Lignin plasticization to improve binderless fiberboard mechanical properties.
Fiberboards [1, 2] are produced in vast quantities all over the world and are an excellent building material with high specific strength and good insulating properties.
New regulations concerning the use of phenolic resins in medium density fiberboards (MDF; produced in dry conditions with a reactive adhesive) are probably going to generate the development of high density fiberboard (HDF), which can be produced without any adhesive. The HDF process is realized in wet conditions. The bonding strength of the board is believed to be due to two different phenomena:
* lignin-lignin and lignin-polysaccharides cross-linking reactions that occur at high temperature.
* deformation of the system under pressure. As wood is a heterogeneous material, the irregularity inherent to the resulting surface will yield little contact area between adjacent wood elements. To produce a good adhesive bond, the wood must deform sufficiently to produce an intimate wood-wood contact. The largest contact area will result when the polymers of the wood are in a physical state to allow maximum deformation under minimum pressure, i.e., the rubbery state.
The "autoadhesion" of binderless fiberboards is generally not sufficient to obtain mechanical properties equivalent to phenolic based fiberboards. We consider the optimization of mechanical properties from the point of view of the cell wall deformation ability.
An ideal fiberboard process should be done at a temperature higher than the glass transition temperature (Tg) of all polymers. In principle, the HDF (wet) process is well adapted: at the beginning of board heating, fibers are water saturated and the softening temperature of wood fibers is low, between 60 and 90[degrees]C in the literature [3-13]. As the heating temperature is generally nearly 200[degrees]C , the sample temperature T could quickly increase above the softening point. But during the HDF process, the softening temperature also increases due to the decrease of the water content. There is a competition between sample drying and sample heating, which limits the T-Tg difference. At the end of the process, the sample reaches 180-200[degrees]C; and the softening temperature of dry fibers is also around 180[degrees]C.
Due to a water gradient in the thickness of the sample (vapor is generally evacuated from the bottom heating plate using a permeable metallic or glass tissue) (Fig. 1) a higher T-Tg difference is observed at the top side.
In this article we will first quantify this T-Tg difference. Then we will study the possibility to plasticize lignins, in order to increase the T-Tg difference.
Kraft lignin was provided by Westvaco, Charleston, SC, USA. Plasticizers of lignin are vanillin (97%, Fluka) and ethylene glycol (EG) (99%, Aldrich). Lignin-plasticizer blends are obtained by evaporating a solution of plasticizers (water for ethylene glycol and ethanol for vanillin).
Reference poplar is supplied by STORA ENSO (France).
Industrial fibers were supplied by a fiberboard manufacturer: UNALIT Society, St. Usage, France. They are constituted of a blend of soft and hardwood of different maturities.
A glass tissue is used on the lower side (Fig. 1; Side A) for the water vapor evacuation. A dispersion of fibers (7 g in 600 ml water) is deposited and manually pressed in a 50-mm cylindrical mold. Fiberboard disks are then processed under pressure at 200[degrees]C (except in Fig. 2, where the temperature effect is studied) using a GRASEBY SPECAC heating press (Fig. 1). The pressure cycle (inspired from classical industrial procedure) is 2.5 minutes under 5.0 MPa + 2.5 minutes under 2.5 MPa. The first step is realized at higher pressure to prevent the formation of macroscopic porosity due to the important vapor evacuation. The sample is then cooled at room temperature in contact with air (no pressure).
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
Plasticized fibers are obtained according to the following procedure: impregnation at room temperature by a solution of plasticizer (10 g fibers, 35 g solution of adequate concentration, 0, 1.2, and 2.4%, for 0, 6, and 12% plasticized fibers, respectively), in appropriate solvent (ethanol for vanillin, water for ethylene glycol). Samples are dried at 80[degrees]C for 12 hours.
Fiberboards from plasticized fibers cannot be processed by the same procedure, as the plasticizer can be extracted during the step of fiber dispersion in water. For this reason, this step is replaced by a sorption of water vapor (1 day conditioning at room temperature/100% relative humidity). The dispersion of the fibers in the mold is not so uniform as in the case of a dispersion in liquid water. A bigger mold is used (20 X 20 cm) in order to improve the manual dispersion of fibers. Around 15 samples are tested for a given series, taken in the center of the board. But a large standard deviation of mechanical properties is still obtained. The mechanical data of plasticized samples are expressed as a function of sample density, which is the main cause of data dispersion. Density is measured from the ratio of sample volume (from the geometric dimensions) to sample weight.
