Uew

PCL

a

ARRAA

KEAPWC

1

idrtci(caepobpe

bŁ

0h

Industrial Crops and Products 55 (2014) 180–186

Contents lists available at ScienceDirect

Industrial Crops and Products

jo u r n al homep age: www.elsev ier .com/ locate / indcrop

nconventional plasticization threshold for a biobased bisphenol-Apoxy substitution candidate displaying improved adhesion andater-resistance

ierre Verge ∗, Valérie Toniazzo, David Ruch, João A.S. Bomfimentre de Recherche Public Henri Tudor, Advanced Materials & Structures, 5 Rue Bommel, Z.A.E. Robert Steichen, Kaërjeng-Hautcharage L-4940,uxembourg

r t i c l e i n f o

rticle history:eceived 22 August 2013eceived in revised form 20 January 2014ccepted 30 January 2014vailable online 12 March 2014

a b s t r a c t

Toxicity concerns over bisphenol A (BPA) based materials like polycarbonates or diglycidyl ether ofbisphenol A (BADGE) lead to their progressive substitution in food packaging and can coating applications,even if the eventual release of BPA in food remains debatable. In this paper, we study the adhesive, waterresistance and thermo-mechanical properties of a bio-based BADGE substitution candidate issued fromCardanol (CardDE). Different formulations with increasing ratio of BADGE substitution were evaluated,

eywords:poxydhesivelasticizationater resistance

NSL

highlighting a real prospect for BPA-based epoxy substitution. During these studies, an unconventionalrelationship between water uptake by the biobased adhesive and property loss due to plasticization wasevidenced. The water resistance of CardDE adhesives was investigated in detail revealing a plasticizationthreshold at elevated water content, attributed to their hydrophobic structure. While the adhesive prop-erties of conventional BADGE networks are prematurely affected, the deterioration of CardDE adhesivesis delayed.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

Bisphenol A [2,2-bis(4-hydroxyphenyl)propane or BPA] is anmportant industrial chemical that is primarily used as an interme-iate in the production of polycarbonate (PC) plastics and epoxyesins (Staples et al., 1998). Products of BPA range from PC bot-les to the inner coating of cans and from dental sealants toell phone casings. Controversy surrounding BPA mainly involvests potential human health risk due to its estrogenic activitiesMaffini et al., 2006; Tsai, 2006). As the release of BPA fromoatings has been evidenced (Nam et al., 2010), a strong urgencyppears to its substitution in sensitive applications by moreco and human friendly compounds. Renewable resources canrovide an interesting sustainable platform to substitute, partiallyr totally, petroleum-based polymers through the design of bio-ased polymers that can compete or even surpass the existingetroleum-based materials on a cost-performance basis with highco-friendliness values (Raquez et al., 2010).

Literature review reveals several attempts to obtain bio-ased epoxy resins from isosorbide (Chrysanthos et al., 2011;ukaszczyk et al., 2011) or from vegetable oils. Many vegetable

∗ Corresponding author. Tel.: +352 42 5991 4943.E-mail address: pierre.verge@tudor.lu (P. Verge).

926-6690/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.indcrop.2014.01.048

oils display high levels of unsaturated fatty acids that can beconverted into epoxy fatty acids by conventional double-bondoxidation reactions. Nowadays, epoxydized vegetable oils arereceiving a great deal of attention as they are obtained fromsustainable and renewable natural resources and are environmen-tally friendly. Nevertheless, as these oils are obtained from foodresources, or as their production would need farmlands other-wise dedicated to growing food resources and despite a potentiallypositive environmental balance, biosourced materials which cancompete with food resources are deemed unethical raw mate-rials (Pimentel et al., 2009). Clearly, truly sustainable, societallyacceptable bio-sourced materials should rely on non-alimentaryresources, focusing as much as possible on waste materials orby-products.

In this paper, we explore the potential of a derivative ofcashew nutshell liquid (CNSL) as an alternative to BPA-derivedepoxy. CNSL, which main constituent is Cardanol – an unsaturatedmeta-alkylphenol, is a by-product from the processing of cashewnutshells and thus does not compete with the production of foodresources (Mele et al., 2010). Few works treat of the use of Cardanol-based products for the conception of thermosets, in spite of the

non-allergenic, non-estrogenic and non-toxic properties of Car-danol and even it is considered for the development of anti-cancerdrugs (Hemshekhar et al., 2012). In (Aggarwal et al., 2007), the anti-corrosive properties of a mono-epoxidized Cardanol chemical is

