Development of High Strength Particleboards from Hemp Shives and Corn Starch

In the current study, high-strength boards for the construction industry were developed from renewable natural resources, fibrous hemp shives, and corn starch. During the research, the influence of the composition of the mixture, the processing of raw materials, and technological parameters on the operational properties of the board were evaluated. The influence of the binding material and the water content on the properties of the molded boards was evaluated. It was established that the rational amount of starch is 15% of the mass of the shives, and the amount of water is 10%. It has been established that with the proper selection of the forming parameters of the board, it is possible to avoid internal disintegration of the structure due to the water vapor pressure, increase the bending strength, and ensure uniform sintering of the board throughout the entire volume. It was found that additional processing of hemp shives can increase bending strength by more than 40%. Furthermore, during the processing of shives by chemical means, soluble substances are washed out, which reduces the density and thermal conductivity of the shives. Selection of a rational level of compression allowed us to increase the bending strength of the boards by 40%. The assessment of all factors made it possible to obtain boards with a bending strength of 40 MPa. The additives used made it possible to reduce the water absorption of the boards up to 16 times and obtain non-flammable boards. The thermal conductivity of the resulting boards varied from 0.07 to 0.095 W/(m·K). The analysis of macrostructure and microstructure allowed us to evaluate the process of the formation of bonds between hemp shives.


Introduction
Most boards in construction and furniture industry are made from poor wood or wood processing waste as filler and bound with phenol formaldehyde-based binders [1,2]. During the last two decades, many different fillers from renewable resources have been proposed in the scientific literature for the production of boards. Cultivated and uncultivated plant straw, different wood parts, plant leaves, etc., can be used as fillers [3][4][5][6]. Although most of these fillers are bound with traditional phenol formaldehyde-based binders, other binders have also been proposed. Starch, sugar, sapropel, lignin, protein, etc., can be used as binders [6][7][8][9].
Industrial hemp probably receives the most attention for the production of fillers. This attention to industrial hemp is primarily due to the large amount of dry mass obtained per year from one hectare [10]. Additionally, the greater the amount of mass grown, the greater the amount of CO 2 that industrial hemp absorbs from the air during growth [11][12][13]. Hemp stalks can be used to obtain fibers and shives. Hemp shives are characterized by a finely porous structure, which allows one to obtain lightweight and low thermal conductivity composites [14,15]. Fibrous hemp has an even greater uniqueness compared to wood: it is characterized by a significantly lower amount of volatile organic compounds [16]. application depends on the required permanency and the desired effect. The Tubiguard 44N was produced by CHT Germany GmbH, Tübingen, Germany.
The formation mixtures were prepared from the raw materials available, the compositions of which are given in Table 1. The quantities of all raw materials were calculated from the mass of dry hemp shives. In this work, untreated and chemically treated hemp shives were used. In a previous work [15], it was determined that the rational amount of starch to form boards is 10%. In this work, we chose corn starch amounts of 5, 10, 15, and 20% because the starch was additionally modified with sodium metasilicate and its influence on starch kleisterization is unknown. The amount of water depends on the parameters of the molding form and the thermal treatment mode. From previous works [9,15], we know that an excessive amount of water affects the disintegration of the structure, while an insufficient amount of water leads to incomplete bonding of the binder. In this work, we varied the amount of water from 7.5 to 15.0%.
Untreated and chemically treated hemp shives were used for the preparation of specimens. For the tests, fractions of 0-5 mm fraction were used. The data on the shives processing regime are presented in Table 2. The binder composition was prepared from corn starch and sodium metasilicate. Raw materials selected in different proportions were mixed with water until a mixture of liquid consistency was obtained. The prepared mixture was applied to two parallel wooden plates, which were compressed with additional weights and hardened for 1 h at a temperature of 160 • C in a drying oven. After curing, the mixture is subjected to a tensile test according to EN 1607 [24] to evaluate the effect of the ratio of raw materials on the adhesive capacity of the mixture. Five specimens were prepared for the tests. The tensile strength of the specimens was calculated according to the Formula (1): where F m is the maximum recorded tensile force, in N; l and b are the length and width of the tested specimen, in m 2 .

Forming the Mat of the Specimens
The prepared mixture was poured into a pre-prepared mold (see Figure 1). The mold was placed on a perforated plate, under which there was a stainless steel woven mesh and a protective galvanized plate for the removal of excess moisture during thermal pressing. The poured mixture was leveled, and other layers were formed in the same way. All the poured mixture in the form was pressed with a special cover to form a stable mixture mat.