Mechanical properties: three-point bending test was performed with a GT TEST Machine; samples were tested after 3 days at 20[degrees]C and 50% relative humidity; deformation rate was 10 mm.[min.sup.-1]. Sample dimensions were 40 X 10 X 3.2 mm, cut in the middle of the disks (3 bars by disk, 5 disks tested).
Water gradient measurements: after a given time of heating treatment, samples were cut into 30 X 10 X 3.2 mm bars and peeled into 30 X 10 X 0.5 mm sheets (6 samples cut within the thickness of initial bars, 3 bars by disk processed, 3 disks tested). The duration of the operation was less than 2 minutes. Sheets were then weighed to determine the water content, which is calculated from the weight loss after 2 hours drying in ventilated ovens at 70[degrees]C.
Gradient of mechanical properties: sheets were cut into 30 X 5 X 1 mm (3 samples cut within the thickness of initial bars, 3 bars by disk, 5 disks tested). We tried to make thinner samples but the mechanical properties were too sensitive to sample preparation. Samples were tested in traction mode after 3 days of equilibration at 50% relative humidity and 20[degrees]C; strain rate was 10 mm.[min.sup.-1] (GT TEST 108.5 KN).
Temperature gradients during heat treatment were measured by inserting a thin thermocouple at different positions in the thickness of the sample.
Dynamic Mechanical Analysis (DMA) measurements were made with a TA Instruments 2980. Samples (35.5 X 12.8 X 3.2 mm) were dried in the DMA (heating rate 3[degrees]C.[min.sup.-1], to 120[degrees]C, isotherm 120 minutes), cooled to 30[degrees]C, and submitted to a second scan up to 200[degrees]C, in a simple cantilever mode, using a 150 [micro]m dynamic strain. Runs are done in duplicate. Loss modulus was used to quantify the relaxation temperature. For better comparison between samples, a relative loss modulus is given (ratio between loss modulus at T and loss modulus at T = 45[degrees]C). The relaxation of lignins and hemicelluloses can be observed in the temperature scale tested [3, 14] but we attributed the relaxation peak to lignins because it is generally the main peak observed by DMA .
Glass transition temperatures of plasticized Kraft lignin were determined by a differential scanning calorimeter (DSC) (TA Instrument 2920). Samples between 30 and 50 mg were placed in high-pressure hermetic pans. The heat cycles applied were -45 to 160[degrees]C at a rate of 5[degrees]C.[min.sup.-1] for the first scan and -45 to 210[degrees]C at the same rate for the second scan. The pans were either opened (0% relative humidity [RH]) or closed (80% RH). Samples tested at 80% RH were preconditioned in a climatic oven (SECASI Technologies). Glass transition temperature is determined from the second run.
Figure 2 shows the effect of process temperature on mechanical properties measured at the end of the process (2.5 minutes at 5 MPa, + 2.5 minutes at 2.5 MPa). An increase of mechanical properties is observed up to 240[degrees]C. The variation of modulus and strength at break is higher between 180 and 220[degrees]C, and a maximum of the properties is reached at 230[degrees]C.
[FIGURE 3 OMITTED]
The decrease of mechanical properties after 240[degrees]C (look at polynomial fits on Fig. 2) corresponds to the beginning of degradation of dry wood polymers, and the glass transition of dry lignin is reported near 200[degrees]C . A relation between the effect of process temperature and polymer characteristic temperatures seems to exist. But all along the process the sample temperature is much lower than the set process temperature. We will see (in the next paragraph) that the maximum temperature of a sample treated at 200[degrees]C is 175[degrees]C. The increase of properties between 180 and 200[degrees]C is therefore not simply connected to the lignin glass transition at 200[degrees]C. The autoadhesion mechanisms result from a complex association of factors: temperature, water content, pressure, and time. With respect to time (Fig. 3), mechanical properties reach a plateau at about 3.3 minutes. This process time corresponds to a temperature well below 175[degrees]C.
