furthermore, oxidation in pure oxygen re- pressurasation time = 10 min.
sulted in uniform and stoichiometric TiO2 surface oxide AgrodynTM atmospheric plasma pre-treatment or
layers with reproducible composition and thickness. “air” plasma allows the pre-treatment of complex-
Plasma spraying involves rapidly heating either, shaped articles under atmospheric conditions without
TiO2 , TiSi2 , MgO or SiO2 powder to a molten or semi- the need for batch processing using vacuum chambers
molten state and then spraying it onto the substrate at required by other forms of plasmas. This treatment can
high velocity. This is another method of surface treat- be carried out using a process described by US-Patent
ment employed for adhesive bonding that involves no 5837958 [181]. Air plasma relies on the use of a plasma
hazardous chemicals or pollutants [2]. Other advan- jet being formed from the nozzle of a rotating sparking
tages of plasma sprying over chemical treatments in- nozzle, which are set up as pairs of nozzles. The sam-
clude [143]: (a) flexibility to design coatings for specific ple is then placed below the nozzles within the plasma
applications, (b) insensitivity to surface contamination, cone and pre-treated in a processing method similar to
(c) indefinite shelf life prior to bonding, and (d) low that used for flame. The process parameters are set to
processing costs. create the greatest surface energy by the use of wetta-
Successful plasma treatments of titanium have been bility inks, which was identified as being in excess of
reported, whereby bond strengths equivalent to the best 120 dyne cm−1 . For example, the process parameters
chemical treatments were obtained [143]. At high tem- for homopolymer polypropylene, HF 133 M were as
perature, the plasma-sprayed titanium joints produced follows [153]: gas = air at atmospheric pressure, in-
superior results. Ramani et al. [147] employed a simi- put pressure = 2 bar, output jet pressure = 13.9 mbar,

27
rotation speed of nozzle = 1500 rpm and speed = Results obtained [146] reveal no significant differ-
10 mm min−1 . ence in single lap shear strength results obtained using
Laser treatment compared to chromic acid anodisation
which augers well for an environmmentally friendly
5.5. Flame treatment surface treatment. Broad et al. [151] have revealed good
This treatment is widely used in the surface mod- durability results using the patented CLP (CIBA Laser
ification of thermoplastics such as polyolefins (see Pretreatment) laser treatment, whereby laser treated ad-
Table III), to improve printability and paintability, by hesive joints to an aluminum alloy substrate have not
introducing oxygen containing functional groups to the failed after 1400 days.
surface [166, 182–185]. An oxidising flame is very sim- Another laser treatment called, the IR (infrared ) laser
ilar to a gas plasma in that it contains excited species surface pretreatment is carried out using carbon black
such as atoms, ions and free radicals, which oxidise the doped on a substrate (for example on the homopolymer
surface of the specimens. The distance from the surface polypropylene, i.e., HF 135 M samples). The carbon
to be treated, the air/gas ratio and the dwell time are all black is added as an IR absorber(i.e., 2% by weight of
critical parameters in producing a successful treatment. the total mass). Although the Excimer lase techniques
By oxidising the surface prior to bonding an increase is described by a number of authors for different sub-
in hydrophylicity and hence bond strength can occur. strates such as PEEK, the use and technique of the IR
In the flame surface pretreatment a row of propane laser has not to date been published [153], but the tech-
burners is passed repeatedly over the substrates in a nique is similar to that used by Frerichs et al. [191]
continuous and uniform manner as demontrated by for the surface pre-treating of PEEK using an Excimer
Sutherland et al. [186]. However, the gas ratio and laser. The parameters can be optimized by the use of
pass speed are kept constant throughout and optimized wettability inks to maximize measured surface energy.
for maximum surface energy, which was found to For example, the process parameters for homopolymer
be 56 dyn cm−1 . For example, the process parame- polypropylene, HF 133 M were as follows [153]: (a)
ters for homopolymer polypropylene, HF 133 M were a diode IR laser was used to give a square beam pro-
as follows [153]: number of passes = 8, speed = file at 940 nm, (b) the samples were pre-treated within
25 m min−1 , propane flow = 3.751 min−1 , air flow = an oxygen atmosphere at an output of 100 W over a
881 min−1 , cone height = 4 mm, burner gap = 13 mm, treatment area of 675 mm2 , and (c) the laser sample
power density = 207.5 mJmm−1 and burner length = gap was set at 139 mm whilst the samples were passed
265 mm. underneath the laser upon a translating track at a speed
of 2.5 m min−1 .
5.6. Laser treatment
A noval method of surface treatment, namely, the laser
treatment [131, 187, 188] has been employed to in- 5.6.1. Comparison of laser surface
crease the bond strength of adhesive joints to polymers preperation with wet pretreatments
and metallic substrates (Tables III and IV). Initial re- Certain metals and their alloys require a surface pre-
sults are promising with respect to durability but more treatment in order to achieve highly durable adhesively
research is necessary. Park et al. [161, 189] has recently bonded joints. This is the case when aluminum and
shown that laser treating polymer composite surfaces its alloys are adhesively bonded and is clearly demon-
prior to bonding can produce high-strength adhesive strated [151, 192] in Fig. 10. For durable adhesively
bonds. Laser treatment results in increase in surface bonded aluminum joints a simple degreasing or grit
roughness due to the ablation of the polymer matrix. blasting is not sufficient. With regard to joint proper-
In the past surface treatment of titanium substrates ties a variety of surface pretreatments for aluminum
has been based on either chemical or electrochemical substrates, such as (a) phosphoric acid anodizing, (b)
processes. Although these treatments have been very
successful, many contain hazardous chemicals such as
chromates. As a result, new methods have been devel-
oped which are environmentally friendly. Recent stud-
ies have shown that Excimer laser treatments result in
an increase in surface roughness and the formation of
a thin oxide layer [146]. The parameters that affect the
degree of surface treatment include wavelength, po-
larisation and intensity. For Excimer laser treatment,
the parameters governing the quality of treatment are
the wavelength, polarisation, power density, beam di-
ameter, speed and pulse length [151, 190]. In particular,
the pulse length or dwell time strongly influences the
material depth that will be affected. Surface cleaning by
Excimer laser also yielded high peel strengths. Surface
Figure 10 Effects of various wet pretreatments (i.e., phosphoric acid
cleaning by this method ablate the natural oxide layer anodized, chromic acid anodized, pickle etched), grit blasted, and de-
on the surface and do not produce macroroughness as greased methods on the adhesively bonded lap shear strengths of alu-
produced by surface modification by Excimer laser. minum alloy substrates [151, 192].

28
chromic acid anodizing or (c) pickle-etching, are well 5.7. Electrochemical reactions
known in the aerospace industry [151]. These surface Electrochemical reactions have been used for various
pretreatments can be classified as “wet pretreatments”. pretreatment purposes including cleaning, etching and
Joints with long-term durability can be achieved by oxidising metal surfaces, where the metal acts as an
applying such pretreatments. Irrespective of their ex- anode or cathode. Anodisation creates an oxide surface
cellent performance regarding the joint properties, wet that is dependent upon the electrolyte, anodisation volt-
pretreatments suffer from certain major disadvantages age, times and temperature [164]. For example, there
[151]: (a) ecological disadvantages (disposal of chem- are various anodisation techniques employed for tita-
icals and waste water), (b) long pretreatment times, (c) nium alloy surface treatment (see Table IVB) including
compatibility problems with short cycle times (e.g., in (a) chromic acid anodisation, (b) sodium hydroxide an-
the automotive industry), (d) limited surface stability, odisation, (c) catodically deposited Aluminum oxide.
(e) pretretment of the whole submerged surface, and Methods of electrochemically generating oxidising
(f) high pretreatment costs. One main disadvantage of and reducing conditions are also used to pretreat a range
wet pretreatments is one of an ecological nature as they of polymers. The methods include reversible anodic and
require disposal of high quantities of chemicals and cathodic generation of soluble redox couples as medi-
contaminated rinsing water [151]. Furthermore, due to ators capable of reacting with polymer surfaces, and
their relatively long process times wet pretreatments directly contacting polymers with polarised electrodes.
approach their limit with regard to economically ac- Environmental conditions play an important role in the
ceptable effort when larger parts have to be pretreated choice of pretreatment methods. This leads to the use
with short cycle times. A typical example could be in of flame, plasma and corona discharge treatments all of
the automotive industry where the pretreatment of large which have a low environmental impact [129]. These
body parts with short cycle times would require very methods are however of limited use for the treatment of
large acid and rinsing baths. This would result in high complex shapes which often require wet methods. Elec-
investments and running costs. The limited stability of trochemical pretreatments offer some advantages over
the wet pretreated surfaces often requires increased lo- purely chemical methods in that the need for the stor-
gistical effort and additional processing steps to protect age of aggressive and toxic reagents is reduced or elim-
the surfaces with a conservation primer. Additionally, inated and by continuously regenerating the spent so-
the wet surface pretreatments generally do not allow lutions further savings may be made by reducing waste
for selective treatment of the area to be bonded. The disposal costs [129].
whole surface of the part which is submerged, has to
be pretreated even when only a small area of it re-
quires bonding. This clearly increases costs of such 5.7.1. Basic principles
processes. Brewis and Dahm [129] have recently reviewed the ba-
Another surface treatment process called Ciba Laser sic prenciples of the electrochemical process. Electro-
pretreatment (CLP) was developed and patented. As chemical reactions take place at the interphase of an
Broad et al. [151] claimed this newly developed and electronic conductor, the electrode and an ionic con-
patented Ciba laser pretreatment avoids the above- ductor i.e., the electrolyte. The anode has been made the
mentioned disadvantages of wet surface pretreatments working electrode (WE) and only oxidation processes
and at the same time facilitates the achievement of good are considered. We can also use the same argument
joint properties for substrates such as aluminum, alu- to study reduction (electronation) reactions by making
minum alloys, and stainless steels. The CLP process the cathod the WE. The cathod is made electron-rich
consists of the following process steps only: (a) primer by pumping electrons into it and when the potential of
application (including evaporation of solvent), and (b) the cathod is sufficiently high these electrons may cross
laser treatment. A suitable laser and optimum parame- the interphase and reduces i.e., electronate a substrate
ters should be chosen according to the substrate and the situated in the vicinity of the cathod. The electrolyte it-
production requirements. Laser technology from small self may be electronated or deelectronated more readily
mobile lasers to high-performance stationary laser units than the substrate itself. In the case of an aqueous elec-
allows for a wide range of CLP-applications, from re- trolyte water may exchange electrons with the electrode
pair to high-volume series production. Pretreatment according to
speeds depend on the power of the laser being used;
for example, for CLP with a high-performance laser 2H2 O + 2e → H2 + 2OH−
it is easily possible to pretreat a width of 4 cm at a
speed of 8 m/min. Therefore, by the choice of suitable [Reduction (electronation) of water] (18)
lasers, compatibility of CLP with short cycle times can 2H2 O − 4e → O2 + 4H+
be achieved. The primer applicator and the laser can be
adopted to industrial robots and in this way CLP can [Oxidation (deceletronation) of water] (19)
be integrated as an in-line pretreatment into the overall
production process. It is therefore frequently necessary to use a non-aqueous
The recent progress in Laser pretreatment applica- electrolyte if really powerful oxidizing or reducing con-
tions on the adhesive bonding of aluminum, stainless ditions are to be obtained.
steels and titanium alloys substrates will be reviewed We can use of these principles for the treatment of
in Section 6. polymers in two ways: (a) The electrode may be brought

