Unlocking the Main Factors Affecting the Strength and Stability of Yellow Poplar (Liriodendron tulipifera) After a Mild and Environmentally Friendly Densification Process

Abstract

The densification of yellow poplar (Liriodendron tulipifera) has emerged as a critical area of research, driven by its desirable properties and broad potential applications. This study investigated the effects of four densification parameters using a 2⁴ full factorial design to evaluate their impact on physical and mechanical properties. Analysis of variance (ANOVA) and Pareto analyses identified the compression ratio as the most influential factor, significantly affecting bending strength, compression strength, hardness, and spring-back behavior. Pressing temperature was the second most significant factor, with higher levels positively impacting mechanical properties. However, increasing the pre-steaming treatment duration from 30 to 60 min at 130 °C had a detrimental effect on strength and spring-back performance, particularly at a 50% compression ratio. Pressing time showed no significant effect on strength properties but contributed positively to hardness and spring-back behavior at higher levels. Several significant factor interactions were observed, further influencing the outcomes. Differences in density profiles were notable across compression ratios, with higher ratios producing more uniform distributions. Under optimal parameters, densification increased compression strength by 117%, bending strength by 60%, and hardness by 154% compared to undensified control samples, demonstrating the potential of densified yellow poplar (Liriodendron tulipifera) for high-performance applications.

1. Introduction

Wood density is generally directly proportional to its strength, durability, and overall performance. Denser wood exhibits superior load-bearing capacity, making it an excellent choice for structurally demanding applications such as beams, columns, and flooring. Under certain circumstances higher wood density can also contribute to improved dimensional stability, but the relationship depends on various factors. As is well known, dimensional stability refers to a wood’s ability to resist changes in size and shape due to moisture content fluctuations. Densification decreases the wood’s porosity, reducing its capacity to absorb and release moisture; this can improve dimensional stability by minimizing swelling and shrinkage [1,2]. Dense wood products, therefore, will be more reliable and long-lasting, which can lower maintenance costs and minimize the need for replacements over time.

Yellow poplar (Liriodendron tulipifera) is a fast-growing hardwood species native to North America, known for its tall, straight trunk and unique, tulip-shaped leaves. Its fine, uniform grain contributes to excellent workability, making it a favored material for furniture, cabinetry, and millwork. Additionally, its lightweight nature and ease of machining make it accessible and versatile for various applications. However, its relatively low density and hardness limit its utility in more demanding structural or wear-resistant applications.

To address these limitations, densification techniques have been employed to improve yellow poplar (Liriodendron tulipifera)’s performance including thermal, mechanical, and chemical processes [1,2,3].

Chemical densification processes often require harsh chemicals that can be harmful to the environment if not managed properly. These substances can produce hazardous waste, release volatile organic compounds (VOCs), and require additional energy for handling and disposal. Chemicals used in densification can pose risks to workers during processing and to consumers if any residues remain in the finished product. Also, a chemical treatment may alter the appearance, texture, and natural feel of the wood.

In contrast, environmentally friendly methods such as thermo-hydro-mechanical (THM) densification combine compression with heat and steam to increase the wood’s density, hardness, and dimensional stability. This method modifies the wood’s microstructure, collapsing cell walls and reducing porosity, which enhances mechanical properties like modulus of elasticity and rupture [1]. Heat treatment at controlled temperatures further stabilizes the compressed wood, reducing spring-back and improving resistance to environmental moisture. A THM method has a lower ecological footprint, aligns better with sustainability goals, and meets stricter environmental regulations, as seen in many countries increasingly banning or limiting VOC emissions and hazardous substances. This makes green densification methods more favorable for manufacturers aiming to adhere to international environmental standards and certifications like the Forest Stewardship Council (FSC) or Leadership in Energy and Environmental Design (LEED) [4]. Environmentally friendly densification methods tend to require less energy and fewer resources compared to chemical processes, which need additional steps for chemical application, curing, and waste management. This can result in cost savings and greater efficiency for manufacturers, especially as eco-friendly technologies advance [5].