The poured mixture was leveled, and other layers were formed in the same way. All the poured mixture in the form was pressed with a special cover to form a stable mixture mat. The main indicator that was taken into account when creating the board was the bending strength. We determined the bending strength according to the methodologies specified in the EN 12089 [25] standard. The test device used for the test consists of two  The main indicator that was taken into account when creating the board was the bending strength. We determined the bending strength according to the methodologies specified in the EN 12089 [25] standard. The test device used for the test consists of two parallel cylindrical roller supports, the lengths of which were greater than the width of the specimen, its diameter is 15 ± 0.5 mm, and a cylindrical loading head with a diameter of 30 ± 0.5 mm was placed parallel to the supports and at the same distance from them ( Figure 2). The test was carried out using a computerized press Hounsfield H10KS and the computer software QMat Professional Ver.3.83 adapted to it. parallel cylindrical roller supports, the lengths of which were greater than the width of the specimen, its diameter is 15 ± 0.5 mm, and a cylindrical loading head with a diameter of 30 ± 0.5 mm was placed parallel to the supports and at the same distance from them ( Figure 2). The test was carried out using a computerized press Hounsfield H10KS and the computer software QMat Professional Ver.3.83 adapted to it.
(a) (b) Figure 2. Determination of bending strength: (a) preparation of specimens for bending strength tests; (b) specimen during testing between press supports.
The bending strength is determined by applying a load to the center of a specimen placed on two supports. The specimen is placed on the supports so that its longitudinal axis is perpendicular to the longitudinal axes of the supports and its center is under the cylindrical loading head. During the test, the loading head moves at a constant adjusted speed so that the maximum load force is reached within 60 ± 30 s. The bending strength was calculated by determining the ratio of the bending moment at the maximum load force Fm and the moment of resistance of the entire cross-section of the specimen, according to Formula (2): where σb is the bending strength, MPa; Fm is the maximum load force achieved, N; l is the distance between the supports' centers, mm; b is the specimen width, mm; and t is the specimen thickness, mm.
Five specimens with dimensions of (20 × t + 50) × 50 × t mm were prepared for the bending strength test. Before the bending test, the prepared specimens were conditioned in an environment with a temperature of 23 ± 2 °C and a relative humidity of 50 ± 5%.
Thermal conductivities were determined for untreated and chemically treated hemp shives as well as for specimens prepared from the boards. Thermal conductivity was determined according to the requirements of the EN 12664 [26] standard by measuring the heat flow and the temperature difference between the surfaces of the specimen. A computerized thermal conductivity device Fox-304 LaserComp was used for the tests. The temperature difference between the measuring plates was 20 °C and they had average temperature of 10 °C. Each specimen was pressed by the device's vertically moving top plate to ensure good contact between the specimen and the measurement plates.
Specimens prepared for thermal conductivity for at least 72 h were kept at a temperature of 23 ± 2 °C and 50 ± 5% relative air humidity. Specimens with dimensions of 300 × 300 × (from 12 ÷ 15) mm were used to test the thermal conductivity coefficient (see Figure 3). The bending strength is determined by applying a load to the center of a specimen placed on two supports. The specimen is placed on the supports so that its longitudinal axis is perpendicular to the longitudinal axes of the supports and its center is under the cylindrical loading head. During the test, the loading head moves at a constant adjusted speed so that the maximum load force is reached within 60 ± 30 s. The bending strength was calculated by determining the ratio of the bending moment at the maximum load force F m and the moment of resistance of the entire cross-section of the specimen, according to Formula (2): where σ b is the bending strength, MPa; F m is the maximum load force achieved, N; l is the distance between the supports' centers, mm; b is the specimen width, mm; and t is the specimen thickness, mm. Five specimens with dimensions of (20 × t + 50) × 50 × t mm were prepared for the bending strength test. Before the bending test, the prepared specimens were conditioned in an environment with a temperature of 23 ± 2 • C and a relative humidity of 50 ± 5%.
Thermal conductivities were determined for untreated and chemically treated hemp shives as well as for specimens prepared from the boards. Thermal conductivity was determined according to the requirements of the EN 12664 [26] standard by measuring the heat flow and the temperature difference between the surfaces of the specimen. A computerized thermal conductivity device Fox-304 LaserComp was used for the tests. The temperature difference between the measuring plates was 20 • C and they had average temperature of 10 • C. Each specimen was pressed by the device's vertically moving top plate to ensure good contact between the specimen and the measurement plates.
Specimens prepared for thermal conductivity for at least 72 h were kept at a temperature of 23 ± 2 • C and 50 ± 5% relative air humidity. Specimens with dimensions of 300 × 300 × (from 12 ÷ 15) mm were used to test the thermal conductivity coefficient (see Figure 3).
The short-term water absorption of the specimens prepared from the flat prototypes was determined by immersing the specimens in water for a period of 24 h, according to the instructions of method A of the EN 29767 [27] standard. After the specified time, the specimens were removed from the water and excess water that has not been absorbed was allowed to drain. For testing, five specimens with dimensions of 50 × 50 × t mm were prepared from each series of composites. Before the start of the test, the specimens were conditioned for at least 6 h in an environment with a temperature of 23 ± 5 • C. After conditioning, the specimens were weighed, and their lengths and widths were measured. The specimens were placed in an empty bath and a load was placed on them to prevent them from rising when filled with water. Water at a temperature of 23 ± 5 • C was poured into the bath with the specimens until its level reached a level of 10 ± 2 mm above the lower plane of the specimen. The immersed specimens were left in water for a period of 24 ± 0.5 h.  The short-term water absorption of the specimens prepared from the flat prototypes was determined by immersing the specimens in water for a period of 24 h, according to the instructions of method A of the EN 29767 [27] standard. After the specified time, the specimens were removed from the water and excess water that has not been absorbed was allowed to drain. For testing, five specimens with dimensions of 50 × 50 × t mm were prepared from each series of composites. Before the start of the test, the specimens were conditioned for at least 6 h in an environment with a temperature of 23 ± 5 °C. After conditioning, the specimens were weighed, and their lengths and widths were measured. The specimens were placed in an empty bath and a load was placed on them to prevent them from rising when filled with water. Water at a temperature of 23 ± 5 °C was poured into the bath with the specimens until its level reached a level of 10 ± 2 mm above the lower plane of the specimen. The immersed specimens were left in water for a period of 24 ± 0.5 h.