29
into direct contact with the polymer surface, the so anodically polarised electrode or to some soluble re-
called direct electrochemical process or, alternatively, active, anodically generated oxidising agent. But there
(b) an indirect approach may be used where a soluble are certain difficulties associated with this approach:
species is converted into a powerful oxidizing or reduc- (a) firstly, the oxidation potentials of saturated poly-
ing agent by electonation or deelectronation at an elec- mers are very high so that the electrolytes used tend to
trode. The redox reagent thus generated then transfers be oxidised instead of the polymer. Electrolytes resis-
electrons to or from the polymer regenerating the orig- tant to oxidation at these very high anodic potenials
inal species which merely acts as an electron carrier or tend to be expensive and extremely unpleasant typ-
mediator transferring electrons between the electrode ically consisting of mixtures of fluorosulphonic acid
and the polymer. Both approches have been used to treat with anhydrous antimony pentafluoride, the so called
the surface of polymers with the aim of improving their super acids [194]. (b) Secondly, it is unlikely that the
bondability. treated polymer surface is an electronic conductor so
that an anodic oxidative process proceeding from a sin-
gle point contact anode can be ruled out. A number
5.7.2. Chromic acid anodisation of water soluble reversible redox couples were con-
This technique produces a surface with significant mi- sidered for the role of mediator including [129, 195]
croroughness and an oxide thickness of 40 and 80 nm Ce(IV)/Ce(III) E ◦ = 1.72 V; Co (III)/Co(II) E ◦ = 1.83–
for the 5 and 10 V treatments, respectively [65]. In addi- 1.92 V and Ag(II)/Ag(I) E ◦ = 1.98 V.
tion, chromic acid anodisation oxides are reported to ex-
hibit remarkable bond durability and provide the target
bond strengths and durability for all other treatments. 5.7.6. Cathodic pretreatments of polymers
Brewis and Dahm [129] have reviewed the possibility
of cathodic pretreatments of polymers. A number of
5.7.3. Sodium hydroxide anodisation fluoro polymers have been shown to react with vari-
Previous researchers investigated the use of sodium ous electron donors. The cathodic reduction of a so-
hydroxide-hydrogen peroxide anodisation and perox- lution of poly(chlorotrifluoroethylene) (PCTFE) in γ -
ide free sodium hydroxide anodisation as methods of butyrolactone containing lithium tetrafluroborate has
surface treating titanium for adhesive bonding [142]. It been reported [196, 197], but the use of the cathodic
was found that specimens prepared by these methods technique as a surface pretreatment for this particular
produced high-strength adhesive bonds and exhibited polymer does not appear to have been investigated. On
good durability when exposed to conditions of heat, the other hand, large increases in failure loads have
moisture and stress. The presence of hydrogen perox- been observed with ethylene-cholorotrifloroethylene
ide was found to be neither necessary nor beneficial and copolymer (ECTFE), and with poly (vinyl) fluoride
in some cases resulted in a decrease in bond strength (PVF) upon treatment with sodium naphthlenide [129].
[2]. In addition, the sodium hydroxide anodising solu- Considerably longer reaction times are required than is
tion could be reused and was considered to present very the case for the treatment of PTFE. It is possible that
low operational and environmental hazards. the naphthalenide ion assumes the role of a base rather
than an electron donor resulting in the polymer surface
5.7.4. Cathodically deposited [129].
aluminum oxide In summary, it appears that electrochemical carbon-
Cathodic depositions of metal oxides from alcohol so- isation of perhalogenated polymers may be achieved
lutions containing inorganic nitrides have shown good by directly contacting the polymer surface with the
wettability and environmental resistance in hot and hu- cathode or by using a mediator whereas partially halo-
mid conditions [2]. A recommended solution is based genated materials may be treated using bases includ-
on 10 g of hydrated aluminum nitride dissolved in ing those generated electrochemically [129]. There are
11 g of isopropyl alcohol [145]. By making titanium, two types of electrochemical attack on solid polymers,
the cathode of an electrolytic cell at 30 V the oxide which depend on the nature of the polymer.
is deposited on the surface. Results have shown joint Polymers containing groups capable of forming
strengths and durability to be better than the VAST and rather stable radical anions or dianions are said to un-
TURCO 5578 processes [145]. dergo accumulative reductive electrochemical attack.
One example of such system is the cathodic reduction of
PET in aprotic solvents containing tetralkylammonium
5.7.5. Polymer surface pretreatments based salts as support electrolyte where initial two electron
on anodic oxidation reactions transfer to the aromatic moities is particularly facile.
The surface pretreatment of polymers usually involve Polymer containing good leaving groups such as
oxidation resulting in the introduction of various func- halogen are said to undergo dissociative electrochem-
tional groups onto the polymer backbone [129]. The fol- ical attack where electron transfer and elimination of
lowing techniques are used to achieve this: (a) exposure halide ion are concerted without the formation of stable
to plasma, (b) UV light, reactive gasses, or (c) conven- charged intermediates. The electrochemical reduction
tional aqueous oxidising agents such as acid dichromate of PTFE by lithium amalgam is quoted as an exam-
or potassium permanganate [193]. ple of such a process as well as the catalytic elec-
However, in principle it should be possible to bring trochemical reduction of poly(chlorotrifluoroethylene)
about surface oxidation by exposing the polymer to an (PCFET) at graphite and carbon black using solutions
30
of lithium and sodium salts in aprotic solvents such as developed system is based on the principle of a hy-
γ -butyrolactone. brid organic/inorganic coating providing a gradient be-
tween titanium substrate and the adhesive [139]. This
system has covalent bonding through the gradient coat-
5.7.7. The direct electrochemical oxidation ing, thereby reducing dependence of Lewis acid base
of polymers and hydrogen bonding interactions for adhesive bond-
The recent observation has shown [198] that it is possibe ing, which in turn increases bond durability in hot wet
to pretreat the surfaces of PP and HDPE by oxidising environments. Researchers have used an acid catal-
the polymer. This can be done by bringing the polymer ysed sol consisting of zirconium alkoxide and gly-
into close contact with the exposed edge of a platinum cidtrimethoxysilane coupling agent in water [140]. In
electrode held at approximately 2 V versus the satu- addition, a pre-bond treatment of solvent wipe and/or
rated calomel electrode (SCE) in dilute nitric acid in grit blasting prior to the sol/gel treatment was used.
the absence of silver ion [129]. Direct electron transfer Drench and spray techniques as well as immersion are
under these relatively mild conditions is extremely un- imployed for the application of the sol/gel. Results ob-
likely. Nitric acide is oxdised [199] at a platinum anode tained compared very favorably with chromic acid an-
in the region of 2 V to yield highly reactive, short-lived odisation results after 2000 h of hot/wet exposure [140].
nitride radicals capable of initiating the oxidation of
low molecular weight alkanes and hence by implica-
5.9. Primers
tion of polyofins by way of hydrogen atom abstraction.
Primers may be applied to titanium substrate surfaces
It was suggested [198] that such a process is likely to be
for one or more of the following reasons [2]: (a) to
strickly less demanding than the concerted mechanism
protect the substrate surface until bonding is carried out,
suggested for oxidation by the AgNO+ 3 ion. In addition, (b) to increase surface wettability, (c) to block pores of
attack at the more reactive tertiary hydrogen atoms of
porous surface thereby preventing adhesive escaping,
PP is likely to be favored which would account for the
(d) as a vehicle for corrosion inhibition, and (e) as a
higher failure loads achieved for PP when compared
coupling agent capable of forming chemical bonds with
with HDPE [129].
the adherend and adhesive.
Coupling agents are believed to form covalent bonds
5.7.8. Silver electrolytic surface between the adhesive and adherend, thereby producing
pretreatment process strong and durable joints. Silanes are widely used cou-
Silver electrolytic surface pretreatment method can be pling agents. Silane coupling agents have the following
carried out following a surface pre-treatment described structure R − Si(OR )3 where R is the functional group
by Brewis et al. [200]. This process is based upon me- that chemically reacts with the adhesive. R is usually
diated electrochemical oxidation by the use of Ag(II) to an ethyl or methyl group. The main advantage of silane
oxidise organic material at ambient temperatures. The coupling agents is to improve durability of adhesive
process involves the reaction of AgNO+ 3 with water to
bonds in the presence of water or water vapor.
produce hydroxyl radicals, which are able to oxidise the γ -APS (aminopropyltriethoxysilane) is a common
surface of an organic material using a “H ” cell. Owing primer used on titanium and titanium alloys [65, 141].
to the recombination nature of the cell, no silver is con- The γ -APS primer is applied by immersing the titanium
sumed, since it is only used as mediator. For example, substrate in 1% γ -APS for 15 min followed by blow-
the process parameters for homopolymer polypropy- ing the excess off in a stream of nitrogen [201]. The
lene, HF 133 M were as follows [153]: dip time = reported increase in the wet strength between Ti-6Al-
60 s, anodic composition = 50 ml of 5.5 M nitric acid 4V and epoxy was 50% [202]. The γ -APS and γ -GPS
containing 3 mM silver nitrate, elctrode = large area (glycidoxypropylmethoxysilane) have been shown to
platinum mesh, current = 1.1 A and voltage = 9 V. It improve the level of durability of grit-blasted specimens
was noted that following pre-treatment the anode solu- to that of sodium hydroxide and catalytic hydrogen per-
tion had become dark brown suggesting transfer of the oxide etch [65]. In addition a primer called BR-127 is
Ag+ ion. recommended by CYTEC the manufacturers of the ad-
hesive FM-300, while BR-127, EC-3960 and EA-9223
are suitable primers for bonding titanium to compos-
5.8. Sol/gel pretreatment methods ites [144]. However, the problem with BR-127 is that it
The term “sol/gel” may be broadly used to describe contains hazardous chromates. Alternative primers rec-
two processes, one starting from a particulate (col- ommended include the non-chromated CYTEC XBR
loidal) inorganic sol (often aqueous) and another using 6757.
organometallic precursors, chiefly metal alkooxides,
usually in solution with alcohol and water. Since many 6. Recent progress in surface
treatments have hazardous pollutants and/or working pretreatment methods
environments, alternative methods of surface treatment 6.1. Effects of mechanical surface
have been developed such as the sol/gel method for pretreatments on the bonding of
titanium alloys substrates (Table IVA); the sol/gel pro- composite-composite and metal-metal
cess was initially used for making pure oxide glasses adherend systems
and ceramics. The sol/gel formation is water based; Kohli [21] has recently studied the effects of surface
hence contains no hazardous pollutants. This recently preparation of thermoplastics and thermoset composite
31
substrates (composite-composite system) bonding with ply followed by the grit blasting offers the best over-
the FM
R
300-2 adhesive film. In this work, effects of all mechanical performance and the dry peel ply is the
humidity, laminate layup, and as using the various me- worst. The effect of the surface on the hinged double
chanical tests including the lap shear and hinged double cantilever beam test for the Ciba 6376 prepreg is shown
cantilaver beam testings.well as surface pretreatments in Table VIB. These test results also show that grit blast
on the durability of the bonded structure were also stud- in addition to peel ply provides the most durable bond.
ied. For thermosetting composites of secondary bond- In the case of the Hercules 3501-6, there was no sig-
ing, eight epoxy/graphite or epoxy/glass prepeg sys- nificant difference between the peel ply versus the peel
tems with cure temperatures ranging from 121◦ C to ply/hand sand surface preparation. Either of these two
177◦ C were selected. Whereas for secondary bonding surface preparations provided good performance under
of thermoplastic laminates APC-2 (PEEK) or KIII, lam- both the dry and the hot/wet conditions. The failure
inates were used. For the Ciba 6376/3 K 70P prepreg modes for most of the lap shear tests ranged from co-
system, the composite surface preparation studied in- hesive in the case of grit blast prepared substrates to
cluded (a) dry Nylon peel ply, (b) Peel ply/hand sand, mostly interfacial between the adhesive and the lami-
and (c) Peel ply/grit blast. The surface preparation for nate for the peel plied surface preparation [21].
the Hercules 3501-6 carbon/epoxy prepreg, on the other In the case of Hercules 8551-7 epoxy/glass and 8551-
hand, included (a) peel ply and (b) peel ply/hand sand. 7 epoxy/graphite laminates the surface preparations in-
For all metal or secondary bonding of composite sub- cluded (a) solvent wipe only, (b) hand sanding/solvent
strates, the following cure cycle was utilized: Heat-up wiping, and (c) sandblast/solvent wipe. No peel ply was
rate 1.7◦ C/min to 121◦ C, 0.28 MPa, hold 90 min at used. Laminate layup was (±45, ±90, 0)2 for both the
121◦ C. Ciba 6376/3 K 70P (an epoxy/graphite prepreg glass and the graphite substrates. Humidity exposures
system, Hercules 3501-6/3 K 70P (an epoxy/graphite were carried out by exposing individual lap shear spec-
fabric prepreg system), and Hercules 8551-7/AS-4 (an imens to 30 days at 71◦ C/100% RH. Test results are
epoxy/graphite fiber tape system) were cured for the shown in Table VII. The data on the secondary bond-
following conditions: 177◦ C for 120 min, 0.35 MPa, ing of Hercules 8551-7/graphite laminates show that
heat-up rate: 1.7◦ C/min, laminate thickness: 3.1 mm sand blast followed by solvent wiping provided the best
(except that the first substrate were heated-up at the overall strength in the bonded structure. Lap shear val-
rate of 1.7◦ C/min). Hercules 8551-7/AS-4 and Hercules ues from the hand sanded specimens were on an av-
8551-7/glass substrates consist of 1.27 mm laminate erage 10–20% lower than the sand blasted specimens.
thickness and (±45, ±90, 0)2 laminate lay-up. The solvent wipe only was found to be the least effec-
The effects of various surface pretreatments, and tive method for surface preparation of Hercules 8551-
hot/wet conditions on the lap shear strength of the sec- 7/graphite laminates. The failure modes for both sol-
ondary bonding of Ciba 6376 Prepreg with FM
R
300- vent wipe as well as hand sanded surface preparations
2 M are shown in Table VIA. These data show that peel were predominantly adhesive to laminate, whereas for
the sand blast prepared specimens, the failure mode
T A B L E V I A Secondary bonding of Ciba 6376 Prepreg with FM
R was mostly laminate failure. The laminate failure was
300-2 M. Effect of substrate surface pretreatment on carbon-carbon lap mostly in the top plies of the ±45◦ layup.
shear strength [21] However, in the case of Hercules 8551-7/glass pre-
Composite Composite
cured laminates, it was more difficult to evaluate the
Test results (MPa)
surface surface effect of surface preparation on the lap shear strength.
preparation conditioning −55◦ C 23◦ C 82◦ C 82◦ C/Weta Due to the laminate lay-up, laminate failure was pre-
dominant failure mode for both sanded and sand blasted
Peel ply Dry 1 h @ 82◦ C 18.5 20.7 22.9 27.9
specimens. Both solvent wipe only and hand sanded
Peel ply/hand Dry 1 h @ 82◦ C 17.4 20.1 29.9 29.1
sand specimens provided higher overall shear values as com-
Peel ply/grit Dry 1 h @ 82◦ C 25.7 30.6 31.5 31.5 pared to the sand blasted specimens.
blast In conclusion, results from this study show that peel
ply followed by grit blast provided the best overall dura-
a Wet conditioning: 30 days at 71◦ C/100% RH.
bility of the bonded structure. Where it is not practical
to grit blast, light hand sanding should be considered
T A B L E V I B Secondary bonding of Ciba 6376 precured substrates in addition to peel ply.
with FM
R
300-2 adhesive film. Effect of composite surface pretreatment Harris and Beevers [52] have recently investigated
on double cantilever beam testing [21]
the process of dry grit blasting to generate different lev-
Composite Test results els of texture, simply by using different particle sizes
surface of nominally similar alumina abrasive media. It was in-
Test preparation −55◦ C 23◦ C 82◦ C/Wet
tended [52] that this should minimize variability, which
Hinged double Peel ply 45.0 35.0 44.0 might otherwise arise from the different surface treat-
cantilever test (MPa) ment processes such as chemical effects during etch-
Crack energy (kg) Peel ply/hand 45.5 39.0 47.9 ing. Main aims of this investigation were to find re-
sand liable and reproducible methods of grit blasting and
Substrate dried 1 h @ Peel ply/grit 73.6 49.0 80.4
surface measurements. They used mild steel CR1 and
180 (MPa) blast
aluminum alloy 5251 (i.e., metal-metal bonding). Three
Note: Wet conditioning: 30 days @ 71◦ C/100% RH. types of alumina oxide blasting media were used with

32
T A B L E V I I Effect of surface preparation on the LSS of FM 300-2 K, 390 gm/m2 adhesive bonded precured Hercules 8551-7 laminates lap shear
strength (MPa) [21]

Hercules 8551-7/ Failure Hercules 8551-7/ Failure Surface

Test temp. graphite modes glass modes preparation

−55◦ C 13 AL 0.8 AL
24◦ C 19.6 AL 16.3 AL Solvent wipe only
82◦ C 21.6 AL 19.0 AL
82◦ C/Wet 15.1 AL 11.6 90% A, 10% C
−55◦ C 18.2 AL 14.8 LF
24◦ C 23.7 AL 17.9 LF Hand sand (320 grit
sandpaper/solvent wipe
82◦ C 25.6 AL 18.6 AL
82◦ C/Wet 17.8 50% A, 50% C 12.8 90% A, 10% C
−55◦ C 22.6 LF 12.7 LF
24◦ C 27.5 LF 14.8 LF Sand blast/solvent wipe
82◦ C 27.5 LF 14.7 LF
82◦ C/Wet 19.9 50% A, 50% C 11.0 40% A, 60% C

Note: Laminate: Hercules 8551-7 Uni-laminate (±45, ±90, 0)2 .

