Wettability of carbon fibres at micro- and mesoscales

Accepted Manuscript Wettability of carbon fibres at micro- and mesoscales

Jian Wang, Carlos A. Fuentes, Dongxing Zhang, Xungai Wang, Aart Willem Van Vuure, David Seveno PII:

S0008-6223(17)30503-1

DOI:

10.1016/j.carbon.2017.05.055

Reference:

CARBON 12031

To appear in:

Carbon

Received Date:

06 March 2017

Revised Date:

09 May 2017

Accepted Date:

15 May 2017

Please cite this article as: Jian Wang, Carlos A. Fuentes, Dongxing Zhang, Xungai Wang, Aart Willem Van Vuure, David Seveno, Wettability of carbon fibres at micro- and mesoscales, Carbon (2017), doi: 10.1016/j.carbon.2017.05.055

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Wettability of carbon fibres at micro- and mesoscales Jian Wang1, 2,*, Carlos A. Fuentes2, Dongxing Zhang1, Xungai Wang3, Aart Willem Van Vuure2, and David Seveno2 1

School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, PR China 2

3

Department of Materials Engineering, KU Leuven, Leuven 3001, Belgium

Institute for Frontier Materials and ARC Future Fibres Hub, Deakin University, Geelong, Vic 3217, Australia

ABSTRACT Physical adhesion between Carbon Fibres (CFs) and polymer matrices as well as the formation of voids at the interface between these two materials are mostly controlled by the wetting properties of the fibres. Due to the hierarchical structure of CF reinforcements, it is essential to study their wetting behavior at different scales: from the single fibre (microscale) to the fabric (macroscale) via the tow scale (mesoscale). Probing the wettability of CF tows, is, however, highly challenging, because it couples the effects of surface chemistry and geometry of the fibre assembly characterized by spontaneous capillary wicking and elasto-capillarity induced aggregation. Therefore, we first developed a new methodology combining a tensiometric method and a synchronized in-situ optical observation technique to better characterize the wettability of CF tows.

We then used it to evaluate the difference in

wettability between tows composed of unsized and sized (T300) CFs. By comparing

* Corresponding author. Tel: +86 451 86281403. 1

E-mail: [email protected] (Jian Wang)

ACCEPTED MANUSCRIPT their wettability at the micro- and mesoscale, we could quantify how the modification of the surface chemistry at the microscale is transferred to the mesoscale. 1. Introduction CFs have attracted great attention as an outstanding reinforcement for polymer composites because of their superior mechanical properties [1, 2]. Nowadays, CFs are widely used in many fields, from civil to industrial [2-5]. However, the final mechanical properties of carbon-fibre-reinforced polymer composites (CFRP) depend not only on the intrinsic properties of the reinforcing CFs and polymer matrices, but also on the adhesion between these components [6-10]. Moreover, during processing, the wettability of CFs plays a crucial role in determining the quality of the final composite interface. Good wettability between fibres and matrices leads to better physical adhesion and fewer defects at the interface [9, 11]. Hence, various methods have been developed for modifying CF surfaces to enhance their compatibility towards polymers, such as polymer sizing, chemical modifications, plasma treatments and nanoscale coatings incorporating carbon nanotubes or graphene [12-17]. However, the lack of good compatibility between CFs and resins is still a current issue limiting the development of advanced composite materials. Moreover, both Intra-tow and inter-tow voids can be observed in CF composites because of incomplete wetting at both single fibre and tow scales [18, 19] evidencing that . During liquid molding processes, resin infiltration within CF tows and textiles is determined not only by the wetting properties of the CF surface, but also by the fibre architecture [20]. Thus, a better understanding of wetting phenomena at multiple 2

ACCEPTED MANUSCRIPT scales is essential. The wettability of CFs with polymers can, for example, be revealed by measuring contact angles made by liquid polymers or probe liquids on fibres. Typically, at the microscale, contact angle measurements are conducted at the microscale, measuring the contact angles made by liquids around one fibre or multiple parallel aligned fibres are obtained using the Wilhelmy balance method [12, 15, 16, 21, 22]., or at the macroscale, measuring the contact angles of liquid droplets on surfaces of CF fabrics by optical methods [23]. A previous study [24] showed the difficulties of precisely measuring contact angles on a single CF due to its micrometer-scale diameter. At the macroscale, contact angles of liquid droplets on surfaces of CF fabrics use measured by optical methods [23]. It is however also very difficult to independently and reliably evaluate the effects of surface chemistry and textile structure from the analysis of drop shapes at the macroscale. CF tows are show an intermediate and interesting case study, as their simple geometry, compared to fabrics, may not totally obscure the role of the surface chemistry. In addition, their large dimensions, compared to a single fibre, should ease the measurement of contact angles and provide for a representative sample of fibre properties. A CF tow can be considered as a porous medium; its wetting behavior is thus not only dependent on the surface chemical properties of the fibres, but also on its pore structure [25]. So far, to the knowledge of the authors, studies dedicated to wetting of CF tows are not available. Furthermore, reports on the wettability of other fibrous materials at the mesoscale are still limited [25-27]. In those studies, a fibrous material 3

ACCEPTED MANUSCRIPT is considered as a bundle of capillary tubes, and the Washburn method [28] is used to determine effective dynamic advancing contact angles assuming that its geometry remains the same during the experiments. However, as shown by Rieser et al. [29, 30] and their work on capillary-induced attraction between vertical cylinders, there exist fluid-mediated attractive forces between partially submerged vertical cylinders. Thus, when a tow is partially dipped in a liquid, capillary forces, competing with the bending stiffness of the fibres, are able to bend the fibres and trigger an elastocapillary aggregation of the porous structure. This leads to a capillary-induced densification phenomenon which was also observed from arrangements of micro- [30] and nanoscale [31] filaments, for instance CNTs, during the elasto-capillary selfassembly process [32]. This inevitable aggregation can influence the reliability of contact angle measurements of fibrous materials by the classical Washburn method, and hence, lead to incorrect estimation of the wettability of CF tows. Direct optical [15, 25, 26, 33, 34] and the Wilhelmy methods are two other well-known approaches used to measure contact angles on fibre shaped solid surfaces. The latter is mostly used to accurately measure contact angles on dense solid (in cylinder or plate shape) with a known wetted perimeter. However, unpredictable perimeter changes due to elasto-capillary aggregation

of CF tows due to the aggregation process and

simultaneous liquid wicking make it extremely challenging to evaluate the wettability of tows by using the Wilhelmy method only interpret such experiments. In this paper, a new methodology, combining a tensiometric method with synchronized in-situ optical observation is proposed to better characterize the 4

