Colloids and Surfaces A: Physicochemical and Engineering Aspects

Transcription

1 Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Contents lists available at ScienceDirect Colloids and Surfaces A: Physicochemical and Engineering Aspects journal homepage: Wetting behaviour and surface properties of technical bamboo fibres C.A. Fuentes a,, L.Q.N. Tran a, C. Dupont-Gillain b, W. Vanderlinden c, S. De Feyter c, A.W. Van Vuure a, I. Verpoest a a Department of Metallurgy and Materials Engineering (MTM), Katholieke Universiteit Leuven, Kasteelpark Arenberg 44, B-3001, Leuven, Belgium b Institute of Condensed Matter and Nanosciences, Université Catholique de Louvain, Louvain-la-Neuve, Belgium c Division of Molecular and Nanomaterials, Katholieke Universiteit Leuven, Leuven, Belgium article info abstract Article history: Received 29 December 2010 Received in revised form 15 February 2011 Accepted 18 February 2011 Available online 2 March 2011 Keywords: Natural fibre Bamboo Wilhelmy Wetting Contact line Molecular kinetic theory Bamboo fibres recently attracted interest as a sustainable reinforcement fibre in (polymer) composite materials, due to specific mechanical properties which are comparable to glass fibres. To achieve good wetting and adhesion of the bamboo fibre with different polymers, the fibre surface needs to be characterized. The wetting behaviour of technical bamboo fibres is studied experimentally by using the Wilhelmy technique, and the results are modelled using the molecular-kinetic theory. A novel procedure, based on an autoclave treatment, allows stable and reproducible advancing contact angles to be measured. In this way, meaningful information on interfacial interactions can be obtained, allowing improvement of the bamboo-polymer interface. Additionally, for comparison, the wetting behaviour of synthetic poly(ethylene terephthalate) (PET) fibre is studied. This article aims at contributing to a better understanding of the complex phenomena occurring during wetting of natural fibres. The results indicate that the high concentration of lignin on the surface of bamboo fibres is responsible for their wetting properties, whereas typical phenomena affecting wetting experiments on plant fibres can be minimized Elsevier B.V. All rights reserved. 1. Introduction Among the many different kinds of natural fibres used in composite materials, bamboo is deemed to have one of the most favourable combinations of low density and good mechanical properties: the specific strength and stiffness of bamboo fibres are comparable to those of glass fibres [1]. However, many natural fibres have several disadvantages such as poor wettability, incompatibility with some polymeric matrices and high moisture absorption by the fibres [2]. A major difficulty is related to the fibre matrix adhesion. Bonding between the reinforcing fibre and the matrix has a significant effect on the properties of the composite since stress transfer and load distribution efficiency at the interface is determined by the degree of adhesion between the components. Using the experimental data obtained from wetting measurements, fibres and matrices can be examined and matched in terms of their surface components in order to improve the interfacial properties; predicting and verifying their compatibility allows more suitable combinations and therefore better composites to be made. Corresponding author. address: Carlos.Fuentes@mtm.kuleuven.be (C.A. Fuentes). There are a variety of techniques for measuring wetting of single fibres. The most common methods include both the Wilhelmy technique and fluid geometry analysis. The former consists in a measurement of the liquid weight lifted in the meniscus by the spreading of the liquid upwards on a fibre, while the latter is concerned with the profile determination of a barrel-shaped volume where the fibre is wetted by a finite volume of liquid (a droplet) [3 5] or of a meniscus in the case of fibre wetting by an infinite reservoir [6,7]. In the case of natural fibres, both methods are hardly applicable due to surface irregularities and perimeter variation. To avoid these complications, the characterization of the wetting behaviour of natural fibres has been reported through the use of the modified Washburn or capillary-rise method [8]. However, this method does not allow the influence of wetting velocity on the determined contact angle to be studied. If surface irregularities are minimized, the Wilhelmy technique represents a good option to study the wetting of solids at different immersion velocities. The interpretation of experimental wetting data depends on wetting theories. These have been derived to describe wetting on an ideal surface wherein complexities in relation to their wetting behaviour such as the viscoelastic response of a polymer surface to a wetting liquid [9], contact angle hysteresis due to surface irregularities or chemical heterogeneity [10,11] are assumed to be absent. Therefore, wetting phenomena can be modelled with some success /$ see front matter 2011 Elsevier B.V. All rights reserved. doi: /j.colsurfa

2 90 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) for synthetic materials, in which the phenomena mentioned above are not expected to play a major role [12]. In order to predict the wetting properties of solids, several theoretical models were developed. Among them, two approaches focus on the location of energy dissipation during the wetting of a solid by a liquid: viscous dissipation in the bulk of the liquid (hydrodynamic model), and dissipation in the close vicinity of the solid near the wetting line (molecular-kinetic theory) [13,14]. These models obtained for synthetic materials revealed the dependency of dynamic contact angle on wetting velocity (depending on both speed and direction of displacement). As wetting velocity approaches zero, the wetting quasi-equilibrium parameters are obtained and may refer to either an advancing (wetting) or a receding (dewetting) contact angle. The difference between advancing and receding contact angles is called hysteresis. However, similar approaches to model wetting behaviour cannot normally be applied to natural fibres due to complex phenomena at their surface. Barsberg and Thygeseny [12] argued that plant fibres may give rise to various complex phenomena during wetting experiments which are typically