Surface & Coatings Technology 236 (2013) 212223

Contents lists available at ScienceDirect

Surface & Coatings Technology

journal homepage: www.elsevier.com/locate/surfcoat

Application and drying at ambient temperature of thick

organicinorganic hybrid coatings on glass
M. Barletta a,, M. Puopolo b, A. Gisario b, S. Vesco a
a
b

Dipartimento di Ingegneria dell'Impresa, Universit degli Studi di Roma Tor Vergata, Via del Politecnico, 1-00133 Roma, Italy
Dipartimento di Ingegneria Meccanica ed Aerospaziale, Sapienza Universit degli Studi di Roma, Via Eudossiana, 18-00184 Roma, Italy

a r t i c l e

i n f o

Article history:
Received 3 August 2013
Accepted in revised form 23 September 2013
Available online 2 October 2013
Keywords:
Adhesion
Scratch resistance
Damage mechanisms
Siliconeepoxy
Coatings

a b s t r a c t
Organicinorganic hybrid coatings have achieved success in scientic environments because they can be
designed using the solgel route to combine the high hardness and chemical stability typical of glass-like or
ceramic materials (i.e., the inorganic side) with the toughness and ductility typical of organic materials
(i.e., the organic side). Nevertheless, organicinorganic hybrid coatings are often very brittle, and they can
collapse if applied as thick lm on rigid substrates because of the shrinkage during the drying/curing process.
However, the manufacturing of a thick coating is compulsory, when stringent requirements for scratch performance and wear endurance must be met. In this respect, the present investigation proposes the design of selfdrying siliconeepoxy resins and the manufacturing of the corresponding thick coatings (approximately
~120 m thick) on as-received and micro-corrugated glass using an automatic drawdown applicator. The scratch
performance of the coatings was tested using progressive- and constant-mode scratch tests, and the wear
resistance was examined using dry sliding linear reciprocating tribological tests. The experimental ndings
demonstrate how the role of the interface is crucial and how micro-corrugation is extremely benecial in increasing the threshold of the maximum thickness beyond that at which coating bulging and delamination occur.
2013 Elsevier B.V. All rights reserved.

1. Introduction
Organicinorganic hybrid coatings have achieved success in scientic
environments because they can be designed via solgel route to
combine the high hardness and chemical stability typical of glass-like
or ceramic materials (i.e., the inorganic side) with the toughness and
ductility typical of organic materials (i.e., the organic side) [1,2]. These
coatings could be protably used in several domains where substrate
protection and/or functionalization by a tailor-made coating material
are pursued [3]. In addition, the potential of these coatings is greatly
emphasized because they offer good adhesion to various substrates [3].
Furthermore, some of these coatings can dry/cure at relatively moderate
temperature (i.e., ~70250 C [4]), making them suitable as topcoats for
a large share of substrates, including some thermolable substrates. Nevertheless, organicinorganic hybrid coatings are often very brittle [57]
and they can collapse if applied as thick lm on rigid substrates due to
shrinkage during the drying/curing process [8,9]. Malzbender and de
With observed that the shrinkage phenomenon related to the drying/
curing process of organicinorganic hybrid coatings can often induce a
large residual stress eld inside the coating material, which can lead to
an early failure of the coating if applied with a thickness over intrinsic

Corresponding author. Tel.: +39 0672597195; fax: +39 062021351.

E-mail address: barletta@ing.uniroma2.it (M. Barletta).
0257-8972/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.surfcoat.2013.09.049

critical values [9]. In particular, when the organicinorganic hybrid coatings are deposited on non-compliant substrates such as glass substrates,
the mismatch at the interface between the shrunk coatings and underlying substrates is strongly emphasized and this mismatch can increase
the possibility of a spontaneous collapse of the coating system, which
can bulge or even delaminate from the substrate even without the application of an external load.
Thus, preventing the bulging and delamination phenomena of
organicinorganic hybrid coatings synthesized via the solgel route is
of utmost interest as it would greatly extend applications of this class
of materials. In this respect, many efforts have been made by scientists
and practitioners to widen the range of coatings that could be designed
and manufactured via the solgel route. However, the proposed
solutions converged toward the implementation of very thin coatings
(i.e., typically a few microns) to reduce the onset of the detrimental
stress eld by material shrinking and toward the adhesion being mainly
improved through the combination of two or more organicinorganic
molecules, some with only mere grafting functionality. Such thin coatings have been used in industrial applications such as for optical, decorative or architectural materials. For example, the large-scale dip coating
process to prepare reective and antireective layers based on Pd
containing TiO2 or SiO2/TiO2 systems has become a well-established
technology (Irox, Amiran, Calorex, Schott Co.) [10]. Similarly, antireective coatings suitable as cover sheets for photovoltaic (PV) cells
and collectors have been commercially available from nearly a decade,
although their preparation involves a relatively high temperature

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

(600 C) ring process [11,12]. Coatings for decorative and architectural

purposes were industrialized by including pigments in the solgel
design with a range of different features [13,14]. In contrast, the
manufacturing of a thin coating excludes any possibility of developing
surface overlying materials that could meet the industrial standard
for anti-scratch or wear resistant barriers. This fact was conrmed by
several experimental results reported in the literature [15]. Han et al.
attempted to verify the scratch performance of a solgel deposited organicinorganic hybrid coating material based on TEOS, 3-(trimethoxysilyl)
propylmethacrylate and urethane acrylate deposited on a oat glass by
ow coating with a thickness of approximately 3.5 m. The ultimate
critical load was nearly 2 N beyond which the coatings were found to
delaminate [15].
Despite the signicant efforts made to produce versatile and industrially sustainable organicinorganic hybrid coating materials, there
are only a handful of practical applications in which these materials
are effectively in service. Thus, the present investigation proposes an
alternative approach to controlling the shrinkage during post-drying
at ambient temperature of thick organicinorganic hybrid coatings
on already in service glass. This approach would be extremely benecial as it would increase the threshold of the maximum thickness
beyond which coating bulging and delamination would occur, thus
expanding the applicability of organicinorganic hybrid coatings in
industrial domains in which their combined high micro-mechanical
performance and wear endurance cannot be ignored. The alternative
approach consists of the development of a highly transparent organicinorganic hybrid material, whose structure is designed on a silicone
skeleton with lateral chains consisting of epoxy groups that are crosslinkable with amine-based organofunctional silane hardeners. After
cross-linking, the hardener confers both high mechanical and tribological resistance to the coating material. Proper formulations of silicone
epoxy resins were designed and applied with a high thickness
(approximately 120 m) using an automatic drawdown applicator on
as-received and sandblasted glass substrates. The process was followed by spontaneous drying to ensure the consolidation of the coatings.
Sandblasting was thought to generate a high specic contact area at
the coatingsubstrate interface through the onset of widespread
micro-asperities on the glass surface. Such micro-asperities were
believed to act as obstacles to the stress eld generated by the strong
shrinkage phenomenon during the drying/cross-linking of the thick
overlaying layer of the siliconeepoxy resin, thus avoiding or delaying
the coating failure by bulging or delamination. Characterization of the
morphology, visual appearance, micro-mechanical and tribological
performance of the coatings was thus performed. The experimental
ndings revealed that the siliconeepoxy coatings deposited on the
sandblasted glass substrate were extremely suitable for protecting
the overlying layer from early failure due to scratches or wear
phenomena. Accordingly, the coating demonstrated high potential to