P. Verge et al. / Industrial Crops and

Sa

eatwaidmtcrf

dfiaaahwsol

rnlmawnp

2

brbDtr

2

cEuta

the thermal resistance is reduced, the observed transition values for

cheme 1. Molecular structures of the tri-unsaturated component of cardanol (I)nd of CardDE (II).

valuated. The mono-functionality of the molecules limits its uses a grafted plasticizer. In (Huang et al., 2012), the authors reporthe synthesis of a Cardanol-based amine – as called phenalkamine –ith good properties. Nevertheless, the use of phenalkamine is just

n option to replace amines, but not diglycidyl ether of BPA. Finally,n (Maffezzoli et al., 2004), the authors report on the use of a Car-anol derivative-Stabilcardo – as a base compound for thermosetsatrix. Nevertheless, the epoxidization relies on the treatment of

he unsaturations of the alkyl chain, hindering the interest of suchhemical structure (Raquez et al., 2010). Formaldehyde-cardanolesins are not discussed here due to the well-known toxicity oformaldehyde (Feron et al., 1991).

In this work we are testing a Cashew Nutshell Liquid basedi-epoxy resin (CardDE, Scheme 1), a commercially available di-unctional glycidyl ether which chemical structure is reportedn Scheme 1, as a BADGE substituent. Its chemical structureffords a higher flexibility (due to C8 alkyl chain separating theromatic groups) and water resistance (due to the C7 pendinglkyl chain) than traditional BPA thermosets epoxy. Moreover, itsydrophobic character is a potential advantage for moisture- orater-exposed applications (such as packaging or outdoor uses)

ince anti-corrosive properties are strongly expected from this kindf coating (Aggarwal et al., 2007), preventing degradation of metal-ic outer layers or leaching of metallic ions.

First, we evaluated the thermochemical, adhesive and wateresistance and properties of neat CardDE and BADGE/CardDE co-etworks with different rate of BADGE substitution by CardDE,

eading to the conclusion that CardDE has a strong potential whenixed with a small part of BADGE, and even alone for some specific

pplications. More particularly, the water resistance of the adhesiveas investigated to shed more light on the water uptake mecha-isms occurring in these systems and its consequences onto theroperties of the adhesives.

. Experimental

CardDE (commercial name NC514) was kindly suppliedy Cardolite. CardDE is a di-functional glycidyl ether epoxyesin (EEW = 190 g/equiv.) with similar reactivity than traditionalisphenol A type resin. Diglycidyl ether of bisphenol A (BADGE,ER332, EEW = 175 g/equiv.) and 1,2-diaminocyclohexane (mix-

ure of isomers) were purchased from Sigma–Aldrich and used aseceived.

.1. Networks and composites preparation

For each composition, the equivalent epoxy weight EEW wasalculated to determine the stoichiometric amount of hardener.ach epoxy mixture was heated and magnetically stirred at 80 ◦C

nder vacuum for 1 h. The hardener was then added and the solu-ion incorporated into a high temperature silicone mold for curingt 160 ◦C for 2 h followed by a post-cure treatment at 190 ◦C for 2 h.

Products 55 (2014) 180–186 181

2.2. Pull-off tests

Pull-off tests (PoT) were performed as described in publishedstandards (D4541-95e1) using a force-controlled, self-alignedDeFelsko AT-A adhesion tester at 0.2 MPa s−1 (Petersen et al., 2012).Adhesive “dolly-plate” bonded assemblies were prepared usingstandard 20 mm Al dollies. Aluminum plates were degreased withacetone, before applying the composition to test. Once the surfacewas dried, a droplet of the adhesive composition was applied andthe dolly set down on it. Assemblies were cured at 160 ◦C for 2 h,and then post-cured at 190 ◦C for 2 h. Each composition was testedten times.

2.3. Water resistance tests

The water resistance tests were adapted from “wet cata-plasm” published ASTM standards. Al “dolly-plate” assemblieswere wrapped in 10 g of dry cotton. 100 ml of deionized water werethen added to the cotton layer and the wet assemblies were placedin a polyethylene pack and sealed. A thermal treatment of 24 h at80 ◦C was performed before placing the sample at −20 ◦C during4 h. The samples were let stand for 24 h before performing Pull-offTests (PoT).

2.4. Differential scanning calorimetry (DSC)

Differential scanning calorimetry thermograms of compositeswere recorded by means of a Netzsch DSC 204 F1 Phoenix apparatusoperating in inert atmosphere (nitrogen), and with a heating rateof 20 ◦C min−1. The samples were weighted (5–10 mg) and placedinto alumina crucibles and were subjected to the following heatingstep: from 30 to 250 ◦C.

