Degradation Kinetics of Cellulosic Fiber in the Alkali Environment of Cement

Keywords: A. Natural fibers; A. Cellulose; B. Chemical properties; D. Microstructural analysis

1. Introduction

In the quest for green and energy-efficient infrastructure, the requirement of renewable, lightweight, low-cost and high-performance construction materials with exceptional durability was proposed. Unfortunately, renewability and durability tend to be mutually exclusive, and the application of natural materials is always a compromise between environmental sustainability and long-term service. A typical representative of this is the application of cellulosic fibers as reinforcement for cement based construction materials to improve their ductility and crack resistance since ancient times (e.g. the straw and horsehair used by Egyptians [1] and biomass stalks and weeds used by Chinese). Since the 1970s [2], following the banned use of asbestos, cellulosic fibers reawakened interests of researchers in the field of infrastructure engineering due to their readily availability in large quantities in many countries and their continuous renewable source. From then on, a systematic evaluation of engineering properties of composites containing these biomass reinforcements was performed. With excellent mechanical properties, sisal fiber is a typical representative of cellulosic reinforcements for cementitious materials.

As shown in Fig. 1, the cell wall of cellulosic fibers is primarily consisted of three components: cellulose, hemicellulose and lignin. Cellulose is a linear polymer made of glucose subunits linked by β-1,4 bonds with a basic repeating unit cellobiose [3]. As a main structural component of fiber, cellulose’s stiffness and durability is supported by its formation of rigid, insoluble micro-fibrils chains, integrated by numerous intra- and intermolecular hydrogen bonds. The individual cellulose fiber-cells are linked together by means of the middle lamella [4]. The polymerization degree of cellulose in sisal fiber is around 25,000, which means its crystalline arrangement may be disrupted by the chain-ends [4, 5]. While the polymerization degrees of hemicellulose and lignin, which are mainly in the primary wall and middle lamella, lies between 50-200 [5, 6] and around 60 [7], respectively. Hemicellulose, which emerges as an immense renewable biopolymer resource, is the second most abundant biopolymer in the plant kingdom, representing about 25-35% of lignocellulose biomass [8-10], only less than cellulose. In contrast to cellulose, hemicellulose represents a type of hetero-polysaccharides with complex structures containing glucose, xylose, mannose, galactose, arabinose, fucose, glucuronic acid, and galacturonic acid in various amounts depending on the source [11]. Its main chain is characterized by a β-1,4-linked-D-xylopyranosyl, which carries a variable number of neutral or uronic monosaccharide substituents [12]. Lignin is the most abundant natural aromatic (phenolic) polymer [13], the main function of which is to cement the cellulose fibers in plants. As a kind of three-dimensional network heteropolymer, lignin has a complex and non-uniform structure with aliphatic and aromatic constituents, primarily resulting from the oxidative coupling of three p-hydroxycinnamyl alcohols (monolignols): p-coumaryl, coniferyl and sinapyl alcohols [14]. The most common linkages formed during lignin biosynthesis are the β-O-4 ether linkages, followed by other types of ether and C–C linkages such as α-O-4, β-β, β-5, and 5-5 [15-17].  These three main components assemble together to form a distinct cell wall structure, with excellent mechanical properties and ultra-low density. This endows cellulosic fibers outstanding reinforcing effect in the matrix of cement, however, their long-term service life is confined due to the easily deterioration behavior of amorphous components and premature damage of cellulose chains. Gram [18] and Filho [19] described the dissolution of lignin and hemicellulose as well as the alkaline hydrolysis of cellulose molecules in cement, which leads to a reduction of tensile strength and degree of polymerisation. B.J. Mohr et al.[20] presented a three-part progressive degradation mechanism for biomass reinforcements: (1) initial fiber debonding from cement, (2) precipitation of hydration products within the interfacial transition zone between cellulosic fiber and the cement matrix and (3) fiber embrittlement due to cell wall mineralization indicated by the lack of recovered toughness. João de Almeida Melo Filho et al. [21] investigated two fiber degradation mechanisms in Portland cement composites: fiber mineralization and degradation of lignin and holocellulose due to the penetration of calcium and hydroxyl ions into fiber cell walls. In the author’s previous work [22], both of the two degradation mechanisms, alkaline hydrolysis and mineralization, were quantified by directly investigating the fiber’s component change, degradation of tensile strength and crystallization indices.

Fig. 1

The widespread uses of cellulosic fibers in construction and their great potential as alternatives to traditional fibers affirm the need for fundamental studies of their aging behavior. In spite of much research on outstanding performance of cement composite reinforced with cellulosic fibers [23-34], the durability of natural fiber-cement composite exposed to various aging conditions [4, 19, 20, 35-43], and the degradation of cellulosic fiber slowed down by replacing cement with pozzolanic materials [4, 39, 44-51] and pretreatment of fibers[40, 52-55], the degradation kinetics of cellulosic fiber in the complex alkaline and mineral-rich environment of cement and the correlations between the hydration of cement and fiber deterioration have not been thoroughly understood. In addition, the model for service life prediction for cellulosic fiber in cement based materials was rarely reported. Few designers are willing to accept the risk of working with inherently unreliable reinforcement, but difficulties could be alleviated if a parameterized model were developed to simulate the fiber’s deterioration rate as a function of service time.

In this study, degradation kinetics of cellulosic fiber in the alkali environment if cement were investigated. Cement hydration over a time span of up to 1000 days and its effect on the deterioration rate of cellulosic fiber were also studied in the present work. Thermogravimetric analysis (TGA), X-ray diffraction (XRD), scanning electron microscopy (SEM), and universal strength testing system were employed to characterize the aging behavior of fibers by means of component change, crystallinity index, microstructure and uniaxial tensile properties, respectively. Inductively coupled plasma optical emission spectroscope (ICP-OES) was used to investigate the development of elements in the pore solutions. The two-parameter Weibull distribution model was employed to deal with the scattered testing data of tensile properties of natural fibers. Alkaline hydrolysis of cellulose in the embedded cellulosic fibers was parameterized to predict service life of the cellulose in the cellulosic fibers.

