Heterogeneous Acylation of Flax Fibers. Reaction Kinetics and

Flax fibers composed mainly of cellulose were subjected to heterogeneous valerylation reaction. The progress of the chemical modification was assessed...
0 downloads 0 Views 200KB Size
Download PDF
Biomacromolecules 2003, 4, 821-827

821

Heterogeneous Acylation of Flax Fibers. Reaction Kinetics and Surface Properties Elisa Zini and Mariastella Scandola* University of Bologna, Department of Chemistry “G. Ciamician” and ISOF, C.N.R., via Selmi 2, 40126 Bologna, Italy

Paul Gatenholm Biopolymer Technology, Department of Materials and Surface Chemistry, Chalmers University of Technology, S-412 96 Goteborg, Sweden Received February 6, 2003

Flax fibers composed mainly of cellulose were subjected to heterogeneous valerylation reaction. The progress of the chemical modification was assessed by transmission FTIR. The heterogeneous esterification reaction followed first-order kinetics, and a plateau was reached already after 30 min. The intensity of the FTIR hydroxyl absorption band (ν ) 3400 cm-1) did not appreciably decrease during the acylation reaction, showing that only a small fraction of the fiber hydroxyls was involved in the reaction. The degree of valerate substitution (DS) at the fiber surface (50 Å thick layer) was evaluated by means of ESCA. Surface valerylation increased with reaction time and leveled off at DS around 1 after 30 min, in agreement with the FTIR data. The chemically modified fibers maintain the Cellulose I crystal structure and the original crystallinity degree up to the longest reaction time investigated (180 min). Dynamic contact angle measurements showed that surface hydrophobicity as indicated by advancing contact angle rapidly increased upon valerylation reaching a plateau after about 10 min. Chemical modification does not appreciably alter fiber thermal stability (by TGA) and morphology (by SEM). Introduction Under the pressure of growing environmental awareness worldwide, scientific research has focused on the development of new materials from renewable sources.1-3 Efforts are made to use agricultural products in nonfood applications. Bast fibers, for example, are intensively investigated as potential reinforcing agents in polymer composites.2,4-7 The main advantages of natural fibers over man-made fibers such as glass, carbon, and aramid are their renewable origin, worldwide availablility, low production energy requirements, reduced equipment wear, biodegradability, and thermal recyclability by combustion. Particularly attracting is the low density of natural fibers that leads to high specific mechanical properties.4 However, natural vegetable fibers, which mainly consist of cellulose, do not usually perform satisfactorily as polymer reinforcement,6,8 owing to their high surface polarity. The efficiency of fiber reinforcement depends chiefly on the ability to transfer the applied stress from matrix to fiber. This is not achieved with cellulosic fibers owing to poor adhesion between the hydrophilic surface of the fiber and the essentially hydrophobic polymer which is used as a matrix. Many attempts have been made to improve adhesion at the fiber-matrix interface. Some researchers have focused their efforts on chemically modifying the surface of cellulose fibers4,9-12 or the matrix.13,14 Alternatively, compatibilisers15 * To whom correspondence should be addressed. Phone: +39 051 2099577. Fax: +39 051 2099456. E-mail: [email protected].

or coupling agents16 can be added during composite processing. Some authors found that acylation of wood fibers significantly improved the ultimate strength of thermoplastic cellulose esters reinforced with wood fibers.7 Efficient processes of fiber modification are thus required for future successful use of cellulosic fibers as reinforcement in composites. In this study, cellulose fibers isolated from a flax plant (Linum usitatissimum) commonly used in the textile industry have been chemically modified. The kinetics of a heterogeneous acylation reaction are described. The changes of the fiber surface characteristics as a function of reaction time were analyzed by scanning electron microscopy (SEM), electron spectroscopy for chemical analysis (ESCA), and dynamic contact angle (DCA). In parallel, bulk properties such as crystal structure, thermal stability, and functional group analysis were investigated by wide-angle X-ray diffraction (WAXS), thermogravimetry (TGA), and transmission Fourier transform infrared spectroscopy (FTIR). Experimental Section Materials. Natural fibers from flax (Linum usitatissimum), bleached with hydrogen peroxide in the presence of NaOH and Na2CO3, were supplied by Linificio & Canapificio Nazionale S. P. A. (BG, Italy). The carbohydrate composition of the fibers was analyzed at STFI (Sweden) using acid hydrolysis and anion chroma-