Short-term water absorption after 24 h of soaking in water was calculated according to Formula (3): where Wp is the short-term water absorption, kg/m 2 ; m24 is the mass of the specimen after partially immersed in water for 24 h, kg; m0 is the initial mass of the dry specimen, kg; and A0 is the area of the lower plane where the specimen is immersed in water, m 2 .
The combustion tests of the manufactured boards were carried out according to the methodologies specified in the LST EN ISO 11925-2:2010 standard [28]. This test determines the propagation of a small flame up the vertical surface of the specimen when the surface or edge of the specimen is exposed to a small flame for an appropriate period of time. From each series of boards, 5 specimens were prepared with a length of 250 ± 2 mm and a width of 90 ± 2 mm. Before the test, the specimens were conditioned until a constant mass was reached, at least 48 h in an environment with a temperature of 23 ± 2 °C and relative humidity of 50 ± 5%. After the specified conditioning time, the specimen was placed in the holder. The adjusted burner was advanced until the flame reaches the intended point of contact with the specimen. The specimen was exposed to the flame for 30 s. After this time, the burner was retracted. After the set duration of flame operation, the Short-term water absorption after 24 h of soaking in water was calculated according to Formula (3): where W p is the short-term water absorption, kg/m 2 ; m 24 is the mass of the specimen after partially immersed in water for 24 h, kg; m 0 is the initial mass of the dry specimen, kg; and A 0 is the area of the lower plane where the specimen is immersed in water, m 2 . The combustion tests of the manufactured boards were carried out according to the methodologies specified in the LST EN ISO 11925-2:2010 standard [28]. This test determines the propagation of a small flame up the vertical surface of the specimen when the surface or edge of the specimen is exposed to a small flame for an appropriate period of time. From each series of boards, 5 specimens were prepared with a length of 250 ± 2 mm and a width of 90 ± 2 mm. Before the test, the specimens were conditioned until a constant mass was reached, at least 48 h in an environment with a temperature of 23 ± 2 • C and relative humidity of 50 ± 5%. After the specified conditioning time, the specimen was placed in the holder. The adjusted burner was advanced until the flame reaches the intended point of contact with the specimen. The specimen was exposed to the flame for 30 s. After this time, the burner was retracted. After the set duration of flame operation, the test data were recorded: whether the specimen ignites; whether the top of the flame reaches a height of 150 mm above the point of flame action and after what time; whether there are falling flaming drops or particles that ignite the filter paper; whether the specimen smolders and for how long; and other observations regarding physical changes in the specimen.
The view of the tests is presented in Figure 4.

Results and Discussions
An image of untreated and chemically treated shives is presented in Figure 5. After chemical treatment, there was not only the loss of leached materials (see Table 2) but also a change in appearance (see Figure 5b), and many fine particles were formed. These fine particles are hemp fibers that have been attached to the shives prior to processing. During chemical treatment, not only were large amounts of soluble substances washed out but so was a large amount of dust, which negatively affected the board production process: it made the boards heavier and increased the amount of binding material. Hemp shives treated with sodium carbonate and kept in solution for 23 h resulted in a loss of 12.5%.

Results and Discussions
An image of untreated and chemically treated shives is presented in Figure 5. After chemical treatment, there was not only the loss of leached materials (see Table 2) but also a change in appearance (see Figure 5b), and many fine particles were formed. These fine particles are hemp fibers that have been attached to the shives prior to processing. During chemical treatment, not only were large amounts of soluble substances washed out but so was a large amount of dust, which negatively affected the board production process: it made the boards heavier and increased the amount of binding material. Hemp shives treated with sodium carbonate and kept in solution for 23 h resulted in a loss of 12.5%.

Results and Discussions
An image of untreated and chemically treated shives is presented in Figure 5. After chemical treatment, there was not only the loss of leached materials (see Table 2) but also a change in appearance (see Figure 5b), and many fine particles were formed. These fine particles are hemp fibers that have been attached to the shives prior to processing. During chemical treatment, not only were large amounts of soluble substances washed out but so was a large amount of dust, which negatively affected the board production process: it made the boards heavier and increased the amount of binding material. Hemp shives treated with sodium carbonate and kept in solution for 23 h resulted in a loss of 12.5%.  processing [29,30]. In Figure 6, it can be seen how the leaching of the boiled shives changes in several stages. According to data presented in the literature, various soluble substances such as sugars, pectin, waxes, paraffin, etc., are washed out with hot water during scallions processing [29,30]. In Figure 6, it can be seen how the leaching of the boiled shives changes in several stages. (1) (2) (3) Figure 6. Leaching of the shives boiled in water: 1-after the fifth wash; 2-after the third wash; 3after the first wash.
