Abstract
Engineered cementitious composites containing polymeric fibers such as poly(vinyl alcohol) show high tensile ductility and toughness. The aim of the current study was to evaluate the effect of crystallinity degree of poly(vinyl alcohol) fibers on fiber debonding from cementitious matrix in the bending test. In this work, two types of poly(vinyl alcohol) fibers and one grade of polypropylene fiber were separately incorporated in cementitious composites. The chemical structure, surface chemistry, roughness, and microstructure of fibers were examined by Fourier transform infrared, attenuated total reflection, atomic force microscopy, and wide-angle X-ray diffraction tests, respectively. The compression and flexural behaviors of cementitious composites were also assessed. Attenuated total reflection results were indicative of similar surface chemistry for both poly(vinyl alcohol) fibers, while the main difference was observed in the case of degree of crystallinity, which plays an important role in the Poisson’s ratio. Finally, the way in which the degree of crystallinity and Poisson’s ratio of fibers can lead to premature debonding was described and confirmed by scanning electron microscopic images.
Keywords
Introduction
Concrete and traditional cement composites are brittle materials with high compressive and yet low tensile strengths. Because of their brittle nature, cracks can appear under very small tensile strains and propagate to cause a total fracture. Therefore, short and long fibers have been used to increase the toughness of cementitious matrixes. It has been found that the presence of fibers in the cementitious composites decreases the rate of crack propagation and increases toughness, ductility, and energy absorption capacity.1,2
In the 1960s, short steel fibers were used in concrete to reduce its brittleness. 3 Various natural and synthetic fibers such as sisal, 4 cellulose, 5 asbestos, 6 glass, 7 carbon,8,9 poly(vinyl alcohol) (PVA),10–12 polypropylene, 13 and polyethylene 14 were applied by numerous researchers to reinforce concrete. In addition, hybrid fiber systems were used to improve the tensile performance of cementitious composites. 15 The fiber reinforced cementitious composites (FRCCs) have been well studied in the last two decades.16–18 Typical FRCCs exhibit softening behavior after the first cracking. High-performance fiber reinforced cementitious composite (HPFRCC) is a class of cement composites, which shows strain hardening behavior accompanied by multiple cracking (leading to a high strain prior to failure) under a tensile load. 19 Multiple cracking and strain hardening behaviors can improve the ductility and the load bearing of the composites after the first crack, respectively. To put it simply, the strength of a strain hardened composite is higher than its first cracking strength. SIMCON (slurry infiltrated mat concrete) and SIFCON (slurry infiltrated fiber concrete) are two known examples of HPFRCCs, 20 which have been designed using a great volume fraction (up to 20%) of fibers. The main drawbacks of this design method include the diminished workability of fresh concrete and the high cost of this procedure.
Li et al.21–23 developed an engineered cementitious composite (ECC) based on the principles of fracture mechanics and micromechanics of short random fiber composites. One of the major advantages of ECC is its small volume fraction of fibers (about 2%), which does not cause a considerable reduction in the workability of fresh mixture. ECC resembles ductile metals showing pseudo strain hardening and multiple cracking properties under tension. 24 Similar to FRCC, ECC consists of water, cement, sand, fiber, and common chemical additives, but the coarse aggregates are eliminated and maximum aggregate size should be lower than 300 µm.
Using a relatively small volume fraction of fibers and its chopped nature eases the mixing process of ECC similar to what is seen for normal concrete. Controlled crack opening (usually less than 100 µm) improves durability, 25 impermeability, 26 and ultimate tensile strain (up to 7%). 27 Furthermore, fine cracks in ECC can be easily self-healed through moisture or water absorption, as a result of hydration reaction in pure remained cementitious material or the growth of old cementitious reactions.28,29 All desired properties of ECC are attributed to the accurate tailoring of matrix, fiber and interface. 30 By changing the ECC formulation, its application can be extended to different areas such as shotcrete ECC, 31 extrudable ECC, 32 and self-consolidating ECC. 33
As mentioned above, ECC is designed based on micromechanics and fracture mechanics. It has shown that the relationship between stress (σ) on crack and crack opening (δ) is a determinant factor for ECC. 34 The σ–δ behavior depends on fiber bridging conditions and explains how the constituent properties lead to tensile ductility.