10.1021/bm034040h CCC: $25.00 © 2003 American Chemical Society Published on Web 04/05/2003

822

Biomacromolecules, Vol. 4, No. 3, 2003

tography coupled with pulsed amperometric detection.17 The carbohydrate content (%) was as follows: arabinose 0.2; galactose 2.8; glucose 92.5; xylose 1.0; mannose 3.5. All chemicals from Sigma-Aldrich (reagent grade) were used without further purification. Fiber Chemical Modification. A total of 5 g of flax fibers (dried overnight at 80 °C under reduced pressure) were placed into a 500 mL flask containing 130 mL of 1,2dichloroethane and 7 mL of pyridine under dry N2 flow; valeryl chloride (11 mL) was added dropwise to the reaction vessel. The reaction was performed at constant temperature (70 °C). After selected reaction times (in the range from 10 to 180 min), the fibers were sequentially washed with 1,2dichloroethane, dichloromethane, and finally with running water. The fibers were then dried overnight at 80 °C. Small fiber amounts for ESCA and DCA analysis were Soxhletextracted with acetone for 1 h to remove waxy components from fiber surface. Instrumental Methods. Transmission Fourier transform infrared spectroscopy (FT-IR) was performed by means of a Nicolet 210 Spectrometer, taking 32 scans for each sample with a resolution of 4 cm-1. The fibers were pounded in a liquid-nitrogen-cooled mortar, and 1 mg of the obtained powder was dispersed in 150 mg of potassium bromide. Both fibers and KBr were dried before dispersion, and the mixed powder was pressed into a disk that was immediately analyzed. SEM observations of the fibers before and after chemical modification were carried out using a Philips 515 scanning electron microscope. The fibers were laid down on aluminum stubs using conductive adhesive tape and were sputter-coated with gold prior to measurements. Dynamic contact angle (DCA) measurements were performed with a Cahn DCA-322 Dynamic contact angle analyzer. The dynamic contact angle between distilled (Millipore) water (surface tension γl ) 72.6 dynes/cm) and fibers was evaluated at room temperature using the Wilhelmy technique at constant rate (21 µm/sec). The depth of fiber immersion was approximately 2 mm. The perimeter of the fiber was calculated from the receding force in water, hexadecane, or ethylene glycol, depending on fiber hydrophobicity, assuming complete wetting, i.e., a contact angle of 0°. The surface chemistry of fibers was investigated by means of a Physical Electronics Quantum 2000 scanning ESCA microprobe. The measurements were performed using a monochromatic Al (KR) X-ray source (20.4 W) and a slit width of 100 µm; the X-ray beam was at 45° to the transmission/imagine lenses axis and the pressure in the measurement chamber was about 10-9 Torr. A small fiber bundle was laid down on the sample stub; a steel mask with a 2 mm diameter hole was placed over the fibers and fixed with two locking screws. The analyzed sample area had a diameter of 500 µm. The build up of a positive charge on the surface was neutralized with a low-energy electron flood gun. Thermogravimetric analysis (TGA) of the fibers was carried out using a TA-TGA 2950 thermobalance. The measurements on samples weighing 6-8 mg were performed at 10 °C/min from room temperature to 600 °C under N2 flow. The fibers were compacted in a press at room temperature before being introduced in the TGA sample pan. Powder wide-angle X-ray diffraction (WAXS)

Zini et al.

Figure 1. FT-IR spectra of unmodified cellulose fibers (F0) and of fibers esterified for 60 min (F60). Arrows: valerate stretching absorptions. Asterisk: C-O stretching vibration of cellulose backbone. The insert shows the method used to evaluate the extent of substitution, r and b (see text).

