1. Introduction
Lignocellulosic biomass derived from many agricultural and forestry residues, such as wood processing or harvested crop by-products, represents one of the most abundant renewable resources on Earth. Among agricultural residues, sugarcane bagasse (SCB) is the fibrous residue obtained after sugarcane juice extraction and is the most important sugar industry by-product, with 0.3 tons of SCB generated per ton of processed sugarcane.
SCB, which is mainly composed of lignocellulosic (cellulose, hemicellulose, and lignin) fibers, could be used for fiberboard production by thermocompression. This type of application has been tested recently with delipidated cake obtained after sunflower oil extraction from whole sunflower plants [
1],
Broussonetia papyrifera [
2], sugarcane leaves [
3], and bamboo [
4]. In addition, SCB-based materials have been produced using thermocompression without any other treatments [
5,
6]. SCB has high lignin (~20%) and sugar (~60%) contents, which could contribute to self-bonding, thereby offering an opportunity to produce cohesive materials without adding any synthetic resin [
7]. However, it was shown that a high hemicellulose content in lignocellulosic biomass was detrimental to the mechanical properties and dimensional stability of the final material in the presence of water [
8]. A correlation even indicated that a reduction in hemicellulose content was linked with a reduction in water absorption in the final material [
9]. In a previous study, hemicellulose degradation during binderless SCB thermocompression improved the water resistance of the final material by allowing internal chemical reorganization [
6]. Nevertheless, the lignocellulosic materials obtained with the best tradeoff between mechanical properties and water resistance had shortcomings in terms of their behavior in the presence of water, with much higher thickness swelling values (41%) than those reported for materials bound with synthetic resin, i.e., <13% [
10].
These results raise the question of the impact of partial hemicellulose removal on the properties of binderless lignocellulosic materials. Molecule extraction from SCB could modify its chemical composition and affect the properties of materials obtained from the remaining solid residue. Partial hemicellulose solubilization during SCB pretreatment could reduce the hydrophilic character of the final material while recovering a liquid filtrate rich in extractable biomolecules. The addition of a pretreatment could also enable modification of the lignocellulosic structure by making the bagasse compounds, e.g., lignin, more accessible, thereby facilitating internal reorganization of the final material during thermocompression.
SCB pretreatment has been widely studied for the purpose of obtaining refined cellulose that could be used to produce bioethanol [
11,
12]. But SCB is also known to contain free sugars and phenolic compounds, i.e., predominately ferulic acid and p-Coumaric acid [
13,
14], that are covalently linked to polysaccharides by ester bonds and to lignin by ester or ether bonds [
15]. Solubilized molecules from SCB could be used to produce high-value-added molecules such as sugars, furfural, and polyphenols, the latter of which are particularly interesting, as they have great potential as antioxidants in the food and nutrition sectors [
16]. Liquid hot water treatment is a very well-known process for molecule extraction from lignocellulosic biomass. Hydrothermal treatment involves the use of pressure to maintain water in a liquid state at elevated temperatures, so it is an attractive approach for biomass fractionation since no chemicals are used [
17]. Aqueous extraction of SCB using this process at 180 °C for 30 min has been shown to release phenolic compounds (14 g/kg of bagasse initial dry matter, DMi), sugars (156 g/kg DMi), acetic acid (55 g/kg DMi), and furfural (5 g/kg DMi) resulting from sugar degradation [
18]. This extraction process was also found to be efficient for hemicellulose recovery (100–150 g of xylose per kg DMi) [
19,
20]. However, the main drawbacks of this type of process are the long extraction times and the large amount of water required to extract the molecules, with L/S ratios > 10 [
21,
22].