Li and Wu
35
developed a model for pseudo strain hardening materials under a tensile load. To attain strain hardening behavior, some conditions must be satisfied; for example, the first cracking strength of matrix (σfc) must be less than the maximum fiber bridging capacity
Otherwise, the initial crack results in the fiber rupture and sudden failure. Furthermore, the steady-state cracking, which was first analyzed by Marshall and Cox,
36
must be dominant. According to this criterion, the crack tip toughness (Jtip) should be less than the complementary energy (Jb) calculated from σ–δ curve as
ECC has been first developed using ultra high molecular weight polyethylene fiber, which is a very expensive material. PVA fibers, which typically have a tensile strength ranging from 1600 to 2500 MPa and high elastic modulus of about 40 GPa, are likely to improve the toughness of brittle matrices such as concrete, mortar, etc. 37
There has been a great deal of research on the effect of PVA fibers on cementitious composites. It has been reported that very strong chemical interactions of PVA fibers with cementitious matrices lead to fiber rupture and prevent strain hardening of the resulting composites. 38 PVA fibers should be coated by some oiling materials to reduce its chemical bonding. 30
In this paper, two types of PVA fibers and one grade of polypropylene (PP) fiber were separately used in fine sand mortar matrices. The compression and flexural behaviors of the prepared composites were evaluated. The bulk and surface chemistry of PVA fibers were also studied by Fourier transform infrared (FTIR) and attenuated total reflection (ATR) techniques. The surface roughness and topography of each intact fiber were studied by atomic force microscopy (AFM). Moreover, the degree of crystallinity of PVA fibers as one of the main physical properties of polymers was assessed by wide-angle X-ray diffraction (WAXD). An attempt is made to draw attention to the role of microstructure of fibers in the performance of composites under flexural loads. Specially, the results illustrated how fiber microstructure may be engineered to satisfy the necessary conditions to achieve strain hardening and multiple cracking behavior.
Materials and methods
Chemical ingredients of cement and silica fume.

XRD diagram of silica fume.
Physical properties of the fiber used.
The notation used to nominate samples contains three parts. The first part determines the type of fiber used (PVA1, PVA2, or PP fiber). Ref. refers to a sample without any type of fibers. The second part states the percentage of fiber volume fraction. To prove repeatability, at least three samples were prepared for each fiber type, referred to as T.1, T.2, and T.3 in the third part.
The mix design of mortar.
X/CM means the weight of X per the weight of cementitious material.
Volume of fibers per total volume.
Compression and flexural tests were performed by a Zwick-1494 Instrument with a rate of 5 mm/min. The specimens for flexural tests were prepared in molds with dimensions of 500 × 100 × 30 mm3, while those for the compression test were casted in cubes with 50 × 50 × 50 mm3 dimensions. Four-point bending test and compression analysis were carried out according to EN 12467 and ASTM C109, respectively. At least three samples were analyzed to confirm the repeatability of each experiment. The modulus of rupture (MOR) can be calculated as follows
The topography and surface roughness of fibers were studied by ARA-AFM, an Iranian made AFM (manufactured by Ara-Research Company in Tehran) in noncontact mode. The study was carried out at room temperature under air atmosphere and a HQ: NSC15/Al BS silicon probe from MikroMasch Company was employed. Moreover, to observe the morphology of the fractured surfaces and pull out fibers, field-emission scanning electron microscopy (FESEM) was performed by means of a Hitachi S4160 instrument using 5 kV secondary electrons. The specimens were mounted and coated with gold before imaging. FTIR and ATR tests were carried out using a Nicolet NEXUS 670 apparatus to analyze the chemical structure of bulk and surface of fibers, respectively. IR characterization tests were conducted at normal temperature (20℃) and pressure (1 atm.). Crystallography of PVA fibers was carried out using a diffractometer with CuKα1 X-ray source and wavelength of 1.5406 Å operating at 40 kV and 30 mA radiation (Inel EQUINOX 3000). Scattered filament current was detected in the range of 2θ = 5°–40°, at a speed of 2°/min.