patterns were recorded from 2θ ) 5° to 60° with a Philips PW 1050/81 diffractometer, equipped with a graphite monochromator in the diffracted beam and using Cu KR radiation at λ ) 0.1542 nm (40 kV, 40 mA). The measurements were conducted on fibers compacted into small mats. Results and Disussion Bulk Properties of Fibers. The heterogeneous reaction applied to flax fibers in the present work aims at acylating the easily accessible hydroxyls with valeryl groups in mild experimental conditions, to maintain the native fiber crystal structure and morphology, which are responsible for the good mechanical properties of flax, unchanged. Transmission FTIR analysis is used to verify the progress of the valerylation reaction. As an example, Figure 1 compares the transmission FTIR spectrum of unmodified fibers (F0) with that of the fibers modified after 60 min of reaction (F60). The spectrum of unmodified fibers well reproduces that of cellulose reported in the literature,18 except for the presence of a weak shoulder at 1745 cm-1 attributed to residual noncellulosic components present in the fibers. The carbohydrate analysis showed the presence of 92.5% glucose, hence more than 7.5% hemicelluloses in the fiber. The typical hemicelluloses present in flax are glucomannans, xylans, and rhamnogalacturonan type structures.19 The most of native hemicelluloses are partially acetylated, and they also can contain uronic acid residues. The weak shoulder at 1745 cm-1 might originate from acetyl in hemicelluloses or uronic acid. The fiber might also contain some vaxes which would also contribute to ester bond peak in IR spectrum. The FTIR spectrum of F60 shows two typical stretching vibration bands of the valeryl group, marked by arrows in Figure 1: ν CdO at 1745 cm-1 and ν C-O at 1165 cm-1. These bands are present in the FTIR spectra of all modified fibers investigated, and their intensity is found to increase with reaction time.

Biomacromolecules, Vol. 4, No. 3, 2003 823

Heterogeneous Acylation of Flax Fibers

Table 1. Thermogravimetric Analysis of Fibers

Figure 2. Dependence of y ) r (9) and y ) b (O) on reaction time for valerylated fibers (from FTIR, see text for definition). Standard deviation on symbols.

It is worth pointing out that in Figure 1 the large band at 3400 cm-1, because of OH stretching, does not appreciably change upon fiber esterification; this observation indicates that only a small fraction of the total hydroxyls of the flax fiber is acylated in F60. The same consideration also holds for the fibers obtained after the longest reaction time applied in this work (F180). Comparison of the intensity of the two ester vibration bands (B at 1745 cm-1 and C at 1163 cm-1, see arrows in Figure 1) with that of an absorption that is not affected by the chemical modification applied, for example the C-O stretching vibration of the cellulose backbone (A at 1058 cm-1, see asterisk in Figure 1), provides a means to quantify the overall extent of substitution of the valerylated fibers. Two parameters are therefore calculated: the ratio (r) between B and A, and the ratio (b) between C and A. The values of r and b are plotted as a function of reaction time in Figure 2. As expected, r and b show the same increasing trend and reach a plateau after similar reaction times. The observed behavior agrees with earlier results on the heterogeneous acetylation of flax fibers9 where it was shown that the reaction follows a first-order law, up to a time that depends on the specific reaction conditions. The general firstorder equation yt ) y∞ + Ae-kt

(1)

is used to fit the experimental results of Figure 2. In eq 1, y∞ is the plateau value at t ) ∞, the term (y∞ + A) is the value of y at t ) 0, and k is the rate constant. The curves drawn in Figure 2 are calculated using y∞ ) 0.32 for r and y∞ ) 0.78 for b. The obtained rate constants are k ) 23 and 14 min-1, respectively. Figure 2 clearly shows that eq 1 satisfactorily fits the trends of both r and b with increasing reaction time up to 120 min. The fact that the reaction follows a first-order kinetic law implies that it proceeds until consumption of the limiting reagent, i.e., the easily accessible hydroxyls of the flax fibers. Such hydroxyls are only a small fraction of the total OH content of the fiber. The large fraction of unreacted hydroxyls that remains when the reaction reaches its plateau is responsible of the intense FTIR absorption band at 3400 cm-1 mentioned above. The result at 180 min indicates an amount of acylation well above y∞.

sample

reaction time (min)

weight loss RT-150 °C (%)

Tmax (°C)