213

preserve the initial efciency of the underlying glass substrate over a

long period.
2. Experimental
2.1. Materials
Flat substrates, 3.8 mm thick, 25 100 mm2 wide and cut from a
2 m2 tempered glass sheet, were coated with a high-solid two-pack
siliconeepoxy resin (SILIKOPON EF, Evonik, Essen, Germany) dilutable
(1:1) by esters, ketones and glycol ethers and cross-linked with bis
(trimethoxysilylpropyl) amine (amine H-equivalent, 335 g/mol)
supplied by Evonik (Ameo, Essen, Germany). Thickeners (Tego
Viscoplus 3030, Evonik, Essen, Germany) and ow promoters (Tego
Flow 370, Evonik, Essen, Germany) were added to the formulation to
obtain a homogenous coating thickness through strict control of the
mixture rheology.
2.2. Manufacturing process
As received glass substrates were degreased by sonication in 95%
EtOH for 30 min. Half of the specimens were sandblasted (6 bar, ne
glass beads, sandblasting equipment, 0580, Fervi Srl, Modena, Italy) to
coarsen the starting morphology and favor coating adhesion. To remove
all glass residuals from the previous process, the substrates were rinsed
thoroughly by sonication in demineralized water. The siliconeepoxy
formulation was then deposited after dilution (1:1) in acetone on both
the as-received and sandblasted glass substrates using an automatic
drawdown bar applicator (Automatic Film Applicator L, BYK-Gardner,
Geretsried, Germany) equipped with a doctor blade. The resulting
coatings were left to cure at ambient temperature for 24 h. After curing,
the coating thicknesses were measured using a digital Palmer caliber
(293-816, Mitutoyo, Kawasaki, Japan) with a resolution of 0.001 m.
All coatings with thicknesses falling outside the prescribed range of
120 20 m were discarded.
2.3. Characterization of the coatings
The morphology of the coatings was recorded using a contact inductive gauge of a CLI proler (TalySurf CLI 2000, Taylor Hobson, Leicester,
UK). An area of 4 4 mm2 was analyzed, and 2000 proles were stored
with a resolution of 1 m. The data were elaborated using TalyMap 3.1
software (Taylor Hobson, Leicester, UK) to extrapolate the roughness
parameters and corresponding 3D maps. The scratch resistance of the
hybrid coating was evaluated by constant-load scratch tests (MicroCombi, CSM Instruments, Peseaux, Switzerland) using two different
Rounded Conical Rockwell C diamond indenters with tip radii of 800
and 200 m and an applied load of 20 N along the 2 mm scratch

Fig. 1. 3D maps of the as-received and coated glasses: (a) as-received glass, (b) sandblasted glass, (c) as-received coated glass, and (d) sandblasted coated glass.

214

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

pattern. The sliding speed of the scratch indenter was varied from 0.2 to
100 mm/min.
Progressive load scratch tests were performed with the same
indenters (200 and 800 m tip radii). The applied load, along the
3 mm pattern, was progressively increased from 0 to 30 N for the indenters with the 800 and 200 m tip radii. The sliding speed was kept
constant at 1 mm/min.
The scratch equipment enabled the measurement of the normal and
tangential forces and of the penetration and residual depths, that is, the
penetration of the indenter tip inside the coating material during the
application of the load and after the load release and recovery in the
elastic eld of the indented material. For this purpose, the initial prole
on which the scratch test was to be performed was measured by probing it at a constant applied load of 0.03 N during the pre-scan and, then,
subtracting the indenter position during the application of the load
(scan), thus determining the penetration depths. The residual depths
were measured during the post-scan by applying the minimal load of
0.03 N and probing the scratched surface again after the recovery of
the material in the elastic eld. The analysis of the scratch pattern
after the load release was also performed using a eld emission gunscanning electron microscopy (FEG-SEM Leo, Supra35, Carl Zeiss SMT,
Inc. Thornwood, NY, USA) and stereomicroscopy (SMZ 745T, Nikon,
Japan).
The wear resistance of the siliconeepoxy coatings was investigated
by performing dry sliding linear reciprocating tribological tests
(Tribometer, C.S.M. Instruments, Peseaux, Switzerland), using a spherical antagonist (6 mm in diameter, 100Cr6 bearing steel). Incremental
sliding distances of up to 500 m were considered. The load during the
tribological tests was set at 1 N, the sliding frequency was set at 3 Hz
and the sliding distance for each stroke was set at 6 mm. The equipment
permitted the measurement of the friction during the test. In addition,
after the load release and recovery of the material in the elastic eld,
the wear pattern was monitored by contact gauge prolometry, SEM
and stereomicroscopy. The prolometry allowed for the 3D rebuilding
of the wear pattern (resolution of 2 m) and thus the estimation of
the amount of abraded material during the tribological tests.