Cure cycle: 1.7◦ C/min to 120◦ C, 60 min at 120◦ C, /28 MPa.
Failure modes: AL = Adhesive to laminate; C = Cohesive; LF = Laminate failure; A = Adhesive.

slightly different composition and shape. The two grit table displays the average departure of the profile from
sizes used to vary the level of surface texture were sieve the mean plane (Ra ), the root mean square of the aver-
mesh numbers 180/220 and 60, having opening sizes age departure from the mean plane (Rq ) and Wenzel’s
of 71–92 and 250 µm, respectively. Prior to grit blast- roughness factor (r ), which is a ratio between the actual
ing, the substrates were solvent wiped with MEK and surface area and the geometric area. The observations
after blasting, a jet of clean dry air was blown across confirm the expected relationships that the finer grit par-
the surface to remove loose grit particles and dust. Sur- ticles produced smoother surfaces and the aluminum al-
face texture was measured using a non-contacting laser loy surfaces displayed higher levels of roughness than
profilometer with a spot size of 1 µm and a stand-off the mild steel when treated with the same size of grit.
of 2 mm. The effect of the grit-blasting media on surface en-
To measure the adhesion properties the single lap ergy as calculated from static contact angle measure-
shear and tensile butt strength tests were used. The ments is shown in Table X, along with the measured
adhesive used to manufacture these joints was a two- contact angle data. It can be seen that the Pink and White
part room temperature curing epoxy (Araldite 2011), grits produced similar surface energy characteristics,
which was cured in ambient conditions for 24 h and but the Brown has a lower total surface energy with
then subjected to a post curing process of 1 h at 60◦ C. higher dispersive and lower polar components that the
The static contact angle data was generated using a other grit types (see Fig. 6). The polar component also
Video Contact Angle (VCA) Analyser. The surface en- shows greater variation between the different surfaces
ergy information was then calculated, in terms of dis- than the dispersive. The effect of the surface rough-
persive and polar or non-dispersive components (see ness on surface energy for aluminum alloy is shown
Equation 8), using the geometric mean equation com- in Fig. 5, with the Pink and White grits the smoother
bined with Young’s equation (see Equation 4) [56]. surfaces exhibited higher surface energy readings than
Surface energy values for the polar and dispersive the 60 treated surfaces but with the Brown grit, surface
components for the liquids used in the calculations are energy values appear to be independent of roughness.
given in Table VIII. Note that the contact angle mea- Harris and Beevers [52] concluded that the mild steel
surements were made with di-ionised water and di- treated substrates produced higher surface energies than
iodomethane taking an average of twenty 20 µm drops the aluminum alloy, although these values of surface
with each type of liquid. energy are significantly lower than values sometimes
The most intermediate effects of grit blasting are seen quoted for metals [203].
as changes in surface texture. The results of surface The analyses of the elemental composition of the sur-
roughness analysis for different grits, based on 3D pro- face before and after treatment for the mild steel and
filometry measurements, are shown in Table IX. This aluminum alloy substrates were also performed using
the XPS (X-ray photoelectron spectrodcopy) method
in this work [52]. This analysis has shown that sodium
T A B L E V I I I Surface energy components (i.e., dispersive and polar; (Na) is present on the White and Pink treated surfaces
according to Equation 8) of wetting liquids [52] with both the mild steel and aluminum alloy substrates.
On the mild steel, the Brown grit seems to have also
Surface energy (mJ/m2 )
introduced magnesium (Mg), in addition to aluminum
p
Liquid type γLV
d γLV γLV (Al) from the alumina grit residues. The elemental con-
centrations of these residue contaminants appear to be
De-ionised water 21.8 51 72.8
Diiodomethane 49.5 1.3 50.8
dependent on the abrasive particle size, with the coarse
grits generally resulting in lower concentrations. It can

33
T A B L E I X Surface roughness results of the surface pretreated mild steel and aluminum alloys substrates using various grit blasting and solvent
wiping [52]

Surface roughness (µm)

Mild steel Aluminum alloys

Surface treatment Ra Rq r Ra Rq R

Solvent wiped 1.2 ± 0.1 1.4 ± 0.0 1.02 0.4 ± 0.0 0.6 ± 0.0 1.01
Brown 180/220 1.3 ± 0.1 1.7 ± 0.1 1.05 1.7 ± 0.0 2.2 ± 0.0 1.07
Brown 60 3.4 ± 0.3 4.5 ± 0.4 1.12 5.1 ± 0.1 6.7 ± 0.1 1.17
Pink 180/220 1.3 ± 0.1 1.7 ± 0.2 1.05 1.8 ± 0.1 2.3 ± 0.2 1.08
Pink 60 3.2 ± 0.2 4.3 ± 0.3 1.13 5.1 ± 0.5 7.0 ± 0.8 1.21
White 180/220 1.3 ± 0.0 1.7 ± 0.1 1.05 1.7 ± 0.1 2.2 ± 0.1 1.07
White 60 3.0 ± 0.1 4.0 ± 0.2 1.11 4.9 ± 0.2 6.5 ± 0.2 1.18

T A B L E X Contact angle (θ ) and surface energy data [dispersive (γ d ), polar (γ p ) companents and total (γ total )] for grit-blasted and solvent wiped
surfaces for mild steel and aluminum alloy substrates [52]

Contact angle (◦ ) Surface energy (mJ/m2 )

Surface material Surface treatment De-ionised water Diiodomethane γd γp γ total

Mild steel Solvent wiped 68 ± 5 37 ± 7 36 ± 4 10 ± 3 46 ± 3

Brown 180/220 52 ± 7 18 ± 4 41 ± 2 16 ± 4 57 ± 3
Brown 60 61 ± 4 21 ± 4 41 ± 2 11 ± 2 52 ± 2
Pink 180/220 8±3 27 ± 3 34 ± 1 40 ± 1 74 ± 1
Pink 60 28 ± 4 30 ± 3 34 ± 1 34 ± 68 ± 2
White 180/220 6±2 26 ± 4 33 ± 1 41 ± 1 74 ± 1
White 60 34 ± 10 25 ± 3 36 ± 2 29 ± 6 65 ± 5
Aluminum alloy Solvent wiped 54 ± 3 45 ± 5 29 ± 3 21 ± 3 50 ± 2
Brown 180/220 76 ± 4 33 ± 3 39 ± 2 5±2 44 ± 1
Brown 60 75 ± 5 30 ± 4 41 ± 2 5±2 46 ± 2
Pink 180/220 34 ± 3 31 ± 3 34 ± 1 30 ± 2 64 ± 2
Pink 60 51 ± 3 35 ± 5 34 ± 2 20 ± 2 54 ± 2
White 180/220 28 ± 6 30 ± 3 34 ± 1 34 ± 3 68 ± 3
White 60 59 ± 3 28 ± 4 38 ± 2 14 ± 2 52 ± 2

also be noted that the appearence of Na and Mg on the

surfaces after grit blasting correlates with the presence
of these elements in the composion of the grit.
Harris and Beevers [52] determined the initial joint
strengths of the bonded Pink treated surfaces for the lap
shear and tensile butt configurations. They showed that
the initial joint strengths were relativley independent of
grit type (although the joints with grit-blasted surfaces
were typically 50% higher than the solvent wiped sur-
faces). They therefore carried out joint durability tests
in order to measure adhesion characteristics produced
by different sizes of pink grit. Bonded lap and butt joints
were immersed in de-ionised water at 60◦ C and tested at
intervals up to 12 weeks. The results in Fig. 11 show that Figure 11 Effects of grit-blasting media on the strength retention charac-
with mild steel substrates, the lap shear joints displayed teristics of the lap shear joint durability for a mild steel and an aluminum
no difference between the two surface textures. How- alloy [52].
ever, with the tensile butt joints the rougher surfaces
showed a higher level of durability. As observed by
Harris and Beevers [52], with the aluminum alloy, both 6.2. Effects of non-mechanical
the lap shear and butt joint configurations displayed the pretreatments on the bonding of
same characteristics with the smoother surfaces pro- metal-composite, metal-metal and
ducing the higher level of strength retention.The failure polymer-polymer substrate systems
modes of the lap shear joints were consistently 100% Molitor and Young [59] have recently investigated alter-
apparent adhesion failure by non-magnified visual in- native techniques for bonding of a glass fiber reinforced
spection for both substrate materials. With the tensile composite structure to a titanium alloy (grade Ti-15-3)
butt joints, however, the failure increased from 30% sheet metal component (i.e., metal-composite system
apparent adhesion failure to 70% exposure for both the bonding). This was an aerospace application where the
mild steel and aluminum alloy substrates. bond would be subjected to cyclic loading in a hot moist

34
environment and furthermore exposed to de-icing flu- in each case was mostly cohesive, thus indicating ade-
ids and cleaning agents. This work was conducted in quate titanium surface treatment. Surface modification
two stages. Firstly, selected surface pretreatments for by Excimer laser did yield peel strengths as high as 138
the titanium adherend were evaluated by 90◦ peel tests. N/25 mm and as low as 78 N/25 mm. The difference
For these tests, thin strips of felexible aluminum alloy between the two techniques was that surface modifi-
(Al-2024-T3) were bonded to rigid strips of the tita- cation was achieved at a fluence of 77 J/cm2 . For this
nium alloy using a phenolic-based film adhesive. The reason, surface cleaning of titanium by Excimer laser
adhesive, recommended for metal/metal bonding, was does offer an attractive industrial option if comparable
only used for these screenings tests and was not used in peel strengths can be achieved at a significantly lower
the metal/composite bonding. The two best performing fluence than surface modification by the same laser.
titanium surface treatments were then evaluated follow- In conclusion, while these surface treatments did pro-
ing exposure to a hot/humid environment. duce good peel strengths, it was pointed out [59] that
Whereas in those tests, which investigated the envi- the same treatments might not produce durable bonds
ronmental durability of the titanium/composite bond- when exposed to a high/humid environment. In partic-
ing, a number of parameters were varied as part of ular, the durability of Excimer laser treatments has not
the study: (a) two types of glass fiber material were been well documented, and this technique may be out-
evaluated—a thermoset epoxy material and a thermo- performed by the tradationally durable chromic acid
plastic material; (b) two film adhesives (approved for anodisation [65].
use in the aerospace industry) were used. The test spec- Lennon et al. [128] have investigated the effects of
imens were manufactured in accordance with the stan- microwave plasma surface treatments of polyamides
dard BS EN 2243; Part 2: 1991 (Test methods for struc- (PA) film, polyamide 12 (Orgasol
R
2002D: PA 12) and
tural adhesives). The rigid adherend material was the polyamide 11 (Rilsen BD30: PA11 as substrates (i.e.,
titanium alloy Ti-15-3 [204]. In addition to Ti, the al- polymer-polymer system), using the two surface pre-
loy also contains V (14–16%), Cr (2.5–3.5%), Al (2.5– treatment processes: the ammonia and nitrogen mix-
3.5%), Sn ( 2.5–3.5%) as well as smaller percentages of tures by changing the O/N mixtures and NH gases.
O, N, C, H, and Fe. The flexible or “peeling” adherend Two epoxy adhesives were used in this study; an epoxy-
was a thin strip of the aluminum alloy Al-2024-T3. anhydride system and an epoxy-amine system. In both
The adherends were bonded with the phenolic-based cases, the epoxy prepolymer is a diglycidyl ether of
film adhesive Redux 775 [2] at 150◦ C for 60 min in an bisphenol-A (DGEBA) of low molecular weight (DER
autoclave at a pressure of 350 kPa. For these tests, a 332). The anhydride hardener used is methyltetrahy-
tensile loading rate of 70 mm/min was applied to the drophthalic anahydride (MTHPA) with a polymerisa-
load cell and the peeling force recorded. tion catalyst methylimidazole. These two adhesive sys-
As a pretreatment the titanium alloy was degreased in tems were chosen [128] because they differ in their
methylthylketone (MEK) for 10 min and rinsed in dis- polymerisation mode, the functional groups present in
tilled water. Table IVB gives the procedures for the six the reactive systems and the ductibility of the networks
selected titanium surface treatments and their effects. (the MTHPA based network was more brittle than the
Prior to adhesive bonding, the aluminum alloy was de- Jeffamine based network at room temperature).
greased with MEK, rinsed with distilled water and pick- Lennon et al. [128] used two treatments (i.e.,
led in a chromic acid solution at 60–65◦ C for 30 min ammonia and nitrogen/oxygen mixtures) in the mi-
[59]. Following pickling, the material was rinsed in run- crowave plasma in order to investigate the wettability
ning water and dried. The peel force was recorded over of polymide fims. They evaluated the effects of nitro-
a stripping length of 150 mm on the titanium surface. gen, power and treatment time on the surface energy.
The average peel force was then determined over five As seen in Fig. 12 the improvement of the wettability
intervals of 25 mm along this length and was expressed
as N/25 mm. These results are given in Table XI. Based
on the mean peel strengths given in Table XI, the best
performing titanium surface treatments were: surface
modification by Excimer laser (97N/25 mm), surface
cleaning by Excimer laser (96 N/25 mm) and sodium
hydroxide anodisation (SHA) (84 N/25 mm). Failure

T A B L E X I 90◦ Peel test results for the surface pretreated titanium

alloy using various surface preparation methods [59]

Peel strength
Surface treatment (N/25 mm)

Untreated 20
Surface modified by Excimer laser 97
Surface cleaned by Excimer laser 96
Plasma sprayed 60 Figure 12 Effect of nitrogen % in the O/N plasma on the surface energy
Pasa Jell 1078 64 of the Polyamide 11 (i.e., PA 11) substrate film [128]: (a) polar compo-
Sodium hydroxide anodization 84 nent () and (b) dispersive component (); Microwave plasma surface
Chromic acid anodization 54 treatment conditions: power = 800 W, gas flow = 100 cm min, treatment
time = 120 s.

35
depends not essentially on the quantity of oxygen in
the mixture but simply on its presence. The surface en-
ergy (i.e., polar and dispersive components) does not
change by treatment with nitrogen alone. Once oxygen
is added, the wettability increases significantly with a
reduction of the water contact angle on the PA11 from
75◦ to approximately 45◦ . However, there was not any
difference between the wettability or the O/C ratios of
the films treated in pure oxygen and those treated with
the gas mixture. From the investigation of the plasma
power and the treatment time on the surface energies
of the polyamides 11 and 12 Lennon et al. [128] con-
cluded that an improved level of wettability can be ob-
tained for a 200 W treatment and raising the power
does not amplify the effects, whereas the the surface Figure 13 Effect [128] of surface pretreatment time on the pull-off test
energies show a rapid increase with treatment and the on the assembly constituted by O/N microwave plasma treated Orga-
energies continue to rise between 100 and 360 s plasma
◦
sol 2002D and DGEB/MTHPA/MIA ( : load at break and : work at
break).
exposure times. Whatever the power and the treatment
time the two kinds of PA have similar behavior. The
surface energy obtained after 360 s of treatment is over
60 mN/m compared to the value of 40 mN/m of the un-
treated films [128]. Optimal treatment conditions with
respect to the wettability and the chemical composition
of the surface can therefore be determined as 100 cm
min, 200 W, 360 s, which can be used to evaluate the
adhesion of the substrate [128].
They used also three mechanical tests for the adhe-
sion measurements between the epoxy networks and
the PA film covering a steel substrate; (a) pull-off, (b)
three-point bending, and (c) lap-shear tests. For the pull- Figure 14 Influence of the nature of microwave plasma surface pretreat-
ment on the adhesive properties of Orgasol 2002D/DGEBA-MTHPA by
off test, an aluminum stud was adhesively bonded to
lap shear test [128] (O: load at break; : work at break).
the PA film substrate and a 4.9 N compression load
applied to each stud. The three-point failure test was
carried out according to the standard NFT 30010. The (b) gas flow, and (c) composition of the gas mixture in
epoxy adhesive was used as a stiffner on the PA coated the case of O/N treatment has lead to determine optimal
substrates (10 × 50 × 0.8 mm) with a contact surface treatment conditions with respect to the wettability: for
of 5 × 25 mm. The ultimate load (F), the ultimate O/N, a composition 20–80%, a power of 200 W with
displacement l and the subtended energy (W ) were a gas flow of 100 cm min during 180 s were chosen
used. The lap-shear test was carried out according to whereas for ammonia treatment the optimal conditions
the standard NFT 76107. In all these three tests, af- were 200 cm min—600 W—180 s.
ter curing, the samples were left in an air-conditioned In conclusion, after plasma treatment and whatever
room (22◦ C, 50% RH) for 24 h before testing. The re- the gas used, a great increase of the adhesion prop-
sults using pull-off test on assembly constituted with erties was noticed whatever the epoxy reactive sys-
O/N plasma treated PA and epoxy/anhydride reactive tem. The propagation zone was located near the in-
system are presented in Fig. 13. It can be seen that the terface PA/primary adhesive showing clearly that the
O/N plasma treatment increases strongly the toughness plasma treatment reinforces the PA/epoxy interface. As
of the PA/adhesive interface. The increase of rupture regards, the adhesion of the NH-plasma-treated films
energy (by a factor of 5.4 for a treatment of 6 min and shows an important increase compared to the untreated
by a factor of 4 for a treatment of 4 min) is a much im- samples that can be attributed to the introduction at
portant as the O/C ratio increases [128]. This behavior the surface of nitrogen species (N/C goes from 0.09 to
is in agreement with other studies for PA [193, 197] and 0.15 after 6 min of treatment). These results are in good
other polymers [205, 206]. agreement with those of Wertheimer et al. [207], who
The comparison with the NH plasma treatment was show that load for crack propagation during peel test
made with the optimum conditions, with the same ex- is slightly higher for N/O plasma treatment than with
posure time (6 min) using the lap-shear test and the NH one in the case of aromatic polyamides. With the
anhydride based system as adhesive. The results are untreated Orgasol
R
2002D, studied in three-point bend-
presented in Fig. 14. Both plasma treatments princi- ing, both the initiation and propagation zone are located
pally lead to an increase of the non-dispersive compo- at the polyamide-epoxy interface. When the polyamide
nent of the surface energy related for O/N treatments is plasma treated (both with NH or N/O ) the crack
to the increase of the O/C ratio and for NH treatments initiation is again located at the interface PA/epoxy
to the introduction at the surface of nitrogen species. but, undoubtedly, the crack enters the polyamide film,
The study of different parameters such as (a) power, near the interface PA/primary adhesive, showing clearly

