Construction and Building Materials 96 (2015) 679–689

Contents lists available at ScienceDirect

Construction and Building Materials

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

Effects of hygrothermal conditioning on epoxy adhesives used

in FRP composites
B. Paige Blackburn 1, Jovan Tatar ⇑, Elliot P. Douglas 2, H. R. Hamilton 3
University of Florida, Department of Civil and Coastal Engineering, Gainesville, FL, USA

h i g h l i g h t s

Six epoxies were tested to determine curing kinetics in hygrothermal environments.

Competing effects of added cure and plasticization were observed in all epoxies.

Significant decrease in Tg from theoretical values was observed in all clear epoxies.

Average Tg of tested epoxies was below the design temperature for some U.S. bridge locations.

a r t i c l e i n f o a b s t r a c t

Article history: Durability of FRP composites and their bond to concrete is essential to structural integrity of an FRP
Received 11 March 2015 repair. Epoxy adhesives are used to form FRP composites, and as a bonding medium between the FRP
Received in revised form 30 July 2015 and concrete substrate. Susceptibility of epoxy to the negative effects of water and high temperature
Accepted 9 August 2015
affects the longevity of FRP repairs in hygrothermal environmental conditions. The presented work
investigates the effects of such environments on the curing kinetics of epoxy. Competing effects of
plasticization and post-cure during accelerated conditioning are discussed; recommendations targeting
Keywords:
researchers, practitioners, and manufacturers are made based on the research findings.
FRP
Epoxy
Ó 2015 Elsevier Ltd. All rights reserved.
Concrete
Bond
Durability
Hygrothermal
Cohesive
Adhesive
Accelerated aging

1. Introduction bond critical environments, the reliability of the FRP strengthen-

FRP composites are deemed a prominent method for strength-
ening and repair of aged structures. However, their application in
harsh environments is still inhibited by the poor resistance of
FRP–concrete bond to such conditions. Due to the changing
mechanical properties of adhesives (loss of stiffness) and their
adhesion properties (loss of chemical bonds), when exposed to
⇑ Corresponding author at: University of Florida, Department of Civil and Coastal
Engineering, 365 Weil Hall, Gainesville, FL 32611, USA.
E-mail addresses: pblackburn182@gmail.com (B.P. Blackburn), jtatar@ufl.edu
(J. Tatar), edoug@mse.ufl.edu (E.P. Douglas), hrh@ce.ufl.edu (H.R. Hamilton).
1
U.S. Air Force, 365 Weil Hall, Gainesville, FL 32611, USA.
2
University of Florida, Department of Materials Science and Engineering, 156
Rhines Hall, Box 116400, Gainesville, FL 32611, USA.
3
University of Florida, Department of Civil and Coastal Engineering, 365 Weil Hall,
Gainesville, FL 32611, USA.
ing/repair scheme is compromised. Effects of environmental expo-
sure, primarily moisture, on the integrity of epoxy–concrete bond
is shown graphically in Fig. 1: in dry ambient conditions external
load is completely transferred into the concrete substrate, allowing
for distribution of damage in the substrate (Fig. 1a) at critical load-
ing; however, when stiffness of epoxy and chemical bonding are
adversely affected by the presence of moisture, full bond capacity
cannot be attained, as the bond fails prematurely along the
epoxy–concrete interface (Fig. 1b). Development of accelerated
conditioning protocols (ACP) to assess the durability of FRP–con-
crete bonded systems is necessary in establishing more reliable
design factors. Due to multiple degradation mechanisms existing
in polymers when exposed to different environments, the selection
of appropriate ACP is crucial in determining the durability of the
bonded system.

