Fast-Curing 3-Layer Particleboards with Lignosulfonate and pMDI Adhesives

Abstract

Currently, the industrial success of bio-based adhesives remains limited, despite the growing interest in these compounds. One example is the use of lignosulfonates (LS), a byproduct from the pulp and paper industry, which requires long pressing times to ensure proper performance for wood-based panel production. This study successfully manufactured particleboards using a low press factor of 7.5 s/mm, commonly used for conventional urea-formaldehyde resins on a lab scale. To the best of our knowledge, lignin-based particleboards have never been reported using such low press factors. Thus, 3-layer boards were manufactured in which the core layer was bonded with polymeric isocyanate (pMDI), and the surface layers were bonded with LS. Propylene carbonate (PC) was used as a solvent for pMDI to improve adhesive distribution. The optimum amounts of adhesive were determined using response surface methodology: 1.3% pMDI with 2.2% PC in the core layer and 15% LS in the surface layers. These boards obeyed the requirements of standard EN 312 for general-purpose boards for use in dry conditions (type P1). Their formaldehyde content, determined through the perforator method, was equal to that of the wood mix at the maximum value set by IKEA for class E0.5.

1. Introduction

Technical lignins are the main byproduct of the wood pulping industry, with 100 million tons being produced annually [1,2]. However, despite the abundance of this natural phenolic compound, only 1 to 2% are used for the production of value-added products [3,4].

The kraft pulping process currently is the leading pulping process worldwide, being responsible for almost 90% of the pulp production. The production of kraft lignin, derived from this process, is also on the rise, having increased by 150% between 2014 and 2018. Purification processes for this lignin, such as LignoBoost, have also been implemented industrially with success. Still, lignosulfonates (LS), the technical lignin derived from sulfite pulping, accounted for about 70% of the total commercial market of lignin in 2023. The market size of this lignin was also estimated at $659.9 million in 2022 and is and expected to grow to $886.4 million in 2030 [1,2,5,6].

Since the start of wood pulping, it has been suggested to use lignin waste for the preparation of adhesives, particularly wood adhesives [7,8]. Extensive research has been conducted with this purpose, driven by the current industrial interest in non-toxic, eco-friendly natural adhesives [8,9]. Unfortunately, most of these lignin-based adhesives have either never reached industrial application or were discontinued after a short period of time, due to technical limitations or economic reasons [8,10].

For example, when spent sulfite liquor (SSL), composed mainly of LS, was tested as a wood adhesive, high pressing temperatures (190 to 210 °C) and high press factors (60 to 40 s/mm) were necessary to ensure proper bonding. Combinations with strong mineral acids were proposed to decrease these values. However, this solution never reached industrial success, as the performance and aging resistance of the resultant particleboards were still not comparable to that of standard phenol-formaldehyde boards. On the other hand, hydrogen peroxide has also been used in combination with SSL for particleboard manufacture but, once again, with limited success, as it resulted in machinery corrosion [9,10,11].

Antov and co-workers applied unmodified magnesium LS as a binder for industrial waste fibers. Composites were produced using a 15% content of magnesium LS (based on the dry fibers), a high pressing factor of 60 s/mm, and a pressing temperature of 210 °C [11]. It must be noted that in lab-scale particleboard production, conventional urea-formaldehyde (UF) adhesives, press factors of 5.6 to 7.5 s/mm, and temperatures of 190 °C are typically used [12,13]. The obtained modulus of rupture and elasticity (MOR and MOE) values obeyed the minimum required for particleboards (PBs) for interior fitments for use in dry conditions (type P2) and standard grade medium-density fiberboard (MDF, dry process fiberboard) boards. However, these boards displayed low water resistance [11].

In a similar approach, Ferreira and co-workers confirmed the adhesive properties of spent sulfite liquor. The final boards had an LS content of 20% and were pressed at 190 °C, considering a press factor of 40 s/mm. The internal bond (IB) values of these boards complied with the requirements of standard EN 312 [14] for general-purpose boards for use in dry conditions (type P1) [9]. In order to improve the performance of the panels, wheat flour was added to the adhesive system. Nevertheless, these boards required a minimum press factor of 32 s/mm. The authors concluded that lower pressing times, as well as higher water resistance, were still needed in order to make these PBs sufficiently interesting for industrial use [15].