The objective is now to look at the characteristics of the material up to this critical step. As the geometry of the process leads to heterogeneous water and temperature distribution in the sample thickness, water and temperature profiles have to be measured.
To avoid degradation mechanisms the process temperature in the following is fixed at 200[degrees]C.
[FIGURE 4 OMITTED]
Figure 4 shows the evolution of temperature as a function of time, at different relative sample thicknesses. Relative sample thickness is defined in Fig. 1: 100% corresponds to the side in contact with the glass tissue; 0% corresponds to the side directly in contact with heating plate. The existence of a temperature profile can be due either to the nature of the materials used between heating plates and sample and/or to the other gradients in the sample. The glass tissue used on side A (Fig. 1) has a lower thermal conductivity compared to metallic tissues used in industry. To decrease the higher thermal conductivity on side B (Fig. 1), we inserted a polytetrafluoroethylene plate (1 mm) between side B and the heating plate. But even in this case, the temperature was higher on side B than on side A. The temperature gradient in the sample is therefore due to pressure and water profiles.
Because of the large dispersion of experimental data (the standard deviation varies from 25 to 41%), linear fits of water concentration profiles were used (Fig. 5). At time zero, the water content is higher on side A. due to a gravity effect. From t = 0.5 minute to t = 2.5 minutes, the water content is 5% higher on side B. This is due to the water pressure gradient: the pressure and the concentration are low on the side where water is released, i.e., near the glass tissue. The inversion of water gradient at the end of the process is probably due to the temperature gradient: at this time, the pressure gradient is close to zero (not influent) and the temperature difference can lead to a faster drying on side B.
Kelley et al.  have proposed a relation (from the Kwei  model) between the glass transition temperature of lignin in wood, and the water content, from in situ DMA measurements. The values measured by Kelley et al.  are certainly overestimated compared to real glass transition temperatures, because:
* the T[alpha] relaxation (from dynamic measurements) is generally higher than Tg, at the usual frequencies used in DMA experiments (more than 0.3 Hz).
[FIGURE 5 OMITTED]
* the Kelley et al.  measurements were realized in dry conditions, from initially hydrated samples. Obviously the water content decreases during the temperature scan.
Nevertheless we choose to use the Kelley et al.  data, as they are the most used in the literature. The variation of calculated glass transition temperature is reported in Fig. 6. Glass transition temperature increases with time, as water content decreases. The increase is faster when water content goes down to 20% (exponential decrease of Tg with water content in the Kelley et al.  relation). The increase is faster on side A, as the water content decreases more quickly on this side. From Tg and T temperature gradients, a T-Tg gradient can be calculated. This calculation was proposed by Wolcott et al.  to evaluate another type of boards (with adhesives). The T-Tg gradient is reported in Fig. 7. T-Tg is higher near side B than near side A. During time, T-Tg reaches a maximum and decreases to around zero. This period is longer near side B: T-Tg falls to zero after 3 minutes of treatment time, while this decrease is observed before t = 2 minutes on side A.
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
Figure 8 shows the profile of mechanical properties. The data could not be measured before t = 3 minutes, as the cohesion of the board was too low. Nevertheless the gradients of properties measured between 3 and 5 minutes show systematically a higher modulus near side B. Strength at break was not reported because the error bars were too high (due to the mode of sample preparation by transversal and longitudinal cuts, which probably initiates sample fractures).
Whatever the position in the sample thickness, the maximum mechanical properties are obtained roughly at t = 3 minutes, i.e., when the sample temperature is around 160[degrees]C. At this time:
* the temperature gradient is not very large and tends to decrease: an homogeneous temperature is obtained at t = 200 seconds
* the distribution of water (and also glass transition) is homogeneous.
[FIGURE 8 OMITTED]
[FIGURE 9 OMITTED]
However, the mechanical properties are not homogeneously distributed in the sample thickness. We conclude that the thermal and moisture history of the sample before t = 3 minutes determines the characteristics of mechanical properties after t = 3 minutes, the density of the board is established before autoadhesion mechanisms and contribute to the future cohesion of the system. The final cohesion of the board is established at high temperature, and is probably connected to cross-linking reactions.
