Formula development of sizing for basalt fiber. Part I: Role of film former

Abstract Sizing is a multi-element liquid system consisting of film former, coupling agent and other functional additives. In this article, Taguchi method with orthogonal array design was used to get an optimal formula of sizing for basalt fiber, and the role of film former in sizing was studied. The results showed film former promoted the sizing to spread on fiber surface, which increased the interaction area between the fiber and sizing. With the assistance of film former in sizing, a uniform layer was formed on fiber surface, which repaired the surface defect on basalt fiber with reduced stress concentration. The sizing with an appropriate concentration of film former had the capability to increase the tensile strength of a single filament with the best effect of forming a film on fiber surface.


Introduction
Basalt fiber (BF), an environment-friendly high-performance filament made from natural basalt rock, has attracted much attention in both academia and industry in recent years (Yan et al., 2021). Besides the high tensile strength and elastic modulus, BF possesses high resistance to alkalis, acids and environmental attacks (Overkamp et al., 2018;Yan et al., 2021). One of the main applications of BF is for fiberreinforced polymers (FRPs). FRPs have been widely used in many fields, such as aerospace, automotive, marine, recreational equipment and construction, due to the advantages of high stiffness to weight ratio and ease of processing (Kocaman et al., 2017;Wan et al., 2019;Zhou et al., 2020), and so on.
Sizing is a kind of surface coating that is applied to nearly all types of artificial fibers during manufacture. Applying a thin layer of sizing on fiber surface can protect the substrate from damage and friction during the production and processing and promote the fiber-matrix adhesion in composites through the coupling effect. Sizing can also regulate the distribution of surface defects, such as scratches and voids, which are commonly found in brittle fibers (Moosburger-Will et al., 2018). Generally, sizing consists of film former, coupling agent, antistatic material, lubricant, and each component has its specific function (Thomason, 2020;Thomason, 2019).
By retrieving the literatures in this field, it was found that massive researches put effort into revealing the function of sizing for BF. Our recent work (Xing et al., 2019) investigated the effects of sizing on the tensile strength of commercial BF. The results showed that sizing, functioning as 'clothing' for fiber, could effectively enhance the strength of BF by reducing the stress concentration in crack tips. It illustrated the importance of such post-treatment for developing high-performance fiber. Ralph et al. (2019) investigated the role of different fiber sizing, including sizing formulated by epoxy and polypropylene, on the properties of BF-reinforced polypropylene composites. They found that sizing changed the topography and chemistry of fiber surface significantly, responsible for the increased mechanical properties and chemical bonding between the fiber and polymer matrix. Wang et al. (2020) prepared starch-based sizing to improve the mechanical properties of BF. The results demonstrated that a uniform and compact coating layer successfully covered the fiber surface. The breaking force of BF was more than 45% higher than that of nonsized fiber. While much progress has been achieved in this field, the function of sizing was not clear in the reported papers, mainly because of the differences in the constituents of reported sizing for BF. In our recent paper (Xing et al., 2021), the effect of type and concentration of constituents in sizing on the mechanical properties of BF was investigated. The results revealed that the sizing brought a positive effect on enhancing the tensile strength of both single and bundle BFs, and the synergy arising from the main film former (Epoxy) and auxiliary one (Polyurethane) in sizing contributed much for such enhancement. Film former, making up the bulk of sizing, is one of the essential components in sizing. It can form a thin layer of film on fiber surfaces to protect the substrate from damage and promote the bonding of single filaments. Unfortunately, there is barely any clear vision of the role of this component on the performance of sizing and corresponding filament. Definitely, this field needs more participation from the composite communities to achieve the high impact it deserves.
In this study, sizing with different components and concentrations was formulated by Taguchi method with orthogonal array design, aiming at optimizing the formulation of sizing for BF. Based on the optimized result, we specifically studied the effect of film former and its content on the mechanical properties of BF, and the role of film former in governing the overall performance of sizing was illustrated, which was the innovation of this article compared with the previous research (Xing et al., 2021

Preparation of sizing and application on fiber surface
E-51 epoxy resin (200.0 g), span-80 (12.0 g) and SDS (6.0 g) were put into a beaker, and the mixture was stirred and dispersed at 2000 rpm using the high speed mixer (HSM-100L, Ross, China) in a 55 C water bath. During this process, water (282.0 g) was added dropwise to the system to obtain epoxy emulsion with a concentration of 40% (Liu et al., 2012). The coupling agent (KH-550) was added into the water with a pH value of 3.5 adjusted by glacial acetic acid and stirred at room temperature for 2 h until the solution was transparent. The lubricant (PEG-MS) and antistatic agent (CTAB) were dissolved with deionized water at 80 C. After the above solutions were prepared, an appropriate amount of deionized water was added to the film former emulsion. Then the antistatic agent, lubricant and coupling agent were added in order and stirred for 1 h. Finally, the solution was thoroughly mixed and the required sizing was obtained.