2.5. Dynamic mechanical analysis (DMA)

DMA tests were performed on a dynamic mechanical analyzerNetzsch DMA 242C. For this study, 20 mm × 8 mm × 1 mm rectan-gular specimens cut using a precision cutting saw from moldedspecimens were used. They were subjected to sample cantilevermode of flexural loading with a maximum displacement of 40 �min the temperature range −50 ◦C to 250 ◦C (rate 10 ◦C min−1) and afrequency of 1 Hz.

2.6. Swelling tests

Samples of 1 cm × 1 cm × 1 mm approximately were cut fromcomposites prepared as described previously. The surfaces wereslightly polished to remove any over oxidation that occurs on thesurface during the thermal treatment. The water uptakes weremeasured by weighting the samples at different time. A particu-lar attention was paid to remove any solvent excess (i.e. water ortoluene) at the surface of the samples.

3. Results and discussion

The thermo-mechanical properties of the networks were evalu-ated by means of dynamic mechanical analysis (Fig. 1). The E′ valuesfor all 5 samples at temperatures far from the Tg are within a 10%range.

As expected, the incorporation of CardDE in BADGE networksleads inevitably to the decrease of the Tg of the networks, as a resultof the higher flexibility of the bio-diepoxy. However, even though

CardDE-containing materials are above room temperature, slightlyso for neat CardDE (40 ◦C) but considerably more for the 70% bio-based formulation (90 ◦C).

182 P. Verge et al. / Industrial Crops and

Fig. 1. Dynamic mechanical analysis curves of BADGE/CardDE {x/100 − x} networks.

Table 1Summary of the thermo-mechanical properties of BADGE substituted networks.

RatioBADGE/CardDE

T˛ (◦C) E′ (0 ◦C)(GPa)

E′T˛ (+30 ◦C)(GPa)

E′ (30 ◦C)(GPa)

100–0 185 ≥4.2 1.2 4.270–30 120 4.5 1.5 4.250–50 115 4.2 1 3.730–70 90 4 1.4 3.8

noaE

iclTmrwsu

materials; indeed, up to 50 wt% substitution, the water absorption

0–100 40 4 1.2 2.7

Regarding the elastic modulus of the samples, it is surprising toote that the presence of CardDE does not really impact the rigidityf the systems in the glassy state, whatever the considered temper-ture range. To help the comparison, the representative values of′ from Fig. 1 are reported in Table 1.

Even if Tg is affected, the elastic moduli remain similar. Fornstance, at 0 ◦C, the elastic modulus reaches 4–4.5 GPa for eachomposition (Table 1, column 3). Above the Tg, the elastic modu-us keeps a very interesting value above 1 GPa (Table 1, column 4).he adhesive properties of the new (partially) bio-based epoxy for-ulations were assessed using the “dolly-plate” assemblies. Fig. 2

eports the evolution of the adhesive strength of BADGE/CardDE

ith different amounts of BADGE substitution. The adhesive

trength was determined by PoT, and the water resistance eval-ation was performed by PoT after wet cataplasm tests.

Fig. 2. Adhesive properties before a

Products 55 (2014) 180–186

For each sample, it is worthy to note that a similar adhesiverupture is observed. Neat BADGE adhesive has a joint strengthof around 3.5 MPa, as it is generally reported in the literature(Zhai et al., 2006). Higher strengths are however observed withneat CardDE networks, reaching a value of 6 MPa. Such behav-ior is deemed a direct consequence of a better surface wettabilitywhen CardDE is used, yielding stronger aluminum-epoxy joints, asconfirmed by contact angle tests of BADGE and CardDE onto anAluminum substrate. From elliptic calculation method, BADGE-Alcontact angle is equal to 66 ± 2◦, while CardDE-Al contact angle isaround 51 ± 1◦, evidencing a better wettability of Cardanol-basedadhesive toward Al. the wettability and the adhesion strength aredirectly linked and explain the better adhesive strength of the Al-to-Al assembly with CardDE. After wet cataplasm tests, the neatbio-sourced adhesive performs even better than neat BADGE. Asthe latter loses 35% of its adhesive strength (dropping to 2.1 MPaits “stress to break”), CardDE adhesive only loses 21% of its initialstrength, with a stress to break around 4.5 MPa, higher than theinitial value of BADGE.

However, a partial substitution of BADGE by CardDE does notlead to a substantial improvement of the adhesive strength, indi-cating that this property is mainly governed by BADGE. Indeed, upto 50 wt% loading of bio-based adhesive, nor the initial adhesionstrength or the water resistance are really modified. However, athigher CardDE loadings, i.e. when 70 wt% is substituted, a remark-able improvement is observed, with only a decrease of 17% of theadhesive strength after the water aging test, while 35% of the adhe-sive strength is lost for neat BADGE networks after the similar tests.