After chemical treatment of the shives, changes in the density and thermal conductivity of the shives were evaluated. The dispersion analysis conducted [31] of experimental studies showed (see Figure 7) that the averages of the thermal conductivity coefficients of the treated and untreated shives are 0.0493 and 0.515 (W/(m × K)), respectively. The statistic of the F criterion is equal to 91.94, p = 0. This indicates a statistically significant difference in the results of the subjects. Consequently, the mean value of the density of the treated shives was (88.0 ± 0.994 kg/m 3 ), and the mean value of the untreated shives was (126.8 ± 1.643 kg/m 3 ). The coefficient of determination of treated and untreated shives R 2 = 0.920 was obtained. The bulk density factor of treated and untreated shives explains 92.0% of the variance of the thermal conductivity coefficient in the specimens. Further studies attempted to equate the bulk densities of treated and untreated shives to assess differences in thermal conductivity. Accordingly, the average value of the bulk density of the shives was (126.2 ± 1.64 kg/m 3 ), and the average value of the untreated shives was (126.8 ± 1.64 kg/m 3 ). The statistic of the F criterion was equal to 200.73, p = 0. After chemical treatment of the shives, changes in the density and thermal conductivity of the shives were evaluated. The dispersion analysis conducted [31] of experimental studies showed (see Figure 7) that the averages of the thermal conductivity coefficients of the treated and untreated shives are 0.0493 and 0.515 (W/(m × K)), respectively. The statistic of the F criterion is equal to 91.94, p = 0. This indicates a statistically significant difference in the results of the subjects. Consequently, the mean value of the density of the treated shives was (88.0 ± 0.994 kg/m 3 ), and the mean value of the untreated shives was (126.8 ± 1.643 kg/m 3 ). The coefficient of determination of treated and untreated shives R 2 = 0.920 was obtained. The bulk density factor of treated and untreated shives explains 92.0% of the variance of the thermal conductivity coefficient in the specimens. According to data presented in the literature, various soluble substances such as sugars, pectin, waxes, paraffin, etc., are washed out with hot water during scallions processing [29,30]. In Figure 6, it can be seen how the leaching of the boiled shives changes in several stages. (1) (2) (3) Figure 6. Leaching of the shives boiled in water: 1-after the fifth wash; 2-after the third wash; 3after the first wash.
After chemical treatment of the shives, changes in the density and thermal conductivity of the shives were evaluated. The dispersion analysis conducted [31] of experimental studies showed (see Figure 7) that the averages of the thermal conductivity coefficients of the treated and untreated shives are 0.0493 and 0.515 (W/(m × K)), respectively. The statistic of the F criterion is equal to 91.94, p = 0. This indicates a statistically significant difference in the results of the subjects. Consequently, the mean value of the density of the treated shives was (88.0 ± 0.994 kg/m 3 ), and the mean value of the untreated shives was (126.8 ± 1.643 kg/m 3 ). The coefficient of determination of treated and untreated shives R 2 = 0.920 was obtained. The bulk density factor of treated and untreated shives explains 92.0% of the variance of the thermal conductivity coefficient in the specimens. Further studies attempted to equate the bulk densities of treated and untreated shives to assess differences in thermal conductivity. Accordingly, the average value of the bulk density of the shives was (126.2 ± 1.64 kg/m 3 ), and the average value of the untreated shives was (126.8 ± 1.64 kg/m 3 ). The statistic of the F criterion was equal to 200.73, p = 0.  Further studies attempted to equate the bulk densities of treated and untreated shives to assess differences in thermal conductivity. Accordingly, the average value of the bulk density of the shives was (126.2 ± 1.64 kg/m 3 ), and the average value of the untreated shives was (126.8 ± 1.64 kg/m 3 ). The statistic of the F criterion was equal to 200.73, p = 0. This indicates a statistically significant difference in the results of the subjects. The coefficient of determination R 2 = 0.962 was obtained.
The main strength Indicator of boards for structural and furniture industry is the bending strength value. The amount of binder is closely related to the bending strength of the boards. In Figure 8, the relationship between the tensile strength of the ratio of starch and sodium metasilicate is shown (see Table 3, Equation (4)). The ratio of starch to sodium metasilicate was evaluated by preparing specimens with different ratios for bending tests. Table 3. ANOVA statistical analysis of results of physical properties.

Equation
Number  We can see that the relationship between tensile strength and the ratio of starch to sodium metasilicate is very strong at 0.947. The analysis of the experimental data showed that the variation in the bending strength is 89.8%, dependent on the ratio of starch and sodium metasilicate. For further research, we chose to use a starch-sodium metasilicate ratio equal to 0.6.
In the next step, we specify the amount of starch required to obtain the maximum bending strength. The experimental studies carried out showed that the bending strength values of the boards with different starch content can be described according to Equation (5) (see Figure 9 and Table 3).