ACCEPTED MANUSCRIPT wettability of CF tows, and is applied to evaluate the difference in wettability between tows composed of unsized CFs and sized (T300) CFs. The combination of these two techniques provides a way to observe the aggregation of the tows and, thus, follow the evolution of the tow perimeter, needed to calculate accurate contact angles according to the Wilhelmy method. The optical technique also provides an independent way to estimate the wettability of CF tows by analyzing the shape of the menisci formed around CF tows in addition to the force measurements, so validating the results obtained from the Wilhelmy method. By comparing the wettability of the unsized and T300 CFs at the micro- and mesoscale, we could finally quantify how the modification of the surface chemistry at the microscale is transferred to the mesoscale. Finally, the Cassie and Baxter [35] theory was applied to link the contact angles measured at micro- and mesoscales.

2. Materials and methods 2.1. Materials The CF tows considered here are composed of two types of untwisted polyacrylonitrile-based CF tows. These two materials are laboratory made unsized and untreated CF tows provided by Deakin University and commercially available sized CF tows named FT300-3000-40A (T300) purchased from Toray CFs Europe S.A, respectively. The unsized CFs were produced at Deakin University from commercial PAN precursor but skipping the sizing step. One T300 CF tow contains 3000 filaments with density of 1.76 g/cm3. The wetted T300 CF and unsized CF tows are around 600μm and 300μm in diameter respectively. 5

ACCEPTED MANUSCRIPT Before the wetting tests, the CF tows were washed in ethyl alcohol and dried at 80ºC for 1h to obtain a clean surface by removing dust and grease without altering the CF sizing. The tows were then cut into pieces (20 mm in length) and each piece was clipped firmly into a holder that could be installed directly in the tensiometer. The main properties of the test liquids (n-hexane: Acros; Deionized water: Millipore Direct Q-3 UV) are listed in Table 1. Table 1 Test liquid properties. Test liquid

𝛾𝐿𝑉 (mN/m)

𝜌 (g/cm3)

Purity

n-hexane

18.4

0.659

99.6%

Deionized water

72.8

0.98

18.2   cm resistivity

2.2. Single fibre contact angle measurement (Wilhelmy method) The method proposed by Qiu et al. [24] for measuring dynamic contact angles was used in this study to characterize the wettability of unsized single CFs with deionized water according to the Wilhelmy method. A high-precision force tensiometer – Krüss K100SF (Krüss GmbH, Hamburg, Germany) was used to measure the forces exerted on a single CF during the dynamic wetting test. The K100SF tensiometer has a claimed weight resolution of 0.1µg with testing velocities ranging from 0.1mm/min to 500mm/min. During the measurements, the fibre is stationary and the vessel holder moves up (advancing cycle) and down (receding cycle). A sketch of the apparatus is shown in Figure 1.

6

ACCEPTED MANUSCRIPT

Figure 1. Schematic of the experimental setup showing the combination of tensiometry and optical methodologies to characterize the wettability of CF tows. Since measuring accurate perimeters of thin fibres is important for calculating precise contact angles, the wetted perimeters of single CFs were measured using n-hexane (99.6%) as a perfectly wetting liquid. Once accurate and reliable fibre perimeters and capillary forces are obtained, the dynamic contact angles at constant advancing and receding velocities can be calculated from the Wilhelmy equation [36]: 𝐹𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 = 𝐿𝑓𝛾𝐿𝑉𝑐𝑜𝑠𝜃𝑑 Equation 1 where Fmeasured is the force detected by the microbalance, 𝐿𝑓 is the perimeter of the fibre, 𝜃𝑑 is the dynamic contact angle. All tests were conducted in a monitored and controlled environment (25°C, 65% relative humidity, and vibration isolation cabinet) so that 𝛾𝐿𝑉 is a constant value depending on the test liquid. Four fibres were tested by this method in case of the unsized CFs (contact angles for sized T300 fibres were taken from [24]. Each fibre was repeatedly dipped in and withdrawn from the liquid vessel three times at a velocity of 3.6 mm/min to measure a 7

ACCEPTED MANUSCRIPT series of dynamic advancing and receding contact angles. Qiu et al. [24] suggested that dynamic advancing contact angles measured below 20 mm/min can be recognized as static advancing contact angles. 2.3. Optical technique - Tow scale 2.3.1. Contact angle measurements A Motic SMZ-171 microscope and Moticam (CMOS) digital microscopy camera was used in combination with the tensiometer to capture in real time (every 500 ms) images of the formation of the meniscus around the tows. This gave access to external contact angles and the variation of the tow diameters at various wetting positions along the sample height. The pixel size was 3.24μm. The optical analysis followed a two-step procedure: first, the pictures were analyzed with Tracker software [37] to determine the local tow diameter 𝑑 and meniscus height 𝑧 by locating the position of two contact points, as shown in Figure 2. The meniscus height is defined as the average value of height differences between these two contact points and the baseline. The contact angle is then calculated using the James equation [38], which models the height of a static meniscus made by a liquid around a cylindrical substrate: 𝑜 𝑧 ≈ 𝑟 𝑐𝑜𝑠 𝜃𝑒𝑥 𝑙𝑛

[(

4𝑎

𝑜 𝑟 1 + 𝑠𝑖𝑛 𝜃𝑒𝑥 )

]

‒ 𝛿 Equation 2

where 𝑟 is the local radius of the tow, 𝑎 = 𝛾𝐿𝑉/𝜌𝑔 the capillary length of the probe liquid (with 𝜌 the liquid density, and g the gravity acceleration), and  the Euler constant (𝛿 ≈ 0.58). This equation predicts the height of a meniscus around a single fibre. This model is therefore not fully adapted to the tow geometry, but should 8