not found for synthetic fibres: liquid sorption/diffusion into the surface layers, diffusion of low-molecular-weight compounds (extractives) from the surface layers into the liquid, different glassy behaviour of the chemical constituents of the surface layers or viscoelastic response of the surface layers to the liquid. As a consequence, the influence of such phenomena on the wetting behaviour of natural fibres may be a possible source of invalidation of typical experimental techniques for measuring wetting. Moreover, the difficulties in the characterization of the wetting properties of natural fibres are increased as a result of the complexity of their overall microstructure which by far exceeds that of synthetic materials. This complexity is due to the natural fibre hierarchical organisation at different length scales and the presence of different materials in variable proportions such as cellulose, hemicellulose, lignin and pectin [15,16]. It is claimed that some liquids can penetrate the natural fibre structure [12,17], allowing wetting properties of natural fibres to be affected by sorption and diffusion. For instance, technical bamboo fibre consists of bundles of more than one hundred elementary fibres. An elementary fibre consists of several layers where crystallized cellulose nano-fibrils are aligned with different angles with respect to the longitudinal fibre axis and are bound together with hemi-cellulose and lignin [18 20]. A schematic illustration of the complexity of bamboo fibre microand nano-structure is shown in Fig. 1. The wetting properties of natural fibres (and solids in general) are determined by molecular interactions between their surface and liquids. If these molecular interactions can be evaluated by means of well described advancing and receding contact angles (quasi-equilibrium parameters), it is then possible to evaluate surface energy components by means of theories such as Owens Wendt and acid base approaches, which are based on the theoretical Young contact angle, assuming that an equilibrium state can be reached. The surface condition and surface constitution of technical bamboo fibres play an imperative role in the interfacial strength of their composites. While information about bamboo fibre microstructure already exists, as mentioned above, the nature of the surface of technical bamboo fibres (used as reinforcement fibre in composites) is still unknown due to the fact that topography, chemical constituents and constituents distribution (mainly lignin and cellulose) are affected by the method of extraction such as steam explosion, retting, chemical extraction, or mechanical processes [21 25]. The aim of the present work is to examine whether technical bamboo fibres can be considered a well defined system for which reproducible and stable advancing contact angles can be measured. Treatment of the fibres was proposed in such a way that typical phenomena known to affect wetting experiments in plant fibres may play a limited role. In this manner, meaningful information on interfacial interactions can be obtained, allowing improvement of the bamboo composite interface. 2. Theoretical basis 2.1. Molecular-kinetic theory The molecular-kinetic theory was developed by Blake [14] to explain the wetting properties of solids. It employs Frenkel and Eyring s explanation of the process that takes place during the momentum transport of liquids, viewed as the movement of a molecule from one local energy minimum to another [26]. This theory in its basic form discards the explicit energy dissipation due to viscous flow and focuses instead on energy dissipation occurring in the immediate vicinity of the moving contact line due to the process of attachment or detachment of fluid molecules from the solid surface [13]. The macroscopic behaviour of the wetting line is explained by the individual molecular displacements occurring within the three-phases contact line [27]. These displacements occur randomly but progressively in the direction of the moving contact line [28]. According to Blake [14], during spontaneous spreading, a liquid drop exhibits a dynamic contact angle that depends on the instantaneous contact line velocity and decreases progressively toward a static contact angle 0 at = 0. In forced wetting, the substrate is moved at constant speed to drive the contact line at a given velocity, forming a stable dynamic contact angle (v). The displacement of the contact line depends on the frequency of forward and backward molecular displacements within the three phases zone, K + and K, respectively. At equilibrium, = 0 and the net rate of displacement is zero, so that K + = K = K 0, where K 0 is the equilibrium displacement frequency. According to this theory, energy is dissipated by dynamic friction associated with the moving contact line. If the driving force for wetting is taken to be the out-of-balance surface tension force (cos 0 cos ), then the relationship between and is given by: [ ] 2 = 2K 0 sin h 2kT (cos 0 cos ) (1) where is the average length of each molecular displacement, k is the Boltzmann constant, T the absolute temperature, and the surface tension of the liquid. The complete derivation of this model is presented by Blake [14]. 3. Materials and methods 3.1. Materials Technical fibres (bundles of elementary fibres, see Fig. 1) were mechanically extracted from Guadua angustifolia bamboo culms in the Department of Metallurgy and Materials Engineering (MTM) at KULeuven. Polyethylene terephthalate (PET) monofilaments (diameter 800 m) from Goodfellow were used to compare the wetting behaviour of synthetic and bamboo fibres. Ultrapure water (18.2 cm resistivity, Millipore Direct Q-3 UV) was used to study the fibres wetting behaviour. Lignin powder Protobind 1000 was supplied by Granit from Switzerland Fibre preparation The technical bamboo fibres that were examined underwent the following preparation procedure. After being selected (by means