3. Results and discussion

3.1. Coating build-up and analysis of the morphological features
The deposition of thick siliconeepoxy coatings on a at glass surface at ambient temperature should necessarily involve the microroughening of the surface to ensure the establishment of a solid bond
between the resin and substrate by gluing [16], whereas the establishment of satisfactory chemical bonds is prevented by the lack of a proper
curing process at higher temperature [4,8]. In fact, the siliconeepoxy
resin can ll the gaps among the asperities of the micro-roughened
surface and adhere strongly via simple mechanical interlocking after
spontaneous drying [17].
Fig. 1a and b show the difference in the surface morphology of an asreceived and sand-blasted substrate. The as-received glasses are very
smooth with an average roughness Ra and ISO 10-point height Rz of
~0.2 and ~3.0 m, respectively. After sandblasting with very ne glass
beads, the resulting morphology of the glass substrates is microcoarsened. The established morphology is the result of superimposed
peaks and valleys, which are generated from each individual impact of
the glass beads on the relatively brittle glass surface. Thus, the nal morphology is spiky but displays a uniform distribution of the asperities,
with the average roughness Ra and ISO 10-point height Rz increasing
to ~3.0 m and ~17.9 m, respectively.
Fig. 1c and d show the morphology of the glass substrates after the
deposition of the siliconeepoxy resin. The resin deposits on the asreceived glass, thus copying the initial morphology of the glass itself.
However, the intrinsic self-leveling ability of the resin formulation
causes a further improvement in the smoothness of the substrate,
with the average roughness Ra and ISO 10-point height Rz of the coated
surface decreasing to 0.06 m and 0.8 m, respectively. In contrast,
when the resin is deposited on the sandblasted glasses, it signicantly
modies their morphology. The resin inltrates the surface irregularities, lling the gaps among them. Thus, the morphology of the
sandblasted glass is smoother after coating with the siliconeepoxy
resin. The deposition process is benecial to both the visual appearance
of the substrate, which becomes fairly smooth again and the coating adhesion, which can be favored by the mechanical anchoring between the
resin layers and underlying micro-corrugated glass as mentioned earlier
[18,19]. However, the micro-corrugation of the sandblasted glass does
not allow the resin to elicit its maximum intrinsic self-leveling capability. The wetting of the sandblasted glass surface differs from that of the
as-received glass. The micro-asperities on the sandblasted glass tend to
retain the resin from owing on the surface and limit its self-leveling
property. Thus, an orange-peel morphology is established on the coating surface. This morphology is essentially composed of the superimposition of soft hills and valleys, which are uniformly sparse on the coating
surface. Accordingly, the siliconeepoxy coating deposited on the
sandblasted glass is characterized by an average roughness Ra and an
ISO 10-point height Rz of 0.3 and 2.9 m, respectively. These roughness
parameters conrm the smoothing of the sandblasted substrates after
the deposition of the resin. However, these values differ considerably
from those achievable by depositing the siliconeepoxy resin on the
as-received glass substrate and, overall, the coatings deposited on the
sandblasted substrates maintain a signicant residual coarseness,
which could certainly affect the interaction with the liquid-driven soils.
3.2. Scratch response

Fig. 2. Siliconeepoxy coatings after the drying/cross-linking process: (a) spontaneous

delamination on the as-received glass substrate due to the shrinkage phenomenon; and
(b) good adherence of the coatingsubstrate on the sandblasted substrate.

The hardness and scratch response of the coatings are ascribable to

both the intrinsic properties of the materials selected and the overall
coating system, particularly the latter. Thus, the hardness and scratch
response may be related to the substrate features, coating thicknesses
and nature of the bonding at the interface between coatings and substrates. For in-service glass substrates, the routes to increasing hardness
and scratch performance should adhere to the following steps: (i) the

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

215

Fig. 3. SEM images of to the residual scratch pattern after constant mode scratch tests. The measurement conditions were: an indenter with an 800 m tip radius and a normal load of 20 N.
The test was performed on the as-received substrates (ae) and sandblasted substrates (fl). The sliding speeds were: (a, f) 0.2 mm/min, (b, g) 1 mm/min (c, h) 5 mm/min, (d, i) 25 mm/min,
and (e, l) 100 mm/min.

selection of a coating material with extreme performance, as the class

of the siliconeepoxy resin cross-linked with amine-based organofunctional silane ensures; (ii) the application of coatings that are sufciently thick to ensure a high micro-mechanical response but without
being overly brittle or stressed because of the shrinkage phenomenon
of the resin during the drying and cross-linking process; and (iii) pretreatments of the glass substrates to increase the potential contact
surface at the interface.
For the rst step, the material formulation investigated here is based
on a siliconeepoxy resin with a substantially inert silicone skeleton on
which lateral chains consisting of reactive epoxy groups lay. The epoxy
groups can react in the presence of amine-based organofunctional

silane hardeners. Cross-linking generates complex networks that are

extremely hard and scratch resistant. High hardness can be achieved
after hardening of siliconeepoxy resin with the amine-based hardeners and spontaneous drying. A pencil hardness of 67H is measured
when depositing the resulting coating system on already in-service
glass. Higher hardness could potentially be achieved by controlling the
reaction of the siliconeepoxy resin with amine-based hardeners.
Nevertheless, drying should be performed at a moderate temperature
of 6070 C. The temperature would increase the motility of each
individual molecule, thus increasing the degree of curing and delaying
gelation. Higher drying temperatures (N 6070 C) are counterproductive as they would favor the evaporation of the hardeners, thus

216

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

Fig. 4. Trends in the friction force during constant-mode scratch testing. The test was
performed using an indenter with an 800 m tip radius at an applied load of 20 N.