To overcome these challenges, studies have proposed that the use of LS as the main adhesive component should be accompanied by a suitable crosslinker. This approach could compensate for the relatively low reactivity of LS. Therefore, systems that combine lignin with other compounds, such as polyethylenimine, furfuryl alcohol, polymeric isocyanate, chitosan, tannins, glyoxal, wheat flour, poly(vinyl alcohol), and hexamine, have been proposed [8,9,15,16,17,18,19,20]. Another option is the adjustment of the production parameters of the boards, such as the press factor and the pressing temperature [9,11,15].

Hemmilä and co-workers also highlighted the need for crosslinker addition when LS is used without modification in PB production. Both bio-based furfuryl alcohol (FOH) and polymeric isocyanate (pMDI) were studied as crosslinkers for ammonium LS. The boards were pressed at 190 °C and a press factor of 22.5 s/mm, with a crosslinker load on dry fibers of 2%. A significant decrease in curing temperature with crosslinker addition was confirmed. The best results were obtained with pMDI. The formaldehyde emissions of these particleboards were at the level of natural wood, while FOH-crosslinked boards emitted beyond this level. The authors concluded that, until higher-performance crosslinkers are found, pMDI is a suitable option [17].

Mansouri and co-workers tested an adhesive composed of glyoxalated calcium-LS and pMDI for producing particleboards. Glyoxal is considered essentially non-volatile when in aqueous solution and has relatively low toxicity when compared to other aldehydes (median lethal dose (LD50) rat > 2960 mg/kg), while for formaldehyde: LD50 rat > 800 mg/kg) [8,21,22]. The optimum operating conditions were as follows: a resin load on dry fibers of 8% (pMDI content of 3%) and a press factor of 28.1 s/mm at a temperature of 195 °C. These PBs yielded good internal bond strength (IB) that comfortably passed those required in international standard specifications for exterior-grade panels. Their performance also matched that of PBs manufactured with formaldehyde-based commercial adhesives [8].

It should be noted that diisocyanates alone, namely pMDI, are already frequently used to produce exterior-grade particleboards. These adhesives offer several advantages such as the following: high bonding efficiency, which allows for low adhesive content; the absence of formaldehyde; and compatibility with urea-formaldehyde resins [23]. However, their application carries some disadvantages: a tendency to stick to the press plates (which may be minimized by using a different adhesive on the board’s outer layers or using release agents), toxicity, and high cost [24]. The toxicity specifically is taken into account during PB production, as repeated overexposure to diisocyanates may cause, for example, allergic sensitization with asthma-type symptoms [25]. Therefore, reducing or partially substituting these adhesives remains a current challenge [26].

Ostendorf and co-workers successfully reduced the amount of pMDI necessary for wood fiber insulation boards by 50% by using propylene carbonate (PC) as a solvent and adding 1% of softwood kraft lignin. The use of PC alone improved the performance of the obtained panels regarding their IB, water absorption, and compressive strength. The authors associated this improvement with an enhanced adhesive distribution on the wood fibers’ surface. This hypothesis was supported by elemental analysis. PC also permitted the addition of dry lignin to pMDI without premature reaction. Depending on the type of lignin and production process, the addition also contributed to an increase in the performance of the boards [26].

This pMDI/PC adhesive system was also previously reported by other authors. Gaul et al. investigated the role of PC in the adhesive system. It was concluded that PC may in fact improve the binder distribution on the wood chip, as well as increase the penetration of the adhesive on the wood chips, making its hydroxyl sites more accessible. This explanation is consistent with that of Ostendorf et al. However, the authors also suspect that PC reacts with the hydroxyl groups of cellulose, lignin, and hemicellulose, once again making them more accessible [23]. This reaction, oxyalkylation, has been well documented in other studies for lignin molecules. However, for temperatures of 170 to 180 °C, reaction times of about two to three hours are reported [27,28,29,30].

