Effect of graphene oxide surface treatment on the interfacial adhesion and the tensile performance of flax epoxy composites

The high stiffness and damping properties of flax fibres promote the integration of biocomposites in structural applications. However, the strength of flax/epoxy composites is still limited compared to glass/epoxy composites. Graphene oxide (GO) has proved to be a promising building block for nanocomposites due to its high toughness, stiffness and tunable interfacial interactions with polymers. This study aims to understand the potential of GO-based surface treatment of flax fibres to modify the interfacial adhesion and tensile performance of flax fibre/epoxy composites. GO-modification improves the interfacial shear strength of elementary flax fibre/ epoxy by 43%. The interfacial improvement is also established by the 40% higher transverse bending strength compared to untreated flax/epoxy composites. The tensile moduli of GO-modified flax/epoxy composites are on average 2 GPa higher than for untreated flax fibre/epoxy composites in all strain ranges. The quasi-static longitudinal tensile strength of unidirectional composites is not affected by GO-modification.


Introduction
Natural fibres, such as flax, with excellent stiffness to weight ratio and damping properties [1] are a green class of reinforcements for composites.Flax fibres can partially substitute glass fibre in weight critical applications.Although flax fibres in the market are more expensive than glass fibres [2], the non-renewable energy required to produce flax fibre mats (9.55 MJ/kg) is significantly lower than glass fibres (54.7 MJ/kg) [3].Also, from the perspective of abiotic depletion, human toxicity, photochemical oxidation, and end of life options, flax fibres are environmentally superior to glass fibres [4].Flax/epoxy and glass/epoxy composites demonstrate similar tensile modulus range [5].
Epoxy matrices are ideal for low-cost manufacturing of large structures such as wind turbine blades with resin infusion based on their intrinsic low viscosity.Also, epoxide and hydroxyl groups of epoxy resin can form hydrogen bonds with free hydroxyl groups on flax fibre that is essential for effective load transfer between fibre and matrix [6,7].The excellent performance of flax/epoxy composites is mainly due to the intrinsic properties of flax fibres.Flax fibres with a high amount of cellulose (60-80 wt%), and low cellulose microfibril angle (10 • ) have the highest strength values (800-1000 MPa) among natural fibres [8].Contrary to other bast fibres such as hemp and jute, retted flax bundles can result in a more extensive content of individual elementary fibres which means a higher surface area for load transfer inside a polymer matrix [8,9].However, the ultimate performance of flax/epoxy composites in various loading conditions is still low compared to glass fibre/ epoxy composites [5].
One of the strategies to enhance the strength of flax/epoxy composites is to improve the interfacial adhesion between fibres and matrix.Research efforts in this area focus on the selective extraction of waxy (lipophilic) compounds from the fibre surface [10,11] and development of coupling agents [12][13][14][15][16][17][18][19].The removal of lipophilic compounds from the flax surface increases the surface roughness and the number of reactive hydroxyl groups leading to better adhesion [6,12,14,20].The objective of functionalisation is to introduce new functional groups on the fibre surface which can promote interfacial adhesion or toughness [14].Graphene oxide (GO), a nanoscale material, can significantly enhance the damage tolerance and interfacial interactions as an additive for polymers or as a surface functionalisation of fibres [21][22][23][24][25][26][27][28].GO is a derivative of graphene with oxygen-rich functional groups such as carboxyl, epoxide, and hydroxyl capable of hydrogen bonding or even covalent bonding [29,30].The tensile modulus of GO sheets consisting of three or more layers (thickness range from 3 nm to 1 µm) is around 40 GPa [31,32].Although in-plane mechanical properties of GO are inferior to those of graphene, the efficiency of load transfer from a polymer matrix to fibres is much higher for GO covered fibres [33][34][35].For instance, Zhang et al. [25] improved the interfacial adhesion and tensile strength of commercially sized carbon fibre/epoxy composites by 36% and 34% with deposition of GO on carbon fibres.Chen et al. [24] reported a 16% improvement in the interlaminar shear strength of GOtreated glass fibre/epoxy composites.Sarker et al. [15] reported significant improvement of 89% in the apparent interfacial shear strength (IFSS) of jute/epoxy by GO-modification of technical jute fibres.In another study by Sarker et al. [13], they also enhanced the tensile strength and the tensile modulus of jute/epoxy composites by 110% and 324% respectively by a combination of alkali lipophilic extraction, combing of jute bundles to individual fibres, and GO-adsorption.Their study is the only work dedicated to the contribution of GO-modification to the mechanical performance of natural fibre/epoxy composites.The current state of literature highlights the need for further research to assess the potential of GO-nanomodification of high-performance bast fibres such as flax to achieve better composites strength.
This work investigates the effect of surface treatment of flax yarns with GO on the tensile behaviour of unidirectional (UD) flax fibre/epoxy composites.The contribution of GO-surface modification to the interfacial adhesion between flax fibres and epoxy is assessed by microbond and transverse bending tests.The transverse cross-section of UD composites is analysed by optical microscopy to understand the effect of GOtreatment on the porosity of composites and individualisation of flax fibre yarns to individual fibres.