Most of the wood densification using a hydrothermal process has been performed at temperatures higher than 140 °C [6,7,8,9,10]. Yin et al. [6] investigated the impact of high-temperature steam treatment on spruce wood cell walls, focusing on changes in chemical and mechanical properties. The study used FT-IR microscopy and nanoindentation to examine modifications in the cell wall’s biopolymer matrix, specifically in celluloses, hemicelluloses, and lignin, at steam treatment temperatures of 140 °C, 160 °C, and 180 °C. According to their findings, steam treatment caused the degradation of hemicellulose components, with significant changes observed at 160 °C and above. There was a breakdown of carbonyl groups in xylan and glucomannan, and a loss of specific chemical groups in lignin. These modifications decreased the wood’s capacity to absorb moisture, enhancing dimensional stability. Also, higher steam temperatures (160–180 °C) reduced the wood’s hygroscopicity by diminishing the accessible hydroxyl groups, making it less prone to water absorption. The authors concluded that steam treatment at high temperatures can enhance wood’s dimensional stability and reduce hygroscopicity but may also compromise its mechanical strength due to the degradation of biopolymer linkages within the cell wall.

Kutnar and Kamke [7] investigated how varying temperature and steam conditions affect the set recovery (spring-back) of densified poplar wood. The study examined densification treatments using saturated steam, superheated steam, and transient conditions at temperatures of 150 °C, 160 °C, and 170 °C, with post-treatment at 200 °C to improve dimensional stability. The study found that densification under saturated steam at 170 °C minimized set recovery, especially with post-treatment at 200 °C for up to 3 min. The best results showed only 6% set recovery after five wet/dry cycles. Higher treatment temperatures generally increased the wood’s oven-dry density and reduced its equilibrium moisture content, contributing to greater dimensional stability. The highest density was achieved with saturated steam treatment at 170 °C, followed by 200 °C post-treatment. Bending tests indicated that the post-treatment at 200 °C for up to 3 min had no significant effect on these mechanical properties, though repeated wet/dry cycles reduced the Modulus of Rupture (MOR) and Modulus of Elasticity (MOE).

The study by Zhou et al. [9] investigated the surface densification of poplar wood (Populus tomentosa Carr.) through a thermo-hydro-mechanical (THM) process, focusing on the effects of processing parameters such as compressing temperature, closing speed, and compression ratio. Their conclusions indicated that a compressing temperature of 140–160 °C, closing speed of 10 mm/min, and compression ratio of 20% were recommended to achieve effective surface densification on their wood samples.

In another work, Xiang et al. [10] explored how superheated steam and densification improve the properties of poplar wood. The study focused on creating two types of sandwich-densified wood, which differ based on the densified layer’s position, through controlled densification and superheated steam treatment. Treatment at 0.1 and 0.3 MPa improved dimensional stability by reducing moisture absorption and swelling. Although superheated steam darkened the wood and slightly reduced its hardness and MOR, the treatment effectively stabilized the wood against shape recovery. The superheated steam-treated sandwich-densified wood exhibited properties like hardwood flooring materials, showing promise for use in solid wood flooring with improved stability and reduced environmental impact.

Shao et al. [11] studied also the effects of thermo-hydro-mechanical treatments on the physical and mechanical properties of poplar wood. They used temperatures of 180 °C, 200 °C, and 220 °C and they determined that higher treatment temperatures and longer durations degraded mechanical properties due to the thermal decomposition of chemical components.

Rahimi et al. [12] investigated the impact of hydrothermal treatments, specifically steam and hot compressed water (HCW) at temperatures of 100 °C and 140 °C, on the physical properties and drying behavior of yellow poplar (Liriodendron tulipifera) heartwood as an alternative to produce chemical-free wood products with enhanced properties. The findings suggested that hot-compressed water treatment at a moderate temperature holds promise for enhancing the properties of wood, providing insights for sustainable and energy-efficient alternatives in wood processing.

Considering the importance of the topic and the preliminary work performed, the primary goal of this study was to investigate the mechanical properties of hydro-thermo-densified yellow poplar (Liriodendron tulipifera) wood under varying compression ratios, pre-steaming times, pressing temperatures, and pressing durations. All factors were tested at two levels using a 2⁴ full factorial experimental design for a comprehensive analysis. In each of the 16 experimental setups, the density profile, compression strength parallel to the grain, modulus of rupture (MOR), hardness, and spring-back behavior were determined. The main and second-order interaction effects were evaluated using analysis of variance (ANOVA) and Pareto analyses, providing a detailed understanding of each factor’s influence and interactions on the measured properties.

2. Materials and Methods

For the experiments, yellow poplar (Liriodendron tulipifera) boards were sourced from the Northeastern Appalachian region in West Virginia. The boards were kiln-dried at the sawmill to an equilibrium moisture content of approximately 7% and planed to dimensions of 23 mm × 178 mm × 3048 mm (T × W × L). From these boards, 224 defect-free specimens with tangential annual ring orientation were prepared, each measuring 305 mm × 19 mm × 19 mm in the longitudinal (L), tangential (T), and radial (R) directions, respectively.