To improve board mechanical properties, two strategies can be envisaged, aiming to increase (T-Tg):
* play on the process, limit the rate of water evaporation and/or increase the temperature heating rate
* play on the intrinsic properties of the fibers, i.e., decrease the glass transition of lignin in lignocellulosic fibers. Considering Fig. 7, a Tg decrease of 30[degrees]C should be sufficient to obtain a positive T-Tg difference at any time and position in sample thickness. We will work on the second strategy.
Sakata and Senju  have studied the plasticization of lignins by different synthetic plasticizers They showed that the plasticization efficiency of tributylphtalate decreases drastically in the presence of water with respect to the efficiency in the dry state. Our objective is to have the same plasticizing efficiency whatever the water content. A complete discussion on lignin plasticization will be published in a later work. In Fig. 9, is given the example of vanillin in Kraft lignin. The plasticizing properties of vanillin are not affected by the presence of water: the decrease of glass transition temperature is roughly the same in dry conditions and at 80% RH (11% water content) (straight line fits are parallel).
The in situ plasticizing effect of vanillin is not as efficient as in isolated Kraft lignin. This can be concluded from Fig. 10, where DMA data of unplasticized and plasticized fibers are shown. The [alpha] relaxation (taken from the loss modulus peak, as the damping peak is not very differentiated from degradation) decreases from 175 to 145[degrees]C on the dry sample, after addition of 12% vanillin; 12% vanillin in wood corresponds to a quantity of more than 36% when expressed on the basis of lignin content (a maximum lignin content of 1/3 is assumed in the wood fiber). This quantity was not tested in Kraft lignin, but a Tg lower than 100[degrees]C can be extrapolated from Fig. 9. The small decrease of Tg after in situ plasticization can be explained by:
[FIGURE 10 OMITTED]
* An imperfect penetration of vanillin in the fiber complex structure.
* A nonnegligible solubility of vanillin in hemicelluloses (even if vanillin was chosen for its structural similarity with lignin).
* A lower plasticizing efficiency in the tridimensional in situ lignin network (complex interactions with other macromolecular constituents).
* The DMA characterization of fiberboard sample does not reflect directly the bulk properties of fibers: the DMA response of a noncompact structure can be more linked to the interfaces between wood particles. This can be deduced from the behavior of wood, which is totally different compared to fiberboards. In wood, the main transition is very closed to the degradation and corresponds to the value extrapolated by Kelley et al. . In fiberboard reference samples, the transition is lower; this can be due to the combination of different factors that occur at the surface: exudation of low molecular weight compounds at the surface, degradation of polymer chains at the surface during the refining of fibers.
The same behavior is obtained from fibers plasticized by ethylene glycol (Fig. 10), which was used by Wennerblan et al.  as an efficient in situ swelling (and plasticizing) agent. Samples were impregnated according the same procedure as for vanillin, but using water instead of ethanol. The reference sample (obtained with a water immersion pretreatment) shows a relaxation temperature at 160[degrees]C. The decrease of 15[degrees]C between ethanol reference sample and water reference sample can be explained by a lower extraction by water of low molecular weight compounds that naturally plasticize lignins.
In Fig. 10, the shift between relaxation temperatures corresponding to the reference sample and the ethylene glycol-treated sample is around 10[degrees]C. The decrease of relaxation temperature by ethylene glycol plasticization is much lower than by vanillin plasticization (around a 30[degrees]C decrease).
The plasticized fibers were used to make fiberboards. The procedure was changed in order to avoid plasticizer loss during the step of fiber dispersion in water (cf. Methods section). As a consequence, these boards have not optimized properties (lower mechanical properties, and higher standard deviation). Two reference samples are displayed on Fig. 11, as the blank (0%) treatments were different for ethylene glycol plasticization (impregnation using aqueous solution) and vanillin plasticization (impregnation using a solution in ethanol). In both cases, the increase of plasticizer content leads to an increase of elastic modulus as well as an increase of strength at break. A "pure" plasticizing effect leads classically to a decrease of these responses. Therefore, the consequence of Tg decrease on the autoadhesion phenomena is more influent than its direct consequence on the properties of individual fibers.