Results and discussion
In this section, the physico-chemical properties of fibers as well as the mechanical properties of the prepared composites will be reported and discussed. Figure 2(a) shows the results of the compression test for all samples. The sample with the highest strength was selected for comparison purposes. As illustrated, all composite samples can bear higher stress levels compared to the reference one, but some negligible differences were observed between the composites. In addition, no sudden failure in fiber reinforced specimens was observed and their uniformity held until the end of experiments. These advantages of FRCCs can bring great benefits to earthquake, fire and explosion spalling safety.25,39–41
(a) Compression test results of various composites, flexural test results for (b) PP, (c) PVA1 and (d) PVA2 contained composite.
Generally, the main effects of fibers can be observed under tensile or bending loads. Here, the bending response of the composites has been studied. Figure 2(b) to (d) shows the effects of different fibers on bending response of the specimens. The composites made by PP fibers exhibit a behavior similar to conventional FRCCs, as the bearable load increased to a maximum point followed by stress softening until a large strain was reached. Although the surface area below the bending curve of the PP fiber composites was large and energy absorption was improved (i.e. a high toughness was achieved), after the first cracking, the tolerable load decreased, loading could not be increased and the cracked specimen was softened. The fluctuations observed in the stress softening region of the curve can be attributed to the successive loading and breaking of bridging fibers when the cracks propagate. The weak interface of PP fibers due to their hydrophobic nature leads to an immediate debonding from cementitious matrix. If the embedded length of fiber was less than its critical value, the fiber would be easily removed from matrix. In contrast, a long PP fiber makes a good physical interlocking between fiber and matrix. The large deflection observed in PP composites in Figure 2(b) can be attributed to the easy long fiber pulling out and the large deformation of fibers, due to the low elastic modulus and poor yield stress of PP fibers.
With the start of the first crack, the load is transferred to the PP fibers, which can be easily pulled out, due to the weak interfacial bond between the PP fiber and cementitious matrix. Moreover, long fiber pull out results in high deflection values. In addition, fibers may elongate, leading to composite softening. Although PP fiber composites were associated with high toughness and deflection, the overall behavior of the composites was still stress softening, as only a few cracks could be observed on the specimen, while multiple cracking was not seen. In other words, composite integration was destroyed. It should be mentioned that the desirable properties of ECC such as self-healing can be only achieved through multiple cracking.
When the fiber–matrix interface bond is too weak, the fibers will pull out resulting in a small σ0. However, in the case of a too strong fiber–matrix interface, the bond cannot stretch; hence, the fibers will not slip and will prematurely rupture leading to a small δ0 (see equation (3)). In either case, the composites will show a small complimentary energy, failing under Griffith cracking. 42
Figure 3 shows a PP composite under flexural load with only one wide crack. Immediately after the first cracking, the composites lost its integrity and only the fibers in the break zone could transfer the load, which was not desirable. Accordingly, this crack propagated to cause complete breakdown. Usually, the multicracks usually could not be seen on the specimen, while several parallel cracks appeared in ECC during flexural loading, demonstrating a proper load transfer through bridging fibers.
The first crack propagation to cause complete breakdown in PP fiber composites.
Strain hardening originates from slip hardening, which refers to a process where an increase in the length of a pulling out fiber leads to an increase in the required force. To put it simply, the force required to pull out the fiber (slipped out) from the matrix directly is directly increased as the pull out fiber increases. Since fibers cannot be moved unless the composite cracks, slip hardening cannot be observed before matrix cracking. After matrix cracking, if matrix strength is lower than the bridging strength of fibers, the increment of stress (as a result of slip hardening) on the composites, i.e. the growth of stress on the bridging fibers, may result in another crack in a weak region of the matrix whereby multiple cracking can take place by repeating this sequence. During crack propagation, stress is increased, since the number of fibers exiting from matrix (with slip hardening mechanism) is increased and thus strain hardening and multiple cracking simultaneously occur.