Prior to densification, the specimens were conditioned in a laboratory environment at a temperature of 23 °C ± 2 °C and relative humidity of 60% ± 3% to ensure a 9% equilibrium moisture content (EMC) across all over the specimens. The moisture content of the samples was measured following the ASTM D4442-20 Method B [13] procedures.

2.1. Densification Method

During the densification process, four factors (A–D) were controlled, and the factor levels used in a factorial design are presented in Table 1. The table contains the factors’ real value, and for the regression analysis we used the factors’ coded values (−1) for the low level, and (+1) for the high level.

The applied compression ratio (CR), which specifies the reduction in specimens’ thickness, was calculated using Equation (1).

(1)$CR=\left[{\displaystyle \frac{\left(t_i-t_f\right)}{t_i}}\right]\times 100 \left[\%\right]$

where t_i is the initial thickness of the specimens, and t_f is the final target thickness.

Sixteen experiments were performed as outlined in Table 2, with fourteen randomly selected specimens used for each densification experiment. Additionally, fourteen pieces with the same dimensions were prepared for the undensified control samples.

The experimental setups according to the four-factor at two levels full factorial experiment design (2⁴) are presented in Table 2. The design was replicated once and all 16 setups were randomly repeated, resulting in two batches of specimens for each setup.

The densification process began with a pre-steaming plastification treatment in an autoclave, using saturated steam at a temperature range of 120–130 °C. Immediately after the steaming, the specimens were placed in a hydraulic hot press. A Carver 38952NE1 hydraulic press was used, with both press plates heated to the specified temperature. As shown in Figure 1, the target thickness of the samples was ensured by placing three metal spacers between the press plates, preventing physical over-compression of the specimens.

After the set pressing time, the specimens were removed from the hot press and cooled down at room temperature.

2.2. Determination and Evaluation of Density and Density Profiles

The density of the specimens was measured before and after the densification process in accordance with the ASTM D2395-17 Test Method B [14] procedures. Mass and volume measurements were conducted at an equilibrium moisture content (EMC) of 9% ± 1% for all specimens to ensure comparability between the initial and final densities. The mass and dimensions (longitudinal, tangential, and radial) of each sample were measured with an analytical balance (±0.001 g) and a vernier caliper (±0.01 mm). The density (${\rho }_M$) in (kg/m³) was calculated using Equation (2):

(2)${\rho }_M={\displaystyle \frac{m_M}{V_M}}$

where m_M is the mass at moisture content M and V_M is the volume of specimens at moisture content M.

The density distribution across the thickness of undensified and densified wood samples was determined using an X-ray densitometer (QMS, Model QDP-01, Quintek Measurement Systems, Inc., Knoxville, TN, USA). The density values were recorded at every 0.02 mm along the thickness.

The following parameters of the density profile were determined (Figure 2): the maximum density in the middle section of the thickness (MD_max), the maximum density in both left and right surface layers (SD_Lmax; SD_Rmax), and the distance of the maximum density peaks from the respective surface (D_L; D_R). Two specimens from each experimental setup were subjected to density profile analysis, along with eleven undensified (control) specimens. For the control specimens, the maximum densities were identified, with MD_max representing the denser latewood density across the thickness. For each specimen, the average maximum density (SD_max.avg.) was calculated using Equation (3):

(3)${SD}_{max.avg}={\displaystyle \frac{{SD}_{Lmax}+{SD}_{Rmax}}{2}}$

The average distance of the maximum density peak (D_avg.) from the respective surfaces was calculated using Equation (4):

(4)$D_{avg.}={\displaystyle \frac{D_L+D_R}{2}}$

Factorial analyses were not performed on the density profile data, since two specimens per experimental setup were insufficient to yield reliable results. Since the compression ratio emerged as the most influential factor for all examined mechanical properties of densified yellow poplar (Liriodendron tulipifera), the specimens were grouped into two categories based on the compression ratio. Specifically, the average maximum densities (SD_max.avg.) and the average distances of the density peaks (D_avg.) from setups #1, #2, #5, #6, #9, #10, #13, and #14 were combined into the 20% compression ratio category. Similarly, data from setups #3, #4, #7, #8, #11, #12, #15, and #16 were combined into the 50% compression ratio category.

2.3. Evaluation of Spring-Back Behavior

After the densification was performed, the thickness of the samples was immediately measured. Spring-back (SB) is defined as the relative difference between the thickness achieved immediately after densification and the target thickness; it was calculated using Equation (5):

(5)$SB \left(\%\right)=\left[{\displaystyle \frac{\left(t_f-t_t\right)}{t_t}}\right]\times 100 \left[\%\right]$

where t_f is the final thickness measured immediately after the pressure release and t_t is the target thickness for the compressed specimens.