According to Table 3, the correlation coefficient is equal to 0.944. This shows that the relationship between the bending strength and the starch content is very strong. The coefficient of determination is equal to 0.892; it can be concluded that the amount of starch within the limits of this experiment affects the bending strength by up to 89.2%. To choose the optimal amount of starch for further research, variance analysis was performed to check whether the difference between the obtained results was significant. The dispersion analysis showed that the difference in the average bending strength values was significant; when the amount of starch varied from 5 to 20%, the statistic of F criterion was 47.8, p = 0.000. When the amount of starch changed from 10 to 20%, the statistic of F criterion was 8.15, p = 0.00581, and the difference in the mean values of bending strength was also significant. Repeated analysis showed (see Figure 10 Figure 11 shows the dependence of the bending strength on the density of the specimens. The density of the boards ranged from ~700 to ~1100 kg/m 3 . On the bases of the experimental data, the relationship between the bending strength and the density of the specimens was determined, the empirical regression equation of which is presented in    Figure 11 shows the dependence of the bending strength on the density of the specimens. The density of the boards ranged from ~700 to ~1100 kg/m 3 . On the bases of the experimental data, the relationship between the bending strength and the density of the specimens was determined, the empirical regression equation of which is presented in    Figure 11 shows the dependence of the bending strength on the density of the specimens. The density of the boards ranged from ~700 to ~1100 kg/m 3 . On the bases of the experimental data, the relationship between the bending strength and the density of the specimens was determined, the empirical regression equation of which is presented in  Figure 11 shows the dependence of the bending strength on the density of the specimens. The density of the boards ranged from~700 to~1100 kg/m 3 . On the bases of the experimental data, the relationship between the bending strength and the density of the specimens was determined, the empirical regression equation of which is presented in (Table 3, Equation (6)). We can see that the relationship between the bending strength and the density of the boards is very strong and equal to 0.944. The analysis of the experimental data showed that the variation in the bending strength depends on the density of the boards by 89.1%.
In addition, based on the experimental data, an attempt was made to evaluate the relationship between the density of the material and the compression level. The analysis of the experimental data showed (see Figure 12 and Table 3, Equation (7)) that the variation of the density of hemp shive boards is 97.0%, dependent on the level of compression.  (Table 3, Equation (6)). We can see that the relationship between the bending strength and the density of the boards is very strong and equal to 0.944. The analysis of the experimental data showed that the variation in the bending strength depends on the density of the boards by 89.1%. In addition, based on the experimental data, an attempt was made to evaluate the relationship between the density of the material and the compression level. The analysis of the experimental data showed ((see Figure 12 and Table 3, Equation (7)) that the variation of the density of hemp shive boards is 97.0%, dependent on the level of compression.  In the initial research, we found that the appropriate amount of water is 7.5-15% by the dry mass of shives. The exact amount of water depends on the curing mode and the perforation of the curing plates. For the first tests, the smallest amount of water was used: 7.5%. When changing the curing mode, it was not possible to form the specimens with 7.5% water-the surface of the specimens is formed well, but the specimens are characterized by high brittleness, low bending strength and the powder of unreacted binder is visible on the surface of the boards (Figure 13a). The boards formed with 15% water were free of visible unreacted binder powder but voids formed in the middle of the specimen, resulting in no or weak contact zones in the middle of the board (Figure 13d). With a water content of 10 and 12.5% (Figure 13b (Table 3, Equation (6)). We can see that the relationship between the bending strength and the density of the boards is very strong and equal to 0.944. The analysis of the experimental data showed that the variation in the bending strength depends on the density of the boards by 89.1%. In addition, based on the experimental data, an attempt was made to evaluate the relationship between the density of the material and the compression level. The analysis of the experimental data showed ((see Figure 12 and Table 3, Equation (7)) that the variation of the density of hemp shive boards is 97.0%, dependent on the level of compression.  In the initial research, we found that the appropriate amount of water is 7.5-15% by the dry mass of shives. The exact amount of water depends on the curing mode and the perforation of the curing plates. For the first tests, the smallest amount of water was used: 7.5%. When changing the curing mode, it was not possible to form the specimens with 7.5% water-the surface of the specimens is formed well, but the specimens are characterized by high brittleness, low bending strength and the powder of unreacted binder is visible on the surface of the boards (Figure 13a). The boards formed with 15% water were free of visible unreacted binder powder but voids formed in the middle of the specimen, resulting in no or weak contact zones in the middle of the board (Figure 13d). With a water content of 10 and 12.5% (Figure 13b,c) on the surface of the boards, there were no spots of unsolidified starch powder or layers in the middle of the board. In the initial research, we found that the appropriate amount of water is 7.5-15% by the dry mass of shives. The exact amount of water depends on the curing mode and the perforation of the curing plates. For the first tests, the smallest amount of water was used: 7.5%. When changing the curing mode, it was not possible to form the specimens with 7.5% water-the surface of the specimens is formed well, but the specimens are characterized by high brittleness, low bending strength and the powder of unreacted binder is visible on the surface of the boards (Figure 13a). The boards formed with 15% water were free of visible unreacted binder powder but voids formed in the middle of the specimen, resulting in no or weak contact zones in the middle of the board (Figure 13d). With a water content of 10 and 12.5% (Figure 13b Figure 14 shows the dependence of the bending strength of the boards of treated and untreated shives on different water content, which varied from ~7.5 to ~15.0%. On the basis of the experimental data, a relationship between the bending strength and different water content was determined, the empirical regression equation of which is presented (see Table 3, Equations (8) and (9)). Experimental studies of untreated shives have shown that the variation in bending strength is 96.8% dependent on different water content. Meanwhile, the experimental studies of the treated shives showed that 92.8% of the variation in bending strength depends on the different water content.   