It should be noted that PC is registered in The Safer Chemistry Program by the United States Environmental Protection Agency as a solvent with a low level of hazard concerning human health and environmental safety [15]. It is also classified as a green solvent [31]. Unlike pMDI, PC displays low acute and chronic toxicity and is not known to be a mutagen or carcinogen [32]. This compound is also biodegradable and used by the cosmetic industry [33].

In our preliminary trials, 3-layer particleboards were produced using pMDI as a crosslinker for spent sulfite liquor in the core layer. In order to study the core layer alone, the proper bonding of the surface layers was ensured with a UF resin. However, the performance of these boards was inferior to that of the control groups manufactured using pMDI alone. In another trial, PC was added to improve pMDI’s distribution on the wood surface, whilst spray-dried LS was added as an additive. However, once again, the addition of LS did not improve the performance of the boards. These results are consistent with those obtained by Ostendorf and co-workers when kraft lignin (Indulin AT) and canola hull were used as additives [26]. It was concluded that pMDI was not an effective crosslinker in both cases. Consequently, an alternative approach is described in this work.

This study proposes a system for a 3-layer particleboard, where LS and pMDI are used as adhesives for the surface and core layers, respectively. As the temperature on the outer surface of the mat increases more rapidly, the cure of the LS is expected to be facilitated. On the other hand, pMDI in the core is quite reactive, even at a lower temperature, thus ensuring good cohesion in all layers, using an expectably relatively low pressing factor. Simultaneously, problems related to the adhesion of the boards to the press plates would be minimized by not using pMDI on the outer layers. PC is used in the core layer so as to allow for minimizing the pMDI content.

2. Materials and Methods

2.1. Materials

Thick spent sulfite liquor (HLS) from the acidic magnesium-based sulfite pulping process of Eucalyptus globulus (hardwood) was supplied by Caima-Indústria de Celulose SA (Constância, Portugal). This sample was used as is without purification. This sample was composed of 33% lignosulfonates, 9% sugars, and 13.8% ashes [34].

1,2-Propylene carbonate with a 99% purity was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Polymeric 4,4′-methylene-diphenylene diisocyanate (NCO-content 30.5%–32.5%) (DESMODUR 1520 A20) was supplied by Covestro (Leverkusen, Deutschland).

Wood particles were provided by Sonae Arauco Portugal, SA (Oliveira do Hospital, Portugal). These comprised a mixture of 30% maritime pine (Pinus pinaster Ait.), 15% eucalypt (Eucalyptus globulus), 25% pine sawdust, and 30% recycled wood.

2.2. Methods

2.2.1. Experimental Planning and Statistical Analysis

Design of experiments (DOE) is a mathematical tool used to acquire insights on how process and product parameters influence response variables such as the physical properties of a product or its performance. Interactions between different process and product variables can also be studied with this approach [35].

Response surface methodology (RSM) comprises various mathematical and statistical techniques for modeling and analyzing problems with the aim of optimizing a chosen response that is influenced by several variables [36].

In order to optimize the performance of the final boards, three factors were studied: the content of pMDI and PC in the core layer and the content of LS in the surface layers. Although Ostendorf et al. successfully applied PC in order to dilute pMDI in equal amounts, the authors pointed out that the load of PC needed further optimization [26]. On the other hand, 4% of pMDI based on dry fiber weight has been reported to be suitable for producing P3 particleboards (EN 312) [17,37]. Regarding the LS content, Hemmilä et al. suggested a load of 12% on dry fiber weight for particleboard production [17]. As this study aims to minimize the pMDI content, lower pMDI contents were attempted. The values for the studied factors and levels are shown in

Table 1. Levels of the studied factors.

Factors	Levels
	−1	0	1
% pMDI (core layer)	1	2	3
% PC (core layer)	1	2	3
% LS (surface layers)	5	10	15

.

For fitting the response surface, a 2³ central composite design (CCD) was used. Two central points were considered, resulting in 16 experiments. Taking Table 1 into account, Minitab 19.1 software was used to generate the matrix of experiments, as well as to analyze the subsequent results [36]. The CCD matrix is shown in

Table 2. Central composite design matrix.