As shown in Figure 4, the strength of the first cracking was reduced in composites containing PVA1 fibers, because cracks could nucleate from PVA1 fibers with 27 µm diameter. In fact, thick PVA1 fibers could act as defects in the matrix. In contrast, the PVA2 fibers with 15 µm diameter were thin enough to prevent cracks from nucleation. Therefore, the first cracking strength of PVA2 composites was approximately equal to that of the reference samples. It is notable that while the number of thinner fibers was more than that of the thicker ones in a constant volume of fibers, low deflection and weak energy absorption were observed in the case of the composites containing PVA2 fibers. In the following, it has been attempted to provide a rationale, where it has been shown how the fiber geometry and microstructure can affect the post cracking behavior of composites under flexural loads.
Typical flexural curves of PVA1 and PVA2 composites.
Although based on the literature, reinforcing cementitious matrix with PVA fibers leads to strain hardening and multiple cracking,
43
where some differences in the post cracking behavior of PVA1 and PVA2 could be observed. To precisely compare the post cracking behavior of the composites, a novel criterion was used. The nonlinear part of each curve was fitted with a line from the onset of cracking to the failure point. The positive or negative slope of the line was considered as a measure of strain hardening or stress softening, respectively. Figure 5 depicts the slope of the curve for each sample along with its averaged value. The averaged slope for PVA1 composite was 0.97 ± 0.2, a positive value that indicated the strain hardening behavior. In contrast, the averaged slope for PVA2 composite was negative (−1.14 ± 0.9) implying stress softening. Since the same matrix was used, the difference between the behaviors of PVA1 and PVA2 composites under flexural load could be attributed to the differences in fiber and interface.
Slope of nonlinear section of bending response of PVA composites.
FTIR spectroscopy of two PVA fibers demonstrates that the bulk material of both fibers were poly (vinyl alcohol), as illustrated in Figure 6. ATR was also used to study the surface chemistry of fibers. Figure 7 presents the ATR spectra of PVA fibers. As shown, ATR spectrum of each fiber was dissimilar to its FTIR spectrum indicating that something might have been coated on the fiber surface. As mentioned, to achieve strain hardening behavior, it is essential to coat oil on PVA fiber regarding the strong chemical and frictional bonds between fiber and cementitious matrix.38,44–46 However, little data have been published regarding coating materials. IR absorptions in the regions ∼1260, 1080, and 1000 cm−1 respectively show Si–CH3, Si–C, and Si–O bonds, which are the characteristic bands of the silicone oil. The sharp peak in region ∼1730 cm−1 can be attributed to the carbonyl end groups of polydimethylsiloxane (PDMS).
47
Therefore, the infrared characteristic peaks show the presence of PDMS on the fiber as coating. The chemical structure of PDMS has been also depicted in Figure 7.
FTIR spectra of both PVA fibers. ATR spectra for both PVA fibers and the chemical structure of PDMS.

In addition, more evidence for coating PDMS on PVA fibers could be provided. It is known that similar surface tension values are required to wet the substrate by a coating. 48 At 20℃, the surface tension of PDMS, PVA and water (as the solvent for PVA) have been previously reported to be 22, 50, and 72.8 mN/m, respectively.49,50 Furthermore, Li and Wang 51 have suggested that paraffinic and silicon oils can be coated on PVA fibers in cementitious composites to improve the fiber pull out and decrease the fiber rupture. Altogether, the comparison of ATR spectra of PVA grades revealed their similar coatings. Therefore, the different surface chemistry of fibers cannot be considered as a reason for their different flexural behaviors.
Since the matrix as well as the volume fraction, bulk and surface chemistry of fibers are analogous, it seems that fibers may be physically different.
The topographical scan of each intact fiber was studied by AFM and the results have been presented in Figure 8(a) and (b). Figure 8(c) presents the height of each point placed on an arbitrary line in the direction of fiber length. In comparison with PVA1, rougher surface has been observed by root-mean-square roughness for PVA2 fiber. RMS roughness, calculated over an area of 25 µm2, was about 2.07 and 11.6 nm for PVA1 and PVA2, respectively. It is worthy to note that the rougher surface observed for PVA2 fiber can justify the higher first cracking strength in Figure 4.
AFM results for (a) PVA1, (b) PVA2 fiber and (c) the height of each point on an arbitrary line in direction of fiber length.