36
that the plasma treatment reinforces the PA/epoxy T A B L E X I I I Roughness values for corona discharge, flame, fluori-
interface. nation, vacuum plasma and air plasma pretreatments for homopolymer
polypropylene, HF 135 M [153]
Green et al. [153] have studied the effects of vari-
ous surface pre-treatments on the polypropylene (PP) Pretreatment RA (nm)
(monomer: CH2 = CH–CH3 ) substrates (i.e., polymer-
polymer substrate). Among the pre-treatment methods Corona discharge 1.325
Flame fluorination 1.173
studied were corona discharge, flame, fluorination, low Vacuum plasma 0.616
pressure vacuum plasma and atmospheric plasma were Air plasma 1.201
the most effective ones. It was found that all these five
surface pretreatment methods showed the highest sur-
face chemical modification of the pretreatments stud-
ied. It was identified that the surface chemistry, con- treatments have been found to incorporate oxygen func-
centration depth and topography varied widely across tional groups into the surface. Vacuum plasma and air
the five pretreatments. However, all have been shown plasma have been found to include both oxygen and
to have similar bond strength with polyurethane adhe- nitrogen functional groups. Similarly, the fluorination-
sives, indicating that a number of significant factors treated PP spectra were found to contain both oxygen
were responsible for bond strength. It is surmised that and fluorine functional groups.
the depth of the functional group concentration is the Using the AFM (atomic force microscopy) mi-
determinant joint strength parameter and the O/C ratio croscopy the characteristics of surface topography of
or surface roughness. the substrate was studied following the pre-treatment.
Among the pretreatment methods studied, corona The roughness values of pre-treated surfaces for various
discharge, flame, fluorination, vacuum plasma and air methods are presented in Table XIII. All the pretreat-
plasma were found to increase surface oxygen content ment methods (i.e., corona, flame, vacuum plasma and
through the incorporation of oxygen groups into the air plasma all involved most oxidised surfaces) exam-
near-surface region, whilst all other pretreatments were ined were found to increase roughness above that of
found to have incorporated no additional oxygen into the original “atomically smooth surface” moulded up
the surface layer (see Table XII). Corona was found cleaved mica. The roughness was described by the RA
to incorporate the least amount of oxygen (6.69 at%) value, which is the ratio of the actual surface length
into the surface of the 5 pre-tretments, whilst vacuum to the surface length of a smooth surface. The RA
plasma induced the most (12.99 at%), following the value is an arithmetic mean of the absolute value of
trend from left to right of decreasing oxygen content the departures of the profile from the mean line and
has been observed: is often quoted as an engineering measurement of sur-
face roughness. Corona discharge was found to pro-
Vacuum plasma > air plasma > fluorination > flame duce a very microrough surface with features and nod-
ules around 100 nm in diameter. Flame treatment was
> corona discharge. found to create a “macrorough” surface. Fluorination
was found to only roughen the surface in a very micro-
It was found that pre-treatments, which physically mod- rough manner with large domed nodules creating a dim-
ified the surface leading to an initial contact angle of less pled surface. Vacuum plasma was found to create an ex-
then 60◦ , gave an increased O/C ratio. It was noted that tremely heterogeneous surface with a very high surface
within the high-resolution carbon spectra, significant density of equally sized and spaced micofine nodules.
peak broadening occurred for substrates with a contact Air plasma was found to induce a “micro-rough” sur-
angle less than 60◦ . This significant peak broadening is face with features of ∼80 nm diameter.
due to multiple types of functional groups being present To evaluate the effects of various surface pretreat-
on the surface. Corona discharge and flame surface pre- ments mentioned in (Table XII) the maximum lap
shear strength of the treated substrates bonded using
polyurethane adhesive was measured. Of all the sur-
T A B L E X I I Effects of various surface pretreatment methods on the
surface chemistry modifications (i.e., changes of C, N, O, Si, F, and O:C
face pretreatments examined it was found that corona
ratio) of homopolymer polypropylene, HF 135 M [153] discharge, flame, fluorination, O and N vacuum plasma
and AgrodynTM atmospheric (air) plasma were the most
C N O Si F effective at chemically modifying surface. It was noted
Pre-treatment (at.%) (at.%) (at.%) (at.%) (at.%) O:C
that the silver electrolysis pretreatment caused little sur-
None 99 0 0.5 0.5 0 0.005 face pretreatment effect, which is contrary to findings of
Corona 93.16 0 6.69 0.15 0 0.072 Brewis et al. [200]. Green et al. [153] concluded that the
Flame 91.75 0 7.58 0.51 0 0.083 bond strength increasses can be correlated to the suc-
Fluorination 79.1 0 10.3 0 10.24 0.130 cessful increase in O:C ratio of the surface chemistry.
Vacuum plasma 84.549 2.46 12.99 0 0 0.154
Air plasma 87 0.84 12.16 0 0 0.140
Broad et al. [151] have investigated the effects of
0.025% MA 100 0 0 0 0 0.000 the Ciba laser pretreatment (CLP) process on the joint
0.05% MA 100 0 0 0 0 0.000 performance of the aluminum and titanium alloys, and
Proprietry 99.1 0 0.9 0 0 0.009 stainless steel (i.e., metal-metal substrate). Fig. 15a
coating shows that using CLP pretreatment process, better
IR laser 98.1 0 1.9 0 0 0.019
Ag(II) 97.2 0 2.7 0 0 0.028
joint strengths with fully cohesive failure modes were
achieved before and after aging for the aluminum alloy
37
 (a)
Figure 16 Test results [151] showing comparison of the time to failure
(at the creep conditions of 5 MPa and between the test temperatures of
42–48◦ C under the 100% relative humidity) of the CLP method with the
degreased and pickle-etched surface preparation processes, indicating
that the joints produced with the CLP method differ greatly from those
simply degreased or pickle-etched.

component epoxy adhesive. The test criterion was the

duration until breaking of the specimens. Even under
these demanding test conditions the joints produced
with CLP differ greatly from those simply degreased or
even pickle-etched (see Fig. 16). Despite using a struc-
tural adhesive, the joints on aluminum surfaces that
(b) have been degreased alone failed after only 90 days.
Figure 15 Comparison [151] of the Ciba laser pretreatment (CLP)
A higher durability of 350 days was achieved with a
method with other surface preparation techniques using the lap shear wet process (i.e., pickle-etching). This suggests an evi-
strength of various substrates joints: (a) an aluminum alloy substrates dence for the effectiveness of CLP to achieve adhesively
joint with with 2-component epoxy adhesive before aging and after 14 bonded joints with good durability [151].
days Cataplasma and (b) an aluminum alloy substrate (contaminated
surface).

substrate. The “degreased only” specimens break at 7. Factors affecting the durability
considerably lower strength levels and with adhesion of adhesively bonded joints
failure. By applying primer alone or treating solely with As pointed out by Davis and Bond [152], the formation
a laser, the joining charactersistics have improved; how- of a suitable surface chemistry is the most important
ever, the joints still fail by adhesion failure after ageing. step in the surface preparation process because the in-
Only the combination of primer application and laser tegrity of this surface directly influences the durability
treatment leads to such a strengthening of the adhesion of the adhesive bond. To study the ability of surface
forces between the aluminum surface and the adhe- pretreatments to promote adhesion and to protect the
sive to give high joint strengths with cohesive failure surface against corrosion, it is usually necessary to run
modes before and after ageing. As shown in Fig. 15b, accelerated aging test to differentiate the surface treat-
the CLP did not necessarily require cleaned surfaces ment quality. Aging tests run under normal conditions
prior to surface pretreatment. The better durable ad- of temperature and humidity after the curing of the ad-
hesively bonded joints can also be achieved for other hesive may not provide results allowing a clear ranking
metallic alloys (i.e., stainless steel and titanium alloy) of different surface pretreatments adhesion promoter,
using the CLP process [151]. Whereas without appli- and anti-corrosion protector abilities. To do so, adhe-
cation of CLP the majority of specimens showed the sively bonded samples have to be exposed to severe
failure modes even for the uncleaned and artificially conditions of temperature and humidity. As pointed
aged surfaces of the alloys mentioned above, in all cases out by Lapique and Redford [208], when aging tests
to cohesive failure [151]. It was claimed [151] that the are run to evaluate the quality of surface pretreatments,
CLP is applicable in combination with various adhesive changes in the mechanical properties of the adhesive
types, i.e., one- or two-component epoxy adhesives as should also be taken into account. They may affect the
well as polyurethane adhesives and sealants. the stress transfer from the adhesive layer to the inter-
Broad et al. [151] have also performed the Stress- face and modify the results of the mechanical testing.
humidity testing on the adhesive bonding using the CLP In order to analyze correctly such experiments, the be-
process. In this test, small lap shear specimens were havior of the adhesive should be studied under aging
subjected to cyclic elevated temperatures at 100% rel- test conditions. Lapique and Redford [208] have re-
ative humidity and a mechanical load of 5 MPa. The cently studied such an aging test (40◦ C in water vapor)
demands on the adhesive were increased by the three on the Araldite 2014 epoxy adhesive samples and the
holes drilled into the overlap area, thereby drastically adhesive properties recorded over a period of 36 days.
reducing to a minimum the distance for diffusion of Araldite 2014 is an adhesive paste that can be cured
moisture into all areas of the joint; note that this test at room temperature. This can be important in com-
specimen is made of aluminum bonded by the one- mercial applications where the substrates to be joint
38
 T A B L E X I V Surface roughness for a mild steel (i.e., BS4360 grade
43A) using various grit-blasting grades, and polished surface [73]

Avrage Average Average root

roughness profile length mean square
(µm) (µm) slope (◦ )

Surface finish Ra Rlo Rdq

Grit-blasted surface
120/180 0.98 ± 0.05 13.44 ± 0.01 12.52 ± 0.59
40/60 2.97 ± 0.18 13.69 ± 0.04 22.70 ± 0.56
Figure 17 Weight increase versus aging time during aging test for sam- 30/40 4.23 ± 0.25 13.84 ± 0.04 24.24 ± 0.89
ples with a thickness of 1.5 mm. The weight increase follows a t 1/2 law 24/30 6.31 ± 0.28 13.95 ± 0.07 25.96 ± 0.82
[208]. Polished surface 0.04 ± 0.02 12.79 ± 0.01 0.13 ± 0.02

do not tolerate high temperature. As seen in Fig. 17,

the measurement of the weight of the samples shows a
weight increase due to an increasing water content in steel specimens using the dimond polishing and grit-
the epoxy with the exposure time. The weight increases blasting with four different sizes (i.e., 120/180, 40/60,
with the square root of time in agreement with the stan- 30/40 and 34/30 mesh) of alumina grit. The material
dard Fickian diffusion. According to Fig. 17 the water used in this study were mild steel to British Standard
diffusion continues even after 36 days and no equilib- BS4360 grade 43A as substrate and structural epoxy
rium state is reached during that period of time. Water adhesive, Araldite 420A/B (Redux 420A/B). After pre-
works usually as plasticizer leading to a softening of treatment, the surface roughness of adherends, mea-
the material. This is also the case for the epoxy adhe- sured as average roughness Ra , linear profile length
sive Araldite 2014. Fig. 18 shows how the mechanical Rlo , and root mean square slope (◦ ) Rdq values are
properties (stress at break, E-modulus) and deforma- presented [73] in Table XIV. Note that Rlo , and Rdq
tion at break) are affected by the exposure time and parameters were defined according to the standards
thus by water content. As Lapique and Redford [208] ISO 4287 (1984) and ISO 4287 (1997), respectively.
concluded, the E-modulus and the stress at break are Fig. 19 shows the relationship between the average
lowered while the deformation at break and the plastic cleavage strength and the Ra values of the adherend sur-
contribution to the total deformation are increased with faces. It can be seen that the cleavage strength appears
increasing aging time. to increase linearly with the Ra value. As suggested
Some form of substrate pretreatment is always neces- by Shahid and Hashim [73], the increase in cleavage
sary to achieve a satisfactory level of bond strength. In strength may be attributed to an increase in surface
order to obtain a strong and stable bond between a sub- area by forming of mini scarf joints on adherend sur-
strate (such as a metal) and the adhesive, the naturally faces at micro level. This finding is consistent with that
formed surface oxide on metal has to be removed and of Sargent [209]. However, Harris and Beevers [52],
replaced with a new, continuous, solid and corrosion re- Thery et al. [210] and Critlow and Brewis [65] found
sistant oxide layer during such a surface pretreatment no appreciable change in joint strength with increasing
process. Almost all treatment methods do bring some adherend surface roughness by mechanical treatment.
degree of change in surface roughness but grit-blasting These contrasting findings may be due to the fact that
is usually considered as one of the most effective meth- each researcher used different set of adherend, adhesive
ods to control the desired level of surface roughness and and joint geometry [73]. Moreover, the overall effect of
joint strength [73]. Grit-blasting does not only remove grit blasting is not limited to the removal of contamina-
weak boundary layers but can also alter the chemical tion or to an increase in surface area. This also relates to
characteristics of the adherends [52, 70]. changes in the surface chemistry of adherends [52, 70]
Shahid and Hashim [73] have studied the effects of and to inherent drawbacks of surface roughness, such
2
surface roughness on the cleavage strength of a mild as void formations and reduced wetting [211]. Rlo was
considered as a measure of the effective surface area
available for bonding. These values were then com-
pared with cleavage strengths as shown in Fig. 20. It

Figure 18 Mechanical properties (i.e., stress at break, E-modulus) dur-

ing aging test measured in tension. Each point is the average of measure- Figure 19 Graph showing variation of cleavage strength with average
ments performed on five samples [208]. roughness, Ra [73].