http://dx.doi.org/10.1016/j.conbuildmat.2015.08.056
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
680 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689
External Load Mechanical key
Proper bond:
damage distributed in formed at the
concrete substrate concrete surface
Epoxy
Concrete Chemical Bonds
(a)
Fig. 1. Failure mode of epoxy–concrete bond: (a) in dry ambient conditions; (b) following exposure to moisture.
2. Motivation
Current industry testing standards for FRP composites used in
civil infrastructure are prescribed by multiple internationally rec-
ognized technical organizations: AASHTO FRPS-1 [8], ACI 440.2M
[10], and ICC AC125 [26]. To determine the durability properties
of FRP, the composite is to be tested in direct tension following a
prescribed ACP. ACP incorporate a range of exposure environ-
ments, among which some of them dictate a temperature of 60 °
C, which can be slightly above the glass transition temperature of
cold-cured epoxies used to form the composite.
In addition to testing the durability of FRP composites, multiple
researchers ([1–3,21,33]) have concentrated their efforts on
three-point bending beam test [24] to determine the durability
properties of FRP–concrete bond. Three-point bending test method
is currently a candidate for standardization, under the auspices of
ASTM and ACI 440 subcommittee K – FRP Material Characteristics.
The work of ACI 440 subcommittee L – Durability was focused
primarily on specification of appropriate ACP to determine the
durability of bond in three-point bending test. The main obstacle
in this process was the lack of information in literature on deter-
mining the appropriate conditioning temperature for this purpose.
While higher temperature is deemed to accelerate the degradation
processes that take place at FRP–concrete bond, the main concern
is that if the conditioning temperature is higher than the T g , the
mechanisms leading to loss of bond may be different than the
mechanisms that occur during field exposure at lower tempera-
tures. The purpose of this study is to provide a better understand-
ing of evolution of degradation mechanisms in epoxy with time
when exposed to hygrothermal ACP at temperatures lower than
and higher than the measured T g of epoxy adhesive.
3. Background
Structural epoxies used with FRP composites are available in
two forms: clear epoxy, and more viscous paste epoxy. The differ-
ence in viscosity between the two types of adhesives comes from
additives (such as silica particles) that are not found in clear epox-
ies. Epoxy adhesives used in construction are generally based on
bisphenol A molecules hosting an epoxide functional group at both
ends, forming the monomer in the epoxy structure [29]. Epoxy
hardener is composed of amines that react with the epoxide groups
to form covalent bonds. The amines bond the monomers together
in a linear fashion to form polymer chains. They also allow for
Decreased External Load Loss of
epoxy stiffness chemical bonds
Epoxy
Concrete
(b)
cross-linking between the polymer chains to form additional cova-
lent bonds. The described process is termed ‘‘curing of epoxy”;
with subsequent cross-linking a non-crystalline hardened molecu-
lar structure is formed [22]. At ‘‘full-cure” no additional cross-
linking between the polymer chains is possible; at this point epoxy
has reached its full mechanical properties. The cross-linking den-
sity is usually described in terms of conversion that takes values
from 0 to 1.0, where a value of 1.0 signifies a fully-cured epoxy
[16].
The bond between epoxy and concrete is established through
chemical interactions and mechanical bonding [30]. While the
exact nature of the chemical bond is not known, it is thought to
consist of mostly hydrogen bonding between the surface molecules
of concrete and epoxy [19]. Integrity of epoxy–concrete bond is
considered to be primarily due to mechanical interlocking
([19,27]; Stewart, 2011; [30]). Stiffness of epoxy matrix, and conse-
quently the integrity of its mechanical bond, can be compromised
by two main degradation mechanisms: its transition to rubbery
state at temperatures higher than the T g , and plasticization of
epoxy matrix.
Once the T g is exceeded, the covalent bonds between the poly-
mer chains are capable of rotating, but remain intact. Therefore,
the general shape of the epoxy structure in glassy state is main-
tained, but its stiffness and strength are reduced. The value of T g
is dependent on polymer chain mobility. The lower the cross-
linking density of the epoxy molecular structure, the less thermal
energy is required for transition from a glassy state to a rubbery
state [22,23]. Amounts of free volume and cross-linking (measured
by conversion) affect polymer chain mobility and consequently
change the epoxy T g . At room temperature conditions for two or
more weeks, the conversion of most commercially available struc-
tural epoxies ranges from 0.8 to 1.0, which means that 80–100% of
possible covalent bonds are formed [34]. Exposure to temperatures
above those experienced during initial curing will cause additional
cross-linking [16]. Cross-linking restrains polymer chain mobility
by interconnecting individual chains together; therefore, the
higher the cross-linking density (more covalent bonds), the less
chain mobility, and the higher T g .
Chain mobility is also affected by the amount of free volume in
the polymer structure. Free volume is the available space within
the polymer chain network on a microscopic level [22,23]. With
increase in free volume, the mobility of polymer chains increases
as there is more available space for their movement; hence, less
thermal energy is required to convert the epoxy into a rubbery
state. Water absorption can lead to increases in free volume in
 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689
the epoxy. Epoxy curing results in the formation of hydroxyl
groups in the network. Upon exposure to water, the water mole-
cules are ‘‘drawn” into the epoxy and interact with the hydroxyl
groups via hydrogen bonding [23,16]. This hydrogen bonding cre-
ates more space between the cross-linked polymer chains, thereby
enlarging the free volume available. Water absorption into epoxy is
referred to as plasticization [25], and will result in a decrease of T g
[23,29,16]. Complicating this issue further, exposing epoxy to
water at temperatures above the T g can cause an increase in the
rate of water absorption due to the greater chain mobility at those
temperatures.
According to Zhou and Lucas [31], two types of water bonding
may exist in an epoxy matrix. One type of bonding is responsible
for plasticization by breaking the inter-chain Van der Wals forces,
while the other water bonding type is deemed to allow for
secondary cross-linking (pseudo cross-linking) by forming
hydrogen bonds between two polymer chains. The former is
associated with lower desorption activation energy (approxi-
mately 10 kcal/mol), and water sorption process that takes place
in low-temperature hygrothermal conditions; the other type of
water bonding in such conditions is considerably less. On the
other hand, presence of the latter type of water bonding is
associated with hygrothermal conditioning at higher tempera-
tures and longer exposure times, and approximate desorption
activation energy of 15 kcal/mol.
Multiple researchers examined the effects of hygrothermal con-
ditioning on structural epoxies used in construction. Moussa et al.
[28] showed that strength and T g of epoxy strongly depends on
curing temperature as well as the duration of curing. They also
found that decreasing the testing temperature increased epoxy
stiffness. Changes in T g , conversion and water absorption of two
construction epoxies, during the 0–28-day hygrothermal condi-
tioning, were reported by Choi [16]. It was found that effects of
plasticization and additional cross-linking during exposure were
in competition with each other causing changes in T g throughout
the exposure. In all test groups, by 28 days of conditioning,
increases in T g ranged from 1 to 14 °C. Stewart [29] found that both
modulus and strength of ambient-cured epoxies decreased due to
hygrothermal conditioning, mostly due to plasticization. Similar
findings were reported by Quan et al. [4], Benzarti et al. [6], and
Lattieri and Frigione [5].
4. Experimental methods
Several techniques are available to determine the T g of polymers, among them
the most commonly employed ones are: dynamic mechanical analysis (DMA),
thermo-mechanical analysis (TMA), differential scanning calorimetry (DSC), and
heat deflection method. There is no particular standard method preferred among
researchers; each method is recognized and covered by a respective ASTM standard.
Glass transition occurs over a temperature region and each method registers a dif-
ferent temperature within the region to indicate the transition. No conversion of T g
values between the methods is officially exercised; manufacturers, engineering
design codes, and researchers employ different test methods for T g , therefore T g test
data must be accompanied by the testing technique and specification of curing
history.
Table 1
Epoxy properties as reported by manufacturers.
Epoxy Category Cure conditions
A Clear 3 days, 60 °C
B Clear 5–14 days, 23 °C, 50% R.H.
C Clear-opaque Time N.S., 20 °C, 40% R.H.
D Clear 56 days, ambient
E Paste 7 days, 23 °C, 50% R.H.
F Paste Time N.S., 20 °C, 40% R.H.
N.S. = Information not specified by manufacturer.
R.H. = Relative Humidity.
*
Heat deflection temperature.
681
An informal survey, conducted by ACI Committee 440, found that DMA and DSC
are the two most commonly utilized methods for T g measurements in the industry
[15]. DSC was selected to determine the T g values of epoxy in this study. DSC does
not depend upon mechanical manipulation to detect thermal transitions; it is the
only method that measures T g using a nonmechanical thermal process. The DSC
purely thermal calculation of T g relates well to the epoxy property state over the
course of exposure, without incorporating effects of mechanical loading. Another
advantage of calculating T g with the DSC is the small sample size. To test samples
in the DSC, only 5–10 mg epoxy specimens can be used. Such small DSC samples are
much more representative of epoxy that forms mechanical interlocks on the con-
crete surface, compared to relatively large bar samples that are used in other test
methods, such as the DMA and TMA. Moreover, in an interlaboratory study on con-
struction epoxies, Bakis et al. [15] noted remarkably high repeatability of T g values
obtained by DSC method.
EXSTAR6000 DSC station from Seiko Instruments Inc. equipped with a 220CU
module and automatic gas cooling unit was used. Epoxy specimens were prepared
by casting epoxy directly into the sample pans to maximize the heat transfer
between the pan and epoxy, as described by Choi and Douglas [17]. T g measure-
ments were obtained from the first run at a heating rate of 10 °C/min. All T g values
in this study are reported as midpoint temperature (T m ) on DSC thermal curve, as
defined in ASTM E1356 [12].
An additional set of samples were simultaneously tested in Fourier Transform
Infrared Spectrometry (FTIR) to determine the conversion and absorbed water con-
tent in epoxy samples. The epoxy thin film samples, ranging from 0.23 to 0.69 mm
in thickness, were cast between glass plates; Teflon sheets were used to prevent
adhesion between epoxy and glass. FTIR Nicolet Magna 760 from Thermo
Electron Cooperation was utilized to record near-infrared spectra over the range
of 3800–6800 cm1 in an inert atmosphere. Scanning in near-infrared range was
conducted because a single absorption band for water exists in this range of wave-
lengths; the absorption band for water in the mid-infrared range is at approxi-
mately 3400 cm1; however, this band is broad and contains contributions from
other hydrogen-bonded species present in the epoxy, making the interpretation
of data challenging. Absorption data for each specimen were collected with 32
scans at a 4 cm1 resolution implementing a white light source, a CaF beam splitter,
and an MCT light detector.
5. Test matrix
DSC and FTIR tests were conducted on five different epoxies
provided by three independent manufacturers, and one model
epoxy system of known chemical composition (epoxy D).
Properties of tested epoxies, as reported by the corresponding
manufacturers, are summarized in Table 1; properties of epoxy D
were determined by Stewart [29]. Main known constituents of
each epoxy are listed in Table 2. Epoxies were categorized into
two groups based on their viscosity and transparency, clear and
paste. Clear epoxy was nearly translucent with viscosity similar
to honey. Paste epoxies are opaque with a significantly higher vis-
cosity. Epoxy C had a viscosity similar to that of clear epoxies, but
was opaque and was categorized as clear-opaque.
Exposure conditions in this study were chosen to match the
work of Stewart [29] on effects of hygrothermal conditioning on
mechanical properties of epoxy, and the study by Tatar and
Hamilton [3] on examination of effects of hygrothermal condition-
ing on FRP–concrete bond. The DSC test matrix presented in
Table 3 summarizes the selected conditions. Immersion samples
were immersed in covered water tanks at the specified tempera-
ture. Humidity samples were placed in the air space above the
water in the water tanks; humidity was monitored with a
Tg (°C) (ASTM) Tensile strength (MPa) Tensile modulus (MPa)
ASTM D638 [32] ASTM D638 [32]
82 (D4065 [13]) 72.4 3180
46 (N.S.) 55 1724
71 (N.S.) 55.2 3034
54 (ASTM E1356 [12]) 74 2440
47* (ASTM D648 [11]) 24.8 4482
50 (N.S.) 27.6 N.S.
682 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689