Test	% pMDI (Core Layer)	% PC (Core Layer)	% LS (Surface Layers)
1	1	1	15
2	1	1	5
3	2	2	5
4	2	2	15
5	2	2	10
6	2	2	10
7	3	1	5
8	1	3	5
9	3	3	5
10	3	1	15
11	1	3	15
12	3	3	15
13	1	2	10
14	3	2	10
15	2	1	10
16	2	3	10

.

In order to find a suitable mathematical model for the relationship between the independent variables and responses, a second-order model was applied:

$$y={\beta }_0+\sum _{i=1}^k{\beta }_i·x_i+\sum _{i<j}^k{\beta }_{ii}·{x_i}^2+\sum _{i<j}\sum {\beta }_{ij}·x_i·x_j+\in$$

(1)

where $y$ is the predicted response, $x_i$ and $x_j$ are the variables that influence the response, ${\beta }_0$ is a constant, ${\beta }_i$ are the linear coefficients, ${\beta }_{ii}$ are the squared coefficients, ${\beta }_{ij}$ are the cross-product coefficients, and $\in$ is the noise or error of the response [36].

In this step, non-relevant effects were identified through the analysis of variance (ANOVA). An F-test at a significance level of 10% was used. If the corresponding determined p-value of the effects or interactions was higher than 0.10, then a reduced model excluding the identified variables was selected [36,38].

2.2.2. Particleboard Production

PBs were produced by spraying the adhesive formulation onto the wood chips (2.0%–3.5% moisture content) in a laboratory glue blender, model LGB100 (IMAL-PAL Group, San Damaso, Italy). For the surface layers with 15% LS content, the wood chips were previously dried to a moisture content lower than 1%.

For the core layer, the adhesive formulation comprised a pMDI/PC mixture which had been stirred for 15 min [26].

The surface and core layers were separately mixed with the adhesives. Then, a three-layer particle mat was hand-formed into a flexible aluminum container (220 × 220 × 80 mm³). Then, these containers were pressed in a parallel plate hot-press, ITALPRESSE GL/90 (ITALPRESS, Bergamo, Italy), and the pressing program was set in order to simulate a typical PB continuous pressing operation. The particle size distribution of the surface and core layers was determined using an analytical sieve shaker, Haver EML 200 Pure (Haver & Boecker, Oelde, Germany), comprised between 0.25 and 1.0 mm, and 1.0 and 4.0 mm, respectively. The upper, core, and bottom layers accounted for 20, 62, and 18% of the total percentages of board mass, respectively. The boards were pressed at 190 °C, considering a press factor of 7.5 s/mm. The target density and thickness of the laboratory-produced particleboards were 650 kg/m³ and 16 mm, respectively. Five panels were produced for each set of operating conditions.

2.2.3. Particleboard Characterization

After pressing, board samples for the formaldehyde content determination were taken and sealed in hermetic bags. The panels were appropriately stored in a conditioned room for 7 days (20 °C, 65% relative humidity). The following physico-mechanical properties were studied in accordance with their respective European standards: density (EN 323), moisture content (EN 322), tensile strength perpendicular to the plane of the board, also known as internal bond strength (IB) (EN 319), surface soundness (SS) (EN 311), thickness swelling (TS) (EN 317), and formaldehyde content (perforator method) (EN ISO 12460-5) [39,40,41,42,43,44]. The latter was determined 24 h after the production of the boards. Minor alterations to the perforator standard were also carried out; as for the calibration procedure, a previously standardized Voluette Formaldehyde analytical standard solution of 4000 mg/L (HACH, Cleveland, UK) was purchased. Therefore, the standardization of the initial formaldehyde solution through iodometric titration was not carried out, as it was not required.

As mentioned previously, five panels were manufactured for each operating condition. For each of these boards, two samples were analyzed for each of the mechanical trials.

Due to limitations related to the small dimensions of the lab scale PBs, the bending strength and the modulus of elasticity were not analyzed (EN 310) [45].