39
 network structure in the near-interphase region but also
far from the substrate.
The pretreatment of the aluminum substrate is essen-
tial in order to prevent or hinder the deterioration of the
mechanical performance of the joints as a result of ex-
posure to a wet and corrosive environment [136, 213].
Degradation of adhesive joints is associated with diffu-
sion of water to the adhesive/substrate interface causing
hydration of the oxide conversion coating and loss of
Figure 20 Variation of cleavage strength with effective area of bonding adhesive strength. A moderate temperature increase,
(experimental results) [73]. which in the absence of water does not have an ad-
verse effect on a structural joint, leads to a much more
pronounced loss in strength. Presence of chloride ions
can be seen that the cleavage strength increases as the is further expected to accelarate the loss of adhesion,
effective surface area increases. It is, however, realis- e.g., by propagation of localized corrosion of the metal
tic to believe that this increase in strength combined beneath the adhesive [228].
with an increase in surface area shall be limited by Lunder et al. [214] have recently investigated the
bulk adhesive strength, i.e., until failure becomes cohe- effects of various surface pretreatment methods on
sive within the adhesive [73]. Shahid and Hashim [73] the surface characteristics and durability of adhe-
have also observed that the specimen prepared by grit- sive bonded AA6060-T6 aluminum alloy joints (see
blasting steel adherends with coarser grit have shown Table XV for various surface pretreatment methods ap-
significantly higher crazing (stress whitening zones due plied in this study). Aging of the lap joints was per-
to micro-cracking) in the initial area of the joint com- formed by exposure for 50 days in climate chamber at
pared to those prepared with the finer grit. In both cases, 82% RH and 40◦ C, corresponding to standard condi-
however, failure was near the interface region and ap- tions employed in filiform corrosion testing (DIN EN
parently in a mixed adhesion/cohesion mode. 3665). Table XV also gives the oxide film thickness
Bockenheimer et al. [212] have studied the effects occurred as a results of these pretreatments. As ex-
of blasting with alumina grit or with glass beads on pected, these films are considerably thinner than the
the topography of an aluminum alloy +3% Mg sub- AC anodized films, particularly the ones given a final
strate. These two pretreatment methods has changed acid etch. Fig. 21 shows the tensile test results (i.e., in
the topography and chemical state of the aluminum sur- terms of maximum load) of the lap joints. The results
face. Unexpectedly, after surface pretreatment the oxide show hardly any degradation of hot AC anodized (pre-
was not in all cases Al2 O3 . Al2 O3 was only found for treatments 1 and 2) and FPL etched (pretreatment 7)
the etched surface, whereas Al2 O2.87 and Al2 O1.3 were specimens as a result of the climate chamber exposure.
found 1 day after blasting with glass beads and alumina A significant reduction in joint strength was observed
grit, respectively. Hence, not all Al atoms possess their for the alkaline etched specimens (pretreatment 3), and
maximum degree of oxidation (+3) in the blasted sur- the relative strength loss was further increased when
faces. These aluminum atoms are expected to be very specimens were desmutted after alkaline etching (pre-
reactive due to their deficit of electron density. There- treatments 4 and 5). Application of the phosphate-
fore, it was concluded [212] that the effect of mechan- permanganate conversion coating (pretreatment 6) af-
ical treatment for aluminum bonding is very complex ter alkaline etching results in still poorer results, both
and comprises toplogical changes and changes in the in terms of initial strength and degree of degradation.
chemical surface state of the adherends. Also, the pre- The analysis of specimen surfaces has shown that the
treated surface not only influenced the formation the alkaline etched surface exhibits a mixture of Al and

T A B L E X V Various surface pretreatment applied on the AA6060-T6 aluminum alloy. This table also shows the metal removal by these pretreatment
methods and approximate thickness conversion oxide films based on gravimetric measurements [214]

Metal removal Oxide thickness

Pretreatment (µm) (µm)

1. AC Anodizing of the as-received extrusion in hot (80◦ C) 15% sulphiric acid at a current density of 0.08 0.10
10 A/dm2 (rms) for 4 s, rinsing under tap then distilled water, and draying in hot air.
2. As pretreatment 1, but with anodizing time of 12 s. 0.24 0.22
3. Degreasing in acetone, then etching in 10 wt% NaOH solution for 50 s at 60◦ C, rinsing under tap 5 0.01
then distilled water, and air drying.
4. As pretreatment 3, but with a 15 s de-smutting in concentrated nitric acid following alkaline etch. 5 0.002
5. As pretreatment 3, but with a 15 s de-smutting in 15% sulphiric acid + 2 vol% nitric acid following 5 0.002
the alkaline etch.
6. Pretreatment 3 followed by conversion coating in phosphatepermanganate solution (100 g/l NaH2 PO4 , 5 0.02
30 g/l KMnO4 and 0.5 g/l NaF adjusted to PH2 with sulphiric) for air drying.
7. FPL etch pretreatment; degreasing in acetone, hot water rinse, immersion in FPL etch solution 1 0.004
(16.3 vol% H2 SO4 , 6 wt% Na2 Cr2 O7 ) for 20 min at 60◦ C, 20 min rinse under tap water, drying with
hot air stream.

40
 Surface contamination is a widespread problem in
the adhesive bonding of some parts such as micro-
assemblies, and therefore calls for a convenient method
to assess the cleanliness of the adherends before intro-
ducing the adhesive to the microsystem. As pointed
out by Woerdeman et al. [112], surface cleaning or
preparation by means of solvents is the most common
process step in many electronic manufacturing opera-
tions. Understanding the link between surface prepara-
Figure 21 Tensile test results obtained for adhesive bonded AA6060-T6 tion and adhesion can lead to a reduction in the time
single lap joints before and after exposure in climate chamber (82% RH,
required for development and testing of cleaning pro-
40◦ C, 50 days). Error bars indicate standard deviation for three repli-
cate specimens [214]. : Reference specimens and : After corrosion cedures. It can also lead to improved bondlines and en-
testing. hanced manufacturability, enabling product realization
for encapsulated components [112]. Molecular model-
ing has been employed to elucidate the behavior of a
Mg, which are removed by desmutting. It was sug-
low molecular weight contaminant near an interface be-
gested [136, 215–217] that a high MgO concentration
tween a polymer adhesive and a substrate (see for exam-
in the oxide layer is unfavorable for the bond durabil-
ple references [218–221]. Depending on the substrate-
ity. Whether Mg is present as an oxide or a hydroxide
polymer and polymer-polymer interactions, as well as
is probably of significance. As pointed out by Lunder
the molecular architecture of the contaminant and poly-
et al. [214], the Mg enriched oxide formed as a result
mer, the contaminant can preferentially diffuse toward
of heat treatment may be more susceptible to hydration
the interface [222]. When this happens, the contaminant
and subsequent delamination than the Mg rich hydrox-
molecules will displace the polymer from the interface,
ide formed as a result of alkaline etching. However,
as illustrated in Fig. 23, resulting in a degradation of the
presence of Mg oxide in the aluminum oxide film is ex-
adhesive bond strength [223]. Another molecular mod-
pected to improve the stability in alkaline environment
eling has also been undertaken by Shanahan [224] to
due to the fact that Mg oxides are thermodynamically
examine the role of diffusion in the wetting of a contam-
passive at high pH [214]. Fig. 22 shows a correlation
inated surface. Results of modeling indicate that a thin
between the joint strength after corrosion testing and
layer of a second liquid present on the substrate (the
the extent of cohesive failure. As might be expected,
contaminant) can penetrate the sessile drop of liquid
the these results indicate a higher degree of adhesive
deposited on the solid surface and gradually change the
failure for the inferior pretreatments. Furthermore, the
effective interfacial free energy between the substrate
degree of adhesive failure is generally increased on all
and the drop [224].
variants as a result of environmental exposure, also on
Davis and Bond [152] have investigated the fac-
the AC anodized joints (pretreatment 2) where a negli-
tors affecting the durability problem for the adhesively
gible strength loss was observed.
bonded structural joints using the “clean surface con-
In summary, pretreatment of the AA6060 material
cept” approach. Most adhesive bond failures can be
by alkaline etching in hot NaOH solution resulted in
attributed to poor processes during fabrication, with
the formation of a scalloped surface with a Mg en-
lack of quality surface preparation being the most sig-
riched hydroxide film. A significant reduction in joint
nificant deficiency [152]. The best joint design in the
strength of about 7% was observed for the specimens as
world will not be durable if the surface preparation is
a result of corrosion testing. Desmutting in nitric or sul-
inadequate. The influence of pretreatment surface con-
phuric acid to remove the Mg enriched hydroxide film
ditions on the short- and long-term strength of adhesive
resulted in still higher losses in strength. Application
bonds are presented in Table XVI. Short term strength
of a phosphate-permanganate conversion coating to the
is relatively easy to obtain with minimal surface prepa-
alkaline etched substrate also had no beneficial effect.
ration. If a simple lap shear test (ASTM D1002) is un-
However, all pretreatments employed are superior to
dertaken shortly after complation of the bonding pro-
acetone degreasing alone.
cess, acceptable lap-shear strengths can be obtained

Figure 22 Correlation between (apparent) mode of fracture and peak

load during tensile testing of corroded lap joints [214]. : Hot AC an- Figure 23 Schematic of a polymer adhesive/substrate interface in the

odized (4 s), : Hot AC anodized (12 s), : NaOH etched, : NaOH absence (left) and presence (right) of surface contamination. In the lat-
•
etch + HNO3 desmut,
: NaOH etch + H2 SO4 desmut, : Phosphate ter case, the contaminant molecules can displace the polymer from the
◦
permanganate, and : FPL etched. interface resulting in a degradation of the adhesive bond strength [112].

41
T A B L E X V I The influence of surface condition on the short- and long-term strength of adhesive bonds [152]

Condition Short-term effect Long-term effect

(a) Contaminated surface – Chemical reactions are inhibited – Weak bond which fails soon after
manufacture
– Bonds are weak
(b) Clean surface without hydration – Short-term bonds are strong – Gradual degradation leads to
resistance interfacial failure,
– The joint strength will decay to zero.
(c) Clean and chemically active surface – Short-term bonds are strong – Bond maintains integrity through its
with hydration resistance service life.

even with minimal preparation. However, long-term the adhesion energy is computed by first substracting
bond strength depends on the resistance of the chemical the area under the loading curve from the area under the
bonds to degradation over time. In many cases, poorly unloading curve, which in this particular case, was fit-
prepared metallic surfaces are due to hydrated oxides ted to two lines, as illustrated in Fig. 8 in reference
which displace the chemical bonds between the adhe- [112]. This quantity, H , is calculated using Equation 3
sive and the surface [152]. The consequence being that in reference [112]. H is subsequently normalized with
the adhesive bond fails along the interface some time respect to a normalization factor, N , where
after fabrication (adhesion failure).
As Woerdeman et al. [112] pointed out, surface anal- N = Pmax x amax
3
(20)
ysis techniques, such as X-ray photoelectron Spec-
troscopy, Auger Spectroscopy, and seconddary ion giving rise to a newly defined “adhesion hystheresis
mass spectroscopy, are routinly employed to charac- parameter,” H , where
terize surface properties of substrates, however these
ultra-high vacuum techniques are not practical for prob- H
ing liquid contaminants. However, the JKR contact me- H= (21)
N
chanics technique has a number of unique advantages
for examining microscale areas in a processing envi- Calculations of H are presented in Table XVII. Two
ronment, as it is relatively inexpensive, versatile, easy replicate runs were made for the clean aluminum sur-
to operate, and can be operated in ambient conditions face with the low modulus epoxy lens. These two values
[115]. are very close to each other and very different from the
Woerdeman et al. [112] have used the JKR method value obtained on the contaminated surface. And with
as a means to assess the level of contamination at an both epoxy systems, H was found to be a factor of
epoxy/aluminum oxide interface (see Section 3.4 for two or three lower in the presence of the contaminant
detail about this method). The aim of this study was as opposed to when the contaminant was absent from
to demonstrate the feasibility of the JKR contact me- the surface. With this quantitative distinction between
chanics for probing organic contamination on metal “clean” and “contaminated” substrates, one could em-
subtrates. This was achieved by conducting JKR ex- prically determine the cutoff point for H (Hc ), below
periments in the presence and absence of an organic which the bond strenth would be adversely affected in
contaminant. The model contaminant was hexadecane, a given application. Success of this method hinges on
a non-polar hydrocarbon fluid of medium molecular the fact that the surface energy difference between the
weight. The hexadecane was chosen because it repli- epoxy elastomer and the model contaminant is large;
cates typical machining fluids, is non-reactive with namely, the surface energy of amino-cured epoxide is
aluminum oxide surfaces, and should not readily dis- 46.2 mJ/m2 [225].
solve into the adhesive system of interest. This was ac-
complished by measuring the interactions between an T A B L E X V I I Values of adhesion hysteresis parameter H for the
epoxy probe and aluminized substrate, both in the pres- interactions between a diglycidyl epoxy elastomer and an aluminum
ence and absence of hexadecane surface contamination. oxide substrate using the JKR contact mechanics approach [112]. Note
Woerdeman et al. [112] performed JKR loading and un- that surface cleanliness can be estimated quantitatively by using the
loading experiments, and of particular interest was the adhesion hysteresis parameter, H
extent of adhesion hystherisis that resulted upon bring- Air-plasma- Hexadecane-
ing the epoxy hemisphere into contact with the sub- cleaned contaminated
strate. To be able to readily compare levels of contami- Epoxy elastomer aluminum aluminum
nation on samples of interest, they devised a scheme to
DER331/ED-900 epoxy: 1.17 0.383
extract the information embedded in the adhesion hys-
High Modulus system
theresis curves in a quantitative fashion. Notwithstand- (K = 9 MPa, from
ing the specific interactions that have evolved inside the JKR theory)
contact zone, the area under the unloading curve will DER332/ED-2003 epoxy: 1.68, 1.73 0.867
depend upon parameters such as the maximum load Low Modulus system
(K = 3 MPa, from
achieved in the loading experiment (Pmax ), as well as
JKR theory)
the bulk modulus of the lens, K . With this in mind,

42
 Therefore, it is concluded that the JKR contact me- sensitive to displacement by diffused water. Thus the
chanics can be used to detect the trace amounts of sur- rougher surfaces with high Na and polar surface energy
face contamination. This approach capitalizes on the values would exhibit more rapid degradation. The in-
difference in surface energy between the epoxy probe ability of the lap shear durability tests to detect these
and organic contaminant. Using this approach, surface changes may be due to the lower sensitivity of the test
cleanliness can be estimated quantitatively by comput- method reported by Beevers [229]. Although the alu-
ing an adhesion hysteresis parameter, H , defined in this minum alloy exhibited similar changes in the surface
approach. energy and surface chemistry, the responses from the
Hart-Smith [17] has investigated the surface assess- durability tests were opposite to those observed with
ment in bonded, co-bonded, and co-cured composite mild steel, as reported by Harris and Beevers [52]. This
structures. As underlined in previous sections, the con- may be due to the different degradation mechanisms
tamination prior to bonding is not merely an issue of between two substrates [52]. It is commonly observed
cleanliness, but one of the reactivity and surface en- that when bonded joints of aluminum alloy are exposed
ergy which relates to the wettability. Mahoney [226] to hot, wet environments, loss of strength results from
has studied this problem. Surface contamination may hydrolisis of the oxide layer rather than through ther-
cause insufficient wetting by the adhesive in the liquid modynamic displacement as in the case of mild steel.
state for the creating of a durable bond [17]. But these As Harris and Beevers suggested, It is possible that the
contaminants are not the only cause of such a condition. surfaces formed by the coarser grit may result in a less
For example, composite laminates cured against an in- stable oxide layer and/or a more rapid route for moisture
ert (inactive) surface are far harder for adhesives to wet ingress is created by the surface texture characteristics,
than are the underlying surfaces exposed by abrasion. for example, by capillary channeling.
The nature of the adhesive also has an effect on the bond Green et al. [153] have studied the effects of 13 sur-
strength. It was observed that a composite peel-ply sur- face pre-treatments (see Table XII) on the homopoly-
face, to which one epoxy film adhesive does appear to mer polypropylene (PP) (monomer: CH2 = CH−CH3 )
bond securely if cured at 177◦ C, is not even wet by substrates using surface chemistry, topography and the
a room-temperature curing paste adhesive [17]. One lap shear test. It was reported that the adhesive bond
reason for this is that the hot-bond adhesive becomes strengths of these pretreatments varied widely follow-
fluid during cure, while the paste adhesive maintains a ing a similar pattern to that shown by the O:C ratios
high viscosity. The paste adhesive, when it is separated (see Table XII). The most effective pretreatments, based
prematurely under a shear load, was embossed by a per- on joint strength, namely, corona, flame, fluorination,
fect replicate of every filament in the peel ply surface to vacuum plasma and air plasma all showed both large
which it had refused to bond. The hot bonded structures, increases in surface chemistry in the form of oxygen
on the other hand, had been used without failure for take-up and bond strengths, respectively. It was also
years. Studying the wettability problem, Schliekelmann found that the bond strength was not related to the sur-
[227] pointed out that it is necessary that the surface ten- face topography of the pretreated samples, with both
sion of the liquid adhesive should be lower than the crit- smooth and roughened surfaces creating joints having
ical surface tension of the adherend. He cited a four fold similar strength (see also Table XIII). It is therefore
reduction in the adhesive energy (work of adhesion) of concluded from this that the micro-roughening effects
carbon-epoxy laminates due to the presence of water, of these pretreatments have little effect on adhesion
explaining thereby why it is then so much more difficult of the joints especially with viscous adhesives such as
for adhesives to bond than when everything is dry. He polyurethanes [153].
concluded [227] that the reliable surface pretreatment It was observed [153] that corona discharge and flame
is abrasion of the CFRP composite surface by some both showed some of the highest bond strengths despite
means of sanding or (fine grit) blasting, citing evidence having the lowest O:C ratios of the five effective pre-
that this was recognized as long ago as 1976 [228]. treatments investigated. However, from depth analysis
A detailed study of the effects of grit blasting with it was observed [153] that the highest concentrations
different alumina grits on the surface characteristics of oxygen content were within the first few nanome-
of a mild steel and an aluminum alloy substrates by ters of the surface and rapidly decrease with depth.
Harris and Beevers [52] has shown that in terms of sur- This result suggests that the total amount of chemi-
face energy the different sizes of Brown grit had no cal modification is small compared to other more pow-
effect but with the Pink and White grits the large grit erful techniques, yet the joint strength is comparable.
sizes generated surfaces with lower polar surface en- The chemical modification was found to be localized
ergy components and these changes correlated with a to the near surface allowing a large number of effec-
reduction in the Na presence in the surface chemical tive bond sites to be created and high bond strengths to
analysis. It was observed [52] that the durability re- be achieved for a low overall functional content [153].
sponse observed with mild steel substrates were evident From the pretreatments examined by Green et al. [153],
from the tensile butt joint tests, which indicated that the new atmospheric plasma (i.e., AgrodynTM ) and the
the surfaces produced by the coarser grit gave higher older fluorination pretreatments have been found to be
durability. This trend would be expected from a consid- the most effective at modifying the surface and cre-
eration of work of adhesion equations (see Section 3.3) ating a number of varying functional types for vari-
(see for example [136, 225]), which suggest that the ous adhesives to bond to. The comparison of the O:C
bonds on high polar energy surfaces would be more ratios may be useful as a guide of the likelihood of a