Table 2
Main constituents of epoxy resin and hardener.

Epoxy Resin Hardener

A DGEBA Polyoxyropylenediamine; Polyetheramine
B DGEBA Polyoxypropylenediamine
C DGEBA, Oxirane, mono[(C12-14-alkyloxy)methyl] derivs., Oxirane, alpha-(2-Aminomethylethyl)-omega-(2-aminomethylethoxy)-poly(oxy(methyl-1,2-
2,20 -[(2,2-dimethyl-1,3-propanediyl)bis(oxymethylene)]bis-; ethanediyl)); 3-aminomethyl-3,5,5-trimethylcyclohexylamine; 1,6-Hexanediamine,
ethylbenzene C,C,C-trimethyl-; Oxirane, 2,20 -[(1-methylethylidene)bis(4,1-
phenyleneoxymethylene)]bis-, homopolymer; Benzyl alcohol; imidazole
D DGEBA 5-phenoxy-2,3-pentadienylamine
E Mineral additives; DGEBA; 1,4-bis(2,3 epoxypropoxy)butane) Quartz; Benzyl alcohol; 3,6-diazaoctanethylenediamin;
dimethylaminomethylphenol; methylamine
F DGEBA, crystalline silica; ethylene glycol 2-piperazin-1-ylethylamine; Phenol, 4-nonyl-, branched; crystalline silica; ethylene
glycol

Table 3
tested in FTIR because it was not feasible to prepare thin film
DSC test matrix. epoxy specimens that were transparent to light.

Exposure Moisture Temp. Exposure Number of

(°C) time (weeks) specimens 6. Results and discussion
Control Atmospheric, 23–25.5 12 6
20–28% RH 6.1. Conversion and water absorption
Immersion Water 30 2 6
4 6 To obtain conversion and water content data, areas under char-
8 6 acteristic peaks of FTIR spectra were analyzed. Natural frequencies
12 6
of interest in this study included characteristic water peak at
60 2 6
4 6 5230 cm1, phenyl peak at 4622 cm1, and the epoxide functional
8 6 group peak at 4530 cm1 (Fig. 2). From the collected FTIR spectra
12 6 conversion can be calculated as:
Humidity 100% R.H. 30 2 6
AðtÞ
4 6 a¼1 ð1Þ
8 6 Að0Þ
60 2 6
4 6 where A(0) is epoxide area at zero cure (conversion of 0.0) sample,
8 6 and A(t) is the epoxide area of a partially cured sample, at time of
curing t. To account for differences in epoxy thin film thickness
(thicker samples absorb more light), area of epoxide peaks was nor-
malized to the phenyl peak (which is consistent throughout sample
Table 4
FTIR test matrix. life, regardless of conversion). This allowed for direct comparison of
results between different samples. An unknown peak was observed
Exposure Moisture Temp. Exposure Number of
at approximately 4560 cm1 (Fig. 2) that overlaps the epoxide peak.
(°C) time (weeks) specimens
Area of this peak, measured on a fully-cured sample, was subtracted
Control Atmospheric, 23–25.5 12 4
from the total area of epoxide peaks to obtain absorbance corre-
20–28% RH
sponding only to the epoxide groups. The base lines under each
Immersion Water 30 2 4
8 4
60 2 4
0.5
Humidity 100% RH 30 2 4
8 4
60 2 4 Phenyl 4622 cm-1
8 4 0.4