It is clear that the chemical structure of CardDE-based moleculescould be considered as the driving force of the better water resis-tance of the resulting adhesive, preventing water from diffusing inthe joint and keeping a more stable assembly toward water. Nev-ertheless, diverging results were obtained from water absorptionmeasurements. These measurements in H2O at 80 ◦C were per-formed to check if the water diffusion was really obstructed by thechemical nature of CardDE. The evolution of these water uptakesis reported in Fig. 3, for bulk materials with progressive BADGEsubstitution by CardDE.

Surprisingly, and despite its hydrophobic nature, CardDE-basednetworks are up taking more water than conventional BADGE based

stays constant around 1 ± 0.2 wt%. Beyond 50 wt% substitution, thewater uptake increases, reaching the value of 5.5 ± 0.2 wt% for neatCardDE networks.

nd after wet cataplasm tests.

P. Verge et al. / Industrial Crops and

F

riaB

sfo

(

arHa

rhato

bd

ig. 3. Swelling tests in H2O at 80 ◦C of BADGE-CardDE {x/100 − x} bulk materials.

It is worthy to note that the water uptake is generally directlyelated to the water resistance of an adhesive joint, and it is surpris-ng to observe that in spite of this significant water uptake, CardDEdhesive still behaves better after wet cataplasm tests than neatADGE networks.

The origin of the good water resistance of CardDE-based adhe-ive joints was then investigated more in details considering theour main phenomena of adhesive properties drop in the presencef water:

(I) Post-cure reaction caused by water at elevated temperatures;(II) Effect of the hydrolytic degradation kinetic (Xiao and

Shanahan, 1997)(III) Effect of the lower cross-linking density at the interface adhe-

sive/metal;IV) Effect of the plasticization of the adhesives by water.

In previous works (reviewed by Choi and Douglas, 2010), somenomalous results were explained by taking into account post-cureeactions promoted by hot water used in plasticization studies.owever, the samples used in this study are cured well above Tg

nd DSC analysis confirms they are 100% cross-linked.As for hydrolytic degradation, wet cataplasm tests are short-

ange. The samples are exposed to water only during 72 h. Theydrolytic degradation of BADGE resins is generally consideredfter a couple of week, even at 90 ◦C (Xiao and Shanahan, 1997) andhus assumption (I) could not explain the better water resistance

f CardDE joints.

It is well-known that the main weakness of adhesive/metalonds relies on their interface, where the crosslinking density isecreased by the chemical anchoring of the adhesives networks

Fig. 4. Water uptake of BADGE (on the left) and NC514 (on the rig

Products 55 (2014) 180–186 183

chains onto the Al surface. This lowering in the cross-linking den-sity promotes the water diffusion within the network meshes, andfinally weakens the properties of the interface.

To study the possible effect of the lower cross-linking den-sity of the adhesives at the interface with metal, severalnon-stoichiometric bulk materials were prepared, with harde-ners content below 1 equiv. (0.95, 0.9 and 0.8 equiv.), and theirswelling behavior in water was investigated. To achieve differ-ent cross-linking densities, modifying the epoxy-hardener ratiowas preferred to applying different cure conditions to avoid prob-lems related to post-cure reactions (Marks and Snelgrove, 2009);moreover, limiting hardener content should best reproduce actualconditions in epoxy adhesive joints (Bockenheimer et al., 2002).Fig. 4 represents the evolution of the swelling rates mt − m0/m0 asa function of the stoichiometric ratio of BADGE networks (on theleft) and CardDE networks (on the right).

In both cases, it can be deducted that decreasing the amount ofhardener leads to a higher permeability of the networks towardwater, since the swelling ratios are increasing. For instance, inthe case of BADGE networks, shifting the stoichiometric ratiofrom 1 to 0.8 leads to an increase in the water absorption, from1 ± 0.2 wt% to 2.2 ± 0.2 wt%, after 115 h of exposure. Nevertheless,it is clear that in the case of CardDE, the water uptake seems tobe strongly more dependent on the network structure than forBADGE. It is worthy to note that decreasing the amount of hard-ener will lead to free epoxy pending groups in both networks (i.e.BADGE and CardDE based networks), which can undergo wateraddition through hydrolysis and 1,2-diols formation. Neverthe-less, regarding the results from BADGE networks, this ring-openingreaction does not lead to an appreciable mass increase, although itcan render non-stoichiometric networks more hydrophilic.