3. Results

Response surface methodology was employed to optimize the solvent and adhesive content of the particleboards. The analyzed product properties were internal bond (IB), surface soundness (SS), and thickness swelling (TS). For some boards, formaldehyde content was also analyzed. The design matrix as well as the obtained results are shown in

Table 3. Design matrix and results obtained. For the highlighted cases, failure took place within the core layer of the board samples; in the rest of the cases, failure occurred within the surface layers.

Test	% pMDI (Core Layer)	% PC (Core Layer)	% LS (Surface Layers)	IB/ N/mm2	SS/ N/mm2	TS/ %	Formaldehyde Content/ mg/100 g Oven-Dry Board
1	1	1	15	0.32	0.44	45.2	3.0
2	1	1	5	0.22	0.22	48.1	3.1
3	2	2	5	0.18	0.21	42.9	3.3
4	2	2	15	0.53	0.56	39.3	3.0
5	2	2	10	0.43	0.35	37.8	3.1
6	2	2	10	0.41	0.30	43.6	2.9
7	3	1	5	0.25	0.24	42.0	-
8	1	3	5	0.23	0.24	49.1	-
9	3	3	5	0.22	0.26	39.9	3.9
10	3	1	15	0.65	0.57	38.2	-
11	1	3	15	0.36	0.51	40.2	-
12	3	3	15	0.59	0.67	37.0	-
13	1	2	10	0.43	0.46	43.5	-
14	3	2	10	0.43	0.45	37.7	-
15	2	1	10	0.44	0.43	45.0	-
16	2	3	10	0.44	0.44	43.7	-

.

After the application of RSM, regression Equations (2)–(4) were obtained, relating the responses to the test variables. As previously mentioned, insignificant effects were identified using ANOVA and excluded from the model. However, in Equation (4), the linear term for PC content was maintained to preserve hierarchy.

$$\begin{gathered} IB=0.4300\left(\pm 0.0156\right)+0.0580\left(\pm 0.0121\right)·pMDI+0.1350\left(\pm 0.0121\right)·LS-0.0750\left(\pm 0.0198\right)·{LS}^2 \\ +0.0675\left(\pm 0.0135\right)·pMDI·LS \end{gathered}$$

(2)

$$SS=0.3969\left(\pm 0.0152\right)+0.1580\left(\pm 0.0192\right)·LS$$

(3)

$$TS=40.800\left(\pm 0.768\right)-3.130\left(\pm 0.572\right)·pMDI-2.210\left(\pm 0.572\right)·LS-0.860\left(\pm 0.572\right)·PC+2.040\left(\pm 0.934\right)·{PC}^2$$

(4)

where $LS$ is the content of lignosulfonates in the surface layers, and $pMDI$ and $PC$ are the content of pMDI and PC in the core layer, respectively.

3.1. Internal Bond Strength

According to ANOVA, Equation (2) was suitable for predicting the IB response (p-value < 0.10 and R² = 0.94). This model includes the factors pMDI and LS content, the interaction between these two factors, as well as a quadratic term for the LS content. Consequently, it was not possible to conclude with a 90% confidence level that the PC content influences the IB response.

This model is easily explained through the failure mode of the IB samples, shown in Figure 1. For most samples, failure occurred within the surface layers. However, for the samples of boards 1, 11, and, in some cases, 10, highlighted in Table 3, the failure took place within the core layer. These results are explained by the superior adhesion properties of pMDI in comparison to those of lignosulfonates. Thus, failure occurred at the surface layers in most cases, even when only 1% pMDI was used in the core layer. Failure within the core layer was only observed in some of the boards that had the highest LS content, 15%, in the surface layers. For boards with high PC content, the improved distribution of pMDI appears to enhance the adhesion within the core layer. The combined effect of pMDI and PC provides very effective bonding within the core layer, where they are used, even for relatively low contents. When pMDI and PC contents are both above 1%, failure is always observed within the weaker surface layers, where binding is provided by LS alone. The high reactivity and crosslinking efficiency of pMDI is well known [46]. Thus, only a small amount of pMDI/PC is necessary to make the mechanical performance of the core layer surpass that of the surface layers. In fact, under these manufacturing conditions, our group found that about 4% commercial urea-formaldehyde resin content is required in the core layer to match the IB of the 1% pMDI boards [13].