Water 5230 cm-1

humidity meter, and was determined to be consistently in 95–

Absorbance

0.3
100% R.H. range. Note that each epoxy had one sample tested for
each exposure condition, yielding a total of 6 test specimens per
condition. Exposure times were selected to correspond to typical
conditioning times used in conjunction with the three-point bend- 0.2
ing tests ([3,21]). Conditioning temperatures were chosen to be
lower than (30 °C) and higher than (60 °C) the typical T g of control Unknown 4560 cm-1
group of tested epoxies. 0.1
FTIR testing (Table 4) utilized the same exposure conditions
Epoxide 4530 cm -1
that were applied to DSC samples, but shorter (2- and 8-week)
exposure times were needed to gain understanding of specimen
0
behavior. Only 2-week samples were tested for conditioning tem- 5400 5200 5000 4800 4600 4400
perature of 60 °C. This decision was based on prior research that Wavenumber (cm-1)
reported plateaus in conversion and water content under this con-
dition after only 4 days of exposure [16]. Paste epoxies could not be Fig. 2. Typical FTIR spectra for epoxy.
 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689 683

peak for area calculation were determined as per Dannenberg [20].
Reliability of the quantitative method for calculation of epoxy cure
for construction epoxies was confirmed by Choi and Douglas [17].
Relative amount of water within each sample was calculated in
a similar fashion as conversion. Area under the characteristic water
peak at 5230 cm1, normalized by area of phenyl peak, provided an
indirect measure of the relative amount of water in the sample.
Choi and Douglas [17] verified the reliability of this method for
tracing the relative changes in water content of conditioned epoxy
samples.
FTIR analysis of epoxy A, B, C, and D conversions are summa-
rized in Fig. 3. Data from a previous study was included too, where
applicable [16]. The plots indicate increases in the degree of poly-
merization throughout the 8 week exposure period despite varia-
tions in moisture and temperature conditioning. These results
were expected since the epoxy samples were initially cured under
standard laboratory conditions (20–24 °C), which was 6–40 °C
cooler than temperatures experienced during accelerated condi-
tioning (AC). A very small negative slope, between 2 and 8 weeks,
in plots corresponding to epoxies A and D for 60 °C samples is
likely due to measurement error. Reduction in cross-linking would
only be possible by burning the epoxy structure [22], which does
not occur during AC. Test results also revealed that 60 °C samples
consistently achieved higher (epoxies A, B, and D) conversion than
30 °C samples at 2 weeks of exposure. Such behavior demonstrates
the important influence of temperature on the curing kinetics of
epoxy adhesives. AC temperature, above that experienced during
the initial cure, provided additional energy for cross-linking within
the epoxy structure, thereby increasing conversion. In addition, if
exposure temperature was above the T g , the increased mobility
of the network allowed additional cross-linking opportunities to
become available between polymer chains that were not accessible
in the glassy state. By the end of 8-week exposure period, all epoxy
systems were close to their fully cured state.
Epoxies A, B, and D had similar conversion values prior to expo-
sure ranging from 0.88 to 0.91. Epoxy C, however, had a larger con-
trol conversion of 0.96, possibly because its curing activation
energy is lower than that of other tested epoxy systems.
Consequently, epoxy C reached a fully cured state, for all ACs, fol-
lowing only 2 weeks of conditioning.
Based on the presented data it appears that the curing kinetics
of epoxy systems are independent of the moisture conditioning
method (immersion vs. humidity). Slight differences that were
observed in 30 °C samples, between immersion and humidity sam-
ples, indicate that at lower temperatures the method of moisture
conditioning may play a role in the rate of polymerization of the
epoxy matrix; however, assuming that conversion measurement
errors are on the order of ±1%, the recorded values could be pre-
scribed to the measurement errors.
The additional data by Choi [16] revealed that all 60 °C samples
reached a conversion value close to 1.0 after only 2 days of expo-
sure. This trend agrees well with the findings from this study,
regardless of whether samples were immersed or exposed to
humidity. Inconsistency between the initial (control) conversion
values from the two studies was noted in epoxy B and epoxy C;
the source of such discrepancy was determined to be a slight dif-
ference in the curing temperatures. Samples from previous
research were stated to have been cured under ambient conditions,

1 1
0.98 0.98
0.96 0.96
0.94 0.94
Conversion

Conversion

0.92 0.92
0.9 0.9
Immersion 30 C
0.88 0.88 Humidity 30 C
0.86 Immersion 30 C 0.86 Immersion 60 C
0.84 Humidity 30 C 0.84 Humidity 60 C
Immersion 60 C Immersion 30 C (Choi)
0.82 Humidity 60 C 0.82 Immersion 60 C (Choi)
B
0.8 0.8
0 2 4 6 8 0 2 4 6 8
Exposure Time (weeks) Exposure Time (weeks)

(a) (b)

1 1
0.98 0.98
0.96 0.96
0.94 0.94
Conversion

Conversion

0.92 0.92
0.9 0.9
Immersion 30 C Immersion 30 C
0.88 Humidity 30 C 0.88 Humidity 30 C
0.86 Immersion 60 C 0.86 Immersion 60 C
0.84 Humidity 60 C 0.84 Humidity 60 C
Immersion 30 C (Choi) Immersion 30 C (Choi)
0.82 Immersion 60 C (Choi) 0.82 Immersion 60 C (Choi)
C D
0.8 0.8
0 2 4 6 8 0 2 4 6 8
Exposure Time (weeks) Exposure Time (weeks)

(c) (d)
Fig. 3. Conversion results from FTIR data for: (a) epoxy A; (b) epoxy B (includes [16]); (c) epoxy C (includes [16]); (d) epoxy D (includes [16]).
684 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689