43
strong joint being created [153]. However, the trends that can be achieved by the use of adhesive bonding is an
within the ratios show that various deep-penetrating important advantage, especially for lightweight struc-
pre-treatments such as low pressure and atmospheric tures. In joining lightweight composites, the adhesive
plasmas induce high levels of deep-positioned oxygen bonding is the most appropriate joining technique.
functional groups. Due to their depth within the sub- Adherend surface pretreatment plays a critical role in
strate they can not be of use in adhesive bonding, lead- developing adhesively bonded joints. For example, un-
ing to the hypothesis that the O:C ratios can not be suitable surface pretreatment of composite substrates,
used as a yardstick to judge the performance of a pre- such as inadequate surface roughening, environmental
treatment under loading conditions due to purely the effects, peel ply chemical contamination, and other fac-
relative increased amount of oxygen of a pretreatment tors (both mechanical and chemical) can prevent adhe-
not increasing bond strength by similar ratio compared sives from bonding properly to composites, resulting in
to other pre-treatment regimes [153]. interfacial failures [89b]. These failures occur at loads
The recently developed and patented Ciba laser pre- well below those of properly bonded joints that fail co-
treatment (CLP) have been studied by Broad et al. hesively. Other failures can occur over time in service,
[151]. As observed by these workers, the CLP process as joints are exposed to harsh environments, including
is a beneficial surface preparation technique for metals, elevated temperature and humidity [84, 88b, 230–237].
which facilitates adhesively bonded joints with better The common misconception in surface preparation
durability. CLP provides also important ecological ad- is that the only requirement for a good adhesive bond
vantages over wet pretreatments [151]. A comparison is a clean surface. A clean surface is a necessary con-
with pickle-etching, considering the required energy, dition for adhesion but it is not a sufficient condition
the pretreatment waste and the emissions, demonstrated for bond durability. Most structural adhesive work as a
the ecological benefits of CLP. In contrast to wet pre- result of the formation of chemical bonds (mainly cova-
treatments, with CLP no bulk pretreatment waste ma- lent, but some ionic and static attractive bonds may also
terials arise. The emmisions observed from pickle- be present) between the adherend surface atoms and
etching are mostly acid containing water vapor. The the compounds constituting the adhesive [152, 203].
emmisions from CLP are mostly lower alcohols, such As pointed out by Davis and Bond [152], these chem-
as ethanol or iso-propanol, and some break-down prod- ical links are the load transfer mechanism between the
ucts. Furthermore, it was claimed [151] that the re- adherends. Solvent degreasing is important, because it
quired energy input for the pretreatment with CLP removes contaminant materials which inhibit the for-
is much lower than that of the energy required for mation of the chemical bonds, and increase wettability
pickle-etching. and surface energy of the substrate. However, solvent
Using the available laser technology, applications of degreasing, whilst providing a clean surface, does not
CLP can range from small repair or small production promote the formation of acceptable surface conditions
applications to high-volume series production. There- for longer term bond durability.
fore, the advantages of the CLP pretreatment process Surface pretreatment should not be termed “clean-
can be summarized as [151]: (a) adhesively bonded ing” nor should the chemicals used be termed “cleaning
joints with good durability, (b) ecological advantages, agents” as these expressions lead to confusion between
(c) short pretreatment times (d) compatible with short the different activities which combine to form the whole
cycle times, (e) easy installation as in-line, (f) stable surface preparation process. Contamination should be
surfaces, (g) selective pretreatment possible, (h) low removed by solvent degreasing as the first step of the
pretreatment costs. surface preparation process. Processing chemicals such
as etchants require contact with the substrate, and their
effectiveness is diminished if the surface is contami-
8. Concluding remarks nated. In many cases, surface-modifying chemicals do
An adhesive is a substance capable of holding ad- not dissolve the contaminants, and an ineffective bond
herends together either by chemical or mechanical at- results if the solvent degreasing step is not performed
traction. The effectiveness of the adhesive bonding adequately. The basic principles of surface pretreatment
is dependent upon many variables including [11], (a) dictate a number of aspects which are often violated by
the polymeric composition of the adhesive, (b) surface process specifications even from reputable manufac-
preparation materials and method of surface pretreat- turers [35, 238]. Davis and Bond [152] considered the
ment, (c) adhesive lay-up procedure, (d) fitting of the following three basic steps for adequate surface pre-
parts to be joined, (d) tooling, and (e) curing process. treatment: (a) the surface must be free of contamina-
Adhesive bonding offers many advantages (i.e., ability tion; removing surface contamination by degreasing,
to join dissimilar materials, higher stiffness, more uni- (b) the adherend surface must be a sufficiently fresh
form load distribution, cleaner lines for aerodynamic and chemically active to enable formation of chem-
benefits, part consolidation, no holes drilled in ad- ical bonds between the adhesive and the adherends,
herends with less stress concentrations, and, generally, typically by chemical etching or surface abrasion, and
less labor cost) over the classical fastening techniques (c) the surface should be chemically modified to pro-
such as welding, riveting and mechanical fastening. The duce an interface resistant to environmental deteriora-
adhesive bonding has a high resistance to fatigue, and as tion in service, especially by hydration. However, most
a consequence the life-cycle maintenance costs are sig- thermosetting polymer matrix composite adherends do
nificantly reduced. The substantial reduction in weight not need the chemical modification process as such a

44
surface is not as susceptible to hydration as metallic sur- mined (see Fig. 25). Fig. 25 shows that the curing state
faces. Whereas the recommended surface preparation can be characterized by three different conversion fac-
for thermoplastic composites is a light aluminum oxide tors (once the plateau-like regime is reached) varying
grit blast in dry nitrogen [152]. The abrasion should just from 74% at 23 to 99% at 64◦ C. The plateau-like regime
remove the surface of the resin without exposing bare was reached after only 4 h at 64◦ C and the curing can be
fibers. This process may be effective on moulded sur- considered as complete. At lower curing temperatures,
faces and corona treated tear film surfaces, and has been the adhesive is not fully cured when the plateau-like
shown to reduce silicone contamination when coated regime is reached. This suggests that in that regime the
nylon tear films are used [160]. Dust should be blown glass transition temperature Tg of the adhesive exceeds
off the prepared surface using a nitrogen gas stream. the curing temperature leading to vitrification.
Because the epoxy surface bonds well to other epoxies, The kinetics of the curing reaction has also been stud-
no chemical modification is required. ied. A plot of the fraction of epoxy remaining with time
Polyolefins are the most frequently surface pretreated showed [208] that the reaction follows a first order law
polymers. The corona treatment is usually used with [239, 240] at all curing temperatures until the transition
polyolefins in film forms, whereas symmetrical mould- to the plateau-like regime (see Fig. 25) is reached. Ar-
ings such as bottles, flame treatment is favored. Corona rhenius plot of the rate coefficient with the inverse of
and flame pretreatments are ineffective with fully fluori- the temperature, the activation energy was determined
nated polymers; with these materials, sodium in liquid to be 34.6 kJ/mol (see Fig. 26). After a temperature
ammonia or sodium naphthalenide in an aprotic sol- dependent curing time, the first order law is no longer
vent are usually used. As pointed out by Brewis and valid.
Dahm [129], for plastics such as nylon-6.6, which pos- There is a strong correlation between the degree
sess suitable functionality, a physical method such as of conversion and the viscosity of the epoxy adhe-
grit blasting is often effective. Grit blasting can be very sive [241]. The viscosity measurement during curing
effective at removing mold release agents such as sili- at 23, 45 and 64◦ C has shown that there was a dramatic
cones which would otherwise have a serious effect on increase of viscosity [208]; this increase was associ-
adhesion, and the roughening will generally enhance ated with gelation. From a molecular point of view,
adhesion. this process is associated with both an increase of the
Before testing the ability of different surface pretreat- molecular weight and an incipient creation of molecules
ments to promote adhesion, a detailed description of the with finite branches until a cross-linked structure is ob-
rheological and mechanical properties of the adhesive tained [208]. Measurements show that the relationship
is needed. Change of adhesive properties (due to in- between the conversion factor β and viscosity η can be
complete curing) can affect the results of the adhesion described by (see Fig. 27)
tests and screen out the effect of the pretreatment. In
 
this context, Lapique and Redford [208] have studied η − ηo
the curing process of a room temperature curing adhe- βαk(T ) ln (22)
ηo
sive paste (Araldite 2014). Fig. 24 shows a differential
scanning calorimetry (DSC) curve of uncured sample
(just after the mixing of the two components). The DSC
curve for the unreacted epoxy shows an exotherm peak
representing 197 J/g. This area is defined as Ao and will
be used as reference. Ao is directly related to the amount
on unreacted reactants. As the curing progresses the am-
plitute of the peak decreases and the area A(t, T ) was
measured. The ratio [1 − (Ao /A(t, T ))] was defined as
the conversion factor β. According to this definition, a
conversion factor of 1 (or 100%) was related to a fully
cured epoxy. Samples were cured at different tempera-
tures (23, 45 and 64◦ C) and the conversion factor deter- Figure 25 Curing of Araldite 2014 at different temperatures [208]. :
•
Curing temperature 23◦ C, : Curing temperature 45◦ C, and : Curing
temperature 64◦ C.

Figure 24 Differential scanning calorimetry (DSC) of scan of an un- Figure 26 Arhenius plot of rate coefficient. The activation energy was
cured sample [208]. determined to be 34.6 kJ/mol [208].

45
 19. R . E . P O L I T I , “Recent Developments in Polyimide and Bis-
maleimide Adhesives,” High Temperature Polymer Matrix Com-
posites, edited by T. T. Serafini (Noyes Data Corp., Park Ridge,
NJ, 1987) p. 123.
20. N . A L L B E E , “Adhesives for Structural Applications” (Advanced
Composites, November–December 1989) p. 42.
21. D . K . K O H L I , Int. J. Adhes. Adhes. 19 (1999) 231.
22. Idem., in Proc. of 10th Int. European Chapter Conference of the
SAMPE 55, 1989, p. 239.
23. Idem., 39th Int. SAMPE Symposium, 11–14 April, 1994.
Figure 27 Conversion factor versus viscosity (log scale) for a sample 24. “Thermoplastics: Materials Engineering,” 2nd ed., edited by
cured at 23◦ C [208]. L. Mascia (Elsevier Applied Science, London, 1989) p. 32.
25. B . G E O R G E , Y . G R O H E N S , F . T O U Y E R A S and
J . V E B R A L , Int. J. Adhes. Adhes. 20 (2000) 245.
26. E . D R A G O N I and P . M A U R I , ibid. 20 (2000) 315.
where ηo is the initial viscosity before the curing reac- 27. I . S K E I S T , “Handbook of Adhesives,” 2nd ed. (Van Nostrand
tion starts. Reinhold, New York, 1997).
Lapique and Redford [208] have also studied the ef- 28. D . B . Y A N G , D . W O L F , T . W A K A M A T S U and
M . H O L M E S , J. Adhes. Sci. Technol. 9 (1995) 1369.
fects of the curing process on the mechanical properties
29. A . M . B R O W S T E I N , US Patent 3,428,614 (1969).
of the adhesive. According to the DSC measurements 30. B . M . M A L O F S K Y , Brit. Patent 1,528,626 (1978).
(see Fig. 25) the curing of the adhesive is completed 31. I . J . B A C C E I , Brit. Patent 1,531,351 (1976).
after 4 h at 64◦ C. The measurement of the mechanical 32. D . S A T E S (ed.), “Handbook of Pressure-Sensitive Adhesives”
properties shows that 28 days at 23◦ C were needed to (Van Nostrand Reinhold, New York, 1989).
reach the same level of cure and that the properties of 33. I . S K E I S T , Adhesive Compositions, in “Encyclopedia of Poly-
mer Science and Technology,” Vol. 1, edited by N. Bikales (Wiley,
the adhesive during that period of time changed dras- New York, 1971) p. 482.
tically. This confirms the assumption that the curing 34. A . Z O S E L , Colloid Polym. Sci. 263 (1985) 541.
process was not stopped once the plateau-like regime 35. C O S T A N T I N O C R E T O N , “Materials Science of Pressure-
was reached in Fig. 24 but only slowed down. Sensitive Adhesives, Materials Science and Technology,” Process-
ing of Materials, Vol. 18, edited by R. W. Cahn, P. Haasen and
E. J. Kramer (Wiley-VCH, Weinheim, Germany, 1997).
36. B . S I L L I O N , Polymides and Other Heteroaromatic Polymers,
References in “Comprehensive Polymer Science,” edited by G. Allen and
1. R . A . F L I N N and P . K . T R O J A N , “Engineering Materials J. C. Bevington (Pergamon, Oxford, UK, 1989) Vol. 5,
and Their Applications,” 4th ed. (John and Wiley Sons, Inc., 1995) p. 449.
p. 608. 37. H . D . S T E N Z E N B E R G E R , Polymer Matrices in “Carbon Fi-
2. P . M O L I T O R , V . B A R R O N and T . Y O U N G , Int. J. Adhes. bres and Their Composites,” edited by E. Fitzer (Springer, Berlin,
Adhes. 21 (2001) 129. Heidelberg, New York, 1985) Chapt. 2.
3. K . K . C H A W L A , “Composite Materials, Science and Technol- 38. “Advanced Composite Materials,” edited by L. A. Pilato and M. J.
ogy,” 2nd ed. (Springer, 1998). Michno (Springer-Verlag, Berlin, Heidelberg, New York, 1994).
4. E . P A U L D E G A R M O , J . T . B L A C K and R . A . K O H S E R , 39. H . R . B R O W N , Annu. Rev. Mater. Sci. 21 (1991) 463.
“Materials and Processes in Manufacturing,” 8th ed. (Prentice Hall) 40. R . R . M Y E R S , Trans. Soc. Rheol. 4 (1960) 43.
Chapt. 38. 41. D . H . K A E L B L E , ibid. 4 (1960) 45.
5. “Modern Manufacturing Process Engineering,” edited by B. W. 42. G . Y A S U D A , K . N I N O M I Y A and S . I D O , Nippon Gomu
Niebel, A. B. Draper and R. A. Wysk (McGraw-Hill, New York, Kyokaishi 36 (1963) 1001.
1989). 43. G . Y A S U D A and S . I D O , ibid. 36 (1963) 1089.
6. W . C . W A K E (ed.), “Synthetic Adhesives and Sealants” (Wiley, 44. “Viscoelastic Properties of Polymers,” edited by J. D. Ferry (John
New York, 1987). Wiley and Sons, New York, 1980).
7. A . H . L A N D R O C K , “Adhesive Technology Handbook” (Noyes, 45. R . P . W O O L , in “Fundamentals of Adhesion,” edited by L. H.
Park Ridge, NJ, 1985). Lee (Plenum, New York, 1991).
8. S . R . H A R T S H O R N E (ed.), “Structural Adhesives: Chemistry 46. G . R . H A M E D , Rubber Chem. Technol. 54 (1981) 576.
and Technology” (Plenum, New York, 1986). 47. A . N . G E N T and G . R . H A M E D , Adhesion, in “Encyclopedia
9. J . S H I E L D S , “Adhesives Handbook,” 3rd ed. (Butterworths, of Polymer Science and Engineering, 2d ed., Vol. 1, edited by
London, 1984). J. Kroschwitz (Wiley, New York, 1985) p. 476.
10. S . L . R O S E N , “Fundamental Principles of Polymeric Materials,” 48. F . J . M A Y E R , Bonding, in “Encyclopedia of Polymer Science
2nd ed. (John Wiley and Sons, 1993). and Engineering,” 2nd ed., Vol. 1, edited by J. Kroschwitz (Wiley,
11. “Fundamentals of Composite Manufacturing: Materials, Methods, New York, 1985) p. 518.
and Applications,” edited by A. B. Strong (Society of Manufac- 49. A . W . A D A M S O N , “Physical Chemistry of Surfaces,” 5th ed.
turing Engineers, Development Reference Publications Division, (Wiley, New York, 1990).
Dearborn, Michigan 48121, 1989). 50. P . C . H I E M E N Z , “Principles of Colloid and Surface Chemistry,”
12. D . K . S H A F F E R , H . M . C L E A R F I E L D and J . S . 2d ed. (Dekker, New York, 1986).
A H E A R A , in “Treatise on Adhesion and Adhesives,” Vol. 7, edited 51. R . F . W E G M A N , “Surface Preparation Techniques for Adhesive
by J. D. Minford, (Marcel Dekker, New York, 1991) p. 437. Bonding” (Noyes, Park Ridge, NJ, 1989).
13. H . M . C L E A R F I E L D , D . K . M cN A M A R A and G . D . 52. A . F . H A R R I S and A . B E E V E R S , Int. J. Adhes. Adhes. 19
D A V I S , “Adhesive Bonding” (Plenum Press, New York, 1990). (1999) 445.
14. J . D . V E N A B L E S , J. Mater. Sci. 19 (1984) 2431. 53. J . H . B R O P H Y , R . M . R O S E and J . W U L F F , “The Struc-
15. J . S . A H E A R A and G . D . D A V I S , in Proceeding Adhesion ture and Properties of Materials,” Vol. II, Thermodynamics of
1987, Third International Conference (Plastics and Rubber Insti- Structure (John Wiley and Sons Inc., New York, 1967).
tute, York, UK, 1987). 54. “An Introduction to Composite Materials,” edited by Derek Hull
16. W . B R O C K M A N , O . D . H E N N E M A N N , H . K O L L E K (University Press, Cambridge, UK, 1981).
and C . M A T Z , Int. Adhes. Adhes. 6 (1986) 115. 55. A . K E L L Y , “Strong Solids,” (Clarendon Press, Oxford, 1973).
17. L . J . H A R T - S M I T H , ibid. 19 (1999) 181. 56. J . C O M Y N , Int. J. Adhes. and Adhes. 12(3) (1992) 145.
18. M E L M . S C H W A R T Z , “Composite Materials Handbook,” 57. A . J . K I N L O C H , “Durability of Structural Adhesives” (Elsevier
2nd ed. (Mc Graw-Hill, 1992). Applied Science, Barking, UK, 1983).