BF was coated with sizing using a continuous coating system, which consisted of an unwinding device, a set of rollers, hot air dryers, consolidating and enwinding devices.
The temperature in the dryers was set at 120 C, as reported in our previous paper (Xing et al., 2021). The final sample with different sizing was dried at 120 C for 24 h to remove the moisture.

Characterization
The tensile strength of a single filament was measured on a fiber tensile tester with a load cell of 1 N (XQ-1A, Shanghai Xinxian, China) according to the ASTM C1557-14 standard. The gauge length of the sample and loading speed were fixed at 25 mm and 2 mm/min, respectively. At least 40 specimens of each fiber sample were tested and the mean value with standard deviation was reported. The tensile strength of the fiber bundle was determined on a universal testing machine with a load cell of 10 kN (C43-104, MTS, US) according to the ISO 3341-2000 standard. The gauge length of the sample and the loading speed were fixed at 500 mm and 200 mm/min, respectively. At least 10 specimens of each fiber sample were tested. The morphology of BF with different sizing was characterized by scanning electron microscopy (SEM, Phenom XL, Thermo Scientific, US). The micro-mechanical properties of sizing after drying on glass substrate were tested by a nanoindenter (NanoTest Vantage, Micro Materials, UK; Zhu et al., 2015). The surface tension of sizing was obtained by a micro-electromechanical balance system (DCAT 25, Data-Physics, Germany) at room temperature. The viscosity of sizing was measured by a rotational viscometer (LVDV-1, Shanghai Fangrui Instrument Co., Ltd, China). The content of sizing on fiber surface was evaluated by measuring the weight change of sample before and after thermal treatment at 500 C for 30 min using an electric furnace (Yingkou Jianke, China) .

Design of BF sizing
The sizing formula was optimized by Taguchi method with orthogonal array design. Most commercial sizing is an aqueous system with 0.05-10% solid and generally consists of several multi-purpose components (Thomason, 2020). Epoxy resin has excellent physical, mechanical and electrical insulation properties, and the material shows excellent compatibility and reactivity with various substrates. KH-550 is a typical silane with amino-terminated groups, which is widely used as a coupling agent for sizing (Thomason, 2019). PEG-MS has prominent properties for emulsification, dispersity, lubricity and thickening effect. CTAB has excellent permeability and antistatic performance. Based on these advantages, in the current study, Epoxy was used as a film former, KH-550 was employed as a coupling agent, PEG-MS was adopted as a lubricant agent, and CTAB was used as an antistatic agent. Table 1 summarizes the factors and levels used in the experiment. An orthogonal array L 16 (4 5 ) was adopted in this study, as shown in Table 2. The number of columns in the table was 5, the horizontal number of each column was 4, and the total number of experiments was 16. Each factor was independent, the above factors were placed in the column position of the orthogonal table, and the level of each factor was set in the row position. In addition to the four columns of factors, there was an error column, and the range value (R) obtained by the column was used an experimental error. With such arrangement for formula development, the lowest and highest total concentrations of sizing were 2.5 wt% (Sample No. 1) and 9.8 wt% (Sample No. 16), respectively. Table 2 also summarizes the tensile strength of single and bundle fiber samples with different sizing. For pristine BF, the strength of single fiber was 1258 MPa, and the value for the bundle was 0.14 N/tex. Sample with 10# sizing showed the highest single fiber tensile strength of 1802 MPa among the 17 samples, which was more than 40% higher than that of BF without sizing (Sample No. 0). The strength of bundle fiber for this sample was almost the highest one with a reasonable variation.