signicantly decreasing the reaction grade. However, thermal crosslinking is impractical when it should be performed on in-service glass
panels. At ambient temperature, although the reagents (i.e., the epoxy
groups on the silicone skeleton and the amine groups on the hardeners)
have considerable time to combine, their motility is limited by the
intrinsic viscosity of the formulation. The viscosity can be regulated by
the solvent concentration. However, the viscosity of the mixture increases rapidly once cross-linking begins. The corresponding motility
of the molecule inside the mixture is largely reduced and gelation
can occur rather quickly, thus limiting the maximum degree of crosslinking and the maximum coating hardness achievable.
In the present investigation, the coating thickness was always
maintained at ~120 m, which is at least one order of magnitude greater
than the thickness selected for the application of common coatings on
glass manufactured via the solgel route [4,20,21]. The deposition of
such a thick coating on an already in-service stiff glass substrate at ambient temperature is troublesome. In fact, it involves a massive shrinkage
of the resin during the simultaneous drying/cross-linking process with
a non-compliant substrate. Thus, shrinkage is one source for the onset
of a major stress eld that is retained inside the coating. The stress is
tensile, acting parallel to the coatingsubstrate interface, and may be sufciently high to lead to spontaneous breakage and delamination at the
interface of the coating from the underlying substrate as shown in
Fig. 2a and reported in the literature [8,9]. Such a stress eld cannot be
relieved when the coating is deposited on a non-compliant substrate
such as stiff glass. However, a proper design of the coatingsubstrate
interface can oppose the action of the stress eld. In particular, this opposition can be performed by increasing the interfacial contact surface
between the coating and substrate by corrugation. Corrugation creates
interruptions on the very smooth glass surface in the form of microasperities. The widespread presence of micro-asperities that are well
distributed on the coating surface acts as an obstacle against the action
of the stress eld generated by the shrinkage phenomenon of the resin
on the stiff glass during the drying/cross-linking process. Thus, such
micro-asperities mechanically oppose the spread of the stress eld,
potentially preventing early failure of the coating (Fig. 2b).
The implementation of the latter step includes the selection
between two methods of increasing the interfacial adhesion on in
service glass: (i) chemical etching of the glass substrate with aggressive (acidic) solutions to promote micro-coarsening of the surface;
and (ii) sandblasting of the glass surface with ne glass-beads to
achieve micro-coarsening due to the impingements of the beads on
the brittle substrate. The former route is substantially impractical

because it is characterized by a massive environmental impact related

to the usage of powerful reactants and the need to remove of them.
The latter route is simpler because it involves only inert materials
that can also be left in the environment after use and can adherently
deposit coating formulation on a micro-roughened surface, as stated
in Section 3.1.
Fig. 3 presents the residual scratch pattern of the siliconeepoxy
coatings deposited on both the as-received (panels on the left side)
and sandblasted (panels on the right side) glass after a constant-mode
scratch test at a 20 N load with a sliding speed of 0.2100 mm/min,
against an indenter with a tip radius of 800 m. The superior scratch
response of the coating on the sandblasted substrate is neat because it
does not exhibit any damage or signicant residual deformation. The
coating deposited on the as-received glass does not exhibit any signicant residual deformation either. However, there is signicant damage
on the coating surface in the form of large transversal and approximately C-shaped fractures, which fall well within the expected scratch pattern, that is, the area of real contact between the sliding indenter and
coating surface. These fractures are compatible with those presented
by Bull et al. for the fragile tensile crack mechanism [22,23]. This fracture mechanism is typical of brittle or partially brittle materials, and it
is characterized by cracks whose shape is neat; visible pile up or other
plastic deformations could not be detected. The failure mechanism by
tensile cracking is ascribable to the compressive stresses accumulated
within the coating material when it comes in contact with the sliding indenter and is thus submitted to the corresponding tangential force.
When a compressive stress eld is generated ahead of the advancing indenter tip, a tensile stress eld is generated at its back. When the tensile
stress exceeds the critical load of the indented material, failure occurs
with the concurrent onset of fractures, which are usually C-shaped
and occur behind the last contact position between the indenter tip
and coating material. Otherwise, the fractures can occur even earlier,
that is, under the application of softer loading conditions, when the
stress eld inside the coating material is sufciently high to overcome
the adhesive toughness at the interface between the coating and glass
substrate. In the present investigation, fractures only occur on the coatings deposited on the as-received glass, whereas coatings deposited on
the sandblasted glasses do not fail. Thus, sandblasting of the glass
substrate is particularly useful for improving the interfacial adhesion,
by generating the mechanical interlocking of the coating material and
underlying glass substrates [18]. The increase in the local specic contact area between the coating and substrate, as well as the inltration
of the resin into the irregularities of the substrate surface, also plays
an important role in improving the scratch performance of the investigated systems, which is consistent with the ndings observed by
[24,25]. The increase in the contact surface between the coating and
sandblasted glass together with the presence of a peak to valley morphology at the interface should create a discontinuity in the tensile
stress eld generated at the back of the advancing indenter. This phenomenon is believed to delay the onset of fractures inside the coating
material because it signicantly increases the adhesive toughness
of the coating to the underlying substrate, as stated in the literature
on different coatingsubstrate systems [18,19]. Another interesting
hypothesis regarding the difference in the onset of fractures in the coatings deposited on the as-received and sandblasted glasses can arise by
examining the trends of the tangential forces (Fig. 4) during constantmode scratch tests at a load of 20 N with the speeds of 0.2 and
100 mm/min and using an indenter with the tip radius of 800 m. The
experimentally measured tangential forces applied by the advancing indenter on the coating materials depend on the intrinsic features of the
coatings. The coatings on the sandblasted glasses are stressed with a
low tangential force, which could experimentally lead to the conclusion
that the lack of fractures on the coatings deposited on the sandblasted
substrates could also be related to the lower tangential forces acting
on them. The reason why the tangential forces should be lower on the
coatings deposited on the sandblasted substrates is difcult to explain.

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

217

Fig. 5. Penetration and residual depths of: (a, c) as-received substrates, (b, d) sandblasted substrates. The test was a constant-mode scratch test using an indenter with an 800 m tip radius
at an applied load of 20 N.

However, the tangential forces enables the indenter tip to move along
the scratch pattern with the nal purpose of maintaining the normal
applied load at the set value of 20 N. Thus, the coating material opposes
resistance to the advancing indenter. The resistance that the indenter is
able to oppose depends largely on the intrinsic features of the material
and the mechanism by which the material deforms during the application of the scratch load. As stated above, the indenter presses the
material ahead of its advancing tip, generating a compressive stress
eld inside the coating and its local plastic deformation. This phenomenon leads to the accumulation of some coating material in the form of a
plastic pile-up in front of the advancing indenter. It could be speculated
that the micro-roughened interface is better able to withstand the

generation of the compressive stress eld inside the coating material.

Accordingly, the plastic deformation of the coating and its accumulation
in front of the advancing indenter tip would be limited. Thus, less resistance would be offered to the advancing indenter by the deformed coating material and lower tangential forces would be necessary for the
indenter to move along the scratch pattern. This process would explain
the lower tangential forces measured during the constant-mode scratch
tests. Similarly, this process would support the lack of any fracture phenomenon on the coatings deposited on the sandblasted glasses, as they
are submitted to less severe loading conditions.
However, the scratch performance, failure route and critical load
were often found strictly correlated to the physical afnity between

Fig. 6. SEM images of the residual scratch patterns after progressive-mode scratch test. The measurement conditions were: an indenter with an 800 m tip radius, an incremental normal
load of 030 N and a sliding speed of 1 mm/min. The test was performed on the (a) as-received substrate and (b) sandblasted substrate.