It is known that the crtitical length of a short fiber is a direct function of fiber diameter as
One of the crucial properties of each polymer is its degree of crystallinity. The WAXD patterns of both PVA fibers have been shown in Figure 9. To estimate the degree of crystallinity Origin software was used, in which the crystallinity content was calculated by the sum of the surface area under the fitted curve per total surface area under the whole curve. PVA1 fiber exhibited a higher degree of crystallinity of about 83% in comparison with PVA2 with about 48% crystallinity. It is well known that crystallinity and density of any semi-crystalline material depend on each other. The degree of crystallinity can be calculated as the following
52
WAXD diagrams of PVA1 and PVA2 fibers.
The densities of PVA1 and PVA2 fibers were calculated to be about 1.332 g/cm3 and 1.305 g/cm3, respectively. Increasing the degree of crystallinity enhanced the density and a higher density, in turn, implies a less occupied volume for a determined mass. The occupied volume of materials includes the volume of rigid atoms plus the free volume between the ingredient atoms. The free volume can be calculated by
The
In other words, the crystalline regions (similar to cross linked points) reduce the mobility of polymer chains due to the lower free volume, which leads to a decrease in Poisson’s ratio.54,55
It is worthy to note that increase in the crystallinity of materials at a constant temperature affects the tensile deformation, equivalent to how decreasing the temperature for a polymer material with a constant crystallinity does. 54 Nitta and Maeda 56 showed similar effects in the creep behavior of high density polyethylene under a fixed true stress. The increase in the weight fraction of amorphous phase enhances the overall molecular mobility or extends the experimental timescale which is equivalent with the enhancement of molecular mobility, due to a rise in temperature.
In addition, the results of a study conducted by Urayama et al. 57 confirm the relation between Poisson’s ratio and free volume of polymers. Assessing the Poisson’s ratio of PVA gels in different conditions revealed that the flexibility level of chains is responsible for the increase in the Poisson’s ratio.
In the current study, PVA2 fibers contained more amorphous regions that led to a high Poisson’s ratio and a high lateral shrinkage, which in turn might have caused fiber debonding and reduced the true contact area between the fiber and the matrix.
Premature fiber debonding and interface weakening as a result of the difference between Poisson’s ratio of matrix and fiber have been pointed out for fiber reinforced materials in previous studies. 58
It is believed that the frictional force depends on the true contact area, which is formed by the asperities of two adjacent surfaces. As the normal force on the moving object increases, the rough regions come into contact and the average area of contact grows, increasing the friction. Zhang et al. 59 have shown that due to the high cement content in ECC (compared to the conventional concrete), a high drying shrinkage strain is produced. The high drying shrinkage exerted in the radial direction increases the clamping stress on the fibers and therefore, the frictional bond strength is enhanced. In this work, the matrix used for both PVA1 and PVA2 composites were similar and the drying shrinkage was constant. According to the above discussion, PVA2 had a minor degree of crystallinity, lower density and more free volume showing higher lateral contraction (greater Poisson’s ratio). When tensile stress is applied, the radius of fibers with higher Poisson’s ratio are significantly decreased and the true contact area which is a fraction of the apparent contact area is reduced. Consequently, the intimate contact between the fiber and the matrix is eliminated and interface weakening and fiber debonding from its tunnel in central section of the fiber is more probable.
After semi-debonding of the fibers, pulling out of contracted fibers can easily occur which reduces the friction between the fiber and its matrix. Therefore, the observed stress softening behavior in PVA2 fiber composites has been due to fiber semi-debonding and the easy exit from the matrix. The FESEM images from the fractured surface of PVA2 composite under flexural loads support this idea (Figures 10 and 11). Both images shown in Figure 10 are related to the fractured surface of PVA2 composite under flexural load. As it can be observed, the diameters of the two parts of fiber are different. The left side image guidelines show the thicker section of the fiber with a diameter of 15 µm similar to the virgin fiber, while the right-side image lines show the thinner part with a diameter of ∼11 µm. Such change in the diameter was attributed to the high Poisson’s ratio of PVA2 fiber. In contrast, the scratched surface of PVA1 fiber implied that the pull out process has been accompanied by friction and energy consumption.