ACCEPTED MANUSCRIPT 𝑜 provide a good estimation of 𝜃𝑒𝑥 . In the case of deionized water, 𝑎 = 2724 µm,

therefore the bond number is 𝐵𝑂 =

𝑟

𝑎 ≈ 0.11 for a T300 CF tow (r ≈ 300 𝜇𝑚). For

a fibre, Clanet et al. [39] reported that the relative error in the height using Equation 2 is less than 1% below the actual value when 𝐵𝑂 ≤ 0.3. Figure 3 shows theoretical James profiles for 600 µm and 300 µm diameter fibres, modelling the wetted T300 and unsized tows respectively. The theoretical radial extension of the meniscus is 3060 μm, which is much less than half of the width of the images (of typically 6000 µm). The whole meniscus is therefore captured by the camera, i.e. the position of the baseline can be precisely detected. In addition, errors in the contact angles caused by the detection of the contact line and baseline were checked. If the local diameter of a tow was 600 μm, a variation of the meniscus height or the CF tow diameter would have only a slight influence on the calculated contact angle values (for contact angles ranging from 40o to 70o). An error of ±2 pixels (≈2% of the height or diameter) involves an error in the contact angles of less than 0.4o, i.e. the effect is limited and will not significantly affect the contact angles measured.

Figure 2. In-situ observation of the meniscus formed between a T300 CF tow and

9

ACCEPTED MANUSCRIPT

b)

a)

water. Figure 3. Theoretical James profiles. a= 2724μm (typical of water) and 𝜃𝑒𝑥 = 40o; 50 o;

60 o; 70 o and 80 o from top to bottom, a) r = 300μm, b) r = 150μm.

2.3.2. Measurement of the non-solid volume fraction Before each test, the tow sample was weighed to obtain its mass 𝑚𝑇𝑜𝑤. The real volume of each CF tow (𝑉𝑓𝑖𝑏𝑟𝑒𝑠) can be calculated based on 𝑚𝑇𝑜𝑤 and the density of CFs. If we assume that a CF tow has a perfect cylindrical shape during the wetting test, the diameter of this cylinder can be measured with the camera and its total volume (𝑉𝑇𝑜𝑡𝑎𝑙) calculated. Hence, the non-solid volume fraction of the CF tows when wetted is given by: 𝑃=

𝑉𝑇𝑜𝑡𝑎𝑙 ‒ 𝑉𝑓𝑖𝑏𝑟𝑒𝑠 𝑉𝑇𝑜𝑡𝑎𝑙

Equation 3

2.4. Tensiometric technique – Tow scale The K100SF Tensiometer was used to measure the forces exerted by the liquid on the tows. During a test, the tows were slowly soaked in the liquid over a length of 1mm and stopped at that position for 500s to make sure the external meniscus around the tow reached a static configuration. The vessel was then moved down until complete withdrawal from the liquid bath. The weight of liquid left in the tow was finally measured. The forces exerted on the tows were detected continuously every 200ms by 10

ACCEPTED MANUSCRIPT the microbalance during the whole procedure (including approaching, wetting and withdrawing from the liquid bath). Six samples were tested for both unsized and T300 CF tows. As illustrated in Figure 4, capillary wicking in the inter-fibre space and the formation of a meniscus around the tow are taking place quasi instantaneously when the tows contact the liquid [33]. The detected forces combine the weights of the imbibed liquid inside the tow and the meniscus formed around the tow. As to which process dominates is further investigated by analyzing the force data following the Washburn and Wilhelmy methods.

𝑓 Figure 4. (a) Schematics of a meniscus formed around a CF tow, 𝜃𝑒𝑥 is the external

apparent contact angle determined by the Wilhelmy method. (b) Schematics of internal capillary wicking between CF filaments. 𝜃𝑎 is the effective internal advancing contact angle determined by the Washburn method [28, 29]. 2.4.1. Internal contact angle measurements (Washburn method) The modified Washburn equation defining the flow of liquid penetrating through a fibrous medium [40] was used to characterize the internal structure of a CF tow approximated as a bundle of capillary tubes: 11

ACCEPTED MANUSCRIPT

2

𝑚 (𝑡) =

[

2 (𝑐𝑟)𝜖2(𝜋𝑅2)2 𝜌 𝛾𝐿𝑐𝑜𝑠 𝜃𝑎

]

2 =𝐶

𝜌2𝛾𝐿𝑉𝑐𝑜𝑠 𝜃𝑎 𝜂

𝜂

𝑡

𝑡 Equation 4

where 𝑚 is the mass of liquid above the liquid level, 𝜃𝑎 the effective advancing contact angle, 𝜂 the viscosity of the liquid and t time (s). 𝐶 is a constant accounting for the tortuous path of the flow in the real system, 𝜖 is the relative porosity and 𝑅 the inner effective radius of the measuring tube. Pucci et al. [26, 41] determined in this way dynamic effective advancing contact angles during capillary rise inside CF tows positioned in a container. Pucci et al. assumed that the interaction between the fibres, the container wall, and the liquid may be neglected; therefore, the forces detected by the tensiometer could be considered as simply the weight of liquid rising and penetrating the pores in-between the fibres. If we assume that the geometry and content of pores is constant, i.e. the fibers fibres are immobile, 𝐶 can be determined [26, 40, 41] by using a perfect wetting liquid like n-hexane. Since n-hexane is able to dissolve the sizing on the surface of T300 CFs, we carried out the n-hexane test after the water wetting test. Hence, the internal contact angle measurements included two steps: the first one (with water) to collect the force data for contact angle calculations and the second one (with n-hexane) to obtain the constant 𝐶. It is worth noting that a clip sample holder was used here instead of a container. Hanging the tow perpendicularly at one end and leaving the other end free, makes it possible to visualize the elasto-capillary densification process and the formation of a liquid meniscus around the tow. 2.4.2. External contact angle measurements (Wilhelmy method) 12