For the comparison of conventional epoxy and CardDE-basednetworks, it has to be considered that the cross-linking densitiesare not similar, the bio-based network mesh size being higherthan BADGE one. The average molar mass between the crosslinkingnodes Mc, representative of the crosslinking density of a polymernetwork were calculated by performing swelling tests in tolueneand by applying the Flory-Rehner equation (Eq. (1)):

Mc = Vmol · (�/2) − �1/3

ln(1 − �) + � + X�2(1)

where Vmol is the molar volume of toluene (106.3 mL/mol); � iscalculated from Eq. (2):

� = 1(1 + (dblend/dtoluene) ∗ ((mswollen/m0) − 1))

(2)

where dblend and dtoluene are the densities of the networks and thesolvent, respectively.

X is calculated from Eq. (3)

X = 0.34 + Vmol

RT(ıtoluene − ıi)2

(3)

ht) networks depending on the stoichiometry of the system.

184 P. Verge et al. / Industrial Crops and Products 55 (2014) 180–186

Table 2Summary of average molar mass between crosslinking nodes Mc , water diffusioncoefficients D and ratios of D and Mc as a function of the nature (i.e. BADGE orCardDE), the stoichiometry of the networks, and the initial mass m0 of the samples.

Mc (g/mol) D (109 cm2/s) D.m0/Mc (109 mol cm2/s)

BADGE1 100 12 5.3 × 10−2

0.95 90 24 2.2 × 10−1

0.9 110 19 7.8 × 10−2

0.8 130 45 2.2 × 10−1

CardDE1 1080 10 2.7 × 10−3

0.95 970 32 1.2 × 10−2

ıf(W

ncoiretMcmc1f0ns

ea

ica

wsa

h

as

D

c11t

0.9 1130 108 5.4 × 10−2

0.8 1440 110 3.4 × 10−2

toluene = 18.3 J1/2 cm−3/2. ıi is the Hildebrand solubility parameteror each network and ıCardDE is approximated to 19.2 J1/2 cm−3/2

Greco et al., 2010) and ıBADGE = 20.52 J1/2 cm−3/2 (Ruseckaite andilliams, 1993).The calculated Mc are reported in Table 2.It is worthy to note that if low Mc are found for BADGE/DAC

etworks, even lower than the molar mass of DGEBA, it must beonsidered that the multifunctional amine hardeners are also a partf the network and thus decrease the average Mc of the result-ng materials. Lower Mc than DGEBA Mn have previously beeneported, for instance in Wang and Schuman (2013) or Amaralt al. (2013) where the amines used have still a higher Mn thanhe 1,2-diaminocyclohexane (DAC) used in this study. As expected,

c increases with decreasing the hardener content, whatever theonsidered network. Nevertheless, it is noteworthy that the Mc areore affected in the case of CardDE-based networks that in the

ase of BADGE networks, since the Mc shifts from 1080 g/mol to440 g/mol for CardDE networks, and from 100 g/mol to 130 g/molor BADGE networks while the stoichiometric rate shift from 1 to.8 (Table 2, column 2). This wide size of the meshes in CardDEetworks is directly responsible of the higher water absorption ataturation.

A way to overcome the effect of the meshes size is to compare thevolution of the diffusion coefficient of water inside the networkss a function of Mc.

It is generally accepted that the behavior of sorption curvesn epoxy conforms to Fickian diffusion in the initial sorption pro-ess, with the diffusion coefficients fitting to the following equationccording to Fick’s second law:

M(t)M∞

= 1 −∑∞

n=0

8

(2n + 1)2�2exp

(−D(2n + 1)2�2t

4e

)(4)

here D (cm2/s) is the diffusion coefficient, e is the thickness of theample (cm) and M∞ is the equilibrium value of the diffusing watert infinite time.

A simplest form of Eq. (4) for values of M(t)/M∞ lower than 0.5as the following form (Frisch, 1968):

M(t)M∞ = 4

e∗√

Dt

�(6)

nd the diffusion coefficients can be directly extracted from thelope a of the plot of M(t) = f(

√(t)):

= �(

a ∗ e

4 ∗ M∞

)2(7)

The values of D are reported in Table 2 (column 3), and indi-

ates two main trends: (i) D increases with Mc (e.g. it shifts from2 cm2/s to 45 cm2/s for BADGE networks while Mc shifts from00 g/mol to 130 g/mol, respectively) and (ii) are similar whateverhe chemical structure of the network (e.g. 12 and 10 × 10−9 cm2/s

Fig. 5. Evolution of the D/Mc ratio of BADGE and CardDE networks as a function ofthe stoichiometric ratio.

for BADGE and CardDE stoichiometric networks, respectively), evenif the increase of Mc is widely more significant in CardDE networkscompared to BADGE networks when decreasing the stoichiometricratio (100 g/mol and 1080 g/mol for BADGE and CardDE networks,respectively).