Methodology
Bcomp®, Switzerland provided unidirectional, non-crimp flax yarn fabric of ampliTex® UD type 5009 with an areal density of 300 g/m 2 and density of 1350 kg/ m 3 .AmpliTex® materials are a high-performance class of flax fibres with hot water lipophilic treatment and broad content of individual elementary fibres [36].Based on the manufacture's datasheet, the tensile modulus, tensile strength and the strain to failure of flax fibres are 61 GPa, 580 MPa, and 1%, respectively.Also, based on the assessment by Bcomp®, 5009 ampliTex (300 g/m 2 ) can replace 500 g/m 2 glass fibre UD fabric to have the same tensile modulus.Their calculations assumed a glass fibre with a density of 2600 kg/ m 3 and tensile modulus of 70 GPa.The reported values for the mechanical properties of flax and glass fibres agree with the literature [8].
For the graphene oxide synthesis [30], a 500 mL round bottom flask was placed in an ice bath (0 • C) on a magnetic stirrer.46 mL of H 2 SO 4 was transferred to the flask and continuously stirred by a magnetic bar. 1 g of NaNO 3 and 2 g of graphite powder was added to the H 2 SO 4 .Then, 6 g of KMnO 4 was gradually added to the reaction maintaining the temperature of mixture bellow 5 • C.After 10 min, the ice bath was removed, and the reaction continuously stirred for 6 h at 23 • C, which finally turned into a viscous paste.Afterwards, 92 mL and 280 mL of distilled water were added to the reaction with 30 min gap in between.Also, 10 mL of hydrogen peroxide was added to the flask.
After the GO synthesis, an aqueous dispersion of GO was sonicated for exfoliation and further centrifuged to remove the acid used in the process and the non-exfoliated GO sheets.The obtained brown suspension was dried in an oven at 50 • C for 24 h to achieve a GO film.A 1.2 wt % aqueous dispersion of GO was prepared by mixing GO film with deionised water with a mortar and further sonication.Flax yarns were dip-coated in this dispersion for 24 h at 23 • C and dried at 60 • C overnight.The reference flax yarns were immersed in de-ionised water for 24 h at 23 • C and dried at 60 • C overnight to account for the potential effect of water.
The X-ray diffraction (XRD) pattern of the synthesised graphene oxide crystals was examined by a Panalytical Empyrean Multipurpose Diffractometer with zero background.X-ray diffraction measurements were prepared by attaching the GO film on an amorphous sample holder.The instrument was operated at 40 kV and 40 mA with a Kα (λ = 0.154 nm) radiation source.The X-ray diffraction measurements were performed from 2θ = 5 The outcome of the GO synthesis (GO crystals) and the surface treatment of flax fibres was studied and compared to untreated material by Fourier transform infrared (FTIR) spectroscopy using a PerkinElmer Spectrum One device with the wavenumbers ranging from 500 to 4000 cm − 1 with a resolution of 0.5 cm − 1 .Three scans were performed for each fibre type.
The tensile performance of untreated and GO-modified elementary flax fibres was measured by Textechno FAVIGRAPH® with 20 cN load cell.Tensile tests were performed with 20 mm gauge length and with a crosshead speed of 2 mm/min.A pretension of 0.6 cN/tex was applied before each measurement.Fibres were stored for one week at a controlled environment of RH 50% and 25 • C.
Microbond tests were performed with a FibroBond® device [37] with 1 N load cell on single elementary fibres of flax.The elementary flax fibres were extracted from flax fabrics by tweezer under an optical microscope.The average diameter of elementary flax fibres was 17 ± 3μm.Epoxy droplets were deposited by a FibDrop® device, which enables to form droplets of controlled size.All droplets were inspected under a microscope before testing, and samples with defects were rejected.To avoid fibre breakage during the microbond test, only droplets in the range of 70μ m to 100μ m were tested.The droplets were cured at 90 • C for 24 h and post cured at 150 • C for 2 h.During the test, microvices sheared the droplets while the fibres were fixed with glue on steel sample holders.Tests were observed with a microscope.The loading rate was 8μ m/s.
The interfacial shear strength denoted as τ d [38] was evaluated from the linear regression (slope) of a dataset of maximum force of individual droplets versus embedded area as [37]: where F max is the measured maximum force, and A emb is the embedded area along the fibre in the resin.In total, 20 individual droplets were tested for each sample.Unidirectional (UD) flax fibre/epoxy coupons with a 10 mm width, 2 mm thickness, 250 mm length and 40% fibre volume fraction were manufactured based on the hand-layup method followed by hot pressing.Spacers with a 2 mm thickness were used in the press to assure the final thickness of the samples.Samples were cured at 90 • C for 24 h and post cured at 150 • C for 2 h.The fibre volume fraction (V f ) was calculated according to the ISO 14127:2008 standard: where m f is the mass of the fibres, V c is the volume of the composite and ρ f is the density of the fibres (1350 kg/m 3 ), assuming a pore-free composite.
To compare the porosity content in flax/epoxy and GO-modified flax/epoxy composites, the transverse area of UD composites was inspected by LEICA DM 2500 M optical microscope on a dark field mode.Samples were embedded in an epoxy resin before polishing.Images were processed using image processing software (ImageJ) to differentiate between porosity, fibre, lumen, and matrix.After thresholding on the grey level, a morphological analysis tool was applied to count the number and size of pores.
To study the interfacial adhesion at macroscale, transverse bending tests of composites were performed based on ASTM D7264.The flexural tests were performed on an Instron 5567 general-purpose mechanical testing machine with a 5 kN load cell.Composites were tested in a fourpoint bending condition with a crosshead speed of 2 mm/min.The strain was measured based on the load cell displacement.
Tensile tests were performed on the Instron 5567 testing machine with a 30 kN load cell and a crosshead speed of 2 mm/min based on ASTM D3039.The strain was measured using a clip-on extensometer (50 mm gauge length).Abrasive papers were placed without glue in between the testing clamps and the samples to avoid slipping.All samples were stored for two weeks before testing in a controlled environment of RH 50% and 22 • C. Average results from five samples per series are reported.
The fracture surfaces of tested composites and microbond test samples were examined with a Zeiss ULTRAplus scanning electron microscope (SEM).A thin carbon or gold coating was used to ensure sufficient conductivity for the samples.