ACCEPTED MANUSCRIPT As described in section 2.2, the weight of the external meniscus around a tow measured by the tensiometer can be analyzed by the Wilhelmy method. The weight of liquid (𝑊𝑟) left inside the tow after complete withdrawal from the water bath was subtracted from the total measured forces to account for the effect of water up-take. It is assumed that this force characterizes the volume freely accessible to water inside the tow, i.e. the tow porosity. As already reported from single capillary tube wetting experiments, it is known that an external meniscus takes much less time than an internal one to reach equilibrium [42]. Therefore, although the external meniscus did already reach its static configuration at the beginning of the test, the liquid still kept wicking through the tow until the inter-fibre space was fully filled by the liquid. Hence, the force data detected after the full wicking process were used to calculate external contact angles [27, 36]. 𝑓 𝐹𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 ‒ 𝑊𝑟 = 𝐿𝑡𝑜𝑤𝛾𝐿𝑉cos 𝜃𝑒𝑥

Equation 5

𝑓 where 𝜃𝑒𝑥 is the external contact angle obtained from the force measurement, 𝐿𝑡𝑜𝑤 is

the wetted perimeter of CF tow obtained by the camera. In addition, the volume fractions of fluid retention, 𝑓𝑟, for each sample can be calculated using the following equation:

𝑓𝑟 = V

V𝑟

𝑟 + V𝑓𝑖𝑏𝑟𝑒𝑠

𝑊𝑟

𝜌 + V 𝑟 𝑓𝑖𝑏𝑟𝑒𝑠

=V

Equation 6

where V𝑟 is the volume of fluid in the tow, 𝑊𝑟 the weight of the liquid retention, V𝑓𝑖𝑏𝑒𝑟𝑠 the volume of CF fibres and 𝜌 the density of the fluid.

3. Results and discussion 3.1. Wettability of CFs – Fiber Fibre scale 13

ACCEPTED MANUSCRIPT Four samples of unsized single CF were tested with both water and n-hexane by the method described in section 2.2. The forces versus positions curve of a typical dynamic contact angle measurement at a velocity of 3.6 mm/min is shown in Figure 5. The exact capillary force can be evaluated from the last two cycles by subtracting the pre-force [24] measured before contact is established between the fiber and the liquid. Figure 6 shows the distribution of the dynamic contact angle results. An average advancing contact angle of 79.0 ± 4.8° was obtained, i.e. unsized single CFs are, as expected, more hydrophobic than the T300 single CF, which showed a static advancing contact angle with deionized water of 65.8 ± 2.9° [24]. Similarly, the receding contact angle of unsized single CF with an average value of 41.7 ± 7.1° is significantly larger than the one obtained for the T300 CF fibre with values between 20° and 0° [24]. The measured receding contact angle value of unsized CF in water is somewhat smaller than the value reported by Bismarck et al. (56.3 ± 3.1°) [43]. Moreover, all the values of contact angles followed Gaussian distributions indicating the reliability of the comparison between the two surfaces (see figure 6Figure 6). Testing CFs with a tensiometer means that per sample two millimeters in length can be evaluated. This gives a detailed view on the wetting process and already permits to evaluate how heterogeneous fibre surfaces are. However, it is still a small scale analysis. Forced wetting experiments like the ones performed by Blake et al.[44] or Vega et al.[45], adapted to CFs would give access to a higher scale analysis as, in principle kilometers of fibres can be tested (steady state process). An advanced optical set-up would however be required (thin fibres dipped and withdrawn from a wetting 14

ACCEPTED MANUSCRIPT bath at high speed).

Figure 5. Typical measured forces versus position (mm) curves for an unsized single CF in water.

Figure 6. Distributions of dynamic contact angle values (unsized CFs with water). 3.2. Effective internal advancing contact angle - CF tows Elasto-capillarity induced deformation of CF tows As shown in Figure 7 a significant deformation of the CF tows could be observed when the tows were partially immersed vertically into water. At the free end, the loose 15

ACCEPTED MANUSCRIPT CF filaments aggregate due to the elasto-capillary effect during the infiltration of water. A cylindrically shaped yarn was formed rapidly. Once in contact with water, capillary-induced attractive forces between partially submerged vertical fibres were generated at the contact line, eventually leading to a densification of the tows. It is important to note that this kind of deformation, starting at the free end of CF tow, changes its pore structure and wetted perimeter during the wetting test. This is not directly detectable by the tensiometer, which only measures the resulting forces. Therefore, it significantly influences the reliability of the contact angle values obtained by the Wilhelmy method. This is why we combined the tensiometric method with synchronized in-situ optical observation to better characterize the wettability of CF tows at the mesoscale.

Figure 7. Images of CF tows captured before (a), side view, (b), bottom view), during (c), side view and right after the wetting test (d), bottom view. 3.3. Effective internal advancing contact angle - CF tows 16

ACCEPTED MANUSCRIPT Figure 8 shows the curves used to determine the effective internal advancing contact angles for the T300 CF tow according to the Washburn approach. Figure 9 shows images captured during the test. As shown illustrated in Figure 7 section 3.2, the free end of the CF tow aggregates when touching fluid (in case the contact angle is smaller than 90°) and an external meniscus is formed very fast, hence there is a weight jump at the start, which is mainly due to the formation of external menisci [42]. This first part cannot be used to determine the tortuosity factor 𝐶 (n-hexane test). The next part of the curves could be fitted with straight lines in reasonable agreement with the Washburn model. However, as shown in Figure 9, the levels of aggregation of the CF tow with n-hexane and water were significantly different. The tow immersed in nhexane packed less densely than in water, so there were larger pore spaces to take up liquid for n-hexane. This leads to an overestimate of the weight of imbibed water and thus to an overestimated value of 𝐶. As illustrated in Equation 4, this overestimation of 𝐶 can lead in turn to an overestimation of 𝜃𝑎. In conclusion, for free hanging CF tows, the calculation of effective contact angles via the Washburn equation resulted in incorrect values. It should be noted that if 𝐶 (from n-hexane measurements) is used to calculate the water contact angle, a value of 84.5° is obtained. This is 18.7 ± 2.9° higher than the static advancing contact angle of a single T300 CF with water. Previous studies [46, 47] also found larger contact angles calculated from the Washburn equation than the one measured directly on smooth surfaces of the same solids. This might be due to an incorrect evaluation of the 𝐶 parameter. Therefore, we propose another approach, where we focus on the external contact angle of the fluids 17

ACCEPTED MANUSCRIPT with the fibre bundle.

Figure 8. Square of the mass of water and hexane intake versus time in T300 CF tows.