2.4. Mechanical Tests

The mechanical tests for control and densified specimens were performed using an Instron universal testing machine with the Bluehill 2 (version 2.6) universal testing software (Instron, Norwood, MA, USA).

2.4.1. Static Bending Test

The bending test was conducted following the ASTM D143-14 standard (Section 8) [15]. The test is represented in Figure 3, using center-point loading at a test rate of 1.3 mm/min. The span value was 254 mm for the densified specimens and 266 mm for the control specimens, ensuring the required minimum span-to-depth ratio of 14. Tests were performed in the densified direction (tangential direction). The failure type was recorded for each specimen, and the modulus of rupture (MOR) in (MPa) was calculated using Equation (6):

(6)$MOR={\displaystyle \frac{3\times P_{max}\times L}{2\times b\times h^2}}$

where P_max is the maximum force (N) at the mid-span, L is the span length (mm), b is the width (mm), and h is the depth (mm).

Eight specimens were used for each experimental setup to determine the static bending strength of densified specimens, along with an equal number of control samples.

2.4.2. Compression Parallel to Grain

Compression parallel to grain was performed according to ASTM D143-14 standard (Section 9). Figure 4 illustrates the setup of the compression tests. The test specimens had a length of 76 mm, and the load was applied at a rate of 0.003 mm/mm of nominal specimen length per minute. For each specimen, the maximum load was recorded. The compressive strength (σ_c) in (MPa) was calculated using Equation (7):

(7)${\sigma }_c={\displaystyle \frac{P_{max}}{A}}$

where P_max is the maximum force (N) and A is the cross-sectional area (mm²).

Eighteen specimens were used for each experimental setup to determine the compressive strength parallel to the grain of densified specimens, along with an equal number of control samples.

2.4.3. Hardness

Hardness was determined following ASTM D143 (Section 13) [15]. This is the standard test method for small clear specimens of timber, using a modified ball test with a ball diameter of 11.3 mm, as shown in Figure 5. The projected area of the ball on the test specimen was 1 cm². Penetrations were performed on the densified (tangential) surface of each specimen. The testing speed was 6 mm per minute, and the load at which the ball penetrated to half its diameter was recorded.

Eighteen specimens were used for each experimental setup to determine the hardness of densified specimens, along with an equal number of control samples.

2.5. Scanning Electron Microscopy (SEM)

Samples were sputter-coated with an ultrathin layer of gold palladium for conductivity using a Denton Desk V sputter, from Denton Vacuum Co., Moorestown, NJ, USA. Then, they were imaged using a JEOL JEM7600F scanning electron microscope (JEOL USA, Inc., Peabody, MA, USA) at 5 KV.

2.6. Statistical Analyses

A 2⁴ full factorial experimental design was applied in this study, combining four factors at two levels, resulting in 16 experimental setups. The design was repeated one time, with additional repetitions within each experimental setup to ensure the reliable results for the analyses.

Factorial analysis of variance (ANOVA) was employed to evaluate the results and identify the significant effects of the factors, while Pareto analysis was used to examine their relative contributions. Both main effects and second-level interactions were investigated. All statistical analyses were performed using TIBCO^® Data Science/Statistica 13 software. ANOVA and the Student’s t-test were conducted with the significance level set at α = 0.05. The response surface methodology (RSM) was used to model mathematically the relationship between the dependent variables (factors) and optimize the response variables’ value.

3. Results and Discussion

3.1. Density and the Density Profile

Table 3 presents the initial and final average density, the density increase, and the average thickness of the specimens across all 16 experimental setups. The density values were calculated using Equation (2).

As explained in Section 2, the samples in the density profile analysis were grouped by compression ratio, while the effects of the other three factors (pre-steaming, press temperature, and pressing time) on the density profile were reflected in the deviations of the results. Figure 6 shows the density profile of (a) two control specimens; (b) two samples from setup #1; (c) two samples from setup #7; and (d) two samples from setup #16.

The presented density profiles of the control samples (Figure 6a) exhibit noticeable differences. Density variations along the cross-section could be observed, which could be due to the density differences between earlywood and latewood. When the 20% compression ratio was applied (Figure 6b), density fluctuations were also observed beyond the densified surface in the middle section. Figure 7a,b show the SEM images of the specimens from setup #6.