Characterisation of graphene oxide (GO)
Fig. 1a shows the FTIR spectra of graphene oxide (GO).The transmittance spectra of GO represents the various characteristic oxygen-rich functional groups.The strong and broad-band at 3426 cm − 1 is related to the -OH stretching vibration of hydroxyl groups [39].The existence of a few CH 2 /CH groups can be noticed from bands at 2922 cm − 1 and 2852 cm − 1 [40].The band around 1728 cm − 1 is related to the C--O stretching vibration of carboxylic functional units [39].Finally, the bands at 1195 cm − 1 (C-OH stretching vibration) and 1046 cm − 1 (C-O stretching vibration) testify the presence of epoxy functional groups [39].Fig. 1b shows the sharp XRD peak at 2θ = 10.6 • which corresponds to 0.84 nm spacing between the graphene oxide sheets [40].XRD pattern also indicates that the synthesised GO is free unreacted graphite (or unexfoliated) [41].Based on our previous measurement with the atomic force microscope, the average thickness of single graphene oxide sheets was 0.84 nm [40].In the same study, we also observed the wrinkled morphology of GO crystals which shows the flexible nature of GO.The results indicate the successful synthesis of GO with oxygen-rich functional groups.

Surface characterisation of fibres
Fig. 2. shows the FTIR transmittance spectra of untreated flax and GO-modified flax fibres.The transmittance spectra of flax fibre show the typical band peaks of cellulose, hemicellulose, and lignin [11].The band at 3340 cm − 1 is related to the -OH stretching of hydroxyl groups mainly in cellulose and lignin [42].The bands at 2849 cm − 1 and 2916 cm − 1 are usually related to the symmetric and asymmetric CH 2 stretching vibrations of cellulose and hemicellulose [42].The band around 1632 cm − 1 is ascribed to the O-H bending due to absorbed water [42,43].The band groups at 1427 cm − 1 , 1320 cm − 1 , and 1024 cm − 1 can be related to the C -H wagging vibration in cellulose and hemicellulose [43], the CH 2 rocking vibration at C6 in cellulose [42,44], and the C -O stretch vibrations of acetyl groups (lignin) [45].The spectrum of the GO-modified flax exhibits a new band around 1737 cm − 1 for the -C--O stretching vibrations of the -COOH groups of the GO [24,46].FTIR results suggest successful adsorption of GO on the flax fibres by hydrogen bonding.
Fig. 3 compares the surface morphology of untreated flax fibres, and graphene oxide treated fibres.Fig. 3 a, b shows the rather smooth surface of flax fibres.Fig. 3 c, d represents the clear morphology difference between flax fibres and GO treated flax fibres.In Fig. 3 d the wrinkled nature of GO coating on the flax fibres is in agreement with our previous observation [40].

Tensile performance of elementary flax fibres
Table .1 reports the average tensile properties of elementary flax fibres and GO treated flax fibres with three different concentrations (0.65 wt%, 1.2 wt%, 2 wt%).The breaking force, elongation at failure,   and tenacity of untreated elementary flax fibres are 18 ± 6 (cN), 3.1 ± 0.9 (%), and 6.3 ± 2.7 (cN/dtex) respectively.The variation from the average value is rather high which can be related to the inhomogeneous nature of natural fibres.The average tensile performance of elementary GO-flax fibres is in the range of untreated flax fibre regardless of the concentration.Also, the linear density of flax fibres is not affected by the graphene oxide surface modification.It can be concluded that there is no meaningful change in the tensile performance of single flax fibres after GO surface treatment.The detailed results are reported in the supplementary data 1.