46
 58. T . G . G U T O W S K I , “The Mechanics of Composite Deformation 94. E . H . A N D R E W S and A . J . K I N L O C H , Proc. R. Roc. London
During the Manufacturing Process. First Conference of Composite A 332 (1973) 385.
Materials” (American Society for Composites, Dayton, OH, 1986). 95. Idem., ibid. A 332 (1973) 401.
59. P . M O L I T O R and T . Y O U N G , Int. J. Adhes. Adhes. 22 (2002) 96. M . T O Y A M A , T . I T O and H . M O R I G U C H I , J. Appl. Polym.
101. 14 (1970) 2039.
60. P . G . D E G E N N E S , Rev. Mod. Phys. 57 (1984) 827. 97. M . S H E R R I F F , R . W . K N I B B S and P . G . L A N G L E Y ,
61. L . L E G E R and J . F . J O A N N Y , Rep. Prog. Phys. 55 (1992) J. Appl. Polym. Sci. 17 (1973) 3423.
431. 98. K . N . G . F U L L E R and D . T A B O R , Proc. R. Soc. London A
62. N . E U S T A T H O P O L O U S , Acta Mater. 46 (1998) 2319. 345 (1975) 327.
63. S . M A S O N , in “Wetting Spreading and Adhesion,” edited by 99. S . J . H I T C H O C K , N . T . C A R O L L and M . G .
J. F. Panday (Academic, New York, 1978). N I C H O L A S . J. Mater. Sci. 16 (1981) 714.
64. G . P A L A S A N T Z A S and J . T H . M . D E H O S S O N , Acta 100. Y . T A M A I and K . A R A T A N I C , J. Phys. Chem. 76(22) (1972)
Mater. 49 (2001) 3533. 3267.
65. G . W . C R I T C H L O W and D . M . B R E W I S , Int. J. Adhes. 101. C . H U H and S . G . M A S O N , J. Coll. Interf. Sci. 60(1) (1977)
Adhes. 15(3) (1995) 173. 11.
66. Y . G I L I B E R T and G . V E R C H E R Y , in “Influence of Sur- 102. F . G . Y O S T , J . R . M I C H E A L and E . T . E I S E N M A N N ,
face Roughness on Mechanical Properties of Joints,” edited by Acta. Metall. Mater. 43(1) (1995) 299.
K. L. Mittal (Adhesive Joints Formation, Characteristics, and Test- 103. J . J . B I K E R M A N , J. Phys. Coll. Chem. 54 (1950) 653.
ing, Plenum Press, New York, 1982). 104. R . N . W E N Z E L , Ind. Eng. Chem. 28(8) (1936) 988.
67. C . W . J E N N I N G S , Amer. Chem. Soc. Div. Org. Chem. 31(2) 105. A . C A R R E and J . S C H U L T Z , J. Adhes. 15 (1983) 151.
(1971) 184. 106. R . S H U T T L E W O R T H and G . L . J . B A I L E Y , Disc. Faraday
68. J . P . S A R G E N T , Int. J. Adhes. Adhes. 14(1) (1994) 21. Soc. 3(16) (1948) 16.
69. T . R . K A T O N A , ibid. 3(2) (1983) 85. 107. A . F A L S A F I , M . T I R R E L and A . V . P O C I U S , Langmuir
70. A . F . H A R R I S and A . B E E V E R S , in Conf. Proc. Structural 16 (2000) 1816.
Adhesives in Engineering V, Bristol, Institute of Materials, 1998. 108. D . L. WOERDEMAN, N. AMOUROUX, N.
71. J . M . S Y K E S , Surface Treatments for Steels, in “Surface Anal- P O N S I N E N T , G . J A N D E A U , H . H E R V E T and L .
ysis and Pretreatment of Plastics and Metals,” edited by D. M. L E G E R , Composite A: Appl. Sci. Manuf. 30(1) (1999) 95.
Brewis (London, Applied Science Publishers, 1982) p. 153. 109. K . R . S H U L L , D . A H N and C . L . M O W E R Y , Langmuir
72. D . E . P A C K H A M , Roughness of Surfaces, in “Handbook of 13 (1997) 1799.
Adhesion,” edited by D. E. Packham (Longman Group (FE) Ltd., 110. M . D E R U E L L E , H . H E R V E T , G . J A N D E A U and
New York, 1992). L . L E G E R , J. Adhes. Sci. Technol. 12 (1998) 225.
73. M . S H A H I D and S . A . H A S H I M , Int. J. Adhes. Adhes. 22 111. H . R . B R O W N , Macromolecules 26 (1993) 1666.
(2002) 235. 112. D . L . W O E R D E M A N , J . A . E M E R S O N and R . K .
74. M . J . D A V I S , A Call for Minimum Standards in Design and Ap- G U I N T A , Int. J. Adhes. Adhes. 22 (2002) 257.
plication Technology for Bonded Structural Repairs. Proc. Symp. 113. M . K . C H A U D H U R Y and G . M . W H I T E S I D E S , Langmuir
on Composite Repair of Aircraft Structures, Vancouver, 1995, 7 (1991) 1013.
p. 4-1, 4-15. 114. D . A H N and K . R . S H U L L , Macromolecules 29 (1996) 4381.
75. J . G U I L L E M E N T , B . B I S T A C and J . S H U L T Z , Int. J. Ad- 115. K . L . J O H N S O N , K . K E N D A L L and A . D . R O B E R T S ,
hes. Adhes. 22 (2002) 1. Proc. Roy. Soc. London Part A 324 (1971) 301.
76. A . N . G E N T , J. Schultz, J. Adhes. 3 (1972) 281. 116. H . H E R T Z , Gesamelle Werke, Leipzig, 1895.
77. D . M A U G I S and B . B A R Q U I N S , J. Phys. D 11 (1978) 1989. 117. P . C H I N , R . L . M C C U L L O U G H and W . L . W U , J. Adhes.
78. H . C H U N and A . N . G E N T , J. Polym. Sci. Part B, 34 (1996) 64 (1997) 145.
2223. 118. J A N G - K Y O K I M and Y U - W I N G M A I , Interfaces in Compos-
79. A . D . R O B E R T S , Rubber Chem. Technol. 52 (1979) 23. ites, Vol. 13, “Structure and Properties of Composites (Materials
80. P . G . G E N N E S , Langmuir 12 (1996) 4497. Science and Technology),” edited by R. W. Cahn, P. Haasen and E.
81. A . A H A G O N and A . N . G E N T , J. Polym. Sci. Phys. Ed. 13 J. Kramer, VCH (Weinheim, New York, Cambridge, 1993) p. 239.
(1975) 1285. 119. A . M A H O O N , edited by A. J. Kinloch (Titanium Adherends in
82. D . M A U G I S , J. Adhes. Sci. Technol. 9 (1995) 1005. Durability of Structural Adhesives, 1983) p. 259.
83. A . N . G E N T and G . R . H A M E D , Polym. Eng. Sci. 17(7) (1977) 120. E . P . P L U E D D E M A N N , in Proc. Int. Conf. Composite Inter-
462. faces II, Interfaces in Polymer, Ceramic and Metal Matrix Com-
84. J . A . M A R C E A U , Y . M O J I and J . C . M cM I L L A N , “A posites, edited by H. Ishida (Elsevier Sci. Publ., New York) p. 17.
Wedge Test for Evaluating Adhesive Bonded Durability,” 21st Na- 121. H . I S H I D A and J . L . K O E N I G , J. Coll. Interf. Sci. 64 (1978)
tional SAMPE Symposium, Los Angeles, CA, 1976, p. 332. 555.
85. M . L . W I L L I A M S , J. Appl. Polym. Sci. 14 (1970) 1121. 122. E . P . P L U E D D E M A N N and G . L . S T A R K , 35th Ann.
86. D . M A U G I S , in “Contact, Adhesion and Rupture of Elastic Tech. Conf. Reinf. Plast./Comp. (Society of Plastic Industries,
Solids” (Springer Verlag, Berlin, 2000). Washington, D.C., 1980) p. 20B.
87. D . W . A U B R E Y , G . N . W E L D I N G and T . W O N G , J. Appl. 123. A . G A R T O N and J . H . D A L Y , Polym. Comp. 6 (1985) 195.
Polymer Sci. 13 (1969) 2193. 124. J . L . T H O M A S O N , ibid. 11 (1990) 105.
88a. M . A . K R E N C E S K I and J . F . J O H N S O N , Polym. Eng. Sci. 125. H . C . T S A I , A . M . A R O C H A and L . W . G A U S E , Mater.
29 (1989) 36. Sci. Eng. A 126 (1990) 295.
88b. W . D . B A S C O M and R . L . C O T T I N G T O N , J. Adhes. 20 126. J . W . W I L L I A M S , M . E . D O N N E L L A N , M . R . J A M E S
(1976) 333. and W . L . M O R R I S , ibid. A 126 (1990) 305.
89a. A . N . G E N T and R . P . P E T R I C H , Proc. Roy. Soc. London A 127. W . J . W H A T L E Y and F . E . W A W N E R , J. Mater. Sci. Lett.
310 (1969) 433. 4 (1985) 173.
89b. L . J . H A R T - S M I T H , “The Curse of the Nylon Ply,” Presented 128. P . L E N N O N , E . E S P U C H E , H . S A U T E R E A U and
at 41st International SAMPE Symposium, 1996, p. 303. D . S A G E , Int. J. Adhes. Adhes. 19 (1999) 273.
90. H . S C H O N H O R N , J. Polym. Sci., Polym. Lett. 2 (1964) 465. 129. D . M . B R E W I S and R . H . D A H M , ibid. 21 (2001) 397.
91. Idem., Macromolecules 1 (1968) 145. 130. B L A C K M A N , K I N L O C K and W A T T S , Composites 25(5)
92. H . S C H O N H O R N and F . W . R Y A N , J. Polym. Sci. A-2 6 (1994).
(1968) 231. 131. J . W I N G F I E L D , Int. J. Adhes. Adhes. 13(3) (1993) 151.
93. H . I S H I D A and P . B U S S I , Morphology Control in Polymer 132. G . K E M P E and H . K R A U S S , Adhesion Bonding Techniques
Composites, in “Structure and Properties of Composites (Mate- for Highly Loaded Parts of Continuous Carbon-Fiber Reinforced
rials Science and Technology),” Vol. 13, edited by R. W. Cahn, Polyetherketone (CF-PEEK/APC02). ICAS-92-6.9.1, 1992.
P. Haasen and E. J. Kramer, VCH (Weinheim, New York, 1993) 133. J . C O M Y N and X I A O M A S C I A , Int. J. Adhes. Adhes. 16 (1996)
p. 339. 301.