2. Materials and methods

2.1 Raw materials

Sisal fiber produced from Tanzania provided by Bast Fibers LLC of Creskill, New Jersey was used in this study as a representative of natural fibers. ASTM Type I Portland cement (PC) with a BET-specific surface area of 1.7 m²/g was used. In the present study, in addition to pure cement matrix, cement blends with 10 wt.% and 30 wt.% substitution by metakaolin (MK10 and MK30) was employed to study the effect of cement hydration on fiber degradation kinetics. MK has an Al₂O₃ and a SiO₂ contact of 42.4% and 51.8%, which are 10.24 and 4.02 times of those in PC, respectively. The median particle size (D50) of cement and MK are 9.71 and 3.79 μm, respectively. The BET-specific surface area of MK (2.93 m²/g) is 72.35% higher than that of cement.

In contrast to previous work reported in the literature [4, 19, 20, 39-41, 56, 57], degradation of the embedded cellulosic fibers were investigated directly through a modified “fiber strand-in-cement” method as introduced in author’s previous work [22, 58]: a 40 mm × 70 mm cementitious paste block of cementitious paste was cast around the middle part of a strand consisting of 10 fibers, with a wire mesh in between and epoxy resin for the external part if fiber strand. The hydration studies were performed on paste samples of the pure PC and blends containing 10 wt. % and 30 wt. % of MK.

2.2 Pore solution investigation

There are four steps to form cement pore solution throughout the entire hydration process: (i) from mixing water to continuous phase (with suspended cement); (ii) then it converts to a rather complex alkali- and sulfate bearing solution; (iii) followed by discontinuous pore solution; (iv) at last the pore solution is formed with continued reaction of cement components [59]. The development of alkalinity of pore solution is primarily attributed to three mechanisms [60]: (i) dissolution of NaOH and KOH (with pH around 13.4), (ii) dissolution of Ca(OH)₂ (pH falls to 12.5), and (iii) dissolution of  C-S-H gels (pH remains above 12.5). The pore solution of hardened samples was obtained by the suspension method [61, 62].  The alkalinity (pH value) of pore solution was measured using a pH electrode. An inductively coupled plasma optical emission spectroscope was employed to determine the concentrations of K, Na, Si and Ca at different ages. The effective saturation indices (ESI) of portlandite, ettringite and monosulfate as functions of hydration time were determined. As an index showing whether water tends to dissolve or precipitate a particular mineral, the ESI is defined as:

ESI = log (IAP/Ksp)                                                          (1)

where IAP is an ion activity product, and Ksp is a solubility product.

IAP and Ksp have the same form: both of them are calculated from the product of free ion, however, Ksp is an equilibrium constant in terms of activities of reactants and products at equilibrium, and IAP involves the actual (measured) activities. If IAP = Ksp, ESI=0, which means the solution is in equilibrium with the mineral; if IAP < Ksp, ESI < 0, the pore solution is undersaturated with the mineral and reaction is proceeding from left to right as shown in Eq. (8-15) (dissolution); for IAP > Ksp, ESI > 0, i.e. the pore solution is supersaturated with the mineral and reaction is proceeding from right to left (precipitation).

2.3 Chemical compositions of fiber

As investigated in the author’s previous work [22], due to the degradation of chemical and physical batteries (lignin, hemicellulose and pectin [63]) and cell wall mineralization, sisal fiber shows a higher moisture absorption capacity in Ca(OH)₂ solution (pH = 12.3) than that in tap water and NaOH (pH = 13.0) solution. It was also found that the moisture absorption of sisal fiber increase with temperature. Serving as reinforcement for cement composites, natural fibers were completely embedded in the cement matrix and contact with pore solution. Therefore, it is necessary to know their chemical compositions and the corresponding role in fiber’s alkaline degradation process. Here, in addition to sisal fiber, the components of jute, coir, bamboo and flax fibers were also investigated.

Cellulose, which is the main structural component of all plant cell walls, consists of approximately 10,000 glucose β-D-glucopyranose residues per average molecule linked together by (1→4)-glycosidic bonds to form long, linear macromolecules with a low degree of polymolecularity [64-68]. Cellulose and hemicelluloses are the major components of the secondary layers of the cell wall in lignocellulosic fibers [64]. The hemicelluloses are present in the surface layer as adhesives to strengthen the bonds between individual fibers in a three-dimensional network of lignocellulosic [69, 70]. Lignin, most of which locates in the middle lamella, is an amorphous, three-dimensional, polyphenolic polymer, built up of phenylpropane units. The remaining part of lignin is located throughout the secondary wall, where it interpenetrates and encrusts the cellulose micro-fibrils and the hemicelluloses. Cellulose, hemicelluloses, and lignin can be classified from a morphological point of view as framework, matrix, and encrustation substances, respectively [71]. More vividly, their relationship is similar as that among ossature, connective tissue and muscle of the human body. The contents of these three components determine the mechanical properties and stiffness of fiber and its degradation rate. The procedure used to determine the content of cellulose, hemicellulose and lignin according to the methods proposed by Xu et al. [72] and Viera et al. [73] was followed. The lignin content was determined using the Klason method [74], and the cellulose content was obtained as the sum of α and β cellulose, while the content of the hemicellulose was determined as the mass sum of hemicellulose A and B.

4. Thermal, morphology and X-ray diffraction analysis

Components of fibers were analyzed using a Perkin Elmer Pyris1 thermogravimetric analyzer at a heating rate of 20 ºC/min from room temperature to 700 ºC in an N2-atmosphere. The micro-analysis of the cellulosic fiber was performed on a Hitachi 4700 scanning electron microscope (SEM) under an accelerating voltage ranging from 10 kV to 20 kV.