Interfacial fibre/matrix adhesion
Fig. 4 presents the contribution of GO-treatment to the interfacial adhesion of elementary flax fibres with epoxy, as measured by the microbond test.The slope-based results (linear regression) in Fig. 4a suggest an interfacial shear strength (τ d ) of 33 ± 3 MPa for 1.2 wt% GOflax/epoxy which is 43% higher than 23 ± 3 MPa for flax/epoxy.Microbond test assessment for 0.65 wt% GO-flax/epoxy and 2 wt% GO-flax/epoxy showed an interfacial shear strength of 25 ± 3 MPa and 20 ± 3 MPa respectively.The detailed results are reported in the supplementary data 2. The lower shear strength of 2 wt% GO-flax/epoxy can be potentially related to the formation of multi GO layers on the flax fibre acting as a weak spot against shear force.Daly et al. [47] showed that GO-to-GO shear strength in thicker multilayer graphene oxide could be as low as 5.3 ± 3.2 MPa.In the upcoming sections, performance of composites is based on 1.2 wt% GO-flax/epoxy which shows the best interfacial adhesion based on the microbond test.
Fig. 5 shows the fracture surfaces after the microbond test.The residual meniscus on the fibres (Fig. 5a, b) represents crack initiation in mode I close to the microvices followed by crack propagation along the interface, as was explained previously [48].All droplets debonded with brittle failure.In Fig. 5b, d minor peelings of fibre cell-walls are evident after debonding.The sliver of fibre cell-walls indicates a better interfacial adhesion with epoxy compared to the internal fibre cohesion.In Fig. 5d resin residuals are visible on GO-modified elementary flax fibres after debonding, suggesting the failure onset locus has moved towards the epoxy matrix.
To further understand the effect of GO-modification on the interfacial adhesion of flax fibre/epoxy, Fig. 6 presents the transverse bending strength results of UD composites.Remarkably, the average transverse bending strength of the GO-flax/epoxy composites is 38 ± 1 MPa which is 40% higher than the corresponding value for flax/epoxy composite.The better transverse strength of GO-treated flax/epoxy composites suggests an improvement in the fibre/matrix adhesion.Better adhesion can be ascribed to the high amount of oxygen-containing functional groups of GO, namely hydroxyl ( -OH), carboxyl (O -C--O), epoxide (C-O-C), and carbonyl (C--O), which can interact with the functional groups of epoxy [13,15,25].C -N bond between amine hardener and GO treated flax fibres by ring opening polymerisation [13,49], and mechanical interlocking between wrinkled GO coating and epoxy is also possible [13].The 43% improvement in the interfacial adhesion of flax fibre/epoxy composites by GO-surface modification, agrees with microbond results.Our findings agree with the previous reports on the effect of GO on the interfacial adhesion of glass and carbon fibres with epoxy, that GO can be used as an efficient interfacial coupler in fibrous composites [26,50].

Tensile behaviour of UD composites
Fig. 7a presents examples of the stress-strain curves related to the uniaxial tensile tests of flax/epoxy composites.Fig. 7b shows the trilinear development of tensile modulus with respect to strain.The nonlinearity at approximately 0.3% strain, is anticipated to be related to the viscoplastic deformation of the amorphous matrix of hemicellulose at the primary walls [51,52].The tensile modulus of untreated flax fibre/epoxy and GO-treated flax fibre/epoxy composites are reported in Table .2 for the initial strain range (until 0.1%) and between 0.6% and 0.8% strain range above the transition point.The tensile moduli (stiffness, E) of untreated flax fibre/epoxy composites at the initial and posttransition strain ranges are 20±1 GPa and 13±1 GPa respectively.In the similar strain ranges, GO-modified flax fibre composites have slightly higher (~2 GPa) modulus values of 23±1 GPa and 15±1 GPa.The difference in the tensile stiffness can be related to the better fibre/matrix adhesion with GO-treatment as reported in Fig. 6 by increased transverse flexural strength of composites.Better adhesion is often also associated with better fibre-matrix impregnation, which will lead to the increased modulus.