suggesting a constant 20 °C room temperature [16], while in the increasing water diffusion into the epoxy network which explains
present research, samples were cured at standard laboratory tem- the quicker saturation of 60 °C samples vs. 30 °C samples.
perature, which proved to range from 20 °C to 25 °C during the Water content of each sample was measured to investigate the
cure period, resulting in higher control conversions. possible correlations between the decrease in T g and plasticization.
Water absorption of specimens was traced by measuring char- Based on the characteristic water content measurements, it can be
acteristic water content from FTIR data. Characteristic water con- assumed that moisture equilibrium was maintained throughout
tents of epoxies A, B, C, and D are summarized in Fig. 4. Data by the 2–8-week exposure period.
Choi [16] is included, where applicable.
All samples exhibited an increase in characteristic water con- 6.2. DSC results
tent during exposure, as Fig. 4 denotes. It appears that, within
the first 2 weeks of exposure, all specimens reached a fully satu- T g of 6 different epoxies was measured for varying ACPs at spec-
rated state despite the epoxy used. Data in Fig. 4 illustrates no sig- ified time increments by DSC. Generally clear epoxies exhibited
nificant difference in water absorption between immersed samples similar behavior over time, while clear-opaque and paste epoxies
and those exposed to humidity. Within 2 weeks of exposure, all had varying behavior. Exemplary DSC curves, corresponding to
samples had reached equilibrium moisture content. Conditioned the model epoxy system (epoxy D), for varying exposure times
samples, on average, had characteristic water content approxi- and conditions are shown in Fig. 5. DSC curves exhibited distinct
mately 2.68 times larger than that of their equivalent control sam- glass transition regions in the majority of tested samples; some
ples. Some samples presented a slight decrease in water content of the samples that were exposed to immersion at 60 °C, however,
between 2 week and 8 week samples. This behavior was unex- had a wider glass transition region. It is recognized that in such
pected and it was accredited to measurement errors, drying of samples the reported midpoint temperature could vary approxi-
the sample prior to the test, or both. mately ±2 °C due to the subjectivity of the analysis process.
Water content data from prior research was included where
applicable in Fig. 4, to illustrate water absorption behavior at rela- 6.2.1. Clear epoxies
tively small time increments between 0 and 2 weeks of condition- Clear epoxies (A, B, and D) all shared similar behavior over the
ing [16]. It is clear that during the first week of exposure, water 12 week exposure (Fig. 6). Conditioning at 30 °C caused insignifi-
absorption occurred at a higher rate in samples conditioned at cant variations in T g from the control value in all epoxies, which
60 °C when compared to those conditioned at 30 °C. Full saturation could be well within the error of the measurement. By the end of
was achieved within 1–4 days of exposure in 60 °C conditions, and the 12 week exposure, the final T g was greater than the original
not until about 2 weeks in 30 °C conditions. Higher conditioning control T g (prior to exposure) for most 60 °C exposure conditions.
temperatures provide more energy to water molecules, thereby Conditioning temperature proved to have a strong influence on

7.5 7.5
Characteristic Water Content

Characteristic Water Content

6.5 6.5

5.5 5.5

4.5 4.5
Immersion 30 C
Humidity 30 C
3.5 Immersion 30 C 3.5 Immersion 60 C
Humidity 30 C Humidity 60 C
2.5 Immersion 60 C 2.5 Immersion 30 C (Choi)
A
Humidity 60 C B
Immersion 60 C (Choi)
1.5 1.5
0 2 4 6 8 0 2 4 6 8
Exposure Time (weeks) Exposure Time (weeks)

(a) (b)
7.5 7.5
Characteristic Water Content

Characteristic Water Content

6.5 6.5

5.5 5.5

4.5 4.5
Immersion 30 C Immersion 30 C
Humidity 30 C Humidity 30 C
3.5 Immersion 60 C 3.5 Immersion 60 C
Humidity 60 C Humidity 60 C
2.5 Immersion 30 C (Choi) 2.5 Immersion 30 C (Choi)
C
Immersion 60 C (Choi) D
Immersion 60 C (Choi)
1.5 1.5
0 2 4 6 8 0 2 4 6 8
Exposure Time (weeks) Exposure Time (weeks)

(c) (d)
Fig. 4. Water absorption results from FTIR data for: (a) epoxy A; (b) epoxy B (includes [16]); (c) epoxy C (includes [16]); (d) epoxy D (includes [16]).
 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689 685

Immersion at 30 C Immersion at 60 C
-4000 -4000
Control Control
2 weeks 2 weeks
-6000 -6000
4 weeks 8 weeks
8 weeks 12 weeks
-8000 12 weeks -8000
Heat Flow ( W)

Heat Flow ( W)
-10000 -10000

-12000 -12000

-14000 -14000

-16000 -16000

-18000 -18000

-20000 -20000
10 30 50 70 90 100 10 30 50 70 90 100
Temperature ( C) Temperature ( C)

Humidity at 30 C Humidity at 60 C
-4000 -4000
Control Control
2 weeks 2 weeks
-6000 -6000
8 weeks 8 weeks

-8000 -8000
Heat Flow ( W)

Heat Flow ( W)

-10000 -10000

-12000 -12000

-14000 -14000

-16000 -16000

-18000 -18000

-20000 -20000
10 30 50 70 90 100 10 30 50 70 90 100
Temperature ( C) Temperature ( C)

Fig. 5. DSC curves for epoxy D.

70 70 70

65 65 65

60 60 60
Tg ( C)

Tg ( C)

Tg ( C)

55 55 55

Immersion 30 C Immersion 30 C
50 50 Humidity 30 C 50 Humidity 30 C
Immersion 30 C
Immersion 60 C Immersion 60 C
Humidity 30 C Humidity 60 C
45 45 45 Humidity 60 C
Immersion 60 C Immersion 30 C (Choi) Immersion 30 C (Choi)
Humidity 60 C Immersion 60 C (Choi) Immersion 60 C (Choi)
40 40 40
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Exposure Time (weeks) Exposure Time (weeks) Exposure Time (weeks)

(a) (b) (c)

Fig. 6. Tg values over 12 weeks of conditioning: (a) epoxy A; (b) epoxy B (includes [16]); (c) epoxy D (includes [16]).