218

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

the coating and substrate [15], with the substrate corrugation often
playing a fundamental role for different systems (for instance, epoxy
on sandblasted carbon laminates [26]).
The SEM images in Fig. 3 also reveal that both the number and size of
the transversal cracks are recorded when varying the sliding speed. In
particular, at the higher sliding speeds of 25 and 100 mm/min, no significant crack damage is visible either on the deposited coatings or asreceived glass. The sensitivity of the coating material to the sliding
speed is likely ascribed to the organic moieties of the resin. The organic
materials are extremely sensitive to the speed at which a load is applied,
that is, to the load rate [27]. Polymeric materials can exhibit a stiffer
response as the load rate is increased [28]. The stiffer deformation
response is typically ascribed to the time-dependent response of the
macromolecules, which have an intrinsic time to respond to an external
load and deform accordingly [29]. When the loading rate is excessive,
the macromolecules that the organic material is composed of have a
very short time to comply with the load and behave stify. In contrast,
when the loading rate is low, the macromolecules have more time to
comply with the load, thus deforming signicantly and behaving with
more compliance. In this case, the contribution of the organic moieties
of the resin cannot be neglected. Thus, when the coating material is
subjected to an increasing load rate, the organic moieties of the resin
behaves stify because of their intrinsic visco-elasticity and the coating
system is better able to withstand the action of the sliding indenter,
with no visible cracks appearing along the scratch pattern, regardless
of the geometrical features of the interface between the coating and
substrate surface.
Fig. 5 presents the trend of the penetration and residual depth
measured on the siliconeepoxy coatings deposited on both the asreceived (panel 2ac) and sandblasted glass (panel 2bd) after
constant-mode scratch tests at a load of 20 N and variable sliding
speed (0.2100 mm/min) against an indenter with a tip radius of
800 m. The extent of deformation under the application of the load is
similar for all coatings regardless of whether they were deposited on
as-received or sandblasted glasses, with an average penetration depth
of ~30 m. In addition, the aforementioned sensitivity of the coating
material to the imposed sliding speed is conrmed, as the penetration
depth increases by several microns when the sliding speed is decreased.
Because the residual depth is nearly negligible for all investigated scenarios (always less than 3.5 m), the deformation response of the coating can be considered visco-elastic. The coatings are considered viscous
because they are sensitive to the imposed sliding speed during the

Fig. 7. Penetration and residual depth in progressive-mode scratch tests performed at an

incremental load of 030 N with a sliding speed of 1 mm/min against an indenter with
an 800 m tip radius.

constant-mode scratch test, with the extent of deformation under the

load increasing when a lower sliding speed is imposed. The coatings
are considered elastic because they recover almost all of the deformation imposed during the application of the scratch load after its release.
The residual depth trends in panels 5c and 5d also enable the detection
of the failure of the coating material after the constant-mode scratch
test and recovery in the elastic eld. In fact, the trend of the residual
depth measured on the coating deposited on the as-received glass is irregular, as expected for the coating surface after failure (Fig. 3a, for
example) [26,30]. In contrast, the trend of the residual depth measured
on the coating deposited on the as-received glass is smooth, as the coating surface is not affected by the scratch indentation in this case (Fig. 3f,
for example) [26,30]. Although the role of the interface is crucial in
determining the scratch resistance of the coatings, the interface only
plays a minor role in determining the extent of deformation under the
application of the load. In terms of the deformation response, a certain
role can be ascribed to the different interface only when the residual
depths are examined. However, the difference largely resides in the
fracture events that characterize the coatings deposited on the asreceived glass and therefore in the corresponding trends of the residual
depth. Thus, the siliconeepoxy coatings on the as-received glass under
the indenter with a tip radius of 800 m belong to the class of viscoelastic brittle materials. Under such loading conditions, the silicone
epoxy coatings on the sandblasted glass only reveal their intrinsic
visco-elastic nature.
Progressive-load scratch tests performed with an incremental load
of 030 N at sliding speed of 1 mm/min and using the indenter with a
tip radius of 800 m yield similar ndings as the SEM images in Fig. 6
and the penetration and residual depth trends in Fig. 7. The coating deposited on the sandblasted glass does not display any damage. The corresponding trends for penetration and residual depth are increasing and
fairly smooth. The penetration depth follows a power-law trend, with a
maximum penetration depth of ~40 m under 30 N load. The residual
depth increases nearly linearly, whereas it decreases at high load. However, the maximum residual depths are ~22.5 m (~5% of the maximum penetration depth) and can thus be neglected. In fact, it is
difcult to discern any residual deformation on the SEM images in
Fig. 6. However, these depth trends are fairly common [31]. In particular, the small decreasing branch at the end of the residual depth trend
is attributed to the material accumulated by plastic deformation in
front of the advancing indenter and, in particular, around the last contact position between the indenter tip and coating material. As shown
before, the coating deposited on the as-received glass reveals certain
C-shaped fracture events after constant-mode scratch tests. In this
case, the route to failure is similar to that observed after the constantmode scratch tests. Therefore, the involved mechanisms should be identical. The trend of residual depth for the coating deposited on the asreceived glass is characterized by a fairly uneven trend at high loads
(N2324 N). Such irregularities in the residual depth trends are attributed to the onset of fractures in the coating, as they were for the
constant-mode scratch tests on similar coating systems.
Increasing the nominal contact pressure during constant- and
progressive-mode scratch tests allows the difference in the scratch performance of the coatings deposited on the as-received and sandblasted
glasses to be emphasized. In particular, the loading conditions of the
constant-mode scratch tests were exacerbated by applying sharper contacts using an indenter with a tip radius of 200 m, setting the load to
20 N and varying the sliding speed from 0.2 to 100 mm/min. The SEM
images in Fig. 8 reveal that the coatings deposited on the as-received
substrate collapse under the scratch indenter. The coatings are
completely detached from the underlying substrates and the rupture
spreads far away from the actual contact between the indenter tip and
coating surface. This result implies that the coating material delaminates on an area that is considerably larger than the actual contact
area with the indenter during the scratch indentation. In addition,
delamination begins at a moderate scratch distance, that is, at the