3 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Fig. 1. Schematic diagram of bamboo fibre structure: (A) transverse section from bamboo internodes [16], (B) a typical cross section of the technical bamboo fibres inside the culm presents an irregular form and consists of bundles of elementary fibres. The cross-section of these elementary fibres is either pentagonal or hexagonal and they are arranged in a honeycomb pattern [18], (C) model of the polylamellate structure of an elementary bamboo fibre. It consists of thick and thin layers of cellulose nano fibrils with different fibrillar orientation [18], (D) nano-fibrils are bound together with hemi-cellulose and lignin [19]. of an optical microscope), the fibres were cleaned, first with warm water for 1 h (90 C), then wiped with ethanol with a piece of cotton tissue before being dried in a vacuum oven at 80 C for 1 h. With the aim to smooth the lignin at the fibre surface, a group of technical bamboo fibres was also further put in an autoclave under 3 bars of pressure at 150 C for 1 h. Finally, all the fibres were wiped with hexane and then conserved under silicagel. To obtain a clean surface for the PET fibres, these were washed with a detergent (RBS-35 from Chemical Products) at a concentration of 4% (v/v) in water during 1 h under magnetic stirring to remove organic residues, and next rinsed in ultrapure water at 90 C for 1 h. The cleaned fibres were then dried under vacuum at 90 C for 2 h and then conserved under silicagel Contact angle measurements Dynamic contact angles were measured with a Krüss K100 tensiometer using the Wilhelmy technique. This method is based on the Laplace equation that describes the pressure exerted by the meniscus. Initially developed for plates, the method was also converted to be used with fibres replacing the perimeter of the plate by the perimeter of a cylinder. The fibre is immersed into the liquid and the microbalance detects a force (F measured ), being the sum of the wetting force (F wetting ), the weight of the fibre (G) and the buoyancy force (F buoyancy ): F measured = F wetting + G F buoyancy = p cos + mg gad (2) where p is the fibre perimeter, m the fibre mass, g the acceleration of gravity, the liquid surface tension, the liquid density, A the fibre cross-sectional area and d the immersion depth. When the weight of the probe is measured beforehand and set to zero on the balance and the force is extrapolated back to zero immersion depth, only the wetting force remains: F measured = p cos (3) The method applied to determine the fibre perimeter is based on the same principle as the tensiometric measurement that was discussed above. In this case, however, instead of the contact angle the perimeter is sought. When a liquid with 0 contact angle is used, Eq. (3) becomes: F measured = p (4) At relatively low speeds, hexane is assumed to have a contact angle of 0 with virtually all substrates. To evaluate the accuracy of this assertion, perimeters of technical bamboo fibres were measured using hexane at 1.5 mm/min before they were put in a resin (to prevent some damage during the cutting procedure) and then cut into four pieces of three millimetres to obtain various cross sections along the fibre length. Subsequently, the perimeters were measured at each cross-section using the scanning electron microscope and the mean values were compared to those of the hexane method. The reproducibility of the contact angle determination for technical bamboo fibres was examined by performing duplicate measurements on the same fibres. After the first measurement, the fibres were dried in a vacuum oven at 80 C for 1 h and then conserved under silicagel. The time interval between two measurements of the same sample was set to one week. Dynamic contact angle measurements at a given speed were performed with a 0.05 mm data sampling step. Accordingly, the data for a single 10 mm length fibre represents 200 values. The average and standard deviation values reported in this study were calculated from the data of all the fibres measured at a given speed; so e.g. for 10 evaluated fibres at a certain speed, this means that 2000 data-points were averaged.

4 92 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Absorption measurement Technical bamboo fibres, treated or not in an autoclave at 150 C, were immersed in ultrapure water. Before the sorption test, the specimens were subjected to the same cleaning and conditioning procedure used for contact angle measurement samples. To evaluate the effect of absorption phenomena during the Wilhelmy experiment, the content of water absorbed by the sample was calculated by the weight difference between the weight of the fibre before immersion, and the weight of the fibre after it was removed at a velocity of 1.5 mm/min from water. Weight was measured with the electronic microbalance option (Krüss K100) to a resolution of 1 g Fitting procedure The parameters of the wetting kinetics are obtained by curve fitting from the correlation plot of experimental dynamic contact angle values and wetting velocity. The procedure followed by Vega et al. [29] was adapted in fitting the data using the molecularkinetic theory. Accordingly, Eq. (1) can be simplified to: = A sin h[b(c cos )] (5) where A =2K 0, B = 2 /2kT, and C = cos 0 are the independent parameters. Eq. (5) was used as the regression model to fit each set of experimental data, maximizing the coefficient of multiple determination by adjusting the independent parameters. The starting values were randomly chosen, and were adjusted during the fitting process to bring the curve close to the data points Surface characterization Atomic force microscopy imaging Imaging was carried out in Tapping Mode TM, with a Multimode TM system (Veeco) operating with a Nanoscope IV TM controller (Veeco) and a type E scanner. Silicon RTESPA cantilevers (Veeco) with resonance frequencies of about 275 khz were used. Images were collected at scan frequencies of Hz and a resolution of pixels. Images were corrected for sample tilt and analyzed with Scanning Probe Imaging Processor software (Image metrology, AS) X-ray photoelectron spectroscopy (XPS) XPS analyses were performed on a Kratos Axis Ultra spectrometer (Kratos Analytical, Manchester, UK) equipped with a monochromatized aluminium X-ray source (powered at 10 ma and 15 kv). One single fibre was cantilevered fixed on a flat stainless steel trough with a piece of double sided isolative tape. This way of mounting insures that the fibre surface only was analyzed but not its surrounding. The troughs, holding each one fibre sample, were then inserted in the multispecimen holder. The pressure in the analysis chamber was about 10 6 Pa. The angle between the normal to the sample surface and the direction of photoelectrons