Figure 14 shows the dependence of the bending strength of the boards of treated and untreated shives on different water content, which varied from~7.5 to~15.0%. On the basis of the experimental data, a relationship between the bending strength and different water content was determined, the empirical regression equation of which is presented (see Table 3, Equations (8) and (9)). Experimental studies of untreated shives have shown that the variation in bending strength is 96.8% dependent on different water content. Meanwhile, the experimental studies of the treated shives showed that 92.8% of the variation in bending strength depends on the different water content.  Figure 14 shows the dependence of the bending strength of the boards of treated and untreated shives on different water content, which varied from ~7.5 to ~15.0%. On the basis of the experimental data, a relationship between the bending strength and different water content was determined, the empirical regression equation of which is presented (see Table 3, Equations (8) and (9)). Experimental studies of untreated shives have shown that the variation in bending strength is 96.8% dependent on different water content. Meanwhile, the experimental studies of the treated shives showed that 92.8% of the variation in bending strength depends on the different water content.  With 10% water, the bending strength results were higher, so we chose 10% water for further tests. It is likely that with 12.5% water, the internal decay of the board structure begins.
The experimental studies conducted showed that the short-term water absorption values of the hemp shives boards can be described according to Equation (10) (see Figure 15 and Table 3). According to the experimental values obtained, the correlation coefficient is equal to 0.994. This shows that the relationship between short-term water absorption and the amount of hydrophobizing additive is very strong. The coefficient of determination is equal to 0.989, and it can be concluded that the amount of hydrophobizing additive has even 98.9% short-term water absorption. In our research, boards were prepared without hydrophobizer and with different amounts of hydrophobizer, which varied from 15 to 117 g per 1 kg of shives in the mixture. Research results show that even small amounts of hydrophobizer have a significant impact on the water absorption of boards. With 15 g of hydrophobizer, the absorption decreases more than four times. As the amount of hydrophobizer is further increased, there is a slow decrease in the absorbency. In our case, the 117 g amount of hydrophobizer made it possible to reduce the absorption to 3%. With 10% water, the bending strength results were higher, so we chose 10% water for further tests. It is likely that with 12.5% water, the internal decay of the board structure begins.
The experimental studies conducted showed that the short-term water absorption values of the hemp shives boards can be described according to Equation (10) (see Figure 15 and Table 3). According to the experimental values obtained, the correlation coefficient is equal to 0.994. This shows that the relationship between short-term water absorption and the amount of hydrophobizing additive is very strong. The coefficient of determination is equal to 0.989, and it can be concluded that the amount of hydrophobizing additive has even 98.9% short-term water absorption. In our research, boards were prepared without hydrophobizer and with different amounts of hydrophobizer, which varied from 15 to 117 g per 1 kg of shives in the mixture. Research results show that even small amounts of hydrophobizer have a significant impact on the water absorption of boards. With 15 g of hydrophobizer, the absorption decreases more than four times. As the amount of hydrophobizer is further increased, there is a slow decrease in the absorbency. In our case, the 117 g amount of hydrophobizer made it possible to reduce the absorption to 3%. The experimental flammability studies conducted showed that the fire spread time values of hemp shive boards can be described according to Equation (11) (see Figure 16 and Table 3).  The experimental flammability studies conducted showed that the fire spread time values of hemp shive boards can be described according to Equation (11) (see Figure 16 and Table 3). With 10% water, the bending strength results were higher, so we chose 10% water for further tests. It is likely that with 12.5% water, the internal decay of the board structure begins.
The experimental studies conducted showed that the short-term water absorption values of the hemp shives boards can be described according to Equation (10) (see Figure 15 and Table 3). According to the experimental values obtained, the correlation coefficient is equal to 0.994. This shows that the relationship between short-term water absorption and the amount of hydrophobizing additive is very strong. The coefficient of determination is equal to 0.989, and it can be concluded that the amount of hydrophobizing additive has even 98.9% short-term water absorption. In our research, boards were prepared without hydrophobizer and with different amounts of hydrophobizer, which varied from 15 to 117 g per 1 kg of shives in the mixture. Research results show that even small amounts of hydrophobizer have a significant impact on the water absorption of boards. With 15 g of hydrophobizer, the absorption decreases more than four times. As the amount of hydrophobizer is further increased, there is a slow decrease in the absorbency. In our case, the 117 g amount of hydrophobizer made it possible to reduce the absorption to 3%. The experimental flammability studies conducted showed that the fire spread time values of hemp shive boards can be described according to Equation (11) (see Figure 16 and Table 3).  According to the experimental values obtained, the correlation coefficient is equal to 0.995. This shows that the relationship between fire propagation time and the amount of flame-retardant additive is very strong. The coefficient of determination is equal to 0.991, and it can be concluded that the amount of flame-retardant additive even affects the speed of fire spread by 99.1%. The amount of flame retardant was calculated per 1 kg of shives. In our studies, the amount of flame retardant in the mixture varied from 10 to 60 g per 1 kg of shives. A small amount of flame retardant made it possible to significantly reduce the spread of flame: 10 g of flame retardant slowed down the flame spread by about four times, while using a larger amount of flame retardant from 50 to 60 g meant that fire spread did not occur in the boards.