F0 F10 F20 F30 F60 F120 F180

0 10 20 30 60 120 180

4.8 4.7 4.4 4.2 3.7 3.2 2.9

364 363 365 364 361 363 375

A very similar behavior, i.e., data diverging from the 1st order law at long reaction times, was previously observed in an extensive study of the heterogeneous acetylation of flax fibers by acetic anhydride using sulfuric acid as catalyzer.9 It was suggested that at long reaction times diffusion mechanisms come into play allowing inner hydroxyl groups to take part in the acylation reaction. The high value of r and b shown by the present fibers after 180 min of reaction (Figure 2) is taken as an indication of the beginning of an analogous reaction kinetics change. Because acetone extraction is applied to the fibers before surface analysis (see the Experimental Section), FTIR measurements are performed in order to verify whether acetone treatment modifies the fiber spectrum. It is found that the relative intensities of bands, and in particular the values of r and b, remain constant after acetone extraction, showing that the valeryl groups are bound to the fibers, not merely absorbed on them. To verify if the chemical reaction applied changes the thermal stability of flax fibers, thermogravimetric analysis has been performed. All fibers exhibit an initial weight loss due to absorbed water followed by thermal degradation of the cellulose main chain. In Table 1, the water loss is quantified as the weight change from room temperature to 150 °C. The reported values clearly show that with increasing reaction time, i.e., with increasing amount of valeryl groups at the surface, the fibers become less hydrophilic and less able to absorb water. The main degradation step of the valerylated fibers is centered at a Tmax that does not significantly change with increasing reaction time (Table 1) and is comparable with that of unmodified fibers F0. It is concluded that the thermal stability of flax fibers is not appreciably influenced by the valerate substituents introduced through the heterogeneous reaction described in this work. This result is important if the chemically modified fibers are to be applied as reinforcement in thermoplastic polymers with high processing temperature. X-ray diffraction measurements have been carried out to compare the crystal structure of the chemically modified fibers with that of native flax. The wide-angle X-ray profile of unmodified fibers shows main reflections at 2θ ) 22.4°, 16.4° and 14.8° and closely matches that of native cellulose (Cellulose I) reported in the literature.20 All valerylated fibers display the Cellulose I WAXS pattern and no appreciable changes of crystallinity are observed upon esterification. This result demonstrates that the heterogeneous reaction applied in this work preserves the Cellulose I crystal structure; that

824

Biomacromolecules, Vol. 4, No. 3, 2003

Zini et al.

Figure 4. Oxygen-to-carbon ratio from ESCA survey spectra as a function of reaction time.

Figure 3. SEM micrographs of unmodified fibers (F0) and of fibers after 180 min of reaction (F180). Arrows indicate the pits location. Table 2. Atomic Concentration from ESCA Survey Scans sample

C 1s %

O 1s %

Si 2p %

Ca 2p %

O/C

F0 F10 F20 F30 F60 F120 F180

52.6 57.4 62.1 65.7 63.9 66.2 63.5

42.4 37.9 35.1 31.5 33.8 31.4 33.6

4.7 3.9 2.8 2.7 1.9 2.4 2.9

0.3 0.2 0.0 0.0 0.1 0.0 0.0

0.81 0.66 0.57 0.48 0.53 0.47 0.53

is, it does not occur within the crystalline fraction of the polysaccharide. Scanning electron microscopy allows comparison of the morphology of flax fibers before and after valerylation. As an example, Figure 3 shows SEM pictures of unmodified fibers F0 and of the fibers after the longest reaction time investigated (F180). Native flax fibers have a rather smooth surface, a diameter in the range from 7 to 40 µm and display the characteristic circular features containing the pits21 (see arrows). Fibers F180 are practically indistinguishable from fibers F0 in the micrographs of Figure 3. This result leads to the conclusion that chemical modification does not appreciably alter fiber morphology. Surface Properties of Fibers. The fiber surface is analyzed in order to follow the changes of both chemical and physical properties during the course of the reaction. Electron spectroscopy for chemical analysis (ESCA) is widely used to investigate polymer surfaces22 and has been earlier applied to study the surface chemistry of cellulose fibers.14,23-25 In this work, the ESCA technique is employed to monitor the changes of chemical composition at the fiber surface after different reaction times. ESCA survey spectra of the fibers before and after chemical modification reveal that the surface constituents are mainly carbon and oxygen, with small contamination by silicon and calcium, not unusual in natural cellulose fibers from agricultural sources.26 Table 2 collects the atomic composition data of the investigated samples derived from the ESCA survey spectra. It is