collection was about 0. Analyses were performed in the hybrid lens mode with the slot aperture and the iris drive position set at 0.5, the resulting analyzed area was 700 m 300 m. The pass energy of the hemispherical analyser was set at 160 ev for the survey scan and 40 ev for narrow scans. In the latter conditions, the full width at half maximum (FWHM) of the Ag3d5/2 peak of a standard silver sample was about 0.9 ev. Charge stabilisation was achieved by using the Kratos Axis Ultra device. The electron source was operated with a filament current between 1.9 and 2.1 A and a bias of 1.1 ev. The charge balance plate was set between 3.3 and 3.9 V. Fig. 2. Advancing dynamic contact angle versus fibre position for water on bamboo fibres that were not autoclaved at 150 C. The following sequence of spectra was recorded: survey spectrum, C1s, O1s, N1s, Ca2p, Si2p, Na1s, P2p and C1s again to check for charge stability as a function of time and the absence of degradation of the sample during the analysis. The C (C,H) component of the C1s peak of carbon was fixed to ev to set the binding energy scale. The data treatment was performed with the CasaXPS program (Casa Software Ltd., UK). Mole fractions were calculated using peak areas normalised on the basis of acquisition parameters after a linear background subtraction, and consideration of experimental sensitivity factors and transmission factors (depending on kinetic energy, analyser pass energy and lens combination) provided by the manufacturer. C1s spectra were decomposed with a Gaussian/Lorentzian (70/30) product function, by constraining the FWHMs of all components to be equal. 4. Results and discussion 4.1. Dynamic contact angle Fig. 2 shows examples of contact angle tests in water for technical bamboo fibre samples that were not autoclaved at 150 C. The experimental data show a big scatter for the non-autoclave treated fibres, showing values of advancing contact angle from 60 to 100. These results are in agreement with literature reviews attributing this large fluctuation to the influence of both chemical and topographical heterogeneity of the fibre surface [12,30 32]; the former deals with the difference in wetting between natural fibre constituents such as lignin, cellulose and hemicelluloses, while the latter is concerned with surface roughness and differences in fibre perimeter. The results of duplicate measurements (after one week) on the same fibres that were not autoclaved at 150 C are shown in Table 1 (left), allowing the reproducibility of contact angle measurements to be evaluated. The unpredictable large variation of contact angle values seems to confirm the statement that plant fibres, and bamboo in particular, do not constitute a well-defined system amenable to wetting studies, as reported before for sisal and coir fibres [12]. However, after autoclave treatment at 150 C, technical bamboo fibres exhibit stable contact angles. As revealed in Table 1 (right), the standard deviation of the contact angle average is reduced to around 3 for the first measurement. Furthermore, the duplicate contact angles show small differences, confirming that autoclave treatment allows a better reproducibility to be achieved. The results indicate a reduction of hysteresis caused mainly by fibre surface irregularities reduction at different length scales due to the autoclave treatment, as will be demonstrated further on.

5 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Table 1 Variation between two different contact angle measurements on the same bamboo fibres. Non-autoclave treated fibres (1.5 mm/min) Autoclave treated fibres (1.5 mm/min) Fibre no. Contact angle ( ) Variation % Fibre no. Contact angle ( ) Variation % First measurement Second measurement First measurement Second measurement ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Average 87.7 ± ± 10.4 Average 68.8 ± ± 2.3 Lin [16] has studied the lignification process of bamboo stems by observing cross-sections and using fluorescence microscopy and it was observed that lignin surrounds the bamboo technical fibre. Hence, it is possible that a layer of lignin remains on the surface of mechanically extracted bamboo fibres. Lignin can be softened, as stated by Tejado [33], who reported glass transition temperatures (Tg) between 90 and 180 C among different underivatized lignin samples; consequently, lignin can be put under pressure to even the fibre surface. Moreover, contact angle values may suggest lignin as the main component at the technical bamboo fibre surface. Indeed, Maximova [34] studied the wetting behaviour of lignin on the surface of cellulose fibres, reporting a water advancing contact angle value just under 70 for a cellulose fibre saturated with adsorbed lignin. In the same fashion, Liukkonen [35] reports a contact angle value of 67 for lignin by observing water microscopic drops in the environmental scanning electron microscope (ESEM). These values are similar to the average advancing water contact angle of 69 obtained for autoclaved bamboo fibres in this study, and, in contrast, far from other natural fibre constituents contact angles previously reported in the literature: cellulose 30 [36] and alpha cellulose 14 [37]. However, determination of the surface chemical composition is needed in order to prove the hypothesis of high lignin concentration on the surface of bamboo technical fibres. This was done using XPS Surface chemical composition: XPS In Fig. 3, the decomposed C1s spectra for lignin from Granit and a non-autoclave treated technical bamboo fibre sample are compared. The C1s peak intensity at 285 ev is related to the presence of lignin [38]. As reported by Johansson [39], cellulose is ideally devoid of aliphatic carbon-carbon bonds (designated as C1) because of its polysaccharide structure; however, in milled wood lignin, 49% of the carbon atoms are C1 type, as shown by Shchukarev [38]. Similarly, the measured Protobind Granit lignin shows 64% of C1 type carbon atoms. The bamboo results give an average of 57% of C1 type carbon among the 10 tested samples. This indicates that there may be various types of lignin which are chemically different or that other compounds containing aliphatic-carbon may be present as well. The differences between the various lignin samples may be explained by the use of different lignin isolation processes in which extractives and