The crystallinity fraction of fiber was determined by means of XRD analysis performed on ground powders using an Inel X-Ray diffractometer in a θ-2θ configuration using CuKα source (λ =1.54 Å) at -40 kV and 35 mA. The XRD patterns of fibers were obtained in the angular range of 5–45º. The crystallinity index (Cr.I.), which expresses the relative degree of crystallinity [75], was calculated from the X-Raw diffraction intensity by using the following equation [76-79]:

Cr.I = (I₀₀₂-I_am)/I₀₀₂                                                                (2)

2.5 Tensile strength of fiber and Weibull distribution

ASTM D 3822 was followed to test the uniaxial tensile strength of single raw and embedded fibers using a micro-force Instron 5948 testing system with a gauge length of 20 mm and a loading rate of 0.2 mm/min. Stress-strain curves were obtained by determining diameter (cross section area) of each fiber using SEM. Due to the non-homogeneous nature of natural fiber, the testing data are scattered and the Weibull distribution method was followed to determine the characteristic strength. The probability function of this double-parameter distribution is given by as follows:

f(x) = m(x)^m-1 exp(-x^m)                                                                 (3)

where f(x) is the frequency distribution of the variable x, and m is the Weibull modulus, which indicates the scatter of test data: as m gets larger the distribution becomes narrower.

By integrating both sides of Eq. (3):

F(x)

=∫x∞f(x)dx=-exp⁡(-xm)                                                    (4)

The survival probability S_p can be calculated by defining the variable x as σ/σ₀ and taking its absolute value:

S_p = exp[-(σ/σ₀)^m]                                                                  (5)

where σ is the measured failure stress and σ₀ is a normalizing parameter defined here as characteristic strength.

By taking the logarithm of both sides twice, it yields:

lnln(1/S_p) = mlnσ₀ – mlnσ                                                          (6)

The Weibull modulus and characteristic strength can be determined by plotting a straight fitting line for the points generated from lnln(1/S_p) v.s. -lnσ. Weibull modulus m and characteristic strength σ₀ are determined from the slope of the fitting line and its intercept at lnln[1/S_p] = 0 (which means S_p = 1/e = 0.36788), respectively.

The survival probability S_p was evaluated from the provable index:

S_p = 1 – R/(N+1)                                                               (7)

where R is the number of fiber broken at stress of σ_R, and N is the total number of fibers have been tested (here N = 20).

3. Results and discussion

3.1 Chemical composition of natural fiber

As shown in Table 1, the chemical composition of natural fiber includes pectin, extractive, lignin, hemicellulose, and cellulose. Cellulose is the major component and the one that provides strength and stability. The content of cellulose in sisal, jute, bamboo and flax fibers is much higher than hemicellulose, lignin and other components. However, coir shows a low cellulose content, which is almost the same as lignin. As analyzed above, cellulose is the main structural component in cell walls, so coir fiber shows a relatively low tensile strength compared to the above four kinds of fiber. Coir has the highest content of lignin, pectin and hemicellulose. Flax fiber contains the lowest amount of lignin. While the smallest amounts of pectin, extractive and moisture were all found in bamboo fiber, which accounts for its high tensile strength. According to the results summarized in Table 1, sisal fiber was selected as a representative for the investigation of degradation kinetics of natural fiber in cement composites. Due to the specific cell walls structure of vegetable fiber as shown in Fig. 1, lignin and pectin are the first two components which encounter degradation in aggressive environments. The degradation of lignin, pectin and hemicellulose determine the stripping degree of cellulose micro-fibrils. Therefore, the contents of these three components can considerably impact the durability of natural fiber in a cement matrix at early age. The specific cell wall components of cellulosic fiber endow it excellent mechanical properties with low density, however, this also one of the main reasons for its poor degradation resistance in the cement matrix and high water absorption in cement pore solution. As shown in our previous work [22], the fiber’s high water absorption ratio in Ca(OH)₂ solution might be attributed to the hydrolysis of amorphous components and penetration-precipitation of minerals rich in Ca.

Table 1

3.2 Characterization of the fiber degradation

In contrast to previously used accelerated aging treatment, the degradation behavior of cellulosic fiber under room temperature up to 2.5 years was evaluated in this study by means of the fiber’s tensile properties, crystallinity property and thermogravimetric analysis. The uniaxial tensile strength, crystallinity index and cellulose content of the aged fibers, obtained through the modified “fiber strand” method, were investigated using micro-force system machine, X2 Scintag X-Ray diffractometer and a Perkin Elmer Pyris1 thermogravimetric analyzer, respectively.

3.2.1 XRD analysis

The measured XRD diffractograms of the raw and part of embedded fibers after 10 and 100 days of hydration are shown in Fig 2a. Steamed bread peaks indicate that cellulosic fiber is a kind of hemi-crystal materials, including both amorphous components (lignin and hemicellulose) and that with a crystalline nature. The primary identified crystalline phase is cellulose: a broad peak between 13º and 18º corresponding to (

1̅10) and (110) lattice planes and a relative sharp peak at 2θ ≈ 22.5º corresponding to (002) lattice plane. Due to the residual cement hydration products on surface of fiber, C-S-H and C-A-S-H are also identified for the fiber embedded in MK modified cement samples. Qualitatively a significant difference can be noticed between the raw and embedded fibers, as well as between the fibers embedded in neat PC and that in MK30. Both of the two cellulose peaks increase with increasing cement hydration time, which were related to a dramatic difference between the maximum intensity of diffraction originating from the lattice peak and the intensity of diffraction contributing to the decomposition of amorphous fraction. According to Eq. (2), this demonstrate the increase in crystallinity degree of the embedded fibers. At each aging stage, the partial replacement of 30% cement by MK results in a significant mitigation of fiber crystallization.