In other words, the diffusion coefficients are more affected bythe decrease of the cross-linking density in BADGE networks thanin CardDE networks, as evidenced when comparing the ratio of Dand Mc-reported to m0 to avoid the effect of the initial mass of thesamples (Fig. 5). Indeed, these ratios are at least 10 times higher forBADGE networks.

This means that at a similar network mesh size, the water diffu-sion, and thus the water absorption, is lower in CardDE networksthan in BADGE networks.

As previously reported by Shanahan et al. (Popineau et al., 2005),the first step of water absorption in an epoxy network is driven bythe polar sites of the polymer chains; the higher the polar sitesamounts, and the faster the water diffusion. It is clear that propor-tionally, the number of polar sites is lower in the case of CardDEnetwork than conventional epoxy, and the results obtained fromthe swelling tests in water match perfectly with this prediction.

Even if these results still do not explain directly the origin ofthe better water resistance of CardDE-based adhesive joints, con-crete assumptions could be drawn thanks to the water swellingmechanisms.

Indeed, by comparing the water diffusion between CardDE andBADGE networks, it can be noted that the water uptake in BADGE isdriven by the hydrophilicity of the network, as previously reportedin the literature (Mensitieri et al., 2006), and such interactions areknown to plasticize a polymer network (Zinck and Gérard, 2008).

It is then reasonable to assume that the water diffusing throughthe BADGE networks would lead to a plasticization at lower ratethan in the case of CardDE networks. It is worthy to note that suchplasticization phenomena, even if partially reversible, are known todegrade the adhesive properties of metal-adhesive-metal assem-blies (Doyle and Pethrick, 2009), and could be a good explanationof the better water resistance of CarDE-based joints.

To confirm this assumption, differential scanning calorimetrymeasurements were performed to evaluate the plasticizing effectof water either on BADGE or CardDE networks. The DSC curves ofstoichiometric networks swollen or not with water are reported inFig. 6. Tg values measured by DSC are summarized in Table 3. It is

worthy to note that the water swelling has been experimented to fitwith the wet cataplasm tests: the water exposure was performedduring 72 h at 80 ◦C, i.e. in the same experimental conditions thanfor the wet cataplasm tests.

P. Verge et al. / Industrial Crops and Products 55 (2014) 180–186 185

Fig. 6. DSC curve of BADGE (on the left) and NC514 (on the right) networks before and after swelling tests in water at 80 ◦C.

Fig. 7. Schematic representation of the interaction and repartition of water mole

Table 3Summary of the Tg before and after swelling tests in water, and the correspondingwater uptakes for BADGE and CardDE networks.

Stoichiometric ratio Tg (dry, ◦C) Tg (swollen, ◦C) Tg (◦C) wt% of water

BADGE1 190 180 10 1.50.9 172 165 7 2.00.8 110 101 9 2.4

CardDE1 48 49 1 60.9 45 44 1 14

lmitac6TtwisatC

tafmuib1

ing to better water resistance, in spite of a significant water uptake.

0.8 40 33 7 21

As expected, BADGE networks are plasticized by water even atow content. For instance, a shift of 10 ◦C is observed for stoichio-

etric BADGE networks swollen by only 1.5 ± 0.2 wt% of water, ast is generally reported in the literature, even if the Tg are higherhan water evaporation temperature (Jelinski et al., 1985; Zincknd Gérard, 2008). On the contrary, no Tg shift is observed for stoi-hiometric NC514 networks, while the water uptake is higher to

± 0.2 wt% (Table 3, row 4, column 8). Even up to 14 ± 0.2 wt%, theg does not seem to be affected by the presence of water. First plas-icization phenomena are just observed for networks swollen byater up to 21 ± 0.2% of their weight (Table 3, row 6, column 8). It

s clear that the impact of water is not similar for both networks. Thehift of the Tg to lower temperature, i.e. from 190 to 180 ◦C, is char-cteristic of the plasticization of the polymer network by water. Onhe contrary, and even if they are swollen with 6 ± 0.2 wt% water,ardDE networks do not present any plasticization by water.