In order to evaluate whether the thermal conductivity coefficient of boards made from differently treated hemp shives differs, a variance analysis was performed. It was checked whether the difference between the averages of the obtained results is significant (see Figure 17). The dispersion analysis showed that the difference between the mean values of the thermal conductivity coefficient is insignificant, and the F criterion statistic was 0.787, p = 0.382. According to the experimental values obtained, the correlation coefficient is equal to 0.995. This shows that the relationship between fire propagation time and the amount of flame-retardant additive is very strong. The coefficient of determination is equal to 0.991, and it can be concluded that the amount of flame-retardant additive even affects the speed of fire spread by 99.1%. The amount of flame retardant was calculated per 1 kg of shives. In our studies, the amount of flame retardant in the mixture varied from 10 to 60 g per 1 kg of shives. A small amount of flame retardant made it possible to significantly reduce the spread of flame: 10 g of flame retardant slowed down the flame spread by about four times, while using a larger amount of flame retardant from 50 to 60 g meant that fire spread did not occur in the boards.
In order to evaluate whether the thermal conductivity coefficient of boards made from differently treated hemp shives differs, a variance analysis was performed. It was checked whether the difference between the averages of the obtained results is significant (see Figure 17). The dispersion analysis showed that the difference between the mean values of the thermal conductivity coefficient is insignificant, and the F criterion statistic was 0.787, p = 0.382. Boards made from treated and untreated hemp shives have no difference in thermal conductivity (see Figure 18). The thermal conductivity coefficients of the boards made from treated and untreated hemp shives were combined and a regression analysis was performed (see Table 3, Equation (12)). Further studies of the experimental data showed that the statistic of the F criterion was 253.9, p = 0. Since the p-value is less than 0.05, it can be assumed that the regression model is significant. According to Table 3 (see Table 3, Equation (9)), the correlation coefficient is equal to 0.949. This shows that the relationship between the thermal conductivity coefficient and the density of the boards is very strong. It can be concluded that the density of the boards affects the thermal conductivity coefficient by as much as 90.1%. Boards made from treated and untreated hemp shives have no difference in thermal conductivity (see Figure 18). The thermal conductivity coefficients of the boards made from treated and untreated hemp shives were combined and a regression analysis was performed (see Table 3, Equation (12)). Further studies of the experimental data showed that the statistic of the F criterion was 253.9, p = 0. Since the p-value is less than 0.05, it can be assumed that the regression model is significant. According to Table 3 (see Table 3, Equation (9)), the correlation coefficient is equal to 0.949. This shows that the relationship between the thermal conductivity coefficient and the density of the boards is very strong. It can be concluded that the density of the boards affects the thermal conductivity coefficient by as much as 90.1%.
In Figure 10 the depicted points can be explained by studies of the boards' macrostructure. Figure 19a presents the contact zone of fibrous hemp shives without the use of starch. Due to the high pressure, the segments are well pressed against each other, but voids are observed at the intersection of the segments. In the places of compression, partial bonding of the sheets occurs because of the activated binding properties of cellulose at high temperature. The existing voids between the boards have all sorts of negative effects on the strength properties of the boards: there remains a large unbound surface area, a large space for the deformations of the boards during loading, and the existing voids do not meet the requirements for surface smoothness. In Figure 10 the depicted points can be explained by studies of the boards' macrostructure. Figure 19a presents the contact zone of fibrous hemp shives without the use of starch. Due to the high pressure, the segments are well pressed against each other, but voids are observed at the intersection of the segments. In the places of compression, partial bonding of the sheets occurs because of the activated binding properties of cellulose at high temperature. The existing voids between the boards have all sorts of negative effects on the strength properties of the boards: there remains a large unbound surface area, a large space for the deformations of the boards during loading, and the existing voids do not meet the requirements for surface smoothness.
After adding 5 or 10% of starch, most of the surface of the boards is bound, there are no longer large voids where deformation would easily occur, and the surface of the board is smoother, i.e., less porous. The starch binder does not completely cover the entire free volume between the slats (Figure 19b,c), but this does not have a major effect on the bending strength of the boards. The existing individual voids prevent the product from easily deforming under loads, so the bending strength changes little.