The extruder could be considered a reactor that combines mechanical, thermal, and chemical actions in a single step, with L/S extraction performed in continuous mode. Twin-screw extrusion is an intensified process that can be used for pretreatment to reduce the L/S ratio during aqueous extraction but also to modify the SCB lignocellulosic structure [
23]. This process has already been used for the extraction of molecules such as hemicelluloses from poplar [
24], wheat straw [
25], steam-exploded corncob [
26] and wheat bran [
27], pectin from apple pomace [
28], vegetable oils from coriander [
29,
30] and sunflower [
31,
32,
33], proteins from alfalfa [
34], phenolic acids from hemp by-products [
35], and flavonoids with biostimulatory properties from sunflower stalks and heads [
36]. As is the case for liquid hot water treatment, subcritical conditions can be achieved in twin-screw extruders to favor the hydrolysis of ether or ester bonds, solubilize hemicelluloses, and decompose lignin into phenolic compounds. The combination of conveying and reverse screw elements positioned near a filtration section results in effective compression of the mixture, dynamic plug formation, and effective liquid–solid separation due to efficient pressure buildup [
30]. L/S extraction via twin-screw extrusion is therefore an interesting solution for recovering molecules of interest from lignocellulosic biomass in a single step. Twin-screw extrusion can also be used without a filtration module to break down the biomass structure and make the cell wall internal components more accessible. It is also expected that this pretreatment would modify the fiber morphology by increasing its mean aspect ratio [
37]. To go further, we propose a new approach where the plant material would be pretreated by twin-screw extrusion for an integrated full valorization, where the solid residue would be used for the production of binderless materials and the filtrate for the production of biomolecules.
The aim of this study was to investigate aqueous pretreatment of SCB by twin-screw extrusion in order to modify its structure and composition, and to gain insight into the effects on the material properties. The solid residue fraction (extrudate) was used to produce bio-based materials by thermocompression, while the liquid fraction (filtrate) was recovered to analyze the extracted biomolecule composition. Twin-screw extrusion configurations with and without filtration were compared to assess their impact on the properties of the materials generated from the extrudate, and to determine whether the molecule extraction process had positive effects regarding the material properties. With filtration, the impact of the L/S ratio (0.65–6.00) was studied in relation to the extraction yield in the filtrate, the chemical composition of the extrudates, and the final material properties.
3. Materials and Methods
3.1. Raw Material Preparation
Air-dried SCB was provided by eRcane (Réunion, France). SCB was ground using a 16 mm grid with an Electra F6 N V hammer mill (Paris, France) to homogenize the particle size of the raw material before pretreatment. Ground SCB before pretreatment was referred to as raw SCB. The SCB was moistened in a drum mixer by water spray to achieve a water content (w/w) so as to promote fine particle aggregation with long fibers, thereby avoiding segregation during extruder feeding. For chemical composition characterization, part of the raw SCB was ground, using a 2 mm mesh grid, with an IKA-Werke MF 10 basic Microfine grinder drive (Staufen im Breisgau, Germany).
3.2. Twin-Screw Extrusion Pretreatment
SCB was subjected to aqueous pretreatment using a Clextral Evolum HT 53 co-penetrating and co-rotating twin-screw extruder (Firminy, France), with a 53 mm screw diameter (D), to produce the extrudates and filtrates. The extruder had eight modular barrels, each 4 D in length, except for module 1, which had a length of 8 D (i.e., 36 D for the total barrel length). Modules 2 to 8 were heated by electric resistance and cooled by water circulation. The SCB was fed at a rate of 20 kg/h DM into the extruder inlet port using a Coperion K-Tron SWB-300-N constant-weight feeder (Niederlenz, Switzerland) in module 1. Water was injected using a Clextral DKM Super K Camp 112/12 piston pump (Firminy, France) at the end of module 2. The screw profile used in this study is provided in
Figure 6. Two extrusion configurations, with and without filtration, were tested. In the filtration configuration, a filter section consisting of six hemispherical grids with a 1 mm diameter mesh was positioned on module 7 to enable filtrate collection.