These conclusions are made especially clear when analyzing the influence of the LS content alone on the IB of the boards, as shown in Figure 2. Thus, it is verified that there is a clear linear relationship between these two variables when two outliers, boards 1 and 11, are excluded. As mentioned previously, these are precisely the only cases where the failure mode occurred within the core layer in all samples. A polynomial fit was also applied to the complete data set; however, it was deemed inadequate (R² < 0.95).

In order to comply with the requirements for particleboard type P2 (EN 312), an IB of at least 0.35 N/mm² must be obtained. Therefore, the response surface was analyzed so as to select the minimum amount of pMDI necessary to fulfill these requirements. The contour map for the IB is shown in Figure 3.

Through the analysis of Figure 3, it is concluded that a minimum of 1.2% of pMDI content is needed to safely ensure that the requirements for P2 boards are attained (IB = 0.4 N/mm²). Under these conditions, a corresponding LS content of 12.5% must be selected. This optimum value is shown as a black dot in Figure 3.

3.2. Surface Soundness

As seen in Equation (3), a linear model, considering only the content of lignosulfonates, is sufficient to describe the surface soundness response (p-value < 0.10 and R² = 0.83). The optimum LS content for maximizing the surface soundness of the PBs therefore corresponds to the highest tested value of 15%.

These results are consistent with the previous conclusions, where the IB of the boards increased linearly with LS content when the failure mode occurred within the surface layers.

Consequently, the pMDI content or the PC content does not seem to significantly influence this response. However, when analyzing the results corresponding to the boards with 15% LS content (Table 3), one is led to a different conclusion. Indeed, when comparing boards 1 with 10, and 11 with 12, we see that the increase in pMDI content resulted in a significant increase in SS ($p<0.10$). The same conclusion can be taken for PC content when comparing the results of boards 10 with 12 and 1 with 11 ($p<0.10$). This is made clear by analyzing the failure mode of the samples for the surface soundness tests, shown in Figure 4. For the 5 and 10% LS samples, the failure mode occurred within the surface layer and between the surface and core layers, respectively. On the other hand, for the 15% LS samples, the failure occurred within the inner core layer, bonded with pMDI.

It should be noted that the SS values remained under the specifications for particleboard type P2 (EN 312) (0.8 N/mm²) for all board formulations. Therefore, the manufactured boards only obeyed the specifications for P1 PBs (EN 312).

Nevertheless, when standard urea-formaldehyde P2 particleboards, with 7.5% resin load on dry fibers in the surface layers, were manufactured under the same lab-scale process, SS values of 0.60 ± 0.15 N/mm² were obtained. Consequently, it was admitted that the SS specifications for P2 PBs were not met due to the limitations of the lab-scale process.

3.3. Thickness Swelling

In the case of the TS response, Equation (4) shows a linear relation with pMDI, PC, and LS content, alongside a quadratic term for PC content (p-value < 0.10 and R² = 0.83). Therefore, this seems to be the only property significantly influenced by PC content. Taking Equation (4) into account, an optimum LS content of 15% was once again selected in order to minimize TS.

It must be noted that after 24 h immersion, the surface layers of several board samples partially disintegrated, thus limiting the accuracy of TS measurements. This was especially true for the 5% LS boards. This behavior was attributed to the hydrophilic nature of the lignosulfonates. Low water resistance is a common issue in boards bound with LS [11,15].

The contour map for TS as a function of PC and pMDI content, shown in Figure 5, was analyzed in order to select the optimum amount of PC in the core layer.

It is concluded that, for the previously selected pMDI content of 1.2%, a PC content of 2.2% should be chosen to minimize TS. However, this response could still be optimized further, at the expense of a higher pMDI content.