To determine if a critical amount of water is necessary to plas-icize CardDE networks, the similar experiments were performedt lower stoichiometric rates (similar materials than those usedor the swelling tests in water). The values of the Tg are sum-

arized in Table 3 (columns 3 and 7). The corresponding waterptakes are also reported in columns 5 and 9 to help the compar-

son. Clearly, there is no plasticization of cashew nutshell liquidased networks before a significant absorption of water (between4 and 21 ± 0.2 wt% water in this case) while an immediate

cules in BADGE networks (on the left) and CardDE networks (on the right).

plasticization occurs in BADGE networks, even at low water content(i.e. 1.5 ± 0.2 wt%).

A reasonable interpretation could be given to schematize at amolecular scale the water interaction in conventional and CardDE-based adhesives (Fig. 7).

In a conventional BADGE network, it is accepted that (i) watersolvates the polar sites, leading to the plasticization of the polymernetwork. (ii) Additional water absorption will lead to an additionalplasticization, if any (i.e. if network plasticization is still possible).In the case of CardDE networks, mesh sizes are broad and particu-larly not affine with water. The typical first step (i), driven by polarinteractions (Jelinski et al., 1985), is then limited due to the poorpolar sites content and plasticization does not seem to occur. Waterabsorption is mainly controlled by the huge free volume inside thebio-based network meshes. Even if water is able to penetrate insidethese networks, it should be kept in clusters – or isolated – withweak interactions with polymer chains and a poor plasticizationeffect. The plasticization of the networks is then reached at highwater content.

4. Conclusion

It was demonstrated in this work that at similar stoichiometryand with the same hardener, BADGE networks are much denserthan CardDE ones, yielding materials with high Tg values, whileCardDE networks outperformed the conventional epoxy in termsof adhesion and water uptake, representing interesting alternativesfor applications close to ambient temperature or in formulations toimprove the performance of BADGE adhesives.

It was also demonstrated that, at similar network mesh size,CardDE leads to lower water diffusion than BADGE networks thanksto its hydrophobic nature. An unconventional plasticization effectwas observed for CardDE materials, with a water uptake thresholdunder which no loss in mechanical properties was observed, lead-

This specificity of water absorption is attributed to the hydropho-bic character of CardDE material, available almost exclusively frombiological sources.

1 s and

A

eAo

R

A

A

B

C

C

C

D

D

F

F

G

H

H

J

Ł

86 P. Verge et al. / Industrial Crop

cknowledgments

We thank the Fonds National de la Recherche – FNR – from Lux-mbourg, for financial support (HiperAdh, Grant number: 1178333,FR-PDR). Cardolite Specialty Chemical is thanked for the supplyf the materials used in this study.

eferences

ggarwal, L.K., Thapliyal, P.C., Karade, S.R., 2007. Anticorrosive properties of theepoxy–cardanol resin based paints. Progress in Organic Coatings 59, 76–80.

maral, C.R., Sanchez Rodriguez, R.J., Garcia, G., £Junior, L.P.B., Carvalho, E.A., 2013.Impact of aliphatic amine comonomers on DGEBA network properties. PolymerEngineering & Science, http://dx.doi.org/10.1002/pen.23751.

ockenheimer, C., Valeske, B., Possart, W., 2002. Network structure in epoxy alu-minium bonds after mechanical treatment. International Journal of Adhesionand Adhesives 22, 349–356.

hoi, S., Douglas, E.P., 2010. Complex hygrothermal effects on the glass transition ofan epoxy-amine thermoset. ACS Applied Materials & Interfaces 2, 934–941.

hrysanthos, M., Galy, J., Pascault, J.-P., 2011. Preparation and properties of bio-based epoxy networks derived from isosorbide diglycidyl ether. Polymer 52,3611–3620.

oil coated metals – Test methods – Part 27: Resistance to humid poultice (Cata-plasm test). vol. CSN EN 13523-272009.

4541-95e1, A., Standard Test Method for Pull-Off Strength of Coatings UsingPortable Adhesion Testers.

oyle, G., Pethrick, R.A., 2009. Environmental effects on the ageing of epoxy adhesivejoints. International Journal of Adhesion and Adhesives 29, 77–90.

eron, V.J., Til, H.P., de Vrijer, F., Woutersen, R.A., Cassee, F.R., van Bladeren, P.J., 1991.Aldehydes – occurrence, carcinogenic potential, mechanism of action and riskassessment. Mutation Research 259, 363–385.

risch, H.L., 1968. In: Crank, J., Park, G.S. (Eds.), Diffusion in polymers. AcademicPress, London and New York, p. 452.

reco, A., Brunetti, D., Renna, G., Mele, G., Maffezzoli, A., 2010. Plasticizer forpoly(vinyl chloride) from cardanol as a renewable resource material. PolymerDegradation and Stability 95, 2169–2174.