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

219

Fig. 8. SEM images of the residual scratch patterns after constant-mode scratch tests. The measurement conditions were: an indenter with a 200 m tip radius and a normal load of 17 N.
The test was performed on as-received substrates (ae) and sandblasted substrates (fl). The sliding speeds were: (a, f) 0.2 mm/min, (b, g) 1 mm/min (c, h) 5 mm/min, (d, i) 25 mm/min
(e, l) 100 mm/min.

beginning of the scratch pattern, within the rst 0.5 mm. This result is
demonstrated by the sudden jump of the penetration and residual
depths to values near the average coating thickness (Fig. 9a and c).
The observed failure mechanism can fall within the mechanism of buckling and brittle spallation, which is consistent with [22]. As stated previously, the coating material is subjected to a load that is greater than its
critical load or, otherwise, greater than its adhesive strength to the
underlying substrate. Thus, a brittle fracture is generated at the interface
between the coating and substrate. The fracture propagates rapidly and
causes the delamination of the coating on a very large area. The SEM
images in Fig. 8ae only allow the delamination of the material to be
discerned. No differences were observed with varying sliding speeds;
furthermore, the fastest load rate causes the complete failure of the

coating. The stereoscope failure image (sliding speed of 1 mm/min,

Fig. 10a) also enables the massive bulging phenomenon of the coating
at the beginning of the scratch pattern to be emphasized. The coating
is already completely delaminated during the rst moment of the
scratch test and it bulges over the substrate, thus involving a very
large zone. The surrounding coating keeps the bulged zone weakly
bonded to the substrate, although there is no adherence at the interface.
Delamination is associated with the complete detachment of the coating from the underlying substrate along the scratch pattern and at a
higher scratch distance (Fig. 10a). The complete detachment of the coating from the substrate is ascribable to the instability generated by the tip
of the advancing indenter on the bulged material, which is consistent
with [32]. The bulged material is overstressed by the pushing indenter

220

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

Fig. 9. Penetration and residual depth: (ac) as-received substrates, and (bd) sandblasted substrates. The test was performed in constant mode using an indenter with a 200 m tip radius
and an applied load of 20 N.

and is completely torn off from the underlying substrate, thus generating an extended delamination zone, that is smaller than the bulged
zone. The coatings deposited on the sandblasted substrates exhibited
a clearly improved scratch response. Although several and repeated
cracks can be observed on the coating surface in Fig. 8fl, there is no
delamination and the coating damage is conned well within the actual
contact area between the scratch indenter and surface. In Fig. 10b (sliding speed of 1 mm/min), the coating is not characterized by any bulging
phenomenon and remains rmly adhered to the underlying substrate
despite the onset of fairly severe cracks. The failure mechanisms of the
coatings deposited on the sandblasted substrates are signicantly
affected by the sliding speed (Fig. 8fl). The damage on the coating

surface is larger at the lowest sliding speed (0.2 and 1 mm/min). The
damage consists of the superimposition of two failure mechanisms:
(i) the aforementioned C-shaped cracks, which imply the presence of
the brittle tensile cracking mechanism; and (ii) linear cracks (i.e., layer
break [20]) that develop along the axis of the scratch pattern related
to a presumable cutting of the coating material by the sharp indenter
tip. The latter mechanism is compatible with the nature of the glass substrate, which is extremely rigid and not compliant when submitted to
an external load. In the present case, as the overlying siliconeepoxy
coatings lay on the undeformable glass substrate, they can also be cut
off along the sliding direction of the indenter when indented with a
sharp tip, beyond being fractured according to the tensile cracking

Fig. 10. Optical microscope images of the residual scratch pattern after constant mode scratch test. The measurement conditions were: an indenter with a 200 m tip radius, a normal load
of 20 N and a sliding speed of 1 mm/min. Samples: (a) coating on an as-received substrate and (b) coating on a sandblasted substrate.

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

221

Fig. 11. SEM images of the residual scratch pattern after progressive-mode scratch test performed on a) an as-received substrate and b) a sandblasted substrate. The measurement
conditions were: an indenter with a 200 m tip radius, an incremental load from 0 to 30 N and a sliding speed of 1 mm/min.

mechanism. The second mechanism disappears at a sliding speed of

over 1 mm/min, and only the C-shaped cracks remain visible on the
residual scratch pattern. In addition, the frequency of the C-shaped
cracks along the scratch pattern diminishes when the sliding speed is
increased, thus conrming the visco-elastic response of the coating
material. The visco-elasticity of the coating material is conrmed by
examining the trends of the penetration and residual depths in
Figs. 9b and 9d, despite the presence of a certain plastic contribution.
In fact, the average penetration depths were observed to increase with
decreasing sliding speed, thus conrming that the coating behaved
stify when higher load rates were applied. The residual depths are no
longer negligible (maximum values of ~10 m at a sliding speed of
0.2 mm/min) and clearly decreases with an increasing load rate.
Thus, an overall visco-elastic and partially visco-plastic but still brittle
(i.e., the fractures spread over the scratch pattern) behavior can be reported for the coatings deposited on the sandblasted glass subjected
to the sharper indenter with a tip radius of 200 m. The trends for the
penetration and, in particular, residual depths also transform from
irregular (i.e., higher number of fracture events) at a lower load rate
(0.2 and 1 mm/min) to smoother (i.e., lower number of fracture events)
at a higher load rate (N 1 mm/min), thus demonstrating the dissimilarity in terms of the scratch resistance of the coatings subjected to the different sliding speeds. This result agrees with the SEM images in Fig. 8fl,
where the disappearance of the second damage mechanisms and the
reductions in the frequency of the C-shaped fracture are observed for
sliding speeds greater than 1 mm/min.
The indenter with a tip radius of 200 m was also used to test the
coatings using progressive-load scratch tests at a sliding speed of
1 mm/min and an incremental load of 030 N. The coating deposited
on the as-received glass substrate exhibited a massive failure with a
signicant delamination, whose area extended beyond the expected
scratch pattern (Fig. 11a). In contrast, the coating deposited on the
sandblasted glass substrate exhibited a better overall scratch response
(Fig. 11b). At low and moderate loads, only damage in the form of Cshaped cracks attributable to the tensile cracking mechanisms could
be detected. At higher loads, this mechanism was displaced by the
onset of linear cracks (i.e., layer breakage and adhesive failure [20])