Figure 9. Images of T300 CF tows captured a) during the n-hexane wetting test and b) during the water wetting test. 3.3. Measurements of non-solid volume fraction Table 2 shows the non-solid volume fraction of the CF tow samples and the volume fraction of fluid retention calculated by the methods described in sections 2.3.2 (volume based) and 2.4.2 (weight based). For T300 CFs, there is a good agreement between the non-solid volume fraction and the volume fractions of fluid retention calculated independently by using the force method and the optical method. This agreement indicates that the porous space inside the tow was fully filled by water, validating at the same time the fluid retention correction method. According to the 18

ACCEPTED MANUSCRIPT optical observation, the 20 mm long tow samples were totally wetted after each test, indicating that the equilibrium capillary rise height in the tortuous path is greater than the length of the fibres and that water infiltrated the whole sample throughout. For the unsized CFs, the agreement is still acceptable taking into consideration the standard deviations. As the unsized CF tows are thinner than the T300 CF tows, the measurements of the non-solid volume fraction and fluid retention become highly sensitive leading to larger deviations from one sample to the other. The hydrophobicity of the unsized tow may also prevent the tow to be completely filled by water with the volume increase of the tow P being larger than the picked up water volume, leaving some remaining porosity. Table 2. Volume fractions of non-solid and volume fractions of fluid retention for each CF tow sample. Volume fraction of non-solid, 𝑃

Volume fraction of fluid retention (weight based)

T300 CFs

54 ± 3%

54 ± 5%

Unsized CFs

46 ± 15%

36 ± 12%

3.4. External contact angles - CF tows 3.4.1. Optical method As is shown in Figure 10, three successive steps may be identified when the tows are immersed in fluid. First, a quasi-instantaneous capillary rise is observedmeniscus forms that is characterized by a drop in contact angle from 90° to 76° (unsized CF tow) and 63° (sized CF tow) within the first 200ms (not monitored neither by tensiometer nor by camera). Then, a longer period is observed (~100s for T300 CF 19

ACCEPTED MANUSCRIPT tows and ~60s for unsized CF tows) corresponding to the time needed for water to completely fill the tow, both in the longitudinal and transverse directions throughout the fibres. During this period, the external contact angle keeps on decreasing as water first wets a composite substrate made of fibres and air, air being then progressively replaced by water. The advancing contact angles made between the water meniscus and the wetted fibres then reach their static values. Pictures taken at 10s, 20s and 100s (Figure 11) illustrate how the contact angle progressively attains its static value during the second period. Finally, after 100s, a steady slight decrease of the contact angles was found for both T300 and unsized CF tows. This behavior was attributed to the evaporation of water. It is identified by a decrease of the liquid height during the test. Figure 12 gives a more quantitative assessment of the evaporation process by comparing the position of the contact line (pinned) and baseline. Figure 12 confirms that the visible contact line was pinned during the entire third stage while the baseline was decreasing very slowly and evenly. Hence, at the beginning of the third stage (from 100s to 200s), external contact angles have had enough time to reach to their static configuration and the effect of water evaporation on the contact angle during these 100s (from 100s to 200s) was small. Figure 12 also shows that the height of the baseline decreases around 5 pixels, which involves an error on the contact angle values less than 1.5°. Consequently, the static advancing contact angles were calculated by averaging the external contact angles obtained during the first 100s of the third stage (from 100s to 200s). After this time, the effect of water evaporation becomes large enough to transform the static advancing contact angle into a receding 20

ACCEPTED MANUSCRIPT one.

Figure 10. External water contact angles versus time (t) for the T300 and unsized CF tows. The black and blue dashed lines show the contact angles obtained from the force method and the red and green solid lines are those obtained from the optical method. The static advancing contact angles were calculated by averaging the angles obtained from 100s to 200s (in-between the two vertical dotted red lines).

Figure 11. Evolution of the water meniscus around a T300 CF tow. Pictures were 21

ACCEPTED MANUSCRIPT taken at 10s, 20s, 100s and 500 after the tow contacted water. The dotted red line gives the position of the contact-line (left contact point) after 10s, the green one the position of the baseline after 10s and the blue one the position of the baseline after 500s.

Figure 12. Positions of the contact points and base line versus time (t) curves for a T300 CF tow. The contact line points on both side of the meniscus as exemplified in Figure 11. 3.4.2. Force method Force

and diameter versus time (t) curves obtained from a typical test in water with

a T300 CF tow are shown in Figure 13. We started to measure forces before the tow contacted the water and stopped measurements 10s after the tow was completely withdrawn from the water. Meanwhile, the evolution of the tow diameters was measured by optical observation over the whole testing time. As shown in Figure 13, a force was measured after the tow was withdrawn from the liquid bath (post-force). The same phenomenon can be observed in each test. Since there was no external meniscus around the tow at that moment, the value of this post-force was considered 22

ACCEPTED MANUSCRIPT to be the weight of water left within the tows (𝑊𝑟) due to wicking. During the impregnation process, the force measured by the tensiometer is composed of internal forces (wicking action) and the external forces (formation of the meniscus around the tow). To estimate the apparent contact angles made by the meniscus around the tow, the force associated with external wetting must be independently measured. Hence, 𝑊𝑟 should be subtracted from the total measured force to eliminate the effect of water up-take on the external contact angles measurements. Then the external contact angles can be calculated according to Equation 5.

Figure 13. Measured force (red line) and diameter (black line) versus time (s) for a T300 CF tow wetted by water. As discussed in the last section, there are three stages during the whole test, which can also be identified in Figure 13. A significant jump of the force at the moment of contact represents the sudden formation of the external meniscus. The diameter of the CF tow drops from 745μm to 590μm at the same time and slightly fluctuates due to 23

ACCEPTED MANUSCRIPT the elasto-capillary effect. Then for the second stage, a slow and gradual increase of the force can be observed. This behavior was attributed not only to a decrease of the external contact angles, but also to the gradual water imbibition into the tow both in the longitudinal and transverse directions. Afterwards, from 80s until the end of the test, the force leveled off showing only a slight increase, due to the contact angle decreasing as a result of water evaporation and pinning of the contact line (see figure 11Figure 11). As shown in Figure 14, comparison between the results obtained from the force and optical methods for both unsized and T300 CF tow shows good agreement indicating that the combination of the force and optical methods successfully quantifies the effects of deformations and water up-take within the CF tows on the external contact angles. Figure 15 shows the average results obtained for the T300 and unsized CF tows (external static advancing contact angles calculated during the first 100s of the third period of each test). Moving from an unsized CF to a T300 CF, the static advancing contact angle decreases by 13.2±5.6°. The external contact angle around CF tows decreases more significantly (19.1±7.0°) . For the same surface modification of CFs, larger contact angle changes can be observed at the mesoscale than at the microscale, which indicates that the wettability of CFs at the mesoscale is not only affected by the surface chemistry properties but also by the structure of the tow.