After liquid injection, a series of BL22-1.0-90° 2-lobe kneading blocks (1 D total length) and two series of BL22-0.5-90° 2-lobe kneading blocks (1 D total length) were placed at the end of modules 3 and 5, respectively, to ensure intimate water diffusion in the SCB. The CF2C-0.5-0.5 reversed double-thread screws (1 D total length) were positioned at the beginning of module 8 to achieve an intense shearing of the mixture for deconstruction and in the presence of the filtration module to form a dynamic plug to ensure pressing of the mixture to recover the filtrate. The screw rotation speed (250 rpm), SCB feed rate, inlet water flow rates, and barrel temperature were monitored with a control panel. The set of barrel temperature values were 25, 25, 75, 100, 100, 100, 100, and 100 °C from modules 1 to 8, respectively. The experimental variables of this study included the extruder configuration (with and without filtration) and the L/S ratio used (0.65 to 6.00). The operating conditions are presented in
Table 1.
Twin-screw extrusion was run for 5 min before each sample collection to stabilize the operating conditions of the process. Once these conditions were steady, the extrudate and filtrate were immediately collected over a 4 min period to avoid any variability in the outlet flow rates. The sample collection time was determined with a stopwatch. For each tested condition, sample collection was carried out in duplicate, and the extrudate and filtrate were then weighed. The extrudate moisture content was also measured immediately after its collection with a Sartorius MA35 infrared weighing balance (Göttingen, Germany). Extrudates were then dried at 50 °C for at least 12 h and then conditioned in an environmental chamber at 25 °C and 50% relative humidity (RH) for 3 weeks until their moisture content was stabilized, prior to their characterization and transformation into materials by uniaxial thermocompression. Otherwise, the filtrates were centrifuged at 10,000× g at 20 °C for 10 min to separate the remaining solid (pellet) from the extract containing the solubilized molecules (supernatant). The pellet containing fine solid particles was dried at 105 °C and then conditioned in an environmental chamber at 25 °C and 50% RH for 3 weeks prior to chemical characterization. The supernatant was stored at 4 °C prior to chemical analysis.
3.3. Analytical Methods
3.3.1. Chemical Composition of the Solid Samples
The chemical compositions of the solid lignocellulosic samples (raw SCB, extrudates, and pellets) were determined in triplicate. The contents are expressed as a percentage of dry matter (DM). The DM content was evaluated by drying the sample at 105 °C until constant weight, and the ash content was measured after mineralization at 550 °C for 12 h [
50].
Chemical characterization was carried out using a procedure based on the laboratory analytical procedure of the National Renewable Energy Laboratory (NREL) [
51], as also described previously [
6]. Briefly, water and ethanol extractable contents were determined after water and then ethanol (96%) extraction using 1 g of sample and 100 mL of boiled solvent for 1 h in a Foss Fibertech FT 122 extraction system (Hillerød, Denmark). Cellulose, hemicellulose, and lignin contents were determined after a two-step hydrolysis process using 72% sulfuric acid from VWR (Radnor, PA, USA) at 30 °C for 1 h and then a 4% sulfuric acid solution with deionized water at 121 °C for 1 h, followed by filtration. The acid-insoluble lignin (AIL) content was measured gravimetrically after mineralization at 450 °C for 12 h. The acid-soluble lignin (ASL) content was determined in the liquid fraction on a Shimadzu UV-1800 spectrophotometer (Kyoto, Japan) at 240 nm using an absorptivity constant of 25 L/g·cm. The liquid fraction was neutralized with calcium carbonate from Merck (Darmstadt, Germany) until reaching neutral pH and filtered on a 0.2 µm cellulose acetate filter before HPLC analysis on a Thermo Ultimate 3000 HPLC system from Thermo Scientific (Sunnyvale, CA, USA). All HPLC standards (i.e., acetic acid, arabinose, glucose, and xylose) were purchased from Sigma-Aldrich (Saint-Louis, MO, USA). A Rezex RHM-Monosaccharide H+ 300 × 7.8 mm column connected to a Rezex RHM-Monosaccharide H+ 50 × 7.8 mm guard column, both from Phenomenex (Torrance, CA, USA), were used with 5 mmol/L of H
2SO
4 as an eluent at a rate of 0.6 mL/min. The injection volume was 50 µL, while the column was maintained at 65 °C and the RI detector at 50 °C.