3.2.2 TGA studies

As shown in Fig. 2b, the DTG curves of the investigated raw and embedded fiber samples after 100 days of cement hydration show peaks due to weight losses related to the thermal decomposition of hemicellulose and cellulose at 350 °C and 425 °C, respectively. As degradation proceed, the amount of hemicellulose decrease considerably and its corresponding peak merges with the peak of cellulose. Due to complex molecular structure, the thermal decomposition of lignin proceeds in a wide temperature ranges consisting of two or three stages [80, 81]. Therefore, there is no identified peaks for lignin on TGA/DTG curves. As shown in Fig. 1, the cell walls of cellulosic fiber primarily consist of lignin, hemicellulose and cellulose from outside to inside in sequence. The roles of lignin and hemicellulose are protective barriers to the infiltration of pore solution and precipitation of portlandite in cell walls of natural fiber. However, these two amorphous components are sensitive to the alkalinity pore solution of the cement matrix, which leads to a more impressionable degradation than that of cellulose. In addition, cellulose fibrils might be exposed to the alkaline pore solution even though the lignin and hemicellulose are only partially decomposed. Therefore, the amounts of lignin and hemicellulose are not as important as cellulose in determining degradation rate of cellulosic fiber. In this study, the content of cellulose, which determines tensile strength of cellulosic fibers, is the sole component investigated using TGA to evaluate degradation level of fibers (Fig. 2b).

It can be noticed that, after 100 days of cement hydration, compared with raw fiber, cellulose content of the fiber embedded in neat PC experienced a considerable decrease by 48.59%. In MK30, the cellulose amount of fiber is 27.60% lower than that of raw fiber but is 40.84% higher than that of fiber embedded in PC. This indicate that fiber’s degradation in the cement matrix was effectively mitigated by incorporating 30 wt. % MK. More XRD and TGA testing data can be found in Section 3.8.

Fig. 2

3.2.3 Tensile behavior of embedded fibers

Fig. 3 shows the tensile strength of the raw and embedded fibers after 50, 70 and 100 days. Weibull distribution was employed to analyze the relatively discrete testing data. Although the raw fiber is not homogeneous material, it shows a most concentrated strength data among the five groups. Tensile strength of the fiber embedded in neat PC decreases significantly with increasing cement hydration time. Their testing data are more scattered than those of fibers embedded in MK30. After 100 days of cement hydration, the tensile strength of the fiber embedded in MK30 is 27.9% lower than that of raw fiber, but is 5.3 times of that of fiber embedded in neat PC at the same stage of aging. This demonstrates that the fiber’s degradation, especially the deterioration of cellulose, was effectively mitigated by partially replacing cement with metakaolin. In addition to strength, the Young’s modulus of fiber also experiences a more dramatic increase in neat PC than that in MK30. This may be attributed to two points: (i) the amorphous components of cellulosic fiber, lignin and hemicellulose, suffer severe in the high alkaline environment of neat PC; (ii) mitigated mineralization of fiber in MK30: with better lignin and hemicellulose integrity, less cement hydration products, especially portlandite, ingress and precipitate into fiber’s cell wall and lumens. Therefore, the corporation of 30 wt.% metakaolin not only mitigated the hydrolysis of cellulose but also arrest fiber’s embrittlement caused by cell wall mineralization.

Fig. 3

3.2.4 Microstructure analysis

Fig. 4 shows the microstructure of the embedded fiber on fracture surface of fiber-reinforced cement mortar after exposed up to various wetting and drying cycles (0 to 50 cycles). These microtopographies of fiber at different aging stage demonstrate the alkaline degradation process of natural fiber in the cement matrix, due to the high alkalinity of cement solid phase and pore solution. As shown in Fig. 4, the alkaline deterioration of cellulosic fiber consists of for steps: (i) degradation of lignin and part of hemicellulose; (ii) deterioration of hemicellulose; (iii) stripping of cellulose micro-fibrils; (iv) alkaline hydrolysis of amorphous regions (containing non-reducing end) in cellulose chains. From Figs. 3 and 4, it can be seen that cellulose is the primary structural component of cellulosic fiber and there is no significant mechanical properties degradation in the first three steps. As the degradation proceeds, the protective barriers, lignin and hemicellulose, were gradually destroyed, and as a result the hydration products, especially soluble C-S-H and portlandite, gradually infiltrate into lumens and space between cellulose fibrils, which leads to mineralization of the fiber. This in turn results in an enhanced deterioration of cellulose, and leads to fiber embrittlement and a severe reduction of fiber strength.

Fig. 4

3.3 Degradation mechanisms

As described above, the degradation of cellulosic fibers in cement pore solutions proceeds gradually from fiber surface to inside of cell walls. In other words, the amorphous protective layers, lignin and hemicellulose, suffer deterioration first resulting in the infiltration of alkaline pore solutions in fibers. Then the exposed cellulose fibrils deteriorate leading to premature failure of the fibers. Here the degradation mechanisms of the three main components of cellulosic fiber, such as lignin, hemicellulose and cellulose, were investigated and summarized.

3.3.1 Aging of lignin

Lignin functions as cuticle or glue in the cell wall of natural fiber giving it structure and protecting the fiber against microbial or chemical degradation of the polysaccharides. Unlike cellulose and hemicellulose, lignin is a non-carbohydrate aromatic heteropolymer or phenolic polymer that is derived from the oxidative coupling of three different phenylpropane building blocks (monolignols): p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol [82, 83].  Lignin and hemicellulose have one characteristic in common:  they are both amorphous, which means lignin is also easy to undergo alkaline hydrolysis. The alkaline hydrolysis of α-O-4 bonds and free phenolic hydroxyls results in the formation of α-alcohols (1-guaiacylethanol, 1-guaiacylpropanol) [84]. The cleavage of β-0-4 ether bond is one of the most important reactions of alkaline delignification [85]. When free hydroxyls are present on neighboring α- or γ-carbon atoms or if there is a CO group in the α-position, structural units of lignin of the nonphenolic β-ether type are capable of undergoing alkaline hydrolysis, and lead to the formation of guaiacol, guaiacylpropane, 3-guaiacylpropanol, and p-hydroxyphenylpropane [86-88]. The alkaline degradation products of lignin are mainly protocatechuic acid, vanillic acid catechol, oxalic, and other simple aliphatic acids [89], as well as a dark, amorphous, humin-like “lignic acid”, which is the cause of the darkening effect of alkaline treated fiber, especially under high temperature. The degradation of lignin results in an exposure of hemicellulose and cellulose to the alkaline environment of the cement matrix. Even though a partial degradation of lignin might provide an express path for pore solution of cement to penetrate into fiber’s cell walls. This leads to an enhanced alkali aging and mineralization of the embedded fibers.