Additionally, Fig. 8 shows that the representative microstructure of GO-modified flax/epoxy composites is clearly less porous compared to   reference flax/epoxy composites.The less porous nature of these GOnanoengineered composites suggests a better fibre/matrix impregnation.The better fibre/matrix impregnation can be related to the improvement in fibre/matrix adhesion with GO treatment as discussed previously (in Section 3.4).Naturally, porosity in flax/epoxy would negatively affect the tensile stiffness [53] as well as ultimate performance.Due to the lower amount of porosity also at the fibre/matrix interface, GO-treated flax/epoxy demonstrates indeed better tensile stiffness (Table .2).However, the tensile strength and the failure strain of both composites are at a similar level in this study.The failure process upon tensile test propagates fast and interfacial properties are not significantly shownyet it is not clear if the GO-modification could not improve long-term dynamic (e.g.fatigue) performance in tensile mode.The fracture surface of the flax/epoxy composite in Fig. 9a demonstrates that local failure in shear occurred along fibre/matrix interphases with long fibre pull-outs.The fracture surface of GO-modified flax/ epoxy in Fig. 9b is essentially similar to the one of flax/epoxy except for the length of the fibre pull-outs.For GO-modified flax/epoxy, the pullout lengths were observed shorter, which is known to indicate better load transfer and adhesion between the fibres and matrix of the composites.Also, Fig. 9c shows the relatively smooth fibre surfaces after flax/ epoxy fracture, whereas in Fig. 9d epoxy residual are evident on GO-modified flax fibres.Epoxy residuals on the fracture surface of GOflax/epoxy composites show improved adhesion and affinity between fibre/matrix.Fig. 9 e, f indicates the brittle nature of the fracture surfaces for both untreated and modified composites.
Regardless of the 43% improvement in the microbond τ d values, and 40% enhancement in the interfacial strength measured by transverse bending, our findings suggest that the extent of improvement in the longitudinal tensile strength of flax/epoxy composites by GO-modification of the high-performance flax fibres is not significant for quasistatic loadings.This observation is in line with the existing data.Merttote et al. [9] reported that in the case of high-performance flax yarns with large content of individual elementary fibres, even a 100% improvement in the interfacial adhesion does not significantly affect the longitudinal tensile strength of flax fibre reinforced composites.Also, a 110% improvement in tensile strength of jute/epoxy composites reported by Sarker et al. [13], was achieved by a combination of alkali lipophilic extraction, combing jute bundles to individual fibres, and GOadsorption.Therefore, it can be argued that a significant improvement of the tensile strength in their work is mainly due to the removal of waxes and individualisation of pristine jute bundles.
Nevertheless, a significant improvement in transverse strength is very relevant to improve off-axis strength in multi-directional composites.In future work, the fatigue testing of flax/epoxy composites will have to be analysed to understand the interfacial effects on the dynamic load range.The fatigue testing of unidirectionally reinforced composites is still under development, due to the challenges related to the test specimen design that could reveal interfacial effects on the results [54,55].