24

ACCEPTED MANUSCRIPT

Figure 14. External menisci formed after 200s and comparisons between the external contact angles obtained at this moment by the force and optical methods for the unsized and T300 tows.

Figure 15. Comparison of static advancing contact angles between unsized CF tows and T300 CF tows with water (static advancing contact angle of T300 single CF is 25

ACCEPTED MANUSCRIPT from reference [24]) 3.5. Relationship between the static contact angles at micro- and mesoscales The difference in static advancing contact angles between single CFs and CF tows indicates that the wettability of CFs changes when moving from the microscale to the mesoscale. As CF tows are considered to be porous media, the surfaces around them can be recognized as a composite surface made of CFs and air or fluid, respectively before and after contact with the fluid. As shown in Figure 16 a) and b), the contact line around the CF tow is not straight, which confirms that the external contact angle characterizes an heterogeneous surface composed of CFs and fluid in between them (after contact with the fluid). The Cassie law [48] is a widely used model[34] describing the contact angles of liquid on chemically heterogeneous surfaces. It gives the contact angle corresponding to the minimum free energy configuration, which is expressed in terms of the contact angles for each pure substrate. According to the modified Cassie–Baxter equation [35]: 𝑐𝑜𝑠 𝜃𝐶𝐵 = 𝑓𝑠𝑐𝑜𝑠 𝜃𝑠 + 𝑓𝑖𝑐𝑜𝑠 𝜃𝑖 = 𝑓𝑠cos 𝜃𝑠 + (1 ‒ 𝑓𝑠)cos 𝜃𝑖

Equation 7

with 𝑓𝑠 the fraction of solid/water interface, when immersed in water. Traditionally, 𝑓𝑖 = (1 ‒ 𝑓𝑠) represents the fraction of air/water interface with then θi = 180°. Here, as the tows are infiltrated by water, air is replaced by water so that fi is the fraction of water/water “interface” with 𝜃𝑖 = 0°. (such that (1 ‒ fs) is the water/water fraction), 𝜃𝐶𝐵 is the calculated “average” external contact angle around a CF tow. 𝜃𝑠 is the contact angle on the single CF, which was measured in Section 3.1. θi is the 26

ACCEPTED MANUSCRIPT water/water contact angle and was fixed to 0°. Figure 16 c) models the cross-sectional view of a triangle-packing arrangement of CFs in a wetted CF tow. If we assume that CFs have a regular triangle-packing arrangement and the area fraction of pores (P’) at the cross-section is equal to the non-solid volume fraction (P), then P’ can \be described as follows: '

𝑆𝐶𝐹𝑠

𝑃 =1‒𝑆

𝑇𝑜𝑡𝑎𝑙

=1‒1

𝜋 2 2𝑟

2(2𝑟

+ 2𝑑')2cos 30°

= 𝑃 Equation 8

Where 𝑑' characterises the distance between two CF filaments, 𝑆𝐶𝐹𝑠 is the crosssectional area of the CFs in a triangularly packed cell, 𝑆𝑇𝑜𝑡𝑎𝑙 is the total area of this triangular cell. Hence, 𝑑' can be calculated by substituting r and P into Equation 8. Then, Figure 16 d) depicts the cell of the outermost layer of CFs exposed to air. The liquid meniscus between two CF filaments is modeled as a straight line where 𝜃𝑠 is an adjustable contact angle. A similar model has already been successfully applied for nanofibre yarns [25]. Therefore, the solid to water fraction 𝑓𝑠 can be calculated by: 𝑓𝑠 = 𝑑

𝑑𝑠

𝑠 + 𝑑𝑖

=

𝑑𝑠 𝑟 + 𝑑'

Equation 9

and 𝑑𝑠 = 𝑟𝑠𝑖𝑛𝜃𝑠 Equation 10 Finally, 𝜃𝐶𝐵 can be evaluated using Equations [7]-[10]. The calculated values of 𝜃𝐶𝐵 (Table 3) are close to our experimental results shown in Figure 15 for both the T300 and unsized CF tows. Therefore, the contact angles of CFs at the meso- and microscales can be linked by using the Cassie–Baxter model and a known non-solid volume fraction. Besides, this model also describes how the structure of the tows (porosities or distances between two CF fibres) can influence the wettability of the CFs at the mesoscale. 27

ACCEPTED MANUSCRIPT Table 3 Calculated external contact angles (𝜃𝐶𝐵) and the parameters used for calculation

tow

Non-solid volume fraction, P

𝜃𝑠

𝑑' (μm)

𝑓𝑠

𝜃𝐶𝐵

T300 CFs

54 ± 3%

65.8 ± 2.9°

1.4 ± 0.2

65 ± 3%

51.9 ± 2.5°

Unsized CFs

46 ± 15%

79.0 ± 4.8°

1.0 ± 0.6

76 ± 11%

67.4 ± 6.5°

Type of CF

Figure 16. a) Picture showing the contact-line between water and a T300 CF tow. b) Schematics of the contact-line on the heterogeneous surface of a wetted CF tow (with liquid in between the CFs) c) Schematics of pore areas in a cross-sectional view perpendicular to the tow axis, where r is the radius of CF and 2d is the distance between two filaments d) Cell of the outermost layer of CFs exposed to air, where 𝜃𝑠 is the static advancing contact angle of liquid with single CF.