3.3.2 Deterioration of hemicellulose

Different from the highly uniform polyglucan structure of cellulose, hemicelluloses represent a type of hetero-polysaccharide with complex structures containing glucose, xylose, mannose, galactose, arabinose, fucose, glucuronic acid, and glacturonic acid in various amounts, depending on the source [11]. As shown in Fig. 1, hemicelluloses do not serve as load-bearing component in cell walls of natural fiber, but they bind and fix cellulose micro-fibrils to make the fiber more stable. Probably no chemical bonds exist between cellulose and hemicellulose, and the mutual adhesion is provided by hydrogen bonds and van der Waals forces [90]. In other words, the content of holocellulose (including hemicellulose and cellulose) determines rigidity and strength of natural fiber. Hemicelluloses are usually divided into four general groups of structurally different polysaccharide types: (1) xyloglycans (xylans); (2) manno glycans (mannans); (3) xyloglucans (XG); and (4) mixed-linkage β-glucans [91]. All hemicelluloses are amorphous and lack the crystalline regions that can be found in cellulose, thus making them less mechanically protected against degradation reactions during chemical pulping [92]. Therefore, hemicellulose is much easier degraded in the alkaline environment of a cement matrix. The deterioration of this binding phase might lead to an enhanced stripping and peeling of cellulose micro-fibrils.

3.3.3 Degradation of cellulose

As shown in Fig. 1, cellulose molecular chains are ordered into strands of approximately 40 cellulose microfibrils, which have a diameter of 35 Å, through inter- and intra-molecular hydrogen bonding [93]. These microfibrils are not homogeneous, consisting of crystalline and amorphous regions. As the vulnerable spots, the amorphous regions are most prone to degradation in the alkaline environment [94, 95]. Fig. 5 shows the microfibrillic structure of cellulose, from which it can be seen that, in amorphous regions, cellulose molecules have a non-reducing end and a reducing one, along with repeating glucose units. The reducing and non-reducing ends feature original C4-OH and C1-OH groups, respectively. The reducing end is a latent aldehyde, like an aldehyde function, responding to both reduction and oxidation processes and it plays a key role in the alkaline degradation [60]. Therefore, the alkali degradation of cellulose mainly depends on the reducing end in amorphous regions.

Fig. 5

The alkali degradation of cellulose comprises four processes: peeling-off reaction, chemical stopping reaction, physical stopping reaction, and alkaline hydrolysis [60, 96-99]. Peeling-off reaction is a kind of endwise chain depolymerization, which is accompanied by the release of glucose units from the cellulose molecules caused by β-elimination at the C-4 carbon atom. The main degradation products generated by peeling-off reaction is 3-deoxy-2-C-(hydro-xymethyl)-pentanoic acid (ISA) [100]. If β-elimination occurs at another positions, two stopping reactions occur: chemical stopping reaction and physical stopping reaction. In the chemical one, hexose units remain attached to cellulose molecules and the reducing end group is converted to an alkali stable end group (metasaccharinic acid), which terminates the depolymerization. In the physical stopping reaction, the reducing end group reaches the crystalline region of the cellulose, which is inaccessible to alkali [94]. Alkaline hydrolysis is accompanied by the cleavage of glycosidic bonds in cellulose, and leads to a production of new reducing end groups, which create favorable conditions for the above three decomposition mechanisms. In principle, this would enable complete degradation of cellulose [100]. The crystalline regions of cellulose are more stable than the amorphous region in alkali environment. Generally, as shown in Fig. 6, the degradation of cellulose can be simply generalized as the disconnection of discrete cellulose nano-crystals caused by decomposition of reducing ends in amorphous regions.

Fig. 6

3.4 Pore solution of cement matrix

The liquid phase in the cement matrix, pore solution, which contains a large number of ions from the soluble substance of cement and with a high alkalinity, is considered to be a major reason for natural fiber’s degradation. This mineral-rich fluid can soak and infiltrate into the middle lamella and lumen of the fiber, and directly corrode cellulose, hemicellulose and lignin. The existence of ions, especially OH⁻ and Ca²⁺, in the pore solution is a major cause of the deterioration of natural organic matter. Compared with NaOH solution with same alkalinity, CH solution causes more severe erosion of natural fibers [101, 102]. In the presence of Ca²⁺, a primary cation in cement pore water, isosaccharinic acid (ISA) will be the main degradation product [103, 104]. The development of PC pore solution chemistry, such as the pH value, ions concentration as a function of hydration time, and the effect of supplementary cementitious materials on pore solution chemistry, has been a focal point of research for several years. The concentrations of each element depend on the dissolution of portlandite, alkali sulphates, aluminate phase and the consumption of calcium sulphate, which involve some chemical states of equilibrium as follows [105-107]:

Ettringite Ca₆Al₂(SO₄)₃(OH)₁₂26H₂O  <=> 6Ca²⁺ + 2Al(OH)₄⁻ + 3SO₄²⁻ + 4OH⁻ + 26H₂             (8)

Portlandite Ca(OH)₂ + 2H⁺ <=> Ca²⁺ + 2H₂O                  (9)