the T g of all clear epoxies. By 12 weeks, all samples exposed to 60 ° at 60 °C allowed more pseudo cross-linking from hydrogen bond-
C temperatures had a greater T g than the corresponding 30 °C sam- ing of water molecules to the network to occur than conditioning
ple. This behavior occurred at 8 and 12 weeks despite the moisture at 30 °C. Moreover, additional cross-linking that occurred in 60 °C
conditioning method. It is possible that hygrothermal conditioning samples due to the increased mobility above the glass transition
686 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689
may have contributed to the relatively higher T g , too. A discrep-
ancy between data reported by Choi [16] and the present study
was noted in epoxies B and D. In epoxy B, this was likely due to
the dissimilar curing histories between the two sets of samples
prior to the exposure that caused initial T g values by Choi [16] to
be lower, resulting in more pronounced competing effects of plas-
ticization and added cure. In epoxy D, the discrepancy in T g evolu-
tion plots is believed to be due to the fact that epoxies came from
different batches, and might have had slightly differing stoi-
chiometries depending on the mixing procedures. The range of
measured T g values, however, is similar between Choi [16] and
the present study for epoxy D. This indicates that both groups of
samples demonstrated a similar response to the effects of added
cure and plasticization.
There is some evidence that samples exposed to humidity were
not affected by plasticization during the first 2 weeks of condition-
ing, as much as immersed samples. Looking at each epoxy, mea-
sured T g of every 2-week humidity sample was higher than the
respective immersed sample. This suggests that effects of post-
cure overpowered the effects of plasticization in humidity samples,
while the opposite was true for immersed specimens. Although
there were significant differences in T g between immersed and
humidity samples at 2 weeks of conditioning, they decreased as
exposure continued. By 8 weeks the difference in T g between
humidity and immersed samples was considerably reduced; this
indicates slight recovery of T g of immersed samples over the
course of the exposure, and effects of post-cure. Interestingly,
water absorption plots for epoxies A, B, and D (Fig. 4) do not appear
to correlate well with changes in T g (Fig. 6). This suggests that
measured water content may not be as proportionally related to
effects of plasticization on the T g , primarily because the amount
of pseudo cross-linking from hydrogen bonding of water molecules
to the network cannot be discerned. A strong correlation of 0.89
was observed between the data corresponding to the two condi-
tioning temperatures at 2-weeks conditioning. For 8-week condi-
tioning, however, no correlation between data corresponding to
the two conditioning temperatures was observed. This suggests
that, early in the exposure, behavior of tested epoxies was gov-
erned by the same mechanism, while the degradation mechanism
changed later in the exposure depending on the conditioning
temperature.
6.2.2. Clear-opaque and paste epoxies
The clear-opaque epoxy C and the paste epoxies (E and F) each
exhibited unique behavior in terms of T g . Changes in epoxy C T g
over the exposure time appeared to be dominated by plasticization
70 70
Immersion 30 C
Humidity 30 C
65 Immersion 60 C 65
Humidity 60 C
60 Immersion 30 C (Choi) 60
Immersion 60 C (Choi)
Tg ( C)
Tg ( C)
55 55
50 50
45 45
40 40
0 2 4 6 8 10 12 0 2 4 6
Exposure Time (weeks) Exposure Time (weeks)
(a)
Fig. 7. Tg values over 12 weeks of conditioning: (a) epoxy C; (b) epoxy E (includes [16]); (c) epoxy F (includes [16]).
early in the exposure, as shown in Fig. 7. At 12 weeks of exposure,
epoxy C recovered to its control T g temperature that existed prior
to exposure (Fig. 3). One of the reasons for such behavior may be
the relatively high conversion in control sample, when compared
to clear epoxies. Since post-cure cure was limited, plasticization
was deemed the controlling factor on T g early in the exposure,
hence, reducing the T g of exposed samples below the control T g .
By 8 weeks of exposure, Epoxy E samples of all ACP converged
to the control T g value. Therefore, over time the competitive effects
of post-cure and plasticization were neutralized for 8 week expo-
sure. Prolonged exposure to water immersion, however, caused
an increase in T g , above the measured control value.
Epoxy F exhibited the largest change in T g of all the epoxy sys-
tems, with a drop of 19 °C from control value, under water immer-
sion conditions (Fig. 7). The presented data indicates extreme
sensitivity of epoxy F to plasticization. Interestingly, the T g of
epoxy F humidity samples stayed at, or slightly above the control
T g . These data suggest that epoxy F was more susceptible to plas-
ticization due to immersion in water than due to humidity. Higher
T g values in humidity samples at 30 °C when compared to samples
conditioned at 60 °C could be due to effects of mineral fillers on the
moisture diffusion under different conditioning temperatures.
Unfortunately, due to their inability to pass light, paste epoxies
were not analyzed in FTIR; therefore, the comparative water
contents of epoxy F samples were not available.
6.3. Master plots
Since changes in T g during hygrothermal exposure are depen-
dent on two different competing influences (post-cure and plasti-
cization), master plots were developed to isolate decrease in T g
caused by the effects of water only [17]. Master plots show the
dependence of T g on the level of conversion, in dry ambient condi-
tions (Fig. 8). With these theoretical curves, the T g of exposed sam-
ples could be compared to that of unexposed samples of equal
conversion, to determine the decrease in T g (DT g ). By plotting T g
based on sample conversion, the effects of post-cure are minimized,
leaving any differences in T g to be the result of plasticization [16].
Theoretical curves were created based on the following equation:
 