other chemical components are removed or, at least, their molecular structure is changed [40]. If we hypothesize that only cellulose and lignin are present, then lignin is certainly predominant on the surface of our technical bamboo fibres. Fig. 4 shows the results regarding surface chemical constituents of both autoclave-treated and non-treated technical bamboo fibres obtained from the decomposition of the high resolution carbon 1s spectrum for each fibre. Lignin content on the fibre surface was analyzed by determining the oxygen-to-carbon atomic ratio, and the relative concentration of the C1 component. The first aspect deals with the fact that oxygen-to-carbon atomic ratios are different for cellulose and lignin, while the second aspect is concerned with the lack of C1 bonds in chemically pure cellulose within the C1s spectrum. The references consist of theoretical values for pure cellulose and milled wood lignin [39], and measured data of lignin powder (Protobind 1000, Granit). The results clearly indicate that technical bamboo surface constituents for the ten tested samples are close to our references for lignin, indicating that bamboo technical fibres may be homogeneously covered with lignin and possibly some other molecules, but not with cellulose. Since spectra are enriched in C1 links more than Fig. 3. XPS high resolution spectra from carbon C1s region: (A) lignin from Granit and (B) non-autoclave treated technical bamboo fibre surface.

6 94 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) that it is possible to measure the technical bamboo fibre perimeter using hexane with reasonable accuracy. However, it is necessary to study the influence of the perimeter variation on the determination of contact angle values as a possible source of invalidation of plant fibres wetting studies. If it is assumed that only the perimeter slightly varies during the contact angle measurement for a bamboo water system, then it is possible to evaluate the contact angle obtained as a function of the perimeter deviation, as can be seen in Fig. 6A (using 69 as the average advancing contact angle). Eq. (3) was transformed to: = arccos ( Fmeasured p ) = arccos ( cos 69 ) a (6) Fig. 4. A correlation graph depicting the percentage of C1/C ratio versus O/C ratio for chemical groups at the surface of non-autoclave treated technical bamboo fibres, the surface of autoclave treated technical bamboo fibres, lignin from Granit, and theoretical values for cellulose and lignin according to Shchukarev [38]. expected for milled wood lignin, the presence of other compounds such as lipids cannot be excluded. Furthermore, the difference in constituents between fibres autoclaved at 150 C or not is imperceptible within the measured depth of 5 10 nm. This does indicate that the autoclave treatment does not change the chemical structure and composition of the technical bamboo fibre surface. Hence, it is presented as a reliable method to stabilize contact angle measurements in technical bamboo fibres by reducing surface irregularities. The irregular lignin material on the technical bamboo fibres surface would be smoothened out (see further) Perimeter and absorption evaluation The perimeters of bamboo fibres, autoclaved at 150 C or not, were determined by wetting the fibres with hexane and measuring the wetting force. Afterwards, these values were compared with the perimeters measured on SEM images at 4 cross-sections along the fibre (see Section 3). As can be seen in Fig. 5, a typical cross section of a technical bamboo fibre presents an irregular form and consists of elementary fibres compactly arranged in a honeycomb pattern (with very small lumens), joined by a thin wall of mainly lignin [16,18]. The results, presented in Table 2, indicate a good agreement between both methods with relative errors less than 8% for both autoclaved and non-autoclaved fibres, and thus confirm Fig. 5. SEM image of a bamboo fibre cross-section with contour perimeter line. where a varies from 0.85 to 1.15, and represents the perimeter variation between 0.15p and +0.15p. The analysis of Eq. (6) shows that the contact angle varies with less than 2.5 for a perimeter variation of 10% in the bamboo water system. As can be seen from the results, the large fluctuation of contact angles values obtained for bamboo in water cannot be explained solely on the basis of perimeter variation. Another difficulty in the characterization of the wetting properties of natural fibres is related to liquid absorption. Since the Wilhelmy method is based on measuring the wetting force, the liquid absorbed by the fibre can distort contact angle measurements. Liquids may penetrate the structure of natural fibres, modifying the force value and so the calculated dynamic contact angle. To evaluate the effect of absorption phenomena during the Wilhelmy experiment, the content of water absorbed by the sample was calculated by the weight difference between the weight of the fibre before immersion, and the weight of the fibre after it was removed at a velocity of 1.5 mm/min. This mass value was then expressed in force units and evaluated as percentage of the wetting force for each specific measurement in order to evaluate the variation of the measured force due to the influence of absorbed water. Table 3 shows the results of contact angle variation for both autoclave and non-autoclave treated fibres. The contact angle fluctuation due to water absorption is analyzed by evaluating the variation in the contact angle for a bamboo water system as a function of water absorption (Fig. 6B). The effect is analyzed as gain of weight only. As in the case of perimeter variation that was discussed earlier, Eq. (3) was transformed to: = arccos ( Fmeasured p ) = arccos (cos 69 b) (7) In this case, the variation of the measured force is evaluated up to 0.15F, hence b varies from 1.00 to The results show that the effect of water absorption is small and may not significantly alter contact angle measurements on bamboo fibres; the average weight effect of the water absorbed by the autoclave treated fibre (6.1%, see Table 3) represents a small variation of less than 1.5, as can be seen in Fig. 6B. For the case of non-autoclave treated fibres, the effect of water absorbed by the fibre is larger (14.0%, see Table 3), however it only represents a contact angle variation of about 3. As it was already presented, technical bamboo fibres are composed of several elementary fibres which are joined with lignin. The mechanical process of extraction can disjoin some elementary fibres (see Fig. 7A), facilitating the penetration of water into the fibre structure. The autoclave treatment at 150 C is not only smoothening the fibre surface, it is also compacting the whole fibre structure, resulting in a diminution of water absorption. Eqs. (6) and (7) are not considering the effect of the structure and stability of the three-phase contact line during its movement through different activation energy barriers represented by the different cross section along the fibre length [41], or the effect of the spontaneous diffusion of liquid molecules into the surface