3.3.2. Chemical Composition of the Supernatant
Total supernatant concentrations were determined by DM content measurement, as previously described. The ash content was determined as for the solid samples. The supernatant was filtered on a 0.2 µm cellulose acetate filter before direct HPLC analysis to determine the acetic acid and monomeric sugar concentrations (i.e., arabinose, glucose, and xylose). One-step hydrolysis was also applied to the supernatants in triplicate to determine the total sugar concentrations (monomers and oligomers) and lignin contents. For this purpose, 10 mL of supernatant were placed in a hydrolysis tube with 362 µL of 72% sulfuric acid to achieve a final 4% sulfuric acid concentration. The solutions were autoclaved at 121 °C for 1 h, and then filtered and neutralized with calcium carbonate until reaching neutral pH. Lignin, hemicellulose, and cellulose contents were then determined according to the procedure described for solid samples in a previous section.
Total polyphenolic compounds were determined using the Folin–Ciocalteu colorimetric method, adapted from Singleton and Rossi [
52], on all supernatants filtered using a 96-well microplate. Phenolic compounds were reacted with Folin reagent, i.e., consisting of a mixture of phosphotungstic acid and phosphomolybdic acid, in a reaction in which phenols were oxidized by tungsten and molybdenum oxide reduction, yielding a blue color. Polyphenol levels were measured by visible spectroscopy using a calibration curve. Gallic acid, used as reference standard; Folin–Ciocalteu reagent; and Na
2CO
3 were purchased from Sigma-Aldrich (Saint-Louis, MO, USA). Each well was successively filled with 20 µL of supernatant, 10 µL of Folin–Ciocalteu reagent, and 170 µL of a solution containing 2.36% Na
2CO
3. Each sample was analyzed eight times. After the addition of Folin–Ciocalteau reagent, the microplate was placed in a Spectrostar-Nano plate reader from BMG-Labtech (Champigny-sur-Marne, France) and shaken for 10 s. After the addition of Na
2CO
3, the plate was again placed in the reader and shaken for 10 s prior to the analysis. Absorbance was measured at a 700 nm wavelength at 45 °C after a 45 min reaction time. The final concentration was expressed in grams of gallic acid equivalent per liter (g GAE/L).
3.4. Physical Characterization of Extrudates
The particle size distributions of the raw SCB and extrudates were determined in triplicate using a Retsch AS 200 vibratory sieve shaker (Haan, Germany). Samples weighing 40 g were placed in the shaker for 10 min at 1 mm amplitude using a series of six sieves with 6.3 mm, 4.0 mm, 2.0 mm, 1.0 mm, 0.8 mm, and 0.2 mm mesh openings and a bottom plate. The amount of extrudate retained on each sieve was weighed and expressed as a weight percentage in proportion to the total weight. The fine particle fraction was defined as the fraction retained <0.2 mm and collected in the bottom plate.
SEM raw SCB and extrudate images were obtained using a FEI Quanta 450 scanning electron microscope (Hillsboro, OR, USA), with 130 Pa water vapor partial pressure in the chamber at high voltage (12.5 kW) without saturation of the samples.
The bulk and tapped raw SCB and extrudate densities were determined using a Granuloshop Densitap ETD-20 volumenometer (Chatou, France) fitted with a 250 mL graduated cylinder. The sample was weighed in the graduated cylinder and the volume was recorded prior to compaction to determine the bulk density. The cylinder was then tapped 500 times on the volumenometer at 3 mm height and 250 taps/min. The volume was measured at the nearest graduation and the operation was then repeated until a constant volume was obtained to determine the tapped density. All measurements were performed in triplicate.
3.5. Preparation of Binderless Materials by Thermocompression
All SCB materials were obtained by thermocompression of 20 g of raw SCB or extrudate using a steel mold. A 50 t capacity heated hydraulic press from Pinette Emidecau Industries (Chalon-sur-Saône, France) was used to produce flat square materials measuring 70 mm × 70 mm. A uniaxial pressure of 102 MPa was applied to the sample, and the mold was then heated to 200 °C. The temperature was maintained for 10 min, and finally the mold was cooled down for 10 min while maintaining the pressure before opening. For each tested extrusion condition, two materials were produced and photographed. The materials were cut into eight 45 mm-long × 10 mm-wide specimens, and then stored in an environmental chamber at 25 °C and 50% RH for 2 weeks until constant weight before assessing the properties.