4. Conclusion 28

ACCEPTED MANUSCRIPT This study aims at characterizing the wettability of unsized and T300 CFs (microscale) and CF tows (mesoscale). At the microscale, the wettability of unsized single CFs was first studied by precisely measuring dynamic contact angles and then comparing them with the wettability of single T300 CFs (from previous work). The static advancing contact angle of water on unsized single CFs at a low test velocity (3.6 mm/min) is 79.0 ± 4.8°, which indicates that the unsized CFs are more hydrophobic then the T300 CFs (contact angle 65.8 ± 2.9°). The receding contact angle of water on unsized CFs (41.7 ± 7.1°) is significantly larger than the one measured for the T300 CFs, with a value between 0° and 20°. At the mesoscale, an elasto-capillary aggregation effect has been observed during infiltration of the tows. The Washburn method has first been used to interpret the force data obtained by the tensiometric measurements. A combined optical microscopy analysis confirmed that the CF tows immersed in n-hexane were less densely packed than in water, which causes an incorrect evaluation of the tortuosity constant. It suggests that the calculation of effective internal advancing contact angles via the Washburn method is not suitable in the case of free hanging CF tows. However, the combination of the force method and optical analysis permitted to better characterize at the same time the non-solid volume fraction of the tow and the external contact angles. This innovative method provided consistent results between the externally optically observed contact angle and the contact angle obtained from the Wilhelmy force, by correcting for the absorbed liquid. For wetted T300 CF tows, the non-solid volume fraction is 54 ± 3% and the external contact angles are 46.0 ± 29

ACCEPTED MANUSCRIPT 6.1° (force method) and 48.8 ± 3.2° (optical method) respectively; and for the unsized CF tow, the non-solid volume fraction is 46 ± 15% and the external contact angles are 65.1 ± 3.5° (force method) and 64.0 ± 3.4° (optical method) respectively. The effect of surface sizing on fibre wettability at mesoscale was therefore assessed, which will enable better prediction and optimization of adhesion between a given polymer matrix and CF tows. Moreover, contact angles of CFs at meso- and microscales have been successfully linked by using the modified Cassie–Baxter model. By taking into account that the observed tow contact angle is determined by the fraction of solid surface versus imbibed liquid surface (for the wetted tow, and very likely by the surface fraction of solid surface versus air for the un-wetted tow), the solid contact angle can be obtained and this contact angle appeared to be very close to the contact angle measured at the microscale. The wettability of CF tows can thus be predicted from the contact angle obtained at the microscale.

Acknowledgments This work was partially supported by the Interuniversity Attraction Poles Programme (IAP 7/38 MicroMAST) initiated by the Belgian Science Policy Office. J. Wang was also supported by the China Scholarship Council scholarships during his stay at KU Leuven.

References [1] S. Chand, Carbon fibers for composites, J. Mater. Sci. 35(6) (2000) 1303-1313. [2] C. Soutis, Carbon fiber reinforced plastics in aircraft construction, Mater. Sci. Eng., A 412(1-2) (2005) 171-176. [3] M. Suzuki, ACTIVATED CARBON-FIBER - FUNDAMENTALS AND APPLICATIONS, Carbon 32(4) 30

ACCEPTED MANUSCRIPT (1994) 577-586. [4] H. Ogawa, Architectural application of carbon fibers - Development of new carbon fiber reinforced glulam, Carbon 38(2) (2000) 211-226. [5] S. Ramakrishna, J. Mayer, E. Wintermantel, K.W. Leong, Biomedical applications of polymercomposite materials: a review, Compos. Sci. Technol. 61(9) (2001) 1189-1224. [6] F. Lagattu, M.C. Lafarie-Frenot, Variation of PEEK matrix crystallinity in APC-2 composite subjected to large shearing deformations, Compos. Sci. Technol. 60(4) (2000) 605-612. [7] H. Nouri, G. Lubineau, D. Traudes, An experimental investigation of the effect of shear-induced diffuse damage on transverse cracking in carbon-fiber reinforced laminates, Compos. Struct. 106 (2013) 529-536. [8] J. Karger-Kocsis, H. Mahmood, A. Pegoretti, Recent advances in fiber/matrix interphase engineering for polymer composites, Prog. Mater. Sci. 73 (2015) 1-43. [9] E. Mader, K. Grundke, H.J. Jacobasch, G. Wachinger, SURFACE, INTERPHASE AND COMPOSITE PROPERTY RELATIONS IN FIBER-REINFORCED POLYMERS, Composites 25(7) (1994) 739-744. [10] S.L. Gao, E. Mader, S.F. Zhandarov, Carbon fibers and composites with epoxy resins: Topography, fractography and interphases, Carbon 42(3) (2004) 515-529. [11] B.Z. Jang, CONTROL OF INTERFACIAL ADHESION IN CONTINUOUS CARBON AND KEVLAR FIBER REINFORCED POLYMER COMPOSITES, Compos. Sci. Technol. 44(4) (1992) 333-349. [12] W. Song, A.J. Gu, G.Z. Liang, L. Yuan, Effect of the surface roughness on interfacial properties of carbon fibers reinforced epoxy resin composites, Appl. Surf. Sci. 257(9) (2011) 4069-4074. [13] N.I. Kim, H.M. Kang, Y.T. Hong, T.H. Yoon, Plasma etching and plasma polymerization coating of carbon fibers. Part 2. Characterization of plasma polymer coated carbon fibers, J. Adhes. Sci. Technol. 16(13) (2002) 1825-1838. [14] L.C. Ma, L.H. Meng, D.P. Fan, J.M. He, J.L. Yu, M.W. Qi, Z.W. Chen, Y.D. Huang, Interfacial enhancement of carbon fiber composites by generation 1-3 dendritic hexamethylenetetramine functionalization, Appl. Surf. Sci. 296 (2014) 61-68. [15] Y.-N. Lee, S.-M. Chiao, Visualization of dynamic contact angles on cylinder and fiber, J. Colloid Interface Sci. 181(2) (1996) 378-384. [16] S. Zhang, W.B. Liu, J. Wang, B.C. Li, L.F. Hao, R.G. Wang, Improvement of Interfacial Properties of Carbon Fiber-Reinforced Poly(phthalazinone ether ketone) Composites by Introducing Carbon Nanotube to the Interphase, Polym. Compos. 36(1) (2015) 26-33. [17] M. Sharma, S.L. Gao, E. Mader, H. Sharma, L.Y. Wei, J. Bijwe, Carbon fiber surfaces and composite interphases, Compos. Sci. Technol. 102 (2014) 35-50. [18] Y.P. Qiu, W. Xu, Y.J. Wang, M.A. Zikry, M.H. Mohamed, Fabrication and characterization of threedimensional cellular-matrix composites reinforced with woven carbon fabric, Compos. Sci. Technol. 61(16) (2001) 2425-2435. [19] A. Endruweit, A.C. Long, Analysis of Compressibility and Permeability of Selected 3D Woven Reinforcements, J. Compos Mater. 44(24) (2010) 2833-2862. [20] M.E. Foley, J.W. Gillespie, Modeling the effect of fiber diameter and fiber bundle count on tow impregnation during liquid molding processes, J. Compos Mater. 39(12) (2005) 1045-1065. [21] L. Liu, C.Y. Jia, J.M. He, F. Zhao, D.P. Fan, L.X. Xing, M.Q. Wang, F. Wang, Z.X. Jiang, Y.D. Huang, Interfacial characterization, control and modification of carbon fiber reinforced polymer composites, Compos. Sci. Technol 121 (2015) 56-72. [22] A. Bismarck, M.E. Kumru, J. Springer, Influence of oxygen plasma treatment of PAN-based carbon 31