in the direction of the advancing indenter. The latter mechanism had

a more severe effect on the coating, which exhibited partial delamination from the underlying substrate (i.e., by buckling and/or brittle
spallation) at the highest load. However, the delamination is conned
around the expected scratch pattern and further conrms the superior
scratch resistance of the coatings deposited on the micro-roughness
glass compared to the as-received glass substrate.
The experimental results allowed ve different critical loads to be
dened for the siliconeepoxy coatings deposited on both the asreceived and sandblasted glass substrates after progressive-mode
scratch tests using the indenter with a 200 m tip radius (Table 1).
Sandblasting enables the increase of the scratch resistance of the siliconeepoxy coatings. The superior scratch performance of the coatings
on the sandblasted glasses was ascribed to the good mechanical
interlocking at the interface promoted by the micro-corrugation. In
addition, the micro-roughened morphology may also be able to oppose
resistance to the stresses generated inside the coatings because of the
shrinkage phenomenon during the resin drying after its application on
the substrate surface. As mentioned previously, the differential shrinkage between the resin and stiff glass generates high residual stresses
inside the resin itself. Such stresses are manifested through the onset
of forces acting parallel to the interface between the coating and substrate. The presence of asperities on the micro-corrugated substrate creates a discontinuity at the interface, opposing the action of the stress
eld. Accordingly, the coatings become better able to withstand external loading conditions because the presence of internal stresses is
mitigated by the corrugated morphological features of the substrate
surface itself. This result is crucial because it enables the deposition of
a thick lm despite the potential onset of high internal residual stresses
due to the resin shrinkage during the drying/cross-linking process. This
result is translated into a large increase in critical loads if the thick coatings investigated here are compared with thinner coatings of similar
composition reported in the literature [15,20,21,33]. In particular, the
onset of coating failure by buckling and/or brittle spallation is increased
by more than one order of magnitude when the contact pressure is held
constant [15,20,21,33].

Table 1
Denition of the critical loads of the siliconeepoxy coatings on the as-received and
sandblasted glass substrates after scratch tests using an indenter with a 200 m tip radius.
Critical loads/progressive, 200 m

Coatings on the
sandblasted glass

Coatings on the
as-received glass

Onset of scratch visibility

Onset of C-shaped cracks
Onset of layer break
Onset of adhesive failure
Onset of buckling/spallation

Not applicable
~2 N
~13 N
Not applicable
~16 N

Not applicable
~7 N
~21 N
~24 N
~26 N

Fig. 12. Wear volumes vs. sliding distance.

222

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

Fig. 13. Details of debris generated during the tribological tests.

3.3. Wear endurance

The wear resistance of organic or hybrid organicinorganic coatings
is minimally affected by the substrate interface. This behavior is conrmed by the coatings deposited on the as-received and sandblasted
glass substrates, with a wear volume that increases similar to the sliding
distance regardless of the intrinsic features of the interface (Fig. 12).
Observing the trends in the wear volume, a distance of at least 100 m
is necessary to activate signicant mass loss under the prescribed
testing conditions. However, the process accelerates rapidly and the
measured wear volume increases once signicant wear is measured.
After a sliding distance of 500 m, the coatings deposited on both the
as-received and sandblasted glass substrates are still anchored on the
underlying substrate, although much of the initial material has been
torn off.
The trends of material loss during the tribological tests suggest the
presence of two interaction mechanisms between the coatings investigated and the counterpart: (i) a rst step during which no material
removal is observed and interaction is essentially ruled by friction;
and (ii) a second step involving material removal and thus the formation of some debris along the contact zone. During the former step,
the interaction by friction can be ascribed to adhesion, that is, to the

breaking of adhesive bonds occasionally formed between the coatings

and counterpart. The interaction in this step can also be attributed to
ploughing, that is, to the resistance originating from the elastic and potentially plastic deformation generated by the action of the counterpart
when it slides on the surface of coatings supported by a stiffer substrate
(i.e., the presumably softer thick siliconeepoxy coatings on the stiffer
glass substrates). Indeed, the coating is subject to signicant deformation, at least, in the elastic eld even when submitted to moderate contact pressure (as is the case in the initial step of the progressive-mode
scratch tests). Accordingly, ploughing can be assumed to be the main
friction mechanism. The mechanisms should be the same during the
latter step, which involves material loss, with both adhesion and
ploughing being potentially able to generate the formation of coating
debris along the sliding contact (see the details of the debris in
Fig. 13). In particular, the removal of material from the coating can be
attributed to both adhesion + fracture and abrasion + fracture in
agreement with what is reported in [34]. In fact, the action of the sliding
counterpart can generate an adhesive lifting or, otherwise, a signicant
deformation in the elastic and/or plastic eld in the coating from which
shear stresses much higher than the intrinsic strength of the coating
material itself can originate and thus cause the formation of surface
cracks and the concurrent onset of debris.
The experimental analysis of the fractures on the coating surface
supports the aforementioned hypothesis and elicits ploughing and, accordingly, abrasion + fracture as the main mechanism responsible for
the removal of material from the coating surface during the tribological
tests (Fig. 14). In fact, despite the wear pattern being achieved by a dry
sliding reciprocating motion of the counterpart, which should hinder
the nature of the damage, the material is primarily removed by the
formation of fractures by brittle tensile cracking regardless of whether
the coatings are deposited on as-received or sandblasted glasses
(Fig. 14). Fractures originate in the form of C-shaped cracks, which
always is directed toward the advancing direction of the counterpart
(i.e., the motion of the counterpart is reciprocating and thus it is possible
to discern the presence of both left- and right-trending C-shaped cracks,
as shown in Fig. 14). In addition, no signicant residual deformation is
visible on the wear pattern. The presence of C-shaped cracks and the
lack of signicant residual deformation on the coating surface underline
the elastic brittle response of the coating material during the tribological
tests. Accordingly, most of the material loss can be attributed to the

Fig. 14. SEM images of the wear pattern of: (a) the as-received substrate after a sliding distance of 5 m, (b) the sandblasted substrate after a sliding distance of 5 m, (c) the as-received
substrate after a sliding distance of 200 m, and (d) the sandblasted substrate after a sliding distance of 200 m.