7 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Table 2 Perimeter evaluation of technical bamboo fibres, based on scanning electron microscopy (SEM) images and on wetting measurements in hexane. Fibre Fibre no. Methods Wetting analysis ( m) Image analysis ( m) Relative error (%) Non-autoclave treated bamboo ± ± ± ± ± ± ± ± ± ± Autoclave treated bamboo ± ± ± ± ± ± ± ± ± ± Fig. 6. Influence of perimeter (A) and force (B) variation due to water absorption on the variation of the calculated contact angle. The contact angle variation is presented as the absolute value of the difference between the average advancing contact angle obtained for a bamboo water system (69 ) and the contact angle obtained from Eqs. (6) and (7). Table 3 Measured values of water absorption (in force units) for non-autoclave treated and autoclave treated bamboo fibres. Non -autoclave treated fibres Autoclave treated fibres Fibre no. Absorbed water (mn) Wetting force (mn) % Fibre no. Absorbed water (mn) Wetting force (mn) % ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Average 14.0 Average 6.1 Fig. 7. (A) Optical microscopy of non-autoclave treated bamboo fibre and SEM image of disjointed elementary fibres, the scale bar in (a) is 10 m. (B) Optical microscopy image of altered PET fibre.

8 96 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) A 110 B Contact Angle ( ) Contact Angle ( ) Position (mm) Position (mm) Smooth PET Altered PET 1 Altered PET 2 Smooth PET Altered PET Bamboo Fig. 8. (A) Advancing contact angle of smooth and altered PET fibres: (1) first measurement and (2) second measurement of the same fibre, and (B) advancing and receding contact angles of smooth and altered PET fibres in comparison with a typical autoclave treated bamboo fibre. The effect of surface topography variation in the receding contact angle of PET fibres is evident, showing a similar behaviour as observed in the case of a representative bamboo fibre (instability of receding contact angle). layers decreasing the solid liquid interface with time [12]. These phenomena are related to perimeter variation and water absorption respectively, and can affect the contact angle measurement. We try to make clear that the big scatter of contact angles values obtained for bamboo in water cannot be explained by means of either perimeter variation or absorption without any relation to the stability of the contact line. In order to evaluate the wetting behaviour of an impermeable solid and compare it with bamboo fibre wetting behaviour, artificial surface defects were introduced in PET fibres using a sharp blade to create perimeter variation (waviness) along the fibre (Fig. 7B), trying to reproduce the wavy surface of a bamboo fibre. The waviness variation of the PET fibres used was random with poor reproducibility, and thus, a detailed connection of the experimental data with the geometry profile of the fibre surface is not possible. A goal of this contribution is to identify possible factors related to waviness or whatever surface irregularity with spacing greater than roughness, which can be affecting contact angle measurements. Similarly to what was found for the untreated bamboo fibres, the results on altered PET fibres indicate that the wetting behaviour becomes unstable when waviness plays a role (Fig. 8A and B). As revealed by Table 4, the smooth PET fibres present an average advancing contact angle of 87 with a small standard deviation of 1.5 ; such steady wetting behaviour has been reported for synthetic materials where complexities leading to non-equilibrium can be neglected or minimized [12,42]. In contrast, the altered fibres show an average advancing contact angle of 93.6 and a standard deviation of 9.8 (Table 4). The latter is about seven times the deviation of an ordinary PET fibre with normal surface defects created at the time of extrusion. In order for the contact angle to reach 94 from 87, the measured wetting force is not only being reduced several times but becoming negative as well. This effect can be explained as a consequence of the introduced surface waviness. Surface tension is a tensor that acts perpendicularly to the contact line, in the plane of the surface [43]. While the three-phase contact line is moving, the wetting force position is constantly changing and so its direction, depending on the radii of curvature at the wetting perimeter position (see Fig. 9). The microbalance is only detecting the forces parallel to the fibre immersion direction. A similar behaviour is described by Czachor [44,45] when he studied the contact angle variation in a wavy capillary geometry; by means of a mathematical model of meniscus movement in a sinusoidally shaped capillary, it was concluded that the calculated contact angle is a strongly increasing function of wall waviness. It is clear from the results presented that the scatter of contact angles on PET fibres increases with rise of waviness (or surface defects that provoke a change in the force direction). Furthermore, the effect on the receding contact angle seems to be comparable to the wetting Table 4 Measured dynamic contact angles for PET fibres with different surface topography. Fibre no. Contact angle ( ) Average contact angle ( ) PET (normal) ± ± ± ± ± ± 1.1 PET (altered perimeter) ± ± ± ± ± ± 6.4 Fig. 9. The radius of curvature of the fibre surface is constantly changing during the movement of the meniscus over the fibre body, and so the direction of the wetting force, r curvature radius, F lv liquid vapour interfacial force. Position (1): smooth fibre and position (2): fibre with surface irregularities.