ACCEPTED MANUSCRIPT fibers on their electrokinetic and wetting properties, J. Colloid Interface Sci. 210(1) (1999) 60-72. [23] C.T. Hsieh, W.Y. Chen, F.L. Wu, Fabrication and superhydrophobicity of fluorinated carbon fabrics with micro/nanoscaled two-tier roughness, Carbon 46(9) (2008) 1218-1224. [24] S. Qiu, C.A. Fuentes, D.X. Zhang, A.W. Van Vuure, D. Seveno, Wettability of a Single Carbon Fiber, Langmuir 32(38) (2016) 9697-9705. [25] C.C. Tsai, Y. Gu, K.G. Kornev, Wetting of nanofiber yarns, Colloids Surf., A 459 (2014) 22-30. [26] M.F. Pucci, P.J. Liotier, S. Drapier, Capillary effects on flax fibers - Modification and characterization of the wetting dynamics, Composites Part A 77 (2015) 257-265. [27] T. Pongprayoon, E.A. O'Rear, N. Yanumet, W.L. Yuan, Wettability of cotton modified by admicellar polymerization, Langmuir 19(9) (2003) 3770-3778. [28] E.W. Washburn, The dynamics of capillary flow, Phys. Rev. 17(3) (1921) 273. [29] J.M. Rieser, P.E. Arratia, A.G. Yodh, J.P. Gollub, D.J. Durian, Tunable Capillary-Induced Attraction between Vertical Cylinders, Langmuir 31(8) (2015) 2421-2429. [30] J. Bico, B. Roman, L. Moulin, A. Boudaoud, Elastocapillary coalescence in wet hair, Nature 432(7018) (2004) 690-690. [31] S.H. Tawfick, J. Bico, S. Barcelo, Three-dimensional lithography by elasto-capillary engineering of filamentary materials, MRS Bull. 41(2) (2016) 108-114. [32] S. Tawfick, Z.Z. Zhao, M. Maschmann, A. Brieland-Shoultz, M. De Volder, J.W. Baur, W. Lu, A.J. Hart, Mechanics of Capillary Forming of Aligned Carbon Nanotube Assemblies, Langmuir 29(17) (2013) 5190-5198. [33] M.J. Vega, D. Seveno, G. Lemaur, M.H. Adao, J. De Coninck, Dynamics of the rise around a fiber: Experimental evidence of the existence of several time scales, Langmuir 21(21) (2005) 9584-9590. [34] U. Stachewicz, R.J. Bailey, H. Zhang, C.A. Stone, C.R. Willis, A.H. Barber, Wetting Hierarchy in Oleophobic 3D Electrospun Nanofiber Networks, ACS Appl. Mater. Interfaces 7(30) (2015) 1664516652. [35] A.B.D. Cassie, S. Baxter, Wettability of porous surfaces, Transactions of the Faraday Society 40 (1944) 546-551. [36] Y. Yuan, T.R. Lee, Contact angle and wetting properties, Surf. Sci. Tech., Springer2013, pp. 3-34. [37] http://physlets.org/tracker/. [38] D.F. James, The meniscus on the outside of a small circular cylinder, J. Fluid Mech. 63(04) (1974) 657-664. [39] C. Clanet, D. Quere, Onset of menisci, J. Fluid Mech. 460 (2002) 131-149. [40] E.W. Washburn, Note on a method of determining the distribution of pore sizes in a porous material, Proc. Natl. Acad. Sci. 7(4) (1921) 115-116. [41] M.F. Pucci, P.J. Liotier, S. Drapier, Capillary wicking in a fibrous reinforcement - Orthotropic issues to determine the capillary pressure components, Composites Part A 77 (2015) 133-141. [42] T. Andrukh, D. Monaenkova, B. Rubin, W.K. Lee, K.G. Kornev, Meniscus formation in a capillary and the role of contact line friction, Soft Matter 10(4) (2014) 609-615. [43] A. Bismarck, C. Wuertz, J. Springer, Basic surface oxides on carbon fibers, Carbon 37(7) (1999) 1019-1027. [44] T.D. Blake, Y.D. Shikhmurzaev, Dynamic wetting by liquids of different viscosity, J. Colloid Interface Sci. 253(1) (2002) 196-202. [45] M.J. Vega, C. Gouttiere, D. Seveno, T.D. Blake, M. Voue, J. De Coninck, Experimental investigation of the link between static and dynamic wetting by forced wetting of nylon filament, Langmuir 32

ACCEPTED MANUSCRIPT 23(21) (2007) 10628-10634. [46] Y.W. Yang, G. Zografi, Use of the washburn-rideal equation for studying capillary flow in porous media, J. Pharm. Sci 75(7) (1986) 719-721. [47] E. Chibowski, R. Perea-Carpio, Problems of contact angle and solid surface free energy determination, Adv. Colloid Interface Sci. 98(2) (2002) 245-264. [48] A. Cassie, Contact angles, Discussions of the Faraday Society 3 (1948) 11-16.

33