3.4. Formaldehyde Content

Due to the major concerns regarding indoor air quality, the emission of carcinogenic formaldehyde from wood-based panels must be minimized. Although no-added-formaldehyde adhesives were used in the production of these boards, this compound can still be derived from the wood mix itself. Hence, its quantification is still relevant [47,48].

Table 4. Formaldehyde content of the wood mix and the obtained PBs.

Sample	Core Layer Wood Mix	Surface Layer Wood Mix	LS/pMDI PBs (Average of All Boards)
Formaldehyde content/ mg/100 g oven-dry board	3.3	2.9	3.2 ± 0.3

shows the formaldehyde content measured for the boards and the wood mixes used for the surface and core layers. The formaldehyde content of the boards is equivalent to that of the wood mixes, showing that the adhesives do not release formaldehyde, as expected, and also do not act as scavengers for formaldehyde released by wood. These results are explained by its high content of recycled wood, which includes rejected particleboards or MDF. It should be noted that a formaldehyde content of 0.2 to 0.6 mg/100 g oven-dried board has been reported for natural wood [49]. Nevertheless, the formaldehyde content of all the PBs is lower than the maximum value set by IKEA for the E0.5 class (<4.0 mg/100 g oven-dried board) [48,50].

3.5. Optimum Board

Taking the previous results into account, the optimum adhesive formulation for the particleboards was determined. In order to safely ensure that the requirements for P2 boards were attained, a target value of 0.4 N/mm² was set for IB, whilst SS was maximized, and TS minimized. The response variables were assigned equal preponderance factors. The optimum composition of the boards was determined to be the following: 1.3% pMDI and 2.2% PC in the core layer and 15% LS in the surface layers. These boards were produced and characterized, and their performance was compared with the predicted values. The results are shown in

Table 5. Predicted and experimental properties of the optimum PBs.

Properties	Optimum
	Predicted	Experimental
IB/N/mm2	0.40 ± 0.02	0.41
SS/N/mm2	0.55 ± 0.02	0.52
TS/%	40.7 ± 1.0	45.6

.

As shown in Table 5, the response surface equations are able to accurately predict the IB response. It should also be noted that failure occurred within the core layer for all these boards. The difference between the predicted and experimental values of SS is also not significant. On the other hand, the TS of the optimum boards was not so successfully predicted by the model. This may be partially explained by the inaccuracy introduced by the low water resistance of the surface layers. However, the heterogeneity of the wood mix may have also contributed to these discrepancies.

Despite this low water resistance, these boards complied with the requirements for particleboard type P1 (EN 312). As mentioned previously, this was achieved using pressing factors commonly used for lab-scale PB production with conventional urea-formaldehyde resins. To the best of our knowledge, this was yet to be achieved for PBs that employ LSs as adhesives. Therefore, this study reveals new possibilities for the future industrial application of LSs as adhesives in particleboard manufacture.

4. Conclusions

In this study, particleboards were produced using no-added-formaldehyde adhesives under a press factor commonly used for conventional urea-formaldehyde resins on a lab scale. These boards comprised three layers, in which the core layer was bonded with pMDI, and the surface layers were bonded with lignosulfonates (LS). Propylene carbonate (PC) was also used as a solvent for pMDI in the core layer, which improved the distribution of the adhesive.

In order to optimize the adhesive and additive content, a response surface methodology was employed. The manufactured boards were evaluated regarding their internal bond, thickness swelling, and surface soundness. The optimum boards consisted of 1.3% pMDI, with 2.2% PC in the core layer and 15% LS in the surface layers. The internal bond and surface soundness properties of these boards were successfully predicted with empirical models and complied with the requirements for particleboard type P1 (EN 312).

The formaldehyde content, determined through the perforator method, of the optimum boards was equal to that of the wood mix and at the maximum value set by IKEA for the E0.5 class.

Future studies should focus on improving the thickness swelling of these boards, namely by adding small amounts of paraffin or through the addition of crosslinkers for LS, such as furfural, glyoxal, or polyethylenimine, in the outer layers. The surface soundness may also be increased further with these crosslinkers. This would help minimize the current issues limiting the industrial implementation of LS adhesives in particleboard production.