9 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Fig. 10. AFM images of technical bamboo fibre surfaces: (A) untreated and (B) autoclave treated at 150 C. behaviour of bamboo fibres during the receding process, resulting in a situation of irregular or zero receding contact angles (Fig. 8B); it is hypothesised that water remains in the spaces between surface waves. Due to the waviness effect or macro-defects (micro-roughness is still not being considered), the measured force is affected and the calculated contact angle using the Wilhelmy formula cannot be related to the real contact angle value. The obtained apparent contact angles (even bigger than 90, which involves negative forces) for non-autoclave treated bamboo fibres and altered PET fibres can be explained by their wavy surface (Fig. 7A and B), since other phenomena like diffusion or absorption provoke the reduction of contact angles values. Wensel and Cassie models cannot be applied in this macro scenario, since they may become valid with small-wavelength roughness [46] and a larger scale of surface irregularities (waviness of fibre surface) is what is relevant here. The micro-roughness effect will be addressed next Fibre micro-roughness and absorbing material A representative AFM image of a technical bamboo fibre that was not autoclaved at 150 C is shown in Fig. 10A where its surface exposes a micro rugged topography. Roughness analysis after plane correction yields a root mean square (RMS) roughness value of 481 nm. In contrast, Fig. 10B shows a far less rugged surface for an autoclave treated fibre with a RMS roughness of 64 nm. So, the bamboo fibres without autoclave treatment appear to be rough (see Fig. 10A). This seems to be another reason of the higher water absorption values measured for non-autoclave treated fibres (disjoining of elementary fibres was discussed before). The lignin on the surface is not well compacted and porous which makes the liquid movement inside the surface easier. The area for the spontaneous diffusion of liquid molecules is also being increased. Both phenomena, absorption and diffusion affect the contact angle measurement. For an absorbing material, the interfacial energy is a time-dependent function which decreases from the instant the material is brought into contact with the liquid [12]. In the case of autoclave treated fibres, absorption and diffusion are being reduced since the surface is smoother and well compacted [Fig. 10B]. However, it should be noticed that neither the state nor the geometry of the phase interfaces in the regions remote from the three phase contact line has any direct effect on the contact angle [46]. Diffusion must be occurring in autoclave and non-autoclave treated fibres but relatively far from the contact line. In the Wilhelmy technique, the wetting force is measured instantaneously and always on a new contact line (or surface area fraction [47]) during the continuous immersion in the fluid. The effect of absorption and diffusion remains as an increment of fibre weight only, which was analyzed before, and was shown to represent just a small variation in the calculated advancing contact angle. However, sorption and diffusion play a role in the irregularity of the receding contact angle, and in this way do not allow a proper full characterization of the wettability behaviour of the bamboo fibre surface. The effect of roughness on contact angles has been well described in literature [43,46,47]. Theories like Wenzel and Cassie predict the apparent contact angle of advancing and receding fronts for surfaces with uniform roughness where the contact line moves from one metastable state to another on representative contact areas of the whole surface [46 48]. Hence, these theories are hardly applicable since the surface of non-autoclave treated fibres presents a non-uniform roughness and contact areas are different along the fibre length. Moreover, this non-uniform roughness and irregularities (much larger than the ones observed in treated fibres) must be creating several energy barriers and hence different metastable states along the contact line, increasing the instability of the contact angle results (Fig. 2). The waviness can explain the large variation of contact angles (even bigger than 90 ), but not the non-reproducibility of the values. As can be seen in Fig. 8A, the contact angle profile of altered PET fibres is almost the same from one measurement to the next. The non-reproducibility of measurements on untreated bamboo fibres can be a consequence of swelling phenomena. Natural fibres are known to swell significantly in water, and a typical Wilhelmy measurement is rather slow, which would allow significant swelling to take place (far from the contact line). The dimensional changes that accompany the shrinking and swelling of non-autoclave treated samples would provoke changes of the measured surface from one wetting measurement to the next, creating new and totally different energy barriers that the contact line needs to pass during a new measurement. This effect would be increased by the rough and noncompacted surface presented in non-autoclave treated fibres. The large contact angle scatter found for non-autoclave treated fibres must be a consequence of the combined effect of waviness (surface irregularities) and roughness. Absorption phenomena, waviness, and roughness have been reduced in autoclave treated fibres. However, they still have surface irregularities and some swelling might be happening since there are still some changes in the profile of the duplicate wetting measurements (see Table 1) which leads to a slightly different contact