Further increasing the amount of starch up to 15% completes the filling of individual voids, and the thickness of the starch binder between the layers increases ( Figure 19d).  In Figure 10 the depicted points can be explained by studies of the boards' macrostructure. Figure 19a presents the contact zone of fibrous hemp shives without the use of starch. Due to the high pressure, the segments are well pressed against each other, but voids are observed at the intersection of the segments. In the places of compression, partial bonding of the sheets occurs because of the activated binding properties of cellulose at high temperature. The existing voids between the boards have all sorts of negative effects on the strength properties of the boards: there remains a large unbound surface area, a large space for the deformations of the boards during loading, and the existing voids do not meet the requirements for surface smoothness.
After adding 5 or 10% of starch, most of the surface of the boards is bound, there are no longer large voids where deformation would easily occur, and the surface of the board is smoother, i.e., less porous. The starch binder does not completely cover the entire free volume between the slats (Figure 19b,c), but this does not have a major effect on the bending strength of the boards. The existing individual voids prevent the product from easily deforming under loads, so the bending strength changes little.
Further increasing the amount of starch up to 15% completes the filling of individual voids, and the thickness of the starch binder between the layers increases ( Figure 19d).  After adding 5 or 10% of starch, most of the surface of the boards is bound, there are no longer large voids where deformation would easily occur, and the surface of the board is smoother, i.e., less porous. The starch binder does not completely cover the entire free volume between the slats (Figure 19b,c), but this does not have a major effect on the bending strength of the boards. The existing individual voids prevent the product from easily deforming under loads, so the bending strength changes little.
Further increasing the amount of starch up to 15% completes the filling of individual voids, and the thickness of the starch binder between the layers increases ( Figure 19d).
The microstructure analysis of the board provides information about the contact zones of starch and hemp shives and their quality ( Figure 20). Figure 20a presents a picture of the structure of a board with a 5% binder. In Figure 20a, voids between the starch binder are observed. It can be assumed that, due to the insufficient amount of starch, the binding material contracted and broke during heat treatment, resulting in the formation of elongated opposing binder zones. A very thin thickness of the binding material is observed at the break points of about 0.1-0.3 μm. Figure 20a presents a picture of the structure of a board with a 5% binder. In Figure 20a, voids between the starch binder are observed. It can be assumed that, due to the insufficient amount of starch, the binding material contracted and broke during heat treatment, resulting in the formation of elongated opposing binder zones. A very thin thickness of the binding material is observed at the break points of about 0.1-0.3 µm.
By further increasing the starch content to 10%, continuous starch films are formed in individual contact zones. From Figure 20b, it can be seen that the thickness of the film is about 1.5 µm. Figure 20c presents an image of the structure of a board with a 15% binder. The figure shows that the film thickness is different in individual zones, but it is obviously thicker than with 10% starch: 2.5-3.5 µm.
Meanwhile, Figure 20d shows that with 20% starch, a 4.5-5.5 µm-thickness film is formed in the structure of the board.
The analysis of the results shows that the use of 15% starch provides the required bending strength, while the structure analysis shows that the 15% starch content in the mixture ensures the formation of reliable contact zones.
The works of other authors show that the performance properties of the boards created with starch binding material, especially the bending strength, can differ several times [17][18][19][20][32][33][34]. This is largely due to the density of the panels. In different research works [18,34,35], the density of boards usually ranges from 300 to 800 kg/m 3 . In our work, it was established that when the density is increased by 1.5 times, the strength increases by about 50%. In this case, when the board density is 350 kg/m 3 , the bending strength should be about 10 MPa. In this way, it can be stated that other factors, such as water content, starch content, and starch-sodium metasilicate ratio-analyzed in our work-can lead to an increase in the bending strength of the panels by approximately two times. Similar results are obtained when modified starch is used [32][33][34]. In this case, the bending strength is significantly higher.

Conclusions
The chemical treatment of hemp shives allowed not only the reduction of the density and thermal conductivity of hemp shives but also the increase of the adhesive interaction between the binder and hemp shives.
Proper selection of the starch-sodium metasilicate ratio allows approximately double the adhesion of the binder. The best starch-sodium metasilicate ratio was found to be 0.6.
It was established that the selection of a rational amount of starch is related not only to the bending strength, but also to the formation of the macrostructure. By choosing a starch content of 15%, the highest bending strength is achieved, and further increasing the starch content does not result in bending strength results. When 15% starch is used, the gaps between the sheets are filled with a binding material, so the surface of the boards is smooth.
The pressing selected during the board production process affects not only the density of the boards but also the strength indicators. If the pressing is increased from 5 t/m 2 to 15 t/m 2 , the density of the boards increases by about 1.5 times and the bending strength is more than 50%.
The selection of the amount of water allows you to create a board without external and internal defects and additionally increase the strength during bending. A change in the water content of 2.5% increases or decreases the bending strength by approximately 40%. The most suitable amount of water has been determined to be 10%.
It has been found that the use of additives can reduce the water absorption several times and obtain non-flammable boards.
In the future, additional research should be carried out and the coating of boards with additives should be done from the outside of already-finished boards. In this way, the additives would be less negatively affected by the high pressing temperature, and it would be possible to reduce the yield of the additives. In addition, the mixture obtained could be tested by printing with a 3D printer, which should help to obtain an even more continuous structure of the product and better operational properties.

Conflicts of Interest:
The authors declare no conflict of interest.