The tensile strength of BF with sizing belongs to the larger-the better quality characteristics because one of the main functions of sizing is to enhance the mechanical properties of BF. In other words, the higher strength of fiber means better performance of sizing. Therefore, the larger the k i value (kᵢ is the arithmetic average of single fiber strength at the level i of 1, 2, 3 or 4 for the factors), the more significant improvement of sizing on the tensile strength of BF (Detailed analysis was presented in Table S1 in Supplementary Material). The influence of four controllable factors on the tensile strength of BF monofilament is shown in Figure 1. The strength of a single fiber increased at first and then decreased with an increasing concentration of film former, and the k i of A3 reached the maximum. In contrast, the strength increased with a higher concentration of coupling agent, and the one with B4 level shows the highest value. The trend of lubricant agent was similar to that of film former, where k i of C3 had the maximum value. Interestingly, the part of antistatic agent was irregular in which the biggest k i was 1680 MPa at the level of D1. It can be concluded that the formulation of optimum sizing for improving the tensile strength of BF monofilament was A 3 B 4 C 3 D 1 , where the concentration of film former was taken at level 3 (6.0 wt%), the coupling agent at level 4 (1.0 wt%), the lubricant agent at level 3 (0.5 wt%), and the antistatic agent at level 1 (0.1 wt%) (Verification of optimal formulation was shown in Fig. S1 and Table S2 in Supplementary Material).

Function of film former in sizing
The content of film former in sizing is the highest one except water. Therefore, the effect of film former was studied to illustrate the role of this critical component in sizing. Based on the optimal sizing formula, only the concentration of film former was changed for discussion, and the role of silane coupling agent will be the subject of our future studies. Table 3 shows the experimental scheme for this purpose.
The concentration of film former is expected to affect the property of sizing, thus dominating the formation of sizing formed on fiber surface. With this consideration, the sizing with different concentrations of film former was coated on a glass slide to study morphology and micro-mechanical properties of sizing, which was convenient for direct testing and analysis. For sample preparation, the slide was dipped in the prepared sizing and then taken out. This process was repeated 15 times to ensure that the thickness of sizing was sufficient for characterization. The glass slide with different sizing was dried in an oven at 120 C for 120 h to ensure the complete drying of sizing. Figure 2 shows the surface and cross-section morphologies of sizing with different concentrations of film former. At a low concentration of film former, the surface of sizing was uneven with irregular large size pits ( Figure 2A1). The film thickness was inconsistent,  and prominent convex and concave structures were noticed by observing the cross-section of sizing ( Figure 2A2). With increasing concentration of film former, the surface of sizing became smooth with loosely distributed particles ( Figure  2B1), and the number and size of holes on the surface and inside of sizing decreased ( Figure 2B2). When the concentration of film former was 6.0 wt% (F3), the holes were absent in the sizing. Instead, many particles were evenly attached ( Figure 2C1), and the film became more serried inside ( Figure 2C2). In the extreme case, saying the concentration of epoxy was 8.0 wt% (F4), the distribution of particles in sizing became more uniform ( Figure 2D1), and the film became denser and compacted ( Figure 2D2). The reason for these observations was that when the concentration of film former was low, the sizing lacked the driving force for spreading, thus resulting in the formation of an uneven surface with the evaporation of water. When the concentration of epoxy was high, the film former drove the whole solution of sizing to further spread out on the substrate and finally formed a uniform, flat and dense film.
The nano-indentation test quantitatively measured the micro-mechanical properties of sizing. In this test, the indentation depth was fixed at 1000 nm, the loading and unloading speed was set at 0.05 mN/s, and the hardness (H, GPa) and reduced modulus (E r , GPa) of the material were calculated according to the following equations (Burghard et al., 2007;Malzbender et al., 2002): Where P max (mN) is the maximum indenter load, A c (nm 2 ) presents the projected area of elastic contact, h c (nm) stands for the elastic displacement of indenter, h max (nm) means the indenter displacement at the maximum load, S suggests the slope of unloading curve at the maximum load, e is a parameter related to the indenter geometry (For Berkovich indenter used in this study, e ¼ 0.75). Figure 3 shows the load-depth profiles of different sizing samples. A general trend was demonstrated that the sizing  with a higher film former concentration was always accompanied by a greater load at the same indentation depth. Specifically, when the indentation depth was 1000 nm, the maximum load of the sample with F1 sizing was about 1.3 mN, and this value increased by more than 100% to 2.7 mN for F4 sizing. During the unloading, the creep occurred due to the viscoelasticity of sizing, resulting in a displacement discontinuity known as the pop-out effect, and this was caused by the relaxation of the molecular chain in the epoxy (de la Rosa-Fox et al., 2007;Liu et al., 2019). The inset in Figure 3 summarizes the obtained H and E r of four sizings. The value of both parameters increased with increasing concentration of film former, indicating the improved ability of sizing to resist mechanical deformation.
The reason behind such observation was that film former promoted the sizing to form a uniform layer on the surface of the glass slide. The higher the concentration of film former, the more uniform and dense film created (As confirmed by the SEM images shown in Figure 2), thus leading to the increased hardness and modulus in the corresponding sizing.