DC p1
ð1  aÞT g0 þ DC p0
T g1
Tg ¼   ð2Þ
DC p1
ð1  aÞ þ DC p0
a
where a is conversion (0–1.0); T g0 and T g1 are the T g values of the
monomer (0% cure) and the fully cured epoxy network,
70
Immersion 30 C
Humidity 30 C
Immersion 60 C
60
Humidity 60 C
50
Tg ( C)
40
Immersion 30 C
Humidity 30 C 30
Immersion 60 C
Humidity 60 C
20
8 10 12 0 2 4 6 8 10 12
Exposure Time (weeks)
(b) (c)
 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689
Fig. 8. Master plot construction.
27
2-week exposure
24 8-week exposure
Reduction in Tg ( Tg)
21
18
15
12
9 design temperature of Tg – 22 °C. Based on the previously pre-
6
3
0 1 requirements. It is apparent that maximum service temperatures
ABCD ABCD ABCD
Immersion Immersion Humidity
30ºC 60ºC 30ºC
Fig. 9. Reduction in Tg for 2- and 8-week conditioning.
respectively; DC p0 and DC p1 are the heat capacity changes at T g0
and T g1 , respectively. This semi-empirical equation was developed
by Couchman and Karasz [18] to establish the relationship between
T g and cross-linking (conversion) in polymers.
To generate the theoretical curves, T g and conversion values
corresponding to the same curing conditions were experimentally
obtained for unexposed (control) conditions. The parameter
DC p1 =DC p0 was determined from the best line fit for experimental
T g values corresponding to increasing conversion. In this study,
only three points were used to fit the theoretical curve; three
points were deemed sufficient given the fact that the area of inter-
est was a very narrow region of the theoretical curve (correspond-
ing to 0.9–1.0 conversion). The approach taken in this study, was
compared to a more rigorous curve fitting approach (based on 19
experimental data points) performed by Choi and Douglas [17]
for the same epoxy system; both approaches yielded similar
results for the narrow region that was of interest in this study.
Fig. 9 shows the decrease in T g of samples after 2 and 8 weeks of
conditioning, from their respective theoretical curves, thus indicat-
ing the effects of plasticization. There appears to be little difference
between the reductions in T g of specimens conditioned in water
immersion and humidity at 30 °C. Conversely, while samples condi-
tioned in water immersion and humidity at 60 °C exhibited similar
conversion levels, the immersion samples experienced significantly
higher decrease in T g . Even though this evidence suggests that 60 °C
687
Maximum Service Temperature (ºC)
80
70 Min. Allowed Service Temperature
Max. Allowed Service Temperature
60
50
40
30
20
10
0
A B C D E F A B C D E F
ACI 440.2R AASHTO FRPS-1
Fig. 10. Maximum and minimum service temperatures calculated from experi-
mental data; shaded region represents typical maximum design temperature range.
immersion samples would likely have higher water absorption than
the humidity samples, that is not the case (Fig. 4); FTIR water
absorption measurements may not be directly proportional to plas-
ticization, as the amount of pseudo cross-linking caused by hydro-
gen bonding of water molecules is not clear.
6.4. Maximum service temperature
ACI 440.2R [9] does not recommend application of composites
at service temperatures that are equal to or higher than Tg – 15 °
C. AASHTO FRPS-1 imposes a more conservative limit on maximum
sented T g values of representative epoxy adhesives used in civil
infrastructure, the allowed maximum service temperature ranges
are presented in Fig. 10, based on ACI 440.2R and AASHTO FRPS-
ABCD determined from measured T g values for epoxies used in this study
Humidity are within, or lower than the typical maximum design range of
60ºC
temperatures for concrete bridges in the U.S. (40–50 °C) [7].
7. Summary and conclusions
A study was conducted to enhance the understanding of how
accelerated conditioning in hygrothermal environments affects
the properties of structural epoxies; throughout the exposure to
water immersion and humidity at temperatures below (30 °C)
and above (60 °C) the Tg of control group, evolution of Tg for six dif-
ferent epoxy formulations were measured by DSC. Additionally,
conversion and characteristic water contents were measured by
means of FTIR on four epoxies that were transparent to light.
Results from this research program yielded the following
conclusions:
1. Clear epoxy samples experienced increase in conversion
throughout accelerated conditioning. This increase was depen-
dent on accelerated conditioning temperature; moisture condi-
tioning (immersion or humidity) had insignificant effect on
conversion.
2. Moisture equilibrium was attained in all epoxy samples after
2 weeks for both immersion and humidity, regardless of expo-
sure temperature.
3. An increase in T g by the end of accelerated conditioning, when
compared to control (unconditioned) samples, was noted in
clear epoxies. Samples conditioned at 60 °C experienced a sig-
nificant increase in T g above the control value, while the
increase in 30 °C samples was insignificant. The increase in T g
is explained by the effects of post-cure, primarily.
688 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689
4. No direct correlation between decrease in T g and characteristic
water content measured from FTIR was observed. Even though
characteristic water content stayed constant throughout the
accelerated conditioning, T g was fluctuating. This is explained
by varying amounts of pseudo cross-linking induced by hydro-
gen bonding of water molecules.
5. Clear-opaque epoxy experienced a significant drop in T g from
the control value early in the exposure; however, it recovered
its initial T g value by the end of conditioning.
6. All clear epoxies experienced a significant decrease in T g from
their theoretical values due to exposure to hygrothermal condi-
tions, ranging from approximately 9–25 °C. The most severe
decrease in T g was observed in specimens conditioned by
immersion at 60 °C.
7. One of the tested paste epoxies demonstrated high susceptibil-
ity to degradation in properties when exposed to water immer-
sion. The T g value of exposed samples was measured at
approximately 25 °C, corresponding to a drop from control T g
of about 19 °C.
8. None of the tested epoxies are suitable for application in
bridges (maximum design temperature range 40–50 °C), as
per ACI 440.2R and AASHTO FRPS-1.
8. Recommendations
8.1. Epoxy–concrete bond tests
Based on the findings from this research, accelerated condition-
ing of FRP-concrete bond specimens shall be performed by water
immersion for 8 weeks at the maximum design temperature
expected in the field. Eight weeks of conditioning is deemed
sufficient to allow the stabilization of competing influences of
post-cure and plasticization. Moreover, 8 weeks was found to be
sufficient to allow the full saturation of FRP–concrete bondline
by previous researchers [14].
Accelerated conditioning of epoxy caused changes in T g , from
the initial value, measured prior to exposure. Even though acceler-
ated conditioning may be conducted at temperatures well below
the T g reported by the manufacturer, it should be confirmed that
T g of exposed samples did not drop below the accelerated condi-
tioning temperature during the conditioning. It was found, in this
study, that decrease in T g due to moisture conditioning can be
quite dramatic.
8.2. Field application