Calcite  CaCO₃ + H⁺ <=> Ca²⁺HCO₃⁻                (10)

Anhydrite CaSO₄ <=> Ca²⁺ + SO₄²⁻                 (11)

Gypsum  CaSO₄H₂O <=> Ca²⁺ + SO₄²⁻ + 2H₂O               (12)

Syngenite K₂Ca(SO₄) 2H₂O <=> 2K⁺ + Ca²⁺ +2SO₄²⁻ +H₂O              (13)

Tobermorite-I&II 5(Ca(OH)₂) 6(SiO₂) 5(H₂O) + H₂O 5Ca²⁺ + 6SiO(OH)₃⁻ +4OH⁻            (14)

Monosulfate 3CaO(Al₂O₃)·(CaSO₄)·12(H₂O) <=> 4Ca²⁺+2Al(OH)₄⁻+SO₄²⁻+4OH⁻+6H₂O           (15)

Owing to pozzolanic reaction between cement and MK, CH was consumed, which in turn affects the ion concentration and pH value of pore solution:

Ca(OH)₂ + MK → C-S-H + crystalline products (C₃AH₆, C₄AH₁₃, C₂ASH₈, etc.)  (16)

To determine elemental concentration levels and their effects on alkalinity and mineralization, the concentrations of five elements, such as K, Na, Ca, Si and OH⁻, of pore solution were investigated. The element concentrations for pore solutions of neat PC and MK30, which developed as functions of cement hydration time are shown in Figs. 7a and 7b, respectively.

From Fig. 7a, it was noticed that in neat PC, all elements remained relatively stable except Ca, since they is controlled by the consumption of CaSO₄ phases and by the formation of portlandite [108]. Caused by the consumption of gypsum through reaction with C₃A [109] and the depletion of calcium sulphates [108], concentration of calcium began to decrease rapidly after 6h. Concentrations of silicon and OH⁻ increased considerably at the same time. Although there were also slight increases, the effects of above-mentioned reactions on concentrations of Na and K were not significant. However, the rapid reactions with calcium hydroxide and sulphates gave rise to the continuous dissolution of C₃A, which led to the increase of OH⁻ and high pH value in pore solution. This in turn reduces the concentration of Ca, due to the reverse reaction as shown in Eq. (9).

Fig. 7

It should be noticed that these changes of element concentrations in pore solution at early age were correlated with the silicate reaction and the second aluminate reaction of cement hydration. The development of pH values of pore solution in PC and MK-30 are shown in Fig. 8a. Corresponding to OH⁻ concentrations, the pH of PC’s pore solution stays at a level of 13.1±0.1 for the first 10h, then rises up to 13.7±0.06. After 10 days, the alkalinity showed a relatively steady development, although it increases slightly with hydration time.

In contrast to that of neat PC, except Si, all initial concentrations of elements in the pore solution of MK30 experienced a reduction. This should be attributed to two reasons: (i) dilution effect by partially replacing cement, (ii) high content of Al₂O₃ and high pozzolanic activity of MK, which contributes to the first aluminate reaction. Due to the strong second aluminate reaction in MK30, the concentration of Ca decreases significantly after 8h and it keeps decreasing even after a long hydration time. The concentration of OH⁻ shows a similar trend, not only due to the decrease of readily soluble alkalis in the matrix caused by dilution effect, but also due to the binding of alkalis in the additional amounts of C-S-H generated by the pozzolanic reaction [110]. By blending MK, both the Ca/Si and Ca/(Si+Al) ratios of C-S-H phase were reduced suggesting that the C-S-H in MK-30 can bind much more alkalis than neat PC [108, 111]. The pH of pore solution in MK-30 blends was 12.9±0.1 during the first 8h, and then decreased to 12.7±0.08 after the deceleration and stabilization periods of cement hydration.

It is anticipated that the corporation of MK, which results in a reduction of CH content in the cement matrix and decreased concentrations of OH⁻ and Ca in pore solution, is beneficial to the durability of natural fiber.

Fig. 8b shows the calculated effective saturation indices (ESI) of portlandite, ettringite and monosulfate as functions of hydration time in neat PC and MK-30 blends. The ESI of portlandite in PC is 0.25±0.09 from the time of mixing to 317 days, which means the pore solution of neat PC is supersaturated at a low level for portlandite over the whole hydration time. The slight oversaturation of portlandite was also found in the pore solution of the MK-30 blend at early age. However, after 7 days, the ESI decreased to be negative, which means the pore solution is undersaturated for portlandite for the remaining hydration age. This might be related to the consumption of CH in the cement-MK blend. Ettringite in the pore solution of neat PC dropped considerably from a high level (3.31) to a slight oversaturation level with time. The same trend was also found in pore solution of MK30, but the oversaturation degree of ettringite was much lower than that of neat PC throughout the experiment. The ESIs show that pore solutions in both PC and MK-30 were oversaturated with respect to monosulphate at early age, and then decreased to an equilibrium level after 1 day.

Fig. 8

3.5 Degradation kinetics of cellulose

The degradation kinetics of natural vegetable organics, especially cellulose, in an alkaline medium have been studied by several investigators. Due to the interaction effect between degradation processes in alkaline degradation, alkaline hydrolysis of cellulose cannot be studied separately from the peeling off reaction because the hydrolysis products new reducing end groups, which initiate a new peeling off reaction [112]. Franzon, O. and Samuelson, O. [113] investigated the alkaline degradation of cellulose under various temperatures and presented that the degraded fraction of cellulose can be described by a logarithmic relationship with the number of glucose units and reaction duration. Lai and Sarkanen [114] arrived at a similar conclusion with the following equation:

ln(Cel) = –k • β • t                                                             (17)

where Cel is the fraction of cellulose left,  k is the pseudo-first-order reaction rate constant, β is an average number of glucose units peeled off, and t is reaction time.

