Castor oil based polyurethane adhesive content on OSSB produced with soybean straw

riented Structural Straw Board (OSSB) panels are composites made from straw originally from agricultural residues and non-formaldehyde based adhesive, whose main application is for construction as a closing and ceiling material. The objective of this study was to evaluate the effect of the polyurethane adhesive content on physical and mechanical properties of OSSB produced with locally available SbS (soybean straw). Four castor oil based polyurethane ratios (6%, 9%, 12% and 15%, by mass of the SbS) were tested. OSSB was evaluated according to European standards for wood-based composites; also, a physical and chemical characterization of the SbS was performed. SbS has an irregular shape, which reduced the OSSB mechanical performance due to the low densification of the composite and the development of transversal cracks on the SbS after the OSSB manufacturing process. The static bending and thickness swelling performance of SbS OSSB was not improved by the increase of polyurethane adhesive content. SbS low ash content and neutral pH extractives indicates that probably its chemical properties should not affect the setting of the adhesive.


Introduction
According to the USDA (UNITED…, 2019), more than 360 million metric tons of soybean oilseeds were harvested worldwide in 2018/2019, and Brazil is the second main soybean producer in the world, after the United States. Lal (2005) and Krisnawati and Adie (2015) determined values ranging from 0.8 to 2.6 of soybean dry straw to grain ratio, demonstrating the large availability of this residue.
Agricultural residues are often considered a problem for farmers due to their low monetary return, but they can be viewed as an economic opportunity, especially for developing countries where they can have a positive impact (ADEL et al., 2016). Some residues (e.g. wheat straw, rice straw and rice husk) receive interest due to their low cost, great availability, reduction of environmental pollution and large number of natural polymers in their composition (ARGENTO et al., 2019;LIMA et al., 2019). Barbu, Reh and Çavdar (2017) indicate that the production of composites with agricultural residues is considered attractive, especially due to the increase in the price of wood as a raw material. According to Väisänen et al. (2016) it is required that these products do not present inferior performance in comparison to currently available wood-based composites, otherwise the selection of the agricultural residue as a raw material will not happen.
It is also important to note that there have not been studies performed to evaluate the thermal conductivity of SbS based composites, even though this property is used to determinate heat transfer and insulation of building constructions, such as cladding, roughing and flooring (TRIPATHI; RICE, 2017;ÇAVUŞ et al., 2019).
This paper aimed to analyze the feasibility of producing a structural composite applying locally available SbS, in order to increase its economic value.
OSSB is a commercially available composite with structural performance that uses rice or wheat straw and had its first manufacturing plant starting operations in China in 2009(HAN et al., 2012. The straw is split longitudinally prior to the production of the board due to the action of two grooved rollers with different tangential speed, so the binder is allowed to have contact with the inside of the split straw (BACH; DOMIER; HOLOWACH, 1999).
OSSB is composed of three layers of straw that are oriented perpendicular to each other, offering better dimensional stability and mechanical performance in both directions of the plane (MARRA, 1992); this same principle is applied to wood-based composites, such as plywood, CLT (Cross Laminated Timber) and OSB (Oriented Strand Board). Image analysis is used in order to evaluate the straw orientation and layers thickness, due to the need of better characterizing and understanding this composite structure.
Another feature of OSSB is the use of formaldehyde-free adhesive (i.e. p-MDI resin) in its composition, due to a better adhesion to straw in general (BACH; DOMIER; HOLOWACH, 1999;HAN et al., 2012;CHENG;HAN;FANG, 2013). Also, the limits of formaldehyde release of adhesives applied to lignocellulosic composites are under focus due to society's demand for more ecological processes and because its contribution to poor indoor air quality (STEFANOWSKI; CURLING; ORMONDROYD, 2017;WANG et al., 2018).
Polyurethane is another type of formaldehyde-free adhesive (polyurethane based) that has a wide range of applications in industry (TANG et al., 2008). This binder has been substantially investigated as a replacement of conventional adhesives in wood-based and other lignocellulosic composites (MERLINI; SOLDI; BARRA, 2011;FIORELLI et al., 2012FIORELLI et al., , 2013BERTOLINI et al., 2013;VALARELLI et al., 2014;CHEN;TAI, 2018;NASCIMENTO et al., 2018;VARANDA et al., 2018;FERRO et al., 2019;MACEDO et al., 2019;SHIROSAKI et al., 2019). The use of castor oil polyurethane adhesive in the composition SbS OSSB in this paper is justified due the abundance of the Ricinus communis L. plant in Brazil, whose seeds are used for the production of the resin (CANGEMI et al., 2006).
The goal of this study was to evaluate the potential of using SbS and polyurethane adhesive for producing composites with structural performance.