M. Barletta et al. / Surface & Coatings Technology 236 (2013) 212223

mechanism of ploughing in the elastic eld, that is, by the abrasion +

fracture wear mechanism, with the friction by adhesion and the corresponding wear mechanism by adhesion + fracture playing a minor
role, if any, in agreement with [34].
4. Conclusions
This study investigated the application of thick organicinorganic
hybrid materials on non-compliant substrates. Alternative solutions
were proposed using a class of epoxysilicone to overcome the common
drawback of the shrinkage which generates large stresses inside the resin
synthesized by the solgel route when deposited on a stiff substrate and
left to dry/cross-link at ambient temperature. Micro-corrugation of the
glass was observed to be the key factor in ensuring good interfacial
adhesion and preventing early delamination of the overlaying coatings.
The micro-asperities act as discontinuities that oppose the forces that,
primarily, take action parallel to the interface and signicantly increase
the micro-mechanical performance of the coating system.
In addition, the following conclusions can be drawn:
sandblasting of the glass substrate improves the interfacial adhesion
between the coating and substrate by promoting their reciprocal
mechanical interlocking;
scratch adhesion of the epoxysilicone coatings deposited on the
sandblasted glasses is improved considerably, as conrmed by the
detection of the critical loads;
the onset of the critical loads during progressive mode scratch tests of
the thick siliconeepoxy coatings is at least one order of magnitude
higher than that of the critical loads observed in comparative coating
systems reported in the literature;
abrasion + fracture were observed as the main mechanism of material loss during the tribological tests; however, the wear endurance
was not found to be related to the interface features.
Lastly, the experimental ndings demonstrate that the silicone
epoxy coatings deposited on sandblasted glass substrates are extremely
suitable for protecting overlying layers from early failure due to
scratches or wear phenomena. Accordingly, the coating demonstrated
high potential to preserve the initial efciency of the underlying glass
substrate over long periods.

223

References
[1] S. Amberg-Schwab, E. Arpac, W. Glaubitt, K. Rose, G. Schottner, U. Schubert, in: P.
Vincencini (Ed.), Materials Science Monographs-High Performance Ceramic Films
and Coatings, Elsevier, Amsterdam, 1991, pp. 203210.
[2] K.H. Haas, H. Wolter, Properties of polymerinorganic composites, Encyclopedia of
Materials: Science and Technology, Second ed., 2001. 75847594.
[3] G. Schottner, Chem. Mater. 13 (2001) 34223435.
[4] D.R. Uhlmann, T. Suratwala, K. Davidson, J.M. Boulton, G. Teowee, J. Non-Cryst.
Solids 218 (1997) 113122.
[5] K.T. Chou, B.I. Lee, Ceram. Int. 19 (1993) 315325.
[6] P. Innocenzi, M.O. Abdirashid, M. Guglielmi, J. Sol-Gel Sci. Technol. 3 (1994) 4755.
[7] F. Mammeri, E. Le Bourhis, L. Rozes, C. Sanchez, J. Mater. Chem. 15 (2005)
37873797.
[8] H. Schmidt, J. Sol-Gel Sci. Technol. 1 (1994) 217231.
[9] J. Malzbender, G. de With, J. Non-Cryst. Solids 265 (2000) 5160.
[10] H. Dislich, J. Non-Cryst. Solids 57 (1983) 371388.
[11] W. Glaubitt, D. Sporn, E. Hussmann, A. Gombert, V. Wittwer, Glas. Sci. Technol. 73
(2000) 193.
[12] M. Kursawe, T.A. Hofmann, in: H.A. Meinema, C.I.M.A. Spee, M.A. Aegerter (Eds.),
Proceedings of the 3rd International Conference on Coated Glass (3rd-ICCG), 681,
Universal Press, Veenendaal, The Netherlands, 2000.
[13] A. Shirakura, Development of Easy-to-Recycle Coloured Glass Bottles, Product Information, Kirin Brewery Co. Ltd., Yokohama, Japan, 2000.
[14] Y. Akamatsu, K. Makita, H. Inaba, H. Nakazumi, T. Minami, J. Sol-Gel Sci. Technol. 19
(2000) 387391.
[15] Y.-H. Han, A. Taylor, K.M. Knowles, Surf. Coat. Technol. 203 (2009) 28712877.
[16] S. Ebnesajjad, Chapter 8: Characteristics of adhesive materials, Handbook of Adhesives and Surface Preparation, 2011. 137183.
[17] W.C. Wake, Polymer 19 (1978) 291308.
[18] A.J. Kinloch, J. Mater. Sci. 15 (1980) 21412166.
[19] V. Bellido-Gonzalez, N. Stefanopoulos, F. Deguilhen, Surf. Coat. Technol. 7475
(1995) 884889.
[20] Y. Bautista, M.P. Gomez, C. Ribes, V. Sanz, Prog. Org. Coat. 70 (2011) 358364.
[21] J. Malzbender, G. de With, Thin Solid Films 386 (2001) 6878.
[22] S.J. Bull, Surf. Coat. Technol. 50 (1991) 2532.
[23] V. Jardret, P. Morel, Prog. Org. Coat. 48 (2003) 322331.
[24] F. Awaja, M. Gilbert, G. Kelly, B. Fox, P.J. Pigram, Prog. Polym. Sci. 34 (2009)
948968.
[25] M. Barletta, D. Bellisario, Prog. Org. Coat. 70 (2011) 259272.
[26] M. Barletta, D. Bellisario, G. Rubino, N. Ucciardello, Prog. Org. Coat. 68 (2010)
100110.
[27] N. Aleksy, G. Kermouche, A. Vautrin, J.M. Bergheau, Int. J. Mech. Sci. 52 (2010)
455463.
[28] C. Gauthier, R. Schirrer, J. Mater. Sci. 35 (2000) 21212130.
[29] S.G. Bardenhagen, M.G. Stout, G.T. Gray, Mech. Mater. 25 (1997) 235253.
[30] A. Krupicka, M. Johansson, A. Hult, Prog. Org. Coat. 46 (2003) 3248.
[31] V. Jardret, H. Zahouani, J.L. Loubet, T.G. Mathia, Wear 218 (1998) 814.
[32] S.J. Bull, E.G. Berasetegui, Tribol. Int. 39 (2006) 99114.
[33] X. Chen, S. Zhou, B. You, L. Wu, Prog. Org. Coat. 74 (2012) 540548.
[34] K. Holmberg, H. Ronkainen, A. Laukkanen, K. Wallin, Surf. Coat. Technol. 202 (2007)
10341049.