According to Franzon at al. [113], k• β has a value of 3.6×10^-10 h^-1 in 1.25 M/L OH- with a constant number of glucose units peeled off for each break (β = 65). According to the linear dependence of k • β on the concentration of OH^– for 0 < [OH^–] < 2 M/L proposed by Lai and Sarkanen [114]:

k_1.25 • β / 1.25 = k_C[OH^–_] • β / C[OH^–]                                                   (18)

k • β under a certain OH^– can be calculated as follows:

k_C[OH^–_]• β = K_1.25• β  •  C[OH^–] / 1.25 = 2.88 ×10^-10  •   C[OH^–]                              (19)

where C[OH^–] is the concentration of OH^–.

In peeling off reaction, the reducing end groups are split off from cellulose chains, resulting in the formation of an intermediate product G_e, which reacts further to give the ultimate degradation products, such as isosaccharinic acid, lactic acid, formic acid, and acetic acid [104, 115, 116]. At a constant OH- concentration, when t = 0, the mole fraction of glucose units eliminated G_e can be calculated as follows [60]:

G_e = G_r0 •  (k / k_t) • (1-e^{-kt • t})                                                                 (20)

where, G_r0 is the initial reducing end group content at t = 0, k_t is the total rate constant for chain termination (k_t= k’ + k_cr, where  k_cr is the formal rate constant of termination caused by inaccessibility and k’ is the pseudo first order reaction rate constant for the chemical stopping reaction.)

As shown in Fig. 9, OH^– concentrations of pore solutions for neat PC and MK30 follow a nonlinear relationship with hydration time of cement:

C[OH^–]_PC (mMol/L) = 561.55 – 437.05 / (1+ t^1.8 / 473.08)                                 (21)

C[OH^–]_MK30 (mMol/L) = 54.69 + 36.73 / (0.873 + t^1.168)                                    (22)

Substitution of Eqs. (21) and (22) in Eq. (19) give:

k_PC • β = 2.88 ×10^-7• [561.55 – 206759.6 / (473.08 + t^1.8)]                                (23)

k_MK30 • β = 2.88 ×10^-7 • [54.69 + 36.73 / (0.873 + t^1.168)]                                  (24)

Substitution of Eqs. (23) and (24) in Eq. (17) give:

ln (Cel_PC) = – 2.88 ×10^-7 • [561.55 – 206759.6 / (473.08 + t^1.8)] • t                         (25)

ln (Cel_MK30) = – 2.88 ×10^-7 • [54.69 + 36.73 / (0.873 + t^1.168)] • t                          (26)

As analyzed above, the alkali degradation of cellulose comprises peeling-off reaction and alkaline hydrolysis. Therefore, the overall degradation of cellulose can be expressed by combing Eqs. (17) and (20) [60]:

Y = (1 – Ge) × Cel                                                                 (27)

where, Y is the fraction of remaining cellulose.

Substitutions of Eqs. (20) and (25) in Eq. (27) give the cellulose left of fiber immersed in neat PC matrix:

Y_PC = {1 – G_r0 • (k₁ / k_t) • [1-exp(-k_t • t)]}v / exp{ 2.88 ×10^-7 • [561.55 – 206759.6 / (473.08 + t^1.8)] • t }  (28)

Substitutions of Eqs. (20) and (26) in Eq. (27) give the cellulose left of fiber immersed in the matrix of MK30:

Y_MK30 = {1 – G_r0 • (k₁ / k_t) • [1-exp(-k_t • t)]} / exp{ 2.88 ×10^-7 • [54.69 + 36.73 / (0.873 + t^1.168)] • t }    (29)

Fig. 9

The initial reducing end group mole fraction (G_r)₀ is defined as the initial number of end groups in cellulose divided by initial number of all glucose units [98]. Donald. W. H and Bjorn. F. H [94] reported the mole fraction (G_r)₀ to be 6.25 ×10^-3 for hydrocellulose I and 2.98 ×10^-3  for hydrocellulose II corresponding to their average degrees of polymerization determined by viscosity. Here, an initial reducing end group mole fraction of 0.004, which means, on average, there is one reducing glucose end group per 250 glucose units, was selected as calculated in Aldrich cellulose macromolecule [117]. Fig. 9 shows the overall degradation of cellulose as a function of time in the matrix of neat PC and MK30 under 25 ºC (with an assumption of k₁ = 0.04 and k_t = 0.0007). It can be seen that cellulose encounters more severe degradation in PC than that immersed in the matrix of MK30. The difference becomes dramatic after the first day. Because of the rapid rise of OH^– concentration in pore solution, cellulose of the fiber immersed in PC experiences a sharp linear decrease after 10 days and completely deteriorates after around 2 years. In contrast, two shoulders are found on the curve for MK30, which means that the cellulose encounters four steps of degradation: primary stage, initial acceleration stage, decelerating stage, and final acceleration stage. In the primary stage (within the first day), only 0.62% of cellulose was decomposed due to the low mole fraction of initial reducing end group and slow peeling-off reaction. After the first day of immersion, the degradation of cellulose shows a slight acceleration until 60 days. In this stage, the rate of peeling-off reaction, the rate of which is 10⁸ times faster than that of alkaline hydrolysis [112], dominates the degradation rate. Between 60 days and 300 days of immersion, the degradation rate of cellulose decelerates. This might be attributed to the inhibited peeling off reaction due to a combination of physical and chemical stopping reactions, which are caused by the crystalline regions and the formation of metasaccharinic acids [118]. This removes the reducing end groups and blocks subsequent peeling of the chain. The alkali hydrolysis of mid-chain scission of amorphous portions of cellulose leads to a release of new end groups, which in turn promotes the alkali hydrolysis of cellulose. As a result, cellulose encounters a second accelerated degradation stage (after 300 days) until the complete deterioration after 27.4 years of corrosion. The slow degradation rate in each aging stage, especially the exist of decelerating stage significantly slows down the deterioration of cellulose of cellulosic fiber in the matrix of cement. After 80 days of immersion, the fiber in PC matrix loses half of its cellulose, while in the matrix of MK30, only 23% cellulose is decomposed. The time needed to completely degrade the cellulose in MK30 is 13.7 times longer than that in the matrix of pure cement, which can also account for the better tensile strength of the embedded fiber in MK30 and the improved durability of cellulosic fiber reinforced MK30 samples.