surprising that the ratio between oxygen and carbon (O/C) of unmodified fibers F0 is very close to the expected value for pure cellulose (0.82), although cellulose is the major but not the only component of the fibers. Table 2 shows the presence of 4.7% silicon, most probably in the form of SiO2, that contributes twice as much of oxygen (9.4%). To yield O/C ) 0.81, in fibers F0, this extra oxygen must be compensated by carbon, likely derived from waxes and lignin. Table 2 reports the O/C values for all valerylated fibers investigated. When valeryl groups are introduced at the fiber surface, it is inferred from the valerate molecular structure that the amount of carbon must increase more rapidly than that of oxygen, leading to a decrease of the O/C ratio. In Figure 4, the experimental O/C values of the valerylated fibers are plotted as a function of reaction time. As expected, the O/C ratio at the fiber surface decreases with increasing reaction time until it reaches a constant value indicated by the solid line in Figure 4. It is worth pointing out that the reaction time required for the attainment of a constant O/C ratio is comparable to the time where the intensity of the CdO stretching band by FTIR levels off (Figure 2). The degree of substitution (number of substituents per glucose ring, DS) in the fiber surface layer sampled by the ESCA measurement (about 50 Å thick) can be calculated from the areas of the deconvoluted peaks in ESCA C 1s spectra. Figure 5 shows the deconvoluted high resolution carbon C 1s spectra of unmodified fibers F0 and of the valerylated sample F30 taken as an example. In principle, the ESCA spectrum of pure cellulose should only show the peaks marked C2 and C3 (characteristic respectively of C-O and O-C-O bonds). The additional presence in the ESCA spectrum of F0 of peaks C1 (relative to aliphatic carbon atoms, i.e., C-C bonds) and C4 (characteristic of ester O-CdO bonds) indicates that the flax fibers used in this work contain noncellulosic compounds, such as pectins, hemicelluloses, and waxes. Compounds containing the carboxyl group have been already observed in the FTIR spectrum of F0 in Figure 1 (shoulder around 1745 cm-1). In Figure 5, the ESCA C1s spectrum of F30 shows stronger C1 and C4 peaks than the spectrum of F0 (consider the relative intensity of C1 and C4 with respect to the “cellulose” signals C2 and C3). This result reflects the presence, at the surface of fibers F30, of valerate, whose ester group contributes to band C4 and the four aliphatic carbons to C1. Table 3 compares the relative peak areas from

Biomacromolecules, Vol. 4, No. 3, 2003 825

Heterogeneous Acylation of Flax Fibers

Figure 6. Dependence of y ) β (O) and of y ) γ/4 (9) on reaction time for valerylated fibers (from ESCA, see text for definition). Equation 1 fits the β data (dashed curve) and the γ/4 data (solid curve).

Figure 5. Comparison of ESCA spectra of unmodified fibers (F0) and of fibers F30. Table 3. Relative Peak Areas from Deconvoluted ESCA C 1s Spectra

a

sample

Ra

βb

γc

γ/4

F0 F10 F20 F30 F60 F120 F180

3.4 3.6 3.8 3.6 3.9 3.4 3.1

0.16 0.69 0.81 0.87 1.50 0.99 1.32

1.8 3.7 3.6 3.8 7.3 5.1 5.1

0.45 0.93 0.91 0.96 1.82 1.27 1.28

C2 to C3 area ratio. b C4 to C3 area ratio. c C1 to C3 area ratio.

the deconvoluted C 1s spectra of the fibers after different reaction times. If the change of surface chemistry in the analyzed fibers is only due to the presence of bound valeryl groups, the ratio (R) between the areas of the “cellulose” signals C2 and C3 should remain constant, whereas the ratio of C4 to C3 (β) and of C1 to C3 (γ) should increase with the amount of valerate on fibers. The R values listed in Table 3 show that, as expected, the C2:C3 ratio does not change significantly with changing reaction time, though the experimental R value is markedly lower than expected for pure cellulose (C2:C3 ) 5:1). Analogous discrepancies have been earlier reported in the literature for cotton and flax fibers.26 With increasing reaction time, the ESCA spectra of the valerylated fibers change, and consequentially, the β and γ ratios vary. The β and γ values collected in Table 3 can be used to estimate the degree of substitution at the fiber surface. As a matter of fact, β (C4:C3) reflects the number of ester O-CdO bonds and γ (C1:C3) the number of aliphatic carbons per cellulose repeating unit. Because the valeryl group contains four aliphatic carbons and one ester carbon,