emshekhar, M., Sebastin Santhosh, M., Kemparaju, K., Girish, K.S., 2012. Emergingroles of anacardic acid and its derivatives: a pharmacological overview. Basic &Clinical Pharmacology & Toxicology 110, 122–132.

uang, K., Zhang, Y., Li, M., Lian, J., Yang, X., Xia, J., 2012. Preparation of a light colorcardanol-based curing agent and epoxy resin composite: cure-induced phase

separation and its effect on properties. Progress in Organic Coatings 74, 240–247.

elinski, L.W., Dumais, J.J., Cholli, A.L., Ellis, T.S., Karasz, F.E., 1985. Nature of thewater–epoxy interaction. Macromolecules 18, 1091–1095.

ukaszczyk, J., Janicki, B., Kaczmarek, M., 2011. Synthesis and properties of isosorbidebased epoxy resin. European Polymer Journal 47, 1601–1606.

Products 55 (2014) 180–186

Maffezzoli, A., Calò, E., Zurlo, S., Mele, G., Tarzia, A., Stifani, C., 2004. Cardanol basedmatrix biocomposites reinforced with natural fibres. Composites Science andTechnology 64, 839–845.

Maffini, M.V., Rubin, B.S., Sonnenschein, C., Soto, A.M., 2006. Endocrine disrupt-ors and reproductive health: the case of bisphenol-A. Molecular and CellularEndocrinology 254–255, 179–186.

Marks, M.J., Snelgrove, R.V., 2009. Effect of conversion on the structure–propertyrelationships of amine-cured epoxy thermosets. ACS Applied Materials & Inter-faces 1, 921–926.

Mele, G., Li, J., Mazzetto, S.E., 2010. Cardanol: from waste of the cashew agro-industryto bio-based catalysts used for the photodegradation of 4-nitrophenol in water.Journal of Biotechnology 150 (Suppl.), 271–272.

Mensitieri, G., Lavorgna, M., Musto, P., Ragosta, G., 2006. Water transport in denselycrosslinked networks: a comparison between epoxy systems having differentinteractive characters. Polymer 47, 8326–8336.

Nam, S.-H., Seo, Y.-M., Kim, M.-G., 2010. Bisphenol A migration from polycarbonatebaby bottle with repeated use. Chemosphere 79, 949–952.

Petersen, J., Fouquet, T., Michel, M., Toniazzo, V., Dinia, A., Ruch, D., Bomfim, J.A.S.,2012. Enhanced adhesion over aluminum solid substrates by controlled atmo-spheric plasma deposition of amine-rich primers. ACS Applied Materials &Interfaces 4, 1072–1079.

Pimentel, D., Marklein, A., Toth, M., Karpoff, M., Paul, G., McCormack, R., Kyriazis,J., Krueger, T., 2009. Food versus biofuels: environmental and economic costs.Human Ecology 37, 1–12.

Popineau, S., Rondeau-Mouro, C., Sulpice-Gaillet, C., Shanahan, M.E.R., 2005.Free/bound water absorption in an epoxy adhesive. Polymer 46, 10733–10740.

Raquez, J.M., Deléglise, M., Lacrampe, M.F., Krawczak, P., 2010. Thermosetting(bio)materials derived from renewable resources: a critical review. Progress inPolymer Science 35, 487–509.

Ruseckaite, R.A., Williams, R.J.J., 1993. Castor-oil-modified epoxy resins as modelsystems of rubber-modified thermosets. 1: Thermodynamic analysis of thephase separation. Polymer International 30, 11–16.

Staples, C.A., Dome, P.B., Klecka, G.M., Oblock, S.T., Harris, L.R., 1998. A review ofthe environmental fate, effects, and exposures of bisphenol A. Chemosphere 36,2149–2173.

Tsai, W.-T., 2006. Human health risk on environmental exposure to bisphenol-A: areview. Journal of Environmental Science and Health, Part C 24, 225–255.

Wang, R., Schuman, T.P., 2013. Vegetable oil-derived epoxy monomers and poly-mer blends: a comparative study with review. eXPRESS Polymer Letters 7,272–292.

Xiao, G.Z., Shanahan, M.E.R., 1997. Water absorption and desorption in an epoxyresin with degradation. Journal of Polymer Science Part B: Polymer Physics 35,2659–2670.

Zhai, L., Ling, G., Li, J., Wang, Y., 2006. The effect of nanoparticles on the adhesion ofepoxy adhesive. Materials Letters 60, 3031–3033.

Zinck, P., Gérard, J.-F., 2008. Polyepoxide–water interactions: influence of thechemical structure of the network. Polymer Degradation and Stability 93,1231–1237.