3.6. Material Characterization
The density of the materials obtained after thermocompression was determined in triplicate using the remaining pieces of material obtained after cutting the bending test specimens. The densities of these samples were assessed using a method based on Archimedes’ principle, with cyclohexane as the immersion liquid. Using a Sartorius hydrostatic balance (Göttingen, Germany) capable of weighing in both air and liquid media, we were able to determine the density of our samples with the following formula:
where
and
are the sample weights measured in air and cyclohexane,
and
are the densities of air and cyclohexane at room temperature, and
is the thrust correction factor due to the submerged wire.
The bending properties of the test specimens were assessed according to the ISO 16978:2003 standard [
53] using a Tinius Olsen universal testing machine (Horsham, PA, USA) fitted with a 500 N load cell and the three-point bending test. The thicknesses and widths were measured at the specimen’s center with a Tacklife electronic digital sliding caliper (Levittown, NY, USA). The testing speed was set at 1 mm/min, with a 40 mm separation grip. Bending properties were characterized by testing the sixteen specimens cut from the materials. The properties evaluated were the flexural modulus and flexural strength at breaking point, determined with the following formulas:
where
is the length between the two supports (40 mm);
is the sample width;
is the sample thickness;
and
are the forces measured for
and
deformations at 10% and 40% of
, respectively; and
is the force measured at the breaking point.
Water resistance was tested by immersing 45 mm-long × 10 mm-wide specimens in water at 25 °C to determine the water absorption (WA) and thickness swelling (TS) in triplicate according to the ISO 16983:2003 standard [
54]. Before soaking, samples were oven-dried at 105 °C until constant weight to ensure a uniform initial condition. The initial thickness and weight were then measured. The samples were subsequently submerged in distilled water at 25 °C for 24 h. The sample weights and thicknesses were measured hourly for the first 8 h and again after 24 h. The thickness of each sample was measured at three points, i.e., at the center and both ends. WA and TS were calculated for each sample at 24 h using the following formulas:
where
and
are the sample weights initially and after 24 h of water immersion, and
and
are the sample thicknesses initially and after 24 h of water immersion.
3.7. Statistical Analysis
WA, TS, and density determinations were conducted in triplicate. For the mechanical properties, 16 samples were tested from each extrusion condition. Means were statistically compared by one-way analysis of variance (ANOVA) with α = 0.05 and Student’s t-test. In tables and figures presenting the results, values with no significant difference are presented with the same letter (a–d).
4. Conclusions
Continuous aqueous pretreatment of SCB was achieved by twin-screw extrusion. In the filtration configuration, increasing the L/S ratio to 2.05 significantly increased the extraction yield. Using a low L/S ratio (<1.25) resulted in stronger mechanical action, leading to increased extrudate density, fiber aggregation, and partial SCB degradation, resulting in a material with lower flexural strength. With an L/S ratio > 1.25, the water present facilitated molecule solubilization, and the materials obtained had higher mechanical properties. An L/S ratio of 2.05 was the best tradeoff between the extraction yield (11.5 g/kg DMi), water consumption, and material properties (1485 kg/m3 density, 6.2 GPa flexural modulus, 51.2 MPa flexural strength, and WA and TS values of 37% and 44%, respectively). Under these conditions, the pretreatment substantially improved the flexural strength by 54% and the flexural modulus by 27% and reduced the WA by 59% and TS by 56% due to fiber deconstruction. Finally, the extrudate obtained without liquid–solid separation yielded materials with properties equivalent to those of the extrudates obtained with liquid–solid separation for L/S ratios of at least 1.25, indicating that the partial removal of hemicelluloses and lignin was limited and did not markedly change the final material properties.