To study the influence of film former concentration on the morphology of BF, fiber samples were characterized by SEM, and the results are presented in Figure 4. The pristine BF exhibited a smooth surface ( Figure 4A). Under higher magnification, it was found that there was no other substance on fiber surface except a few small contaminated particles (inset in Figure 4A). After the application with sizing, the surface of BF changed significantly with the variation on the concentration of film former in sizing. At a low concentration of film former ( Figure 4B), a large number of irregular flakes and granular agglomerates were attached to fiber surface, making the fiber surface rough. It can be seen from the section morphology that the content of sizing on fiber surface was low and the film was not continuous (inset in Figure 4B). As the concentration of film former increased ( Figure 4C), the size of aggregates on fiber surface decreased, some areas became flat. At this time, the continuous film of sizing on fiber surface was thinner (inset in Figure 4C). When the film former concentration was 6.0 wt% ( Figure 4D), the surface and section morphologies of BF showed that the sizing had the best spreading effect on fiber surface with a uniform film, and the particulate material on fiber surface was further refined, making the sizing appear smooth without obvious agglomerates (inset in Figure 4D). This was because, with the assistance of film former, the spreading ability of sizing on fiber surface was improved, and the whole components were more uniformly dispersed on fiber surface to form a thin layer of sizing.  However, when the film former concentration continued to increase ( Figure 4E), the fibers were wrapped by a thick layer of sizing, and the bonding phenomenon occurred between the individual filaments. When the single fiber was separated from the bundle, the fiber surface became irregular and even sizing was peeled off from the filament to form a discontinuous film. Definitely, this was not the case that an ideal sizing should be accounted. Table 4 summarizes the properties of sizing and tensile strength of corresponding fiber. With an increasing concentration of film former, the surface tension of sizing was increased. For example, the surface tension of F1 was 24.76 mN/m, and this value was increased by 10% to 27.38 mN/m for F4. This was caused by the higher amount of epoxy, which was a polar component in sizing (Abbasian et al., 2004). The viscosity of sizing with different concentrations of film former changed marginally, and specifically, the viscosity of all sizing was in a low range of less than 2.5 mPaÁs, suggesting the effect of sizing viscosity on the properties of BF could be neglected (Wang et al., 2017, Xing et al., 2021. From this table, it could be observed that the diameter of BF with different sizing changed marginally, about 15 mm, indicating that the variation on the film former concentration had little effect on the diameter of BF. The fiber with F3 sizing exhibited the highest tensile strength of 1967 MPa, with an improvement of 45.60% when compared with that of pristine BF. It seemed that the higher concentration of sizing for BF brought a negative effect on enhancing the strength of the filament. SEM image showed that with excessive content of film former (Figure 4E), the fiber was severely covered by sizing. The whole sample was more like a composite structure, thus limiting the enhancement of tensile strength for the filament.
The relationship between the sizing content on fiber surface and the concentration of the film former is shown in Figure 5A. It demonstrated the content of sizing increased with an increasing concentration of film former. The tensile strength of single fiber with F3 sizing was the highest in the experimental scheme. Whereas the content of sizing was 3.94%, not the case that the higher the sizing applied to the fiber, the better mechanical performance of fiber demonstrates. Therefore, moderate content of sizing on fiber surface played a significant role in promoting the mechanical properties of BF, and these results were in good agreement with our previous findings (Xing et al., 2021). Figure 5B schematically shows the distribution of defects on BF surface with different concentrations of film former. Film former made the mixture of sizing adhered to the fiber surface evenly, and it covered more than 80% content of sizing except for water. Therefore, the higher the concentration of film former was used, the more sizing content on fiber surface was formed. When the concentration of film former was low (Condition I in Figure 5B), the film of sizing was thin, the physical and chemical interactions between BF and sizing were weak, and a small number of the surface defects could be repaired, such as the BF with F1 or F2 sizing. As a result, the mechanical properties of fibers were improved slightly. When the content of the film former was proper (Condition II in Figure 5B), like the BF with F3 sizing, the tensile strength of BF with sizing was enhanced dramatically, which was contributed from the reparation of defects on fiber surface and decreasing stress concentration. When the content of film former was too high (Condition III in   Figure 5B), like the fiber with F4 sizing, the tensile strength of BF with sizing also increased, but the improvement effect was limited due to the inherent poor mechanical performance of sizing (a mixture of polymer) on fiber surface. It is known that the mechanical properties of fiber generally outperform that of the matrix, and the introduction of excessive polymer onto fiber in the form of sizing contributes marginally to the strength modification of filament. Hence, it is important to select an appropriate concentration of film former to improve the mechanical properties of fibers.