Conclusions
This paper presents the potential of graphene oxide (GO) surface treatment to improve the interfacial adhesion and the tensile behaviour of flax fibre/epoxy composites.A GO-treatment is performed on a highend flax fabric intended for structural applications rather than on green or unprocessed flax yarns to highlight the pure effect of GO-treatment.The microbond results indicated a 43% improvement in the interfacial shear strength of the elementary flax fibre/epoxy by GO-surface modification.The micrographs of the GO-treated flax fibres after the microbond test revealed a shift of the failure locusthe onset had shifted towards the matrix.The transverse strength of GO-modified flax fibre/ epoxy composite outperformed the untreated flax fibre/epoxy by 40%.The better transverse strength of GO-flax/epoxy composites was ascribed to the enhanced interactions between oxygen-containing functional groups of GO adsorbed into flax fibres and epoxy (interface).A significant improvement in transverse strength of composites is very relevant to improve off-axis strength in multi-directional composites.The tensile stiffness of GO-modified flax fibre/epoxy composites was on average 2 GPa higher than untreated flax/epoxy composites for all strain ranges.The quasi-static longitudinal tensile ultimate strength and the failure strain were similar for the reference and modified version.Our findings suggest that the longitudinal tensile strength of UD flax/epoxy composites (with large content of individual elementary flax fibres) is not sensitive to the improvement in the fibre/matrix interfacial adhesion.

F
.Javanshour et al.

F
.Javanshour et al.

Fig. 4 .
Fig. 4. Effect of GO-surface modification on the interfacial adhesion of elementary flax fibre/epoxy.(A) experimental microbond test results, (B) the shape of deposited epoxy droplets on the elementary flax and GO modified flax fibres.

Fig. 7 .
Fig. 7. Tensile behaviour of unidirectional flax/epoxy and GO-modified flax/epoxy.The graph A shows representative stress-strain results and the graph B shows the instantaneous stress-strain slope (stiffness) of the composite coupons for all strain ranges.

Fig. 8 .
Fig. 8.The representative transverse cross-sectional micrographs of UD composites showing the microstructure of embedded fibre bundles of untreated flax fibre/ epoxy (A, B) and GO-modified flax fibre/epoxy (C, D).The histograms show the number and size distribution of pores in each micrograph.

F
.Javanshour et al.

Table 1
Summary of the tensile properties of elementary flax fibres.

Table 2
Summary of the tensile properties of composites.