Vanillin could be suspected to play the role of a reactive plasticizer, as it can be implied in the same mechanisms as lignin-lignin cross-link reactions. Therefore, the improvement of mechanical properties is not only connected to plasticization. The objection cannot be done with ethylene glycol, which improves mechanical properties and cannot be implied in cross-linking reactions.
As a "dry" method is used to process plasticized boards, fibers are not ideally dispersed. This leads to an heterogeneity of the board density, which explains the distribution of properties. It is also clear that the mechanical data (Fig. 11) of plasticized and unplasticized samples are located on the same master curves as a function of density. This means that mechanical properties are only governed by density (i.e., contact surface between fibers); plasticization is only a way to obtain higher densities. Lignin reactivity probably does not depend on plasticization.
[FIGURE 11 OMITTED]
A processing procedure of binderless fiberboard was completely characterized. The time variation of the Tg thickness gradient was compared to the time variation of the temperature gradient. The fibers are at a rubbery state during a limited period that depends on the position within the sample thickness: the period is shorter near the side where water vapor is evacuated, due to lower water content at this location within the thickness. Mechanical properties are poorer on this side.
The in situ plasticization of lignin increases the final mechanical properties. Further work is focused: 1) on the improvement of the in situ plasticization and 2) on the in situ characterization of lignins (differentiation of surface and bulks lignins).
A. Lemaitre is acknowledged for technical work. Mr. Duparez (UNALIT) is acknowledged for useful discussions.
1. N. Laemsak and M. Okuma, J. Wood Sci., 46(4), 322 (2000).
2. M.P. Wolcott, F.A. Kamke, and D.A. Dillard, Wood Fiber Sci.: J Soc. Wood Sci. Technol., 22(4), 345 (1990).
3. S.S. Kelley, T.G. Rials, and W.G. Glasser, J. Mater. Sci., 22, 617 (1987).
4. I. Sakata and R. Senju, J. Appl. Polym. Sci., 19, 2799 (1975).
5. D. Atack, Philos, Mag. A, 43(3), 619 (1981).
6. H. Becker and D. Novack, Wood Sci. Technol., 2, 213 (1968).
7. H. Hoglund, U. Sohlin, and G. Tistad, Tappi, 59(6), 144 (1976).
8. T. Sadoh, Wood Sci. Technol., 15, 57 (1981).
9. C.A. Wert, M. Weller, and D. Caulfield, J. Appl. Phys., 56(9), 2453 (1984).
10. K. Takahashi, T. Morooka, and M. Norimoto, Wood Res., 85, 78 (1998).
11. W.E. Hillis and A.N. Rosa, Holzforshung, 32(2), 68 (1978).
12. H. Funakoshi, N. Shiraishi, M. Norimoto, T. Akoi, H. Hayashi, and T. Yokota, Holzforshung, 33(5), 159 (1979).
13. G. Ostberg, L. Salmen, and J. Terlecki, Holzforshung, 44(3), 223 (1990).
14. L. Salmen and A.M. Olsson, J. Pulp Paper Sci., 24(3), 99 (1998).
15. M. Westin, R. Simonson, and B. Ostman, Holz Roh-Werkst-off, 58(6), 393 (2001).
16. T.K. Kwei, J. Polym. Sci. Polym. Lett., 22, 307 (1984).
17. I. Sakata and R. Senju, J. Appl. Polym. Sci., 19, 2799 (1975).
18. M. Wennerblan, A.M. Olson, and L. Salmen, Nordic Pulp Paper Res. J., 4(11), 279 (1996).
J. Bouajila, A. Limare, C. Joly, P. Dole
Institut National de la Recherche Agronomique, UMR (INRA/URCA) FARE, CPCB Moulin de la Housse, BP 1039, 51687 Reims Cedex 2, France
Correspondence to: P. Dole; e-mail: firstname.lastname@example.org
|Printer friendly Cite/link Email Feedback|
|Author:||Bouajila, J.; Limare, A.; Joly, C.; Dole, P.|
|Publication:||Polymer Engineering and Science|
|Date:||Jun 1, 2005|
|Previous Article:||Recycling of PVC/PU waste and reuse in PVC formulations: structure-property relationship.|
|Next Article:||High-density polyethylene/cycloolefin copolymer blends. Part 1: phase structure, dynamic mechanical, tensile, and impact properties.|