To reveal the mechanism behind the enhanced strength of fiber with sizing, we examined the cross-section area of fractured fiber using SEM. For a brittle material, the fiber was stretched at a constant speed until the breakage, and the fractured morphology of the fiber was composed of several regions, known as local failure, mirror, mist and hackle (Lund & Yue, 2008). This process is schematically shown in Figure 6A. Under low stress condition, the main crack of the fiber slowly expanded through a single crack to form a flat and smooth semi-circular mirror area. As the applied stress gradually increased, non-dangerous sub-weak points in the material became dominant, and mist region was formed with expended crack. Since the primary and secondary cracks did not usually expand in the same plane, the region was no longer as smooth as a mirror. When entering the stage of rapid propagation, the rough area of the hackle was shaped by simultaneously expanding many secondary cracks on the uneven plane. With the increase of tensile force, the stress concentration appeared at the defect with a large size first, followed by the rapid propagation and the brittle failure when the stress exceeded the critical value. The relationship between fracture strength r (MPa) and crack size a (mm) can be expressed by the following two equations (Lund & Yue, 2008): where A m is the mirror constant related to the material property, it is generally believed that the mirror radius r m (mm) near the crack initiation point and the strength of the fiber r follow the empirical formula (6): E (MPa) is the elastic modulus of the material, c (mN/m) is the surface energy of the material, Y is a geometry parameter of the material (Y ¼ 1.39 for elliptical surface flaw, and Y ¼ 1.13 for circular internal pores). By combining the above two formulas, the following one was expressed, which described the relationship between the crack size a and mirror radius r m : In this study, the elastic modulus of BF was remained at around 75 GPa due to the inherent nature of the material, and the surface energy of pristine BF was 55.45 mN/m   (Xing et al., 2021). The surface tension of four sizings was about 26 mN/m, as summarized in Table 4, so the surface energy of pristine filament was much higher than that of BF with different sizing, and this was also reported in a previous study (Chen & Huang, 2021). The fiber strength can be qualitatively expressed in Equation (7), in which E and Y are constant, and c was reduced due to the introduction of sizing on fiber surface. The r would be improved only when the a was reduced. The variation in crack size a on fiber surface can be directly correlated with the mirror radius r m , as expressed in Equation (8). Results arising from the cross-section area of fractured fiber showed that the average r m of pristine BF was 3.07 mm ( Figure 6B), and this parameter was reduced to 1.85 mm with F3 sizing ( Figure 6C). It seems that the content of film former played a leading role in governing the value of r m , and the value of the latter was saturated when the concentration of epoxy was 6.0%, as shown in Figure  6D. The reduced r m led to a smaller a on fiber surface, thus the r of fiber was enhanced according to Equation (8). This can be qualitatively expressed as a healing effect of sizing for fiber samples.
To reveal the mechanism of improving fiber strength caused by sizing, the steady-state stress distribution of the two-dimensional fiber tensile model was simulated by using the solid mechanics module of COMSOL software. One was the partial original BF with a surface crack, another was the model that the crack on BF surface was filled by sizing. The upper end of BF was fixed, and the lower end was applied force to simulate the tensile process of BF. The parameters used in the simulation are summarized in Table 5, and the results are presented in Figure 6E. It can be seen that when pristine BF was subjected to tensile force, the stress concentration occurred at the tip of crack on fiber, and the force dispersed to the periphery like butterfly wings from the tip of the crack. After applying the sizing, the stress concentration at the crack tip was decreased, and both ends of fiber with sizing had the capability to bear more stress. The force for BF was more evenly dispersed, which was the reason that BF with sizing exhibited a higher tensile strength.

Conclusions
In summary, Taguchi method with orthogonal array design was implemented to formulate the optimum sizing for BF, and the effectiveness of this methodology was confirmed by measuring the tensile strength of the individual fiber. The results showed that sizing brought a positive effect on enhancing the tensile strength of BF. Additionally, our comparative data demonstrated that film former in the sizing drove the mixture to spread on fiber surface to form a uniform film. The sizing diffused and formed a film by filling the defects on fiber surface. Under such circumstances, the defects were repaired, resulting in reduced stress concentration and improved mechanical properties of BF. Sizing with a suitable concentration of film former had the best healing effect on rectifying the mechanical properties of BF.

Disclosure statement
No potential conflict of interest was reported by the authors.