Following hygrothermal conditioning the T g of epoxy E reduced
to 25 °C (77 °F), and in the other epoxies it fell in the range
between 45 and 54 °C (113–129 °F). These T g values are well
within the temperature range experienced in many field applica-
tions. It is evident that structural epoxies with T g values near or
lower than the design temperature conditions should not be used
where water immersion and/or high levels of humidity are
expected. Designer should not rely on the T g values reported by
the manufacturer, as it was shown that T g is very sensitive to the
conditioning environment. The design T g value is to be determined
experimentally for epoxy samples cured in conditions representa-
tive of the actual design environment.
References
[1] R.A. Atadero, D.G. Allen, O.R. Mata, Long-term Monitoring of Mechanical
Properties of FRP Repair Materials, Colorado Department of Transportation,
Report No. CDOT-2013-13, 2013.
[2] Z. Karim, D. Mela, M. Di Benedetti, A. Nanni, Durability Evaluation of Shear
Bond Strength of Carbon and Glass FRP Laminates Using Small Plain Concrete
Beams, The Composites and Advanced Materials Expo (CAMX 2014), Orlando,
FL, 2014.
[3] J. Tatar, H.R. Hamilton, Bond durability factor for externally bonded CFRP
systems in concrete structures, ASCE J. Compos. Constr. (2015), http://dx.doi.
org/10.1061/(ASCE)CC.1943-5614.0000587.
[4] Y. Quan, G. Xian, V.M. Karbhari, Hygrothermal ageing of and epoxy adhesive
used in FRP strengthening of concrete, J. Appl. Polym. Sci. 107 (2008) 2607–
2617.
[5] M. Lattieri, M. Frigione, Effects of humid environment on thermal and
mechanical properties of a cold-curing structural epoxy adhesive, Constr.
Build. Mater. 30 (2012) 753–760.
[6] K. Benzarti, S. Chataigner, M. Quiertant, C. Marty, C. Aubagnac, Accelerated
ageing behavior of the adhesive bond between concrete specimens and CFRP
overlays, Constr. Build. Mater. 25 (2) (2011) 525–538.
[7] AASHTO, LRFD Bridge Design Specifications, American Association of State
Highway and Transportation Officials, Washington, D.C., 2015.
[8] AASHTO, Guide Specifications for Design of Bonded FRP Systems for Repair and
Strengthening of Concrete Bridge Elements, AASHTO FRPS-1, American
Association of State Highway and Transportation Officials, Washington, D.C.,
2010.
[9] ACI Committee 440, Guide for the Design and Construction of Externally
Bonded FRP Systems for Strengthening Concrete Structures, ACI 440.2R-08,
American Concrete Institute, Farmington Hills, MI, 2008.
[10] ACI Committee 440, Specification for Carbon and Glass Fiber-Reinforced
Polymer (FRP) Materials Made by Wet Layup for External Strengthening of
Concrete and Masonry Structures, ACI 440.8M-13, American Concrete
Institute, Farmington Hills, MI, 2013.
[11] ASTM Standard D648, Standard Test Method for Deflection Temperature of
Plastics under Flexural Load in the Edgewise Position, ASTM International,
West Conshohocken, Pa, 2007.
[12] ASTM Standard E1356, Standard Test Method for Assignment of the Glass
Transition Temperatures by Differential Scanning Calorimetry, ASTM
International, West Conshohocken, Pa, 2008.
[13] ASTM Standard D4065, Standard Practice for Plastics: Dynamic Mechanical
Properties: Determination and Report of Procedures, ASTM International, West
Conshohocken, Pa, 2012.
[14] C. Au, O. Buyukozturk, Peel and shear fracture characterization of debonding
in FRP plated concrete affected by moisture, J. Compos. Constr. 10 (1) (2006)
35–47.
[15] C.E. Bakis, L.A. Bisby, M.M. Lopez, S.E. Witt, T. Alkhraji, Interlaboratory
evaluation of Tg of ambient-cured epoxies used in civil infrastructure, in:
Proceedings of The 7th International Conference on FRP Composites in Civil
Engineering (CICE 2014), International Institute for FRP in Construction (IIFC),
Vancouver, British Colombia, Canada, 2014.
[16] S. Choi, Study of Hygrothermal Effects and Cure Kinetics on the Structure–
Property Relations of Epoxy-Amine Thermosets: Fundamental Analysis and
Application, Master’s Thesis. University of Florida, Gainesville, FL.
[17] S. Choi, E.P. Douglas, Complex hygrothermal effects on the glass transition of
an epoxy-amine thermoset, Appl. Mater. Interfaces 2 (3) (2010) 934–941.
[18] P.R. Couchman, F.E. Karasz, A classical thermodynamic discussion of the effect
of composition on glass transistion temperatures, Macromolecules 11 (1)
(1978) 117–119.
[19] F. Djouani, C. Connan, M. Delamar, M.M. Chehimi, K. Benzarti, Cement paste–
epoxy adhesive interactions, Constr. Build. Mater. 25 (2) (2011) 411–423.
[20] H. Dannenberg, Determination of functional groups in epoxy resins by near-
infrared spectroscopy, SPE Trans. 3 (1963) 78–88.
[21] C.W. Dolan, J. Tanner, D. Mukai, H.R. Hamilton, E. Douglas, Design Guidelines
for Durability of Bonded CFRP Repair/Strengthening of Concrete Beams,
NCHRP Web-Only Document 155, 2008.
[22] E. Douglas, Introduction to Materials Science and Engineering: A Guided
Inquiry, Pearson Higher Education Inc, Upper Saddle River, NJ, 2013.
[23] M. Frigione, M.A. Aiello, C. Naddeo, Water effects on the bond strength of
concrete/concrete adhesive joints, Constr. Build. Mater. 20 (2006)
957–970.
[24] A. Gartner, E. Douglas, C. Dolan, H. Hamilton, Small beam bond test method for
CFRP composites applied to concrete, J. Compos. Constr. 10 (6) (2011) 52–61.
[25] E.H. Immergut, H.F. Mark, Principles of plasticization. Plasticization and
plasticizer processes, Adv. Chem. 48 (1965) 1–26.
[26] International Code Council (ICC), Acceptance Criteria for Concrete and
Reinforced and Unreinforced Masonry Strengthening using Externally
Bonded FRP Composite Systems, ICC AC125, ICC Evaluation Service, Whittier,
CA, 2012.
[27] W.S. Kim, I.H. Yun, J.J. Lee, H.T. Jung, Evaluation of mechanical interlock effect
on adhesion strength of polymer–metal interfaces using micro-patterned
surface topography, Int. J. Adhes. Adhes. 30 (6) (2010) 408–417.
[28] O. Moussa, A.P. Vassilopoulos, D. Castro, T. Keller, Mechanical recovery of
epoxy adhesive subsequent to exceeding glass transition temperature, in: 4th
International Conference on Durability of Composites for Construction (CDCC
2011), University of Sherbrooke, Quebec, Canada, 2011.
[29] A. Stewart, Study of Cement-Epoxy Interfaces, Accelerated Testing, and
Surface Modification, Ph.D. Dissertation, University of Florida, Gainesville, FL,
2012.
[30] J. Tatar, P. Blackburn, C. Weston, H.R. Hamilton, Direct Shear Adhesive Bond
Test, in: Proceedings of the 11th International Symposium of Fiber Reinforced
Polymer for Reinforced Concrete Structures (FRPRCS-11), University of Minho,
Guimaraes, Portugal, 2013.
 B.P. Blackburn et al. / Construction and Building Materials 96 (2015) 679–689 689

[31] J.M. Zhou, J.P. Lucas, Hygrothermal effects of epoxy resin. Part I: The nature of
water in epoxy. Part II: Variations of glass transition temperature, Polymer 40
(20) (1999) 5505–5522.
[32] ASTM, Standard D638, Standard Test Method for Tensile Properties of Plastics,
ASTM International, West Conshohocken, Pa, 2014.
[33] S. Choi, A.L. Gartner, N.V. Etten, H.R. Hamilton, E.P. Douglas, Durability of
concrete beams externally reinforced with CFRP composites exposed to
various environments, J. Compos. Constr. 16 (2012) 10–20.
[34] B.P. Blackburn, Effects of Hygrothermal Conditioning on Epoxy Used in FRP
Composites. M.S. Thesis, University of Florida, Gainesville, FL, 2013.