SEM images of PVA2 showing the shrinkage of fibers under tensile load (a portion of fiber is thinner). Ruptured PVA2 fiber in the tunnel of matrix after shrinkage of fiber under tensile load.

On the other hand, the high radial shrinkage of PVA2 fibers (with high Poisson's ratio) under tension load and reduction of the intimate contact between fiber and its tunnel can lead to some repercussions. First of all, the load transfer mechanism cannot be conducted by fibers and multiple cracking is prevented. In other word, when the diameter of PVA2 fibers was reduced, the friction force i.e. the physical bond between fiber and matrix was reduced and therefore, bridging strength of fibers decreased to lower levels in comparison with the first cracking strength of matrix, i.e. σfc > σ0. Under these conditions, the first criterion for strain hardening was violated and steady-state cracking could not be dominant. Secondly, diminishing the friction force between fibers and the matrix decreased the consumed energy and contributed to sudden fiber pull out.
Radial shrinkage of fiber under tensile load and the lack of an intimate contact between the fiber and tunnel surface as well as the presence of some voids in this zone have been depicted by a white oval in Figure 11.
In fact, due to the high degree of crystallinity and low Poisson’s ratio, PVA1 fiber slips out from cementitious matrix with a slip hardening mechanism. The slip hardening mechanism has been addressed in other studies resulting in peeling and scratching as illustrated in Figure 12.30,37,45 In general, the ideal conditions for a good strain hardening behavior are attained when the fibers can slightly slip out from the cementitious matrix without debonding. The friction remains approximately constant or even increases and fiber movement remains as an energy consumer process.
Scratched and peeled PVA1 fiber exited with friction from cementitious matrix.
In composites containing PVA1 fibers, as fibers almost retain their primary diameter under tensile load, interlocking of fibers and matrix is maintained and physical bonding prevents interface debonding of PVA1 fiber. Consequently, debonding can occur in the interphase zone. The observed scratches on the surface of pull out PVA1 fibers under bending load confirm the interphase debonding. Also, matrix micro spalling is an alternative mechanism for interphase debonding in ECC, which has been described in other studies.45,60
Conclusions
The current study aimed at evaluating the effect of crystallinity of PVA fibers on the bending behavior of cementitious composites. All fibers increased the compressive strength of composites, preventing the sudden failure of the samples. Although same oil coating and similar bulk chemistry were proved by ATR and FTIR techniques for both PVA fibers, some differences were seen in the flexural performance of cementitious matrix. The nonlinear section of the bending curve was fitted by a line, the slope of which was considered as a novel criterion to distinguish between strain hardening and stress softening behaviors. A positive slope for composites containing PVA1 implied the strain hardening behavior. In contrast, PVA2 composites showed a negative slope (stress softening) in post cracking section of the bending curve.
It was also pointed out that different degrees of crystallinity and consequently Poisson’s ratios can result in different debonding modes of fibers. PVA1 fiber with a high degree of crystallinity and low Poisson’s ratio retained the primary diameter during the fiber pull out. So, interlocking of fibers and matrix was maintained, being indicative of debonding at the interphase. In this regard, scratching of fibers could be observed in SEM images. Conversely, PVA2 fiber with high Poisson’s ratio radially contracted under tensile load and the intimate contact between fiber and its tunnel was declined. It could lead to some repercussions, first the load could not be transferred by fibers and multiple cracking was prevented. The friction between fiber and matrix reduced and fibers could easily exit.
In conclusion, the most appropriate behavior was observed in the case of PVA1 composite. Self-healing, safety, and corrosion resistance, which can be attained by multiple microcracks, are some of its advantages. PVA1 fiber composite behaved similar to ECC, which can be used in many applications such as earth retaining walls, dams, bridges, etc.
Although PVA2 fiber is chemically similar to PVA1, its composite did not show a strain hardening behavior and cannot be considered as an ECC. The overall post cracking behavior of PP composites was similar to conventional FRCCs. PVA2 and PP fiber can be proper candidates for nonstructural applications, because their load bearing is decreased after first cracking. Furthermore, for applications without an increasing load, such as drying shrinkage, PP and PVA2 may be utilized.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