Methods
SbS was collected two days after the soybean harvest in the region of Itapeva-SP, in Brazil. The soybean clone used was M5917IPRO in a clayey soil with 14 plants per linear meter. SbS was physically and chemically characterized prior to the production of the strawboard. A commercially available polyurethane adhesive (KEHL ® ) with 100% solid content was used, which was obtained by reaction of polymeric isocyanate with castor oil as a polyol in the mass ratio of 50:50.

SbS physical characterization
Straw density was determined according to the method described by Lam et al. (2008). SbS was considered a hollow cylinder and had its mass measured with a laboratory scale and dimensionsdiameter, wall thickness and lengthmeasured with a caliper in order to determine the straw volume and density, as shown on Equation 1: Where: S.D. = straw bulk-density, in kg . m -3 ; Mi = mass of the internode section, in g; d = internode section diameter, in mm; t = wall thickness of the internode section, in mm; and l = length of the internode, in mm.
SbS had its mass determined in a laboratory scale before and after being dried in an air-forced oven at 103 ºC for 24 hrs. Straw moisture content (S.M.C.) was determined using the initial mass (M 1 ) and the dried mass (M2), as described at Equation 2: Where: S.M.C. = straw moisture content, in %; M1 = initial mass, in g; and M2 = dried mass, in g.
SbS is composed of the plant's stem and petioles, each of them presenting different dimensions. A comparison of number of internodes, average diameter, average wall thickness, average internode length and total length for both parts of the soybean plant was carried out.

SbS chemical characterization
SbS was initially reduced to particles using a hammer mill (TE-650, TECNAL. Brazil) until reaching dimensions of 40-to 60-mesh. Total Extractive Content (T.E.C.) and Hot Water Extractive Content (H.W.E.C.) were determined according to TAPPI 204 method. T.E.C. was determined as the difference between masses of the particles before (mb) and after (ma) a three stage extraction process, as shown in Equation 3: Where: T.E.C. = total extractive content, in %; mb = mass before extraction, in g; and ma = mass after extraction, in g.
The first stage of extraction used 1/3 ethanol and 2/3 benzene solution for 5 h, the second stage used ethanol solution for 4 h, and in the last stage the extraction used boiling water for 1 h. H.W.E.C. followed the same principle, except that there was only one stage of extraction in boiling water for 3 h. The resulting solution of the H.W.E.C. test had its pH measured.
Ash Content (A.C.) of SbS was determined according to EN 1755-01 (AMERICAN…, 2015) method. Two grams samples were heated in a laboratory furnace at 575 ºC for 3 h. A.C. was considered as the ratio between the sample mass after three hours at the furnace (mf) and initial mass (mi), as described in Equation 4: . . = • 100 Eq. 4 Where: A.C. = ash content, in %; mf = final mass, in g; and mi = initial mass, in g.