the number of substituent chains per glucose ring is given either directly by β or by γ/4 (i.e., one forth of the total number of C-C bonds revealed by the deconvoluted ESCA spectrum). Figure 6 shows a plot of β and γ/4 as a function of reaction time. The two parameters show the same behavior (increase and attainment of a constant value). It is not clear why the results for fiber F60 are markedly off the trend of all remaining samples. The curves drawn in Figure 6 represent the best fit to the β and γ/4 data by a first-order equation analogous to that used to fit the FTIR data in Figure 2 (eq 1). In the fit of Figure 6, by assigning at the limiting plateau the values of 1.1 (for β) and of 1.3 (for γ/4), the rate constants obtained are 13 and 16 min-1 respectively. The initial value of β and γ/4 is different, owing to the different contribution to peaks C4 and C1 of the mentioned noncellulosic impurities present in the untreated flax fibers. If the initial values of β and γ/4 in Figure 6 are ideally shifted to zero (i.e., if the contribution of noncellulosic impurities is eliminated), it is easily seen that both parameters level off at a value around 1, which represents the degree of substitution of the valeryl group at the fiber surface sampled by ESCA measurements. The dynamic contact angle technique is used to investigate the changes of hydrophilicity of the fiber surface during the course of the acylation reaction. The dynamic contact angle between fibers and water is evaluated from DCA tensiograms such as those reported in Figure 7, where the curve of unmodified fibers (F0) is compared with that of the fibers after 10 min of reaction (F10). In Figure 7, the force acting on the fiber is shown as a function of the immersion depth. The amplitude in the tensiograms reflects surface roughness, which affects fiber perimeter.27 The advancing and receding curves are different, showing the well-known phenomenon of contact angle hysteresis. The subject has been extensively debated, and many studies on this topic are available in the literature.28-30 Several factors concurring to generate contact angle hysteresis have been identified, including chemical and geometrical heterogeneity of the fiber surface, time-dependent interactions of the liquid with the solid surface, surface reorientation of functional groups, etc.30 From the tensiograms of Figure 7, it is clear that for the two analyzed fibers the average value of the receding force is the same while the advancing force is very different. The hysteresis effect is amplified in modified fibers F10.

826

Biomacromolecules, Vol. 4, No. 3, 2003

Zini et al.

Figure 8. Advancing contact angle (θ) between fiber and water as function of reaction time. Standard deviation on symbols.

min (F10) shows an advancing force very close to zero (Figure 7), which corresponds to a contact angle near 90°. This result indicates that fiber hydrophobicity increases abruptly upon acylation, being quite high already at the shortest reaction time investigated (10 min). When the reaction time is extended, i.e., moving from F10 to F180, no further changes of the wetting force are observed: F remains close to zero, θ is approximately 90°, and the same hysteresis as in F10 (Figure 7) is found. The effect of reaction time on the advancing contact angle is graphically shown in Figure 8. Like in the FTIR results of Figure 2 and in the ESCA data of Figure 6, a first order equation fits the experimental contact angle data. The contact angle for the longest reaction time (fiber F180) is slightly higher than the plateau value, a result that might suggest an increase of fiber surface area. However, SEM observations do not evidence an appreciable roughness increment in sample F180 compared with the original unmodified F0 fibers (Figure 3). The DCA data in Figure 8 show that the advancing contact angle reaches a constant value after a reaction time shorter than that required by the other investigated properties to attain a plateau (compare with Figures 2 and 6). This arises from the strong effect of even small amounts of hydrophobic components on the advancing contact angle.25,31

Figure 7. DCA tensiogram in water of fibers F0 and F10. Table 4. Advancing Contact Angle between Fibers and Water sample

contact angle (°)

standard deviation

F0 F10 F20 F60 F120 F180

54.5 84.0 86.6 87.5 86.9 94.4

4.7 2.9 6.2 3.2 3.5 2.8

Conclusions The contact angle θ can be determined from measurement of the advancing wetting force F by means of the Wilhelmy formula F ) πdγlv cos θ

(2)

where γlv is the surface energy of the liquid wetting a fiber of diameter d. The buoyancy term can be neglected for thin fibers (radius