The experimental data of cellulose amount obtained from TGA analysis after 10, 50, 100, 200, 500 and 1000 days of cement hydration are also presented in Fig. 9. It can be seen that the testing data and model curves agree with each other well, which demonstrates that the analyzed degradation kinetics of cellulose in cellulosic fiber embedded in the cement based matrices is tenable. In addition to the fraction of cellulose left, another degradation kinetics, mineralization, was also investigated and was characterized using fiber’s crystallinity index after up to 1000 days of cement hydration. As shown in Fig. 10a, the raw fiber shows a crystallinity index of 0.71. After 10 days of immersion, this Cr.I increases by 13.17% and 10.92% in neat PC and MK30, respectively. It can be noticed that, as cement hydration time increase, Cr.I of the embedded fiber experiences a dramatic increase in the PC matrix. This indicates that the amorphous components of cellulosic fiber, lignin and hemicellulose suffer severe alkaline attack. By replacing 30 wt.% cement with MK, increase rate of Cr.I. was moderated, demonstrating a mitigated deterioration of both cellulose and non-cellulose components and an arrested mineralization of fiber cell walls. This may be primarily attributed to the reduced alkalinity of pore solution and to the low amount of CH in MK30. Tensile strength of the raw and embedded fibers in PC and MK30 up to 1000 days is shown in Fig. 10b. After 10 days, the fiber embedded in MK30 yield a similar strength as the raw fiber, but a reduction of 33.05% was found for the fiber embedded in neat PC. Due to the severe alkaline hydrolysis of both amorphous components and cellulose, the tensile strength of cellulosic fiber suffered a dramatic decrease in PC. After 500 and 1000 days, the fiber has already been deprived of its strength and the corresponding reinforcing effect. In the matrix of MK30, degradation of fiber strength was mitigated significantly. After being embedded for 200 days, the fiber in MK30 still yields a higher tensile strength than the fiber in PC after 10 days. This confirm the important structural role of cellulose in cellulosic fiber again and agree well with the conclusion that the corporation of MK is a convenient, scientific and effective way to enhance cement hydration, to reduce the alkalinity of cement matrix and to mitigate the fiber’s degradation in the cement matrix. As a corollary, the durability of cellulosic fiber-reinforced cement composites was improved significantly.

Fig. 10

4. Conclusions

Degradation kinetics and service life of cellulosic fiber in the alkaline environment of cement composites was investigated. Influence of cement substitution by 10 wt.% and 30 wt.% metakaolin on degradation behavior of cellulosic fiber in cementitious systems were studied. The following conclusions were obtained.

Degradation behavior and kinetics of cellulosic fiber was studied directly by means of tensile testing, crystallinity characterization, and components analysis on the aged fibers. The tensile strength of fiber embedded in neat PC suffered the most severe reduction. However, in MK30, the fiber’s aging was effectively arrested. Due to decomposition of the amorphous phases, the crystallinity of fibers was increased. Compared to fibers embedded in PC, the fibers in MK30 showed greater amounts of lignin, hemicellulose and cellulose and a smaller crystallinity index. Due to the remnant of lignin and hemicellulose, which serve as binding phases and protective barriers, the precipitation of portlandite in fiber cell walls was also effectively restrained.

By blending 30wt.% MK, the concentration of Si of pore solution was increased, however the concentrations of K, Na, Ca, Si and OH⁻ experienced a significant reduction. The ESI results indicated that the pore solution of neat PC was supersaturated for portlandite, while the pore solution of MK-30 became undersaturated for CH after 7 days of hydration. Alkali degradation of cellulose showed a strong dependence of alkalinity of the cement matrix. The degradation of this phase mainly depends on the reducing end in its amorphous regions and can be simply generalized as the disconnection of discrete cellulose nano-crystals caused by decomposition of reducing end. Lignin and hemicellulose are amorphous and easily undergo the alkaline hydrolysis. As the degradation proceeds, the hydration products, such as soluble C-S-H and portlandite, gradually infiltrated into the cell wall, which in turn led to mineralization and embrittlement of the fibers.

A service life prediction model was developed by determining correlations between cellulosic fiber degradation and hydration of Portland cement. Based on OH^– concentrations of pore solutions the overall degradation of cellulose in the cellulosic fiber embedded in PC and MK30 was estimated by involving two processes: peeling-off reaction and alkaline hydrolysis. Cellulose of the fibers immersed in neat PC showed a linear decrease after 10 days until complete deterioration after about 2 years of immersion. Four steps of fiber degradation were found in the curve of cellulose degradation embedded in MK30: primary stage, initial acceleration stage, deceleration stage, and final acceleration stage. The low degradation rates, especially the existence of deceleration stage, slow down the deterioration of cellulose in fibers significantly. According to the model, the time needed to complete degradation of cellulose was extended by 13.7 times by incorporating 30 wt.% MK. This model is in a good agreement with the experimental observations.

The results demonstrate that pore solution’s alkalinity is the primary factor impacting natural fiber’s alkaline hydrolysis and embrittlement in the matrix of Portland cement. The modification of cement hydration effectively modifies the degradation kinetics of the natural fiber. The proposed degradation model indicates the effect of alkalinity of cement pore solution on the decomposition of cellulose and showed a good agreement with the experimental observations. Therefore, it can be used to predict the failure of natural fiber’s cellulose in the cement matrix. Further investigation of natural fiber degradation in cement matrices by considering effect of temperature, humidity and concentration of Ca²⁺ is recommended.