OSSB production process
OSSB was produced in laboratory scale according to the method described by Bach, Domier and Holowach (1999) and Shirosaki et al. (2019). SbS was manually split along its main axis to expose its inner regions, and then was dried in a laboratory oven at 103 ºC for 24 h before the production of the boards. Dried split SbS was mixed with castor oil based polyurethane adhesive and the mattress was formed inside a retaining square box with 0.42 m side.
The mattress was pressed in room temperature for 10 min with pressure of 0.4 MPa. Then the board was pressed in a heated hydraulic press (PHH 80T, Hidralmac. Brazil) at 100 ºC in three cycles of 180 s (there was a pressure release of 3 s between cycles) with pressure of 4 MPa, following the Ferreira, Silva and Campos (2017) method. The board had a nominal thickness of 15 mm and target density of 650 kg . m -3 and was composed of three perpendicular layers in the proportion of 30 : 40 : 30 by mass. Each layer had the straw oriented longitudinally in a preferential direction. Four adhesive contents (ranging from 6% to 15% in relation to the dried straw mass) were tested, in order to determine its optimal amount in the composition of OSSB.

OSSB physical characterization
The determination of density of the boards was based on EN 323 standard (EUROPEAN …1993a).
Specimens ( Moisture content (M.C.) determination method was based on EN 322 standard (EUROPEAN…, 1993b). Specimens with at least 0.02 kg had their initial mass (mi) and final mass (mf) measured with a precision scale. The final mass was defined as the mass of the specimen dried in a laboratory oven at 103 ºC until reaching a constant valueless than 0.1% variation between consecutive measurements taken every six hours. The moisture content of the boards was defined according to Equation 6: Where: M.C. = moisture content, in %; 27 mi = initial mass, in g; and mf = dried mass, in g.
Thickness Swelling (T.S.) after immersion in water for 2 and 24 h were determined according to EN 317 standard (EUROPEAN…, 1993c). Square specimens (0.05 m x 0.05 m) had their thickness measured with a caliper before (ti) and after (tf) immersion in water at room temperature and neutral pH for 2 h and 24 h, as shown in Equation 7: . . = ( − ) • 100 Eq. 7 Where: T.S. = straw moisture content, in %; ti = initial thickness, in mm; and tf = final thickness, in mm.
The optical micrographs of the specimens were taken by a Lupe (M80, Leica. Brazil) with a coupled camera (DFC295, Leica. Brazil) connected to a computer using Leica QWin Standard V 3.5.1 software (2008) for image processing. Conductivity tests were performed with a Thermal Properties Analyzer (KD2 Pro, Decagon Devises, Inc.) in an acclimatized chamber (SM-3.5S, THERMOTRON) at 20 ºC.

OSSB mechanical characterization
Mechanical tests were performed using a universal testing machine (DL 30000, EMIC. Brazil). The threepoint bending test was based on EN 310 standard (EUROPEAN…, 1993d), where specimens (350 mm x 50 mm) were tested using a span of 300 mm with a crosshead speed of 7 mm/min. MOE and MOR were determined according to Equations 8 and 9, respectively. Internal Bonding (IB) test was performed according to EN 319 standard (EUROPEAN…, 1993e). Square specimens (50 mm x 50 mm) were bonded to wood braces using fast bonding contact adhesive. Crosshead speed of 0.5 mm/min was applied until failure of the samples, and IB was determined according to Equation 10.

IB = Internal bond, in MPa;
Pmax = maximum load, in N; w = width of the specimens, in mm; and l = length of the specimens, in mm.
Resistance to screw withdrawal was performed according to EN 320 standard (EUROPEAN…, 2011), as seen in Figure 1. Specimens (75 mm x 75 mm) had their edge and face screw withdrawal strength measured by the maximum tension applied before the failure of the sample.

Statistical analysis
The average interval of confidence for the physical and chemical evaluated properties of the SbS was determined, tested with p < 0.05. The factor studied was the polyurethane adhesive content in OSSB production, totaling four treatments with six replications for each test. The Tukey test was performed by ANOVA analysis, tested with p < 0.05. The software used to interpret the results was R version 3.2.3.