10 98 C.A. Fuentes et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 380 (2011) Fig. 11. (A) Theoretical curve obtained by nonlinear regression of experimental data using Eq. (1) (each point represents the average contact angle of 10 PET fibres, the maximum standard deviation was 1.4 for a given velocity). The angles were measured at 11 different speeds, (B) dynamic contact angle as a function of wetting velocity for water on bamboo technical fibre. The dynamic contact angles were measured at speeds ranging from 0.15 to 500 mm/min. The theoretical curve through the data was obtained by nonlinear regression of experimental data using Eq. (1) according to the fitting procedure. = nm, K 0 = s 1. The angles were measured at 9 speeds ranging from 0.15 to 500 mm/min. (Each point represents the average contact angle of 8 different technical bamboo fibres.) angle in the second measurement. The micro-roughness of autoclave treated fibres with a RMS of 64 nm is comparable with the roughness of a synthetic fibre [29]; however, non-uniform surface irregularities remain and make it more difficult to study their surface in order to apply some theories of contact angles on rough surfaces. For the purpose of this article we were satisfied with the strong reduction in surface roughness achieved by the autoclave treatment. However, further optimization of the autoclave parameters in future could be useful to further fine-tune the surface topography Dynamic contact angle as a function of wetting velocity In the case of natural fibres, a direct measurement of the contact angle is problematic, and so their wetting behaviour is difficult to study. The Wilhelmy technique represents a reliable method (for the reasons given before) to study the wetting behaviour of bamboo fibres at different immersion speeds. We believe that the unpredictable wetting behaviour of bamboo fibres is related more to surface topography than to some kind of an unpredictable nature of bamboo fibre surface material. If surface waviness, roughness and liquid penetration are minimized, then the bamboo fibre surface represents a well defined system and so its wetting behaviour can be studied and a meaningful interpretation of wetting data is ensured. For this to be the case, contact angles of bamboo fibres must show an expected dependence to immersion velocity. The molecular kinetic theory has already been used to interpret the dynamic contact angle data and to model the wetting phenomena for several synthetic fibres [14,27,49,50]. More specifically, the advancing contact angle as a function of wetting velocity was studied previously by Blake [13] for water on PET. Fig. 11A shows wetting experiments on PET fibres that were conducted here with two intentions: comparing the wetting behaviour of a synthetic fibre with our natural bamboo fibre and having a reference material to evaluate our experimental and fitting procedures. Our experimental wetting data seem to conform well to the prediction of the MKT. The quality of the fit to Eq. (1) is good with R 2 = 0.94 (Fig. 11A). Furthermore, the value of 1.16 nm obtained for the characteristic length is similar to the value of 1.10 nm reported previously by Blake [13] for water on PET monofilament. However, our equilibrium displacement frequency K 0 of s 1 is decreased by a factor of 1.7 in comparison with s 1 obtained by Blake [14]. Concerning this variation, the literature suggests that this value is less stable [14,29]. Blake [14] reported the difference between the values of K 0 and obtained with high and low speed for water on PET fibres; while was almost in the same order of magnitude, K 0 was some 4 orders of magnitude bigger at high speed, and there is also the fact that different PET fibres were used. Accordingly, the surface roughness, crystallinity, and chemical composition might be slightly different. The theoretical curve calculated by inserting determined mean values of K 0 and for water on autoclave treated bamboo fibres into Eq. (1), reveals good agreement between the experimental and the calculated values of the dynamic contact angle over the entire experimental speed range, as can be seen in Fig. 11B. The experimental data provide a satisfactory fit with R 2 = 0.90, which is not as good as in case of PET with R 2 = This difference is probably a consequence of the still remaining irregularities on the technical bamboo fibre surface. Although the autoclave treatment successfully reduces the bamboo fibre irregularities, these are still bigger than those present on a synthetic fibre. The obtained jump frequency K 0 = s 1 is low if it is compared with published values for water on other materials [13,14,29]. While K 0 would suggest a more polar surface (which seems to fit in with the XPS results), the advancing contact angle is a bit high suggesting a largely non-polar surface. Since the measured contact angles represent just the advancing front (which is possibly underestimating the polar content), the small K 0 value would also suggest a small receding contact angle for a bamboo water system, which could not be measured for the reasons given before. The molecular-kinetic equation provides a reasonable fit to the data with acceptable values of K 0 and, as shown in Fig. 11B, confirming the expected immersion velocity dependence, reproducibility and stability of the advancing contact angle in a bamboo water system; indicating that the measured contact angle is the true advancing contact angle. However, as stated before, surface irregularities and probably diffusion contributes to the instability of the receding contact angle. 5. Conclusions The high concentration of lignin on the surface of technical bamboo fibres, as concluded from XPS results, seems to be responsible for their wetting properties. Furthermore, the experimental results reveal that the large fluctuations during wetting between various bamboo fibres of the same batch may be due more to the surface topography irregularities of the fibres than to any other type of unpredictable phenomena. These irregularities are largely reduced by autoclave treatment and subsequent smoothening of the lignin surface layer, as confirmed by AFM results.