47
134. G . W . C R I T C H L O W and D . M . B R E W I S , ibid. 15(3) (1995) 165. Pasajell Data Sheets SEMCO Divison Products Research and
161. Chemical Corporation 5454 San Francisco Road Glendale CA,
135. G . K . A . K O D O K I A N and A . J . K I N L O C H , J. Mater. Sci. 1997.
Lett. 7 (1988) 627. 166. J . M . H I L L , E . K A R B A S H E W S K I and A . L I N , J. Adhesion
136. A . J . K I N L O C H , “Adhesion and Adhesives” (Chapman and Sci. Technol. 9(12) (1995) 1575.
Hall, London, 1987). 167. M . S T R O B E L , M . J . W A L Z A K and J . M . H I L L , ibid. 9(3)
137. A . B A A L M A N N , K . D . V I S S I N G and A . G R O S S , J. Ad- (1995) 365.
hesi. 46 (1994). 168. K . L . M I T T A L and A . P I Z Z I (eds.), “Adhesion Promotion
138. J . R . A R N O L D , C . D . S A N D E R S , D . L . B E L L E V O U , Techniques; Technical Applications” (Marcel Dekker, New York,
A . A . M A R T I N E L L I and G . B . G A S K I N , in Proceeding of 1999).
the 29th International SAMPE Technical Conference, Vol. 1, 1997, 169. I . S U T H E R L A N D , R . P . P O P A T , D . M . B R E W I S and
p. 345. R . C A L D E R , J. Adhes. 46 (1994) 79.
139. J . R . A R N O L D , C . D . S A N D E R S , D . I . B E L L E V O U , 170. H . K . Y A S U D A , “Plasma Polymerisation” (Academic Press,
A . A . M A R T I N E L L I and G . B . G A S K I N , in Proceedings of USA, 1985).
the 29th International SAMPE Technical Conference, Vol. 1, 1997, 171. V . B A R R O N and M . B U G G Y , in Proceeding of the 23rd Annual
p. 345. Adhesion Society Conference Myrtle Beach SC, 2000, p. 540.
140. K . Y . B L O H O W I A K , J . H . O S B O R N E , K . A . K R I E N K E 172. B . O . A R O N S S O N , J . L A U S M A N and B . K A S E M O ,
and D . F . S E K I T S , in Proceedings of the 28th International J. Biomed. Mater. Res. 35(1) (1997) 49.
SAMPE Technical Conference, 1996, p. 440. 173. R . F O E R C H , J . I Z A W A , N . S . M C I N H Y R E and
141. F . J . B O E R I O and R . G . D I L L I N G H A M , in “Adhesive Joints D . H . H U N T E R , J. Appl. Polym. Sci. 46 (1990) 415.
Formation, Characteristics, and Testing,” edited by K. I. Mittal 174. J . P. BRADEY, E. ESPUCHE, D. SAGE,
(Plenum Press, New York, 1991) p. 541. B . C H A B E R T , Y . J U G N E T and C . B A T I E R , Trans.
142. British Aerospace Aircraft Group Weybridge-Bristol Division, Minh. Duc. Polymer 37(8) (1996) 1377.
1984, Report No. AL/MAT/3864. 175. L . L A V I E L L E , J . S H U L T Z and J . N A K A J I M A , J. Appl.
143. G . D . D A V I S , G . B . G R O F F and R . A . Z A T O R S K I , Surf. Poly. Sci. 42 (1991) 28.
Inter. Anal. 25 (1997) 366. 176. R . F O E R C H , J . I Z A W A and G . S P E A R S , J. Adhes. Sci.
144. Lockheed-California Company, Prepared for National Air Devel- Techn. 5(7) (1991) 549.
opment Center, 1983, Report No. NADC-84124-60. 177. G . S . S H E U and S . S . S H Y U , ibid. 8(5) (1994) 1027.
145. A . M A H O O N , in “Durability of Structural Adhesives,” edited by 178. M . S H O B E L , M . J . W A L Z A K , J . H I L L , A . L I N , E .
A. J. Kinloch (Applied Science Publishers, London, 1983) p. 255. K A B A S H E W S K I and C . L Y O N S , ibid. 9(3) (1995) 365.
146. E . S A N C A K T A R and E . Z H A N G , ASME 79 (1994) 65. 179. R . H . H A N S E N , J . V . P A S C A L E , T . D E B E N E D I C H S
147. K . R A M A N I , W . J . W E I D N E R and G . K U M A R I , Int. J. and P . M . R E N T Z E P I S , J. Polym. Sci. Part I, 3 (1993) 991.
Adhes. Adhes. 18(6) (1998) 401. 180. F . M . F O W K E S , Ind. Engng. Chem. 12 (1964) 40.
148. J . L . C O T T E R and A . M A H O O N , ibid. 2 (1982) 47. 181. Methods and Apparatus for Treating the Surface of a Workpiece
149. C . I N G M A N and K . R A M A N I , ibid. 17 (1997) 39. by Plasma Discharge, US Patent 5837958 (1998).
150. J . R . A R N O L D , et al. “A Study of Titanium Surface Pretreat- 182. M . W . P O T T S , in Proceedings of the Polymer, Laminations and
ments for Bonding with Polyamide and Epoxy Adhesives,” in Pro- Coatings Conference (1993) p. 444.
ceedings of the 29th International SAMPE Technical Conference, 183. J . D I G I A C O M O , J. Proceedings Finishing’91 Conference Cin-
1997. cinatti, Ohio, 1991, p. 376.
151. R . B R O A D , J . F R E N C H and J . S A U E R , Int. J. Adhes. Adhes. 184. F . D E M B L O N , Vide Couches Minces 272(Suppl 5) (1994) 394.
19 (1999) 193. 185. D . B R I G G S , J. Mater. Sci. 14 (1979).
152. M . D A V I S and D . B O N D , ibid. 19 (1999) 91. 186. I . S U T H E R L A N D , D . M . B R E W I S and R . J . S H E N G ,
153. M . D . G R E E N , F . J . G U I L D and R . D . A D A M S , ibid. 22 J. Surface Interface Anal. 17 (1991) 507.
(2002) 81. 187. E . W U R Z B E R G , A . B U C H M A N , E . Z Y L E R S T E I N ,
154. D . K . S H A F F E R , H . M . C L E A R F I E L D and J . S . Y . H O L D E N G R A B E R and H . D O D I U K , J. Adhes. Adhes.
M I N F O R D (eds.), “Treatise on Adhesion and Adhesives” (Marcel 10 (1987) 337.
Dekker, New York, 1991) Vol. 7, p. 437. 188. A . B U C H M A N , H . D O D I U K , M . R O T A L and
155. V . B A R R O N , PhD thesis, University of Limerick, 1998. J . Z A H A V I , Int. J. Adhes. Adhes. 11 (1991) 3.
156. J . R . A R N O L D , C . D . S A N D E R S , D . L . B E L L E V O U , 189. J . K . P A R K and K . M U K H E R J E E , Mater. Manufactr. Process
A . A . M A R T I N E L L I and G . B . G A S K I N , Proces. of the 29th 13(3) (1998) 359.
Int. SAMPE Technical Conference, Vol. 1, 1997, p. 345 190. A . B U C H M A N and H . D O D I U K - K E N I G , in “Laser Surface
157. A . J . K I N L O C H , “Adhesion and Adhesives” (Chapman and Treatment to Improve Adhesion,” edited by K. L. Mittal, Adhesion
Hall, London, 1987) p. 123. Promotion Techniques, Technical Applications, Marcel Dekker,
158. D . A R N O T T , P . P E A R C E , A . W I L S O N , R . C H E S T E R , New York.
A . C A M I L L E R I and J . V A N D E R B E R G , The Effect on 191. H . F R E R I C H S , J . S T R I C K E R , D . A . W E S N E R and
Mechanical Properties of Void Formation During Vacuum Bag E . W . K R E U T Z , Appl. Surface Science 80 (1995) 405.
Processing of Epoxy Film Adhesives, in Proc. Int. Aerospace 192. A . J . K I N L O C H , The Service Performance of Structural Adhe-
Congress. PICAST 2-AAC 6, Melbourne, 1995, p. 811. sive Joints. Haftung als Basis für Stoffverbunde Oberursel 1982.
159. A . W I L S O N , M . K I N D E R M A N N and D . A R N O T T , 193. D . M . B R E W I S , J. Adhes. 37 (1992) 97.
Void Development in an Epoxy Film Adhesive During Vacuum 194. J . P . C O L E M A N and D . P L E T C H E R , J. Electroanal Chem.
Bag Cure. Proc. Int. Aerospace Congress, PICAST 2-AAC 6, 87 (1978) 111.
Melbourne, 1995, p. 625. 195. J . B R I N G M A N N , K . E B E R T , U . G A L L A and H .
160. L . J . H A R T - S M I T H , G . R E D M O N D and M . J . D A V I S , S C H I E D E R , J. Appl. Electrochem. 25 (1995) 846.
The Curse of the Nylon Peel Ply. 41st Int. SAMPE Symp. and 196. I . M A T H I E S O N , D . M . B R E W I S , I . S U T H E R L A N D and
Exhib. Brea, CA, Anaheim, 1996. R . A . C A Y L E S S , J. Adhes. 46 (1994) 49.
161. J . K . P A R K , J . W . C H E N and K . M U K H E R J E E , in Pro- 197. D . M . B R E W I S , I . M A T H I E S O N , I . S U T H E R L A N D R .
ceedings of the American Society for Composites 14th Technical A . C A Y L E S S and R . H . D A H M , Int. J. Adhes. Adhes. 16
Conference (Dayton, OH, 1999) p. 661. (1996) 87.
162. P . D . M I L L E R , R . A . J E F F R E Y S and H . A . P R A Y , Metal 198. D . M . B R E W I S , R . H . D A H M and I . M A T H I E S O N ,
Progress Report, 1974, p. 61. J. Adhes. 72 (2000) 373.
163. J . A . F I L B E Y and J . P . W I G H T M A N , in “Adhesion,” edited 199. R . T O M A T and A . R I G O , J. Appl. Electrochem. 16 (1986) 8.
by K. W. Allen, 12 ed. (Applied Science Publishers, London, 1987) 200. D . M . B R E W I S , R . H . D A H M and I . M A T H I E S O N ,
p. 17. Polym. J. Mater. Sci. Lett. 16 (1997) 93.
164. W . B R O C K M A N N , O . D . H E N N E M A N , H . K O L L E K 201. W . A . L E E S , in Proc. of the Bonding and Repair of Composites
and C . M A T Z , Int. J. Adhes. Adhes. 6 (1986) 115. Seminar, Birmingham, UK, 1989, p. 17.

48
202. F . J . B O E R I O , C . A . G O S S E L I N , J . W . W I L L I A M S , of the Impact of Cleaning on the Adhesive Bond and the Process
R . G . D I L L I N G H A M and J . M . B U R K S R A N D , in Proc. Implications, SAND 2000-1042, May 2000.
of the Symp. on Polymer Composites, Seattle, Washington, USA, 224. M . E . R . S H A N A H A N , J. Coll. Interf. Sci. 229 (2000) 168.
1983, p. 171. 225. J . C O M Y N , “Adhesion Science” (The Royal Society of Chem-
203. A . J . K I N L O C H , “Adhesion and Adhesives, Science and Tech- istry, RSC Paperbacks, 1997) p. 98.
nology” (Chapman and Hall, London, 1987) p. 30. 226. C . L . M A H O N E Y , Fundamental Factors Influencing the
204. L . E . R A N T Z , Proper Surface Preparation, Bonding Critical Performance of Structural Adhesives. Internal Report, Dex-
First Adhesive Age, May 1987. ter Adhesives and Structural Materials Division, The Dexter
205. W . W A D E , R . M A M M O N E and M . B I N D E R , J. Appl. Polym. Corporation.
Sci. 43 (1991) 1589. 227. R . J . S C H L I E K M A N N , Adhesive Bonding and Composites,
206. P . R U M E A U , Thesis, no. 81-95, Lyon I, 1995. 4th Int. Conf. on Composite Materials, ICCM-IV, Tokyo, Japan,
207. M . R . W E R T H E I N E R and H . P . S C H R E I D E R , J. Appl. 1982, p. 63.
Polym. Sci. 26 (1981) 2087. 228. L . W . C R A N E , et al., Surface Treatment of Cured Epoxy
208. F . L A P I Q U E and K . R E D F O R D , Int. J. Adhes. Adhes. 22 Graphite Composites to Improve Adhesive Bonding, SAMPE J.
(2002) 337. 1976, Vol. 6.
209. J . P . S A R G E N T , ibid. 14(1) (1994) 21. 229. A . B E E V E R S , A Critical Appraisal of Durability Test Methods,
210. S . T H E R Y , A . L E G R O S and P . B A L L A D O N , Study of SAE IV, Bristol, 1995, p. 137.
Parametrs Influencing the Mechanical Behavior of and Damage 230. J . C O G N A R D , J. Adhes. 22(2) (1987) 97.
to Steel-Polymer Interfaces, in “Mechanics and Mechanisms of 231. Idem., ibid. 20(1) (1986) 1.
Damage in Composites and Multi-Materials, ESIS 11,” edited by 232. P . B . C R O S L E Y and E . J . R I P L I N G , J. Test. Evalu. 19(1)
D. Baptiste (Mechanical Engineering Publications, London, 1991) (1991) 24.
p. 339. 233. E . J . R I P L I N G , S . M O S T O V O Y and C . B E R S C H , J. Adhes.
211. S . J . H I T C H C O C K , N . T . C A R O L L and M . G . 3 (1971) 145.
N I C H O L A S , J. Mater. Sci. 16 (1981) 714. 234. F . S L O A N , J. Comp. Mater. 27(16) (1993) 1606.
212. C . B O C K E N H E I M E R , B . V A L E S K E and W . P O S S A R T , 235. W . J O H N S O N and L . M . B U T K U S , Fatigue and Fract. Engin.
Int. J. Adhes. Adhes. 22 (2002) 349. Mater. and Struct. 21(4) (1988) 465.
213. L . K O Z M A and I . O L E F J O R D , Mater. Sci. Technol. 3 (1987) 236. L. J. H A R T - S M I T H , “A Peel-Type Durability Test
860. Coupon to Assess Interfaces in Bonded, Co-Bonded, and
214. O . L U N D E R , B . O L S E N and K . N I S A N C I O G L U , Int. J. Ad- Co-Cured Composite Structures,” McDonnell Douglas Paper
hes. Adhes. 22 (2002) 143. MDC 97K0042, presented to MIL-HDBK-17 Meeting, Tucson,
215. A . J . K I N L O C H and N . R . S M A R T , J. Adhes. 12 (1981) 1997.
23. 237. R . A . J U R F , “Environmental Effects on Fracture of Adhesively
216. A . J . K I N L O C H , H . E . B I S H O P and N . R . S M A R T , ibid. Bonded Joints,” Adhesively Bonded Joints: Testing, Analysis, and
14 (1982) 105. Design, ASTM STP 981, p. 276.
217. M . J . R I J K H O F F , R . B L E E K E R and J . B O T T E M A , in 238. “Starch and Dextrins in Prepared Adhesives,” Chapter 20, edited by
Proc. of the Int. Symp. on Aluminum Surface Science and Tech- H. M. Kennedy and A. C. Fischer, p. 607; “Starch: Chemistry and
nology, Antwerpen, 1997, p. 125. Technology,” 2nd ed., edited by R. L. Whistler, J. N. Bemiller and
218. K . S . S C H W E I Z E R and J . G . C U R R O , Phys. Rev. Lett. 58 E. F. Paschall (Academic Press, Orlando, San Diego, New York,
(1987) 246. 1984).
219. J . G . C U R R O and K . S . S C H E W E I Z E R , Macromolecules 239. J . W . S I N C L A I R , J. Adhes. 38 (1992) 219.
20 (1987) 1428. 240. P . W . A T K I N S , “Physical Chemistry,” 2nd ed. (Oxford
220. Idem., J. Chem. Phys. 87 (1987) 1842. University Press, 1982) (ISDN 0-19-855151-7).
221. K . S . S C H W E I Z E R and J . G . C U R R O , Adv. Chem. Phys. 98 241. J . J . I M A Z , N . M A R K A I D E and M . J . J U R A D O , Eur.
(1997) 1. Polym. J. 28 (1992) 299.
222. B . K . F I N K and R . L . M cC U L L O U G H , Composites A: Appl.
Sci. Manuf. 30(1) (1999) 1.
223. J . A . E M E R S O N , T . R . G U E S S , J . G . C U R R O , E . Received 27 February
D . R E E D Y , E . P . L O P E Z and P . L E M K E , Investigation and accepted 12 August 2003

49