SbS physical and chemical properties
The physical and chemical properties determined for SbS are shown in  , 2014). Although SbS degradation analysis was not the focus of this paper, mold formation was observed after a couple of weeks when the residue was collected on the day of the soybean harvest, justifying the need of a preliminary drying.
It was determined that more than 90% of SbS extractives were soluble in hot water and did not alter their pH. The T.E.C. amount determined for Sb was higher than those reported for rice straw (10%) and wheat straw ( Table 1 is in the range of values (2% -5%) determined by Reddy and Yang (2009); in contrast to T.E.C., wheat straw and rice straw had higher A.C. than SbS, 6.49% and 9.2%, respectively (MOUSAVI et al., 2013).  High A.C. usually changes the pH of the raw material and accelerates the setting of the adhesive, impairing the composite mechanical performance due the reduction of adhesion between the reinforcement and the matrix (IWAKIRI, 2005). This was not likely to happen since SbS had low A.C., compared to rice and wheat straw, and neutral pH as shown in Table 1, indicating that its chemical properties probably had no negative effect on the OSSB mechanical performance.
The average geometrical dimensions of the soybean stem and petioles are shown in Table 2, as well as the coefficient of variation and the minimum and maximum measured values. It was observed that the soybean stem and petioles differed in almost every parameter measured, except for the total available length. The coefficients of variation were high, indicating a low performance of the adopted procedure.
The number of internodes represents the amount of divisions on SbS that are expected to break during the straw processing for OSSB production. Bach, Domier and Holowach (1999) and Han et al. (2012) describe that a raw material with the least number of internodes (i.e. longer internode length) and a controlled moisture content are desired in order to reduce the generation of fines during the manufacturing process. The effect of these internode divisions is shown in Figure 2 as an irregularity on the OSSB surface, which could difficult or even prevent the successful application of coatings on the board.
Nevertheless, the wide range of results indicates the necessity of a different industrial process than the one applied to wheat straw for the production of OSSB. This can be inferred, as the minimum value obtained of the external diameter is smaller than the maximum value of the wall thickness, due to the variation of the dimensions of the plant along its height. In addition, some plants do not show petioles on harvest season, which are represented by the "zeros" on the number of internodes and on the total available length shown in Table 2.

SbS OSSB physical properties
SbS OSSB physical properties (i.e. density, moisture content and thickness swelling) are shown in Table 3. OSSB density reached 631 kg . m -3 in composites produced with 15% resin content, close to the results reported by Cheng, Han and Fang (2013) and Sitz and Bajwa (2015) for OSSB and MDF (Medium Density Fiberboard) produced with wheat straw and pMDI resin, respectively. The OSSB density increased with higher amounts of polyurethane adhesive content, which can also be observed in Figure 3, where the composition of 15% shows SbS to be more compacted. Bertolini et al. (2013) and Varanda et al. (2018) studied wood composites produced with polyurethane obtaining higher densities (from 800 to 1100 kg . m -3 ). On the other hand, Valarelli et al. (2014) achieved close values to the 15% polyurethane composition SbS OSSB density in the bamboo particleboard while Chen and Tai (2018) obtained inferior values in bamboo charcoal composites produced with PU adhesive.
The density of the lignocellulosic material used in PU adhesive matrix affects the final density of the composite, so it was expected that SbS OSSBs densities were at least the same of the SbS, but only the 12% and 15% polyurethane adhesive content boards were. This was due to the empty spaces caused by the geometry of SbS and the lower compaction of the 6% and 9% polyurethane adhesive content treatments, as seen on Figure 2 and Figure 3, respectively.
30   The M.C. determined for all tested OSSB fullfiled the requirements of EN 300 standard (EUROPEAN…, 2006), which must range from 2% to 12%. The values of M.C. were slightly reduced by the increase of PU adhesive content showing a decrease in water affinity, also observed in the results of T.S. for 2 h (see Table  3). Nevertheless, over 50% of T.S. 24h was achieved in the first 2 hrs of water immersion and more than 9% of PU content did not reduce T.S. 24 hrs significantly.

Figure 2 -Surface imperfection in OSSB due to SbS internode division (magnification of 100X)
The maximum value allowed for T.S. 24 hrs according to EN 300 standard (EUROPEAN…, 2006) is 20%, for load bearing boards on dry conditions, less than half the average values obtained for SbS OSSB. T.S. 24h for wheat straw composites -OSSB and MDFwith pMDI binder ranged from 9.2% to 20.6% (CHENG; HAN;FAN, 2013;SITZ;BAJWA, 2015); and lignocellulosic particleboards with PU adhesive T.S. 24h ranged from 8.17% to 22.69% (FIORELLI et al., 2012(FIORELLI et al., , 2013VALARELLI et al., 2014). The low performance of SbS OSSB is justified due the low adhesion of SbS with PU adhesive, as shown on Table 4 for I.B. tests results.
Specimens showed a thermal conductivity ranging from 0.193 to 0.202 W . m -1 . K -1 , which characterizes this material as a thermal insulator (TRIPATHI; RICE, 2017). Çavuş et al. (2019) determined values between 0.090 and 0.197 W . m -1 . K -1 in their study with 31 different wood species and Binici, Aksogan and Demirhan (2016) found values between 0.075 and 0.1588 W . m -1 . K -1 for corn straw boards. The conductivity of SbS OSSB is slightly higher due to its high content of SbS extractives, which probably increased the thermal conductivity despite the low density of the SbS and the composite.

SbS OSSB mechanical properties
The static bending MOE and MOR in both directions of the board, internal bonding and screw withdraw tests results are shown on Table 4. SbS OSSB MOE performance did not increase for PU content higher than 9% of binder was used in its composition and MOR reached its maximum value with 12% PU, similar to T.S. 24 hr results. EN 300 standard (EUROPEAN…, 2006) HAN;FANG, 2013;SITZ;BAJWA, 2015). The low static bending performance of OSSB is due the presence of transversal cracks on SbS (see Figure 4) which reduced the MOE and MOR values. These fragile cracks were developed because SbS was compressed to form the OSSB, and due to its irregular shape (as showed on the SbS physical and chemical properties subsection) the SbS was bent and twisted (see Figure 3) in the manufacturing process.
The effect of adhesive content of SbS OSSB was clearly seen on I.B. test results, as the 15% PU content composition performance was significantly higher than the other compositions tested, showing values close to those required for the EN 300 standard (EUROPEAN…, 2006) for load bearing boards in dry conditions. The I.B. performance obtained is also close to other straw based composites produced with 3% to 4% pMDI binder and 14% urea formaldehyde binder (CHENG; HAN; FAN, 2013; BEKHTA; KORKUT; HIZIROGLU, 2013; SITZ; BAJWA, 2015), but inferior to PU bonded composites (FIORELLI et al., 2012 and. The rupture of I.B. specimens occurred on the composite interface (see Figure 5), which created an irregular separation plane due to the heterogeneous dimensions of SbS (described in the SbS physical and chemical properties subsection).

Conclusions
SbS has heterogeneous dimensions according to its measured parameters, which were: internode length, external diameter and wall thickness. 6% and 9% castor oil based polyurethane adhesive content OSSB had lower density than the SbS itself due to the raw material geometry. On the other hand, higher binder content boards (12% and 15%) showed a better densification, but developed transversal cracks on SbS due to its irregular shape and the OSSB manufacturing process impairing the boards' mechanical performance.
Some properties of SbS OSSB were clearly affected by the PU adhesive content, such as density, moisture content, internal bonding and screw withdrawal strength, but static bending MOE and MOR and thickness swelling did not improve with higher adhesive content. High SbS moisture content indicates the necessity of its drying before the manufacturing process. Due to the neutral extractives' pH and the lower ash content it can be concluded that the chemical properties of SbS did not affect the setting of the PU adhesive.
It was verified that the boards exhibited good performance as an insulating material, but not as a structural element. Therefore, they can be applied in construction as vertical or horizontal closing elements.