In Figure 7a, it can be observed that beyond the densified zone, the cell morphology did not change, suggesting that the density differences between earlywood and latewood caused these fluctuations. In addition, close to the specimens’ surface, a non-densified zone can be observed, which also appears on the density profile. Figure 7b captures the densified zone, where the extent of deformation and the bending of cell walls indicate that the wood exhibited viscoelastic behavior during compression. However, the presence of open cells suggests that the 20% compression ratio did not fully close the cell structure.

When the 50% compression ratio was employed, the M-shaped density gradient became the most dominant. Kutnar et al. [16] observed a similar M-shape in the densification profile of hybrid poplar wood with a 63% degree of densification. At a 50% compression ratio, the density in the middle section became more uniform while achieving an elevated density value.

Figure 8 illustrates the relative differences between the average maximum surface density and both the average initial density and the maximum density in the middle section of the sample. The initial density of the specimens was calculated using Equation (2), and the average maximum surface density (SD_max.avg) was computed using Equation (3).

For the 20% compression ratio, the relative difference between the middle section and surface densities is 37.3%, which is significantly larger compared to the 50% compression ratio, where this difference is only 9.1%. This indicates a more uniform density distribution in the cross-section for the higher compression ratio studied. Additionally, the relative difference from the average initial density to the surface density (SD_max.avg) is 64% for a 20% compression ratio, compared to 102.3% for the 50% compression ratio, reflecting the greater surface density achieved at the higher compression ratio.

Figure 9 shows the average maximum density for samples from both compression ratios and the control group. For the densified samples, the average maximum density represents the average value of the density peak near the surface (SD_max.avg) and was calculated using Equation (3). For the control samples, the maximum density represents the latewood density, which appears as a periodic peak on the density profile (MD_max).

The densification process significantly increased the density of the samples compared to the control ones. The control samples exhibited an average maximum density of 554 kg/m³. With a 20% compression ratio, the average density increased by 37% (i.e., to 761 kg/m³), while at a 50% compression ratio, it rose by 76% (i.e., to 976 kg/m³). The 50% compression ratio resulted in a 28% increase in density compared to the 20% ratio.

The distance between the average maximum density and the specimens’ surface (Figure 2), grouped by compression ratio, is presented in Figure 10.

For the 20% compression ratio, the average distance between the density peak and the surface was 1.15 mm. At the 50% compression ratio, the density peak shifted deeper into the wood, reaching 1.35 mm, representing a 17% increase in distance. However, as presented in Figure 10, there is no significant statistical difference between these two distances.

A similar phenomenon was reported by Rautkari et al. [17], where the densification rate of Scots pine surfaces increased from 16.6% to 31.8%. Additionally, further density peak shifting was observed in specimens subjected to hydrothermal post-treatment. Laine et al. [18] also documented a density peak shift when Scots pine sapwood was compressed at a 25% compression ratio, with varying temperatures and pressing times. They attributed the deeper density peak to the viscoelastic behavior of wood.

Another possible explanation is the degradation of the wood surface and the hydrolysis of hemicelluloses, particularly at higher compression ratios. These factors, combined with the migration of heat and moisture deeper into the wood, likely contribute to the further density peak shifting into the specimen.

3.2. Spring-Back

Immediately after the press was opened, the thickness of each sample was measured to evaluate spring-back. Spring-back provides information about the immediate elastic recovery caused by the shape memory effect of the cells. This memory effect can pose a significant issue when densified wood is exposed to water or high-humidity environments. Figure 11 presents the spring-back results across the 16 experimental setups, categorized by the densification ratio.

The compression ratio influenced the spring-back significantly, leading to differences of up to one order of magnitude in some cases. A higher compression ratio, especially in the case of volume densification rather than surface densification, causes greater set-recovery or spring-back, as reported in several previous studies [19,20,21,22].

To achieve reliable statistical analysis, the original 2⁴ full factorial design was divided into two separate 2³ full factorial designs, excluding the compression ratio. This separation allowed for the independent analysis of the effects of pre-steaming time, press temperature, and pressing time within each compression ratio.

Using the low-level (20%) compression ratio, setups #5, #6, #13, and #14 produced an average spring-back with negative values, indicating that the samples shrank due to the densification process. Across these four setups, only one factor—the pressing temperature—was set to the same level (high level). Among the 20% densified samples, setup #14 achieved the lowest average spring-back of −1.7%, whereas the highest value of 2.1% was observed with the factor parameters of setup #1. Since the target thickness was ensured by solid metal spacers in the press, once the press plates closed and touched the spacers, the samples began to dry out from the sides. This drying could result in a thinner final thickness than the target. Similarly, Balasso et al. [20] reported this phenomenon during the densification of Eucalyptus nitens, which resulted in an average spring-back value of −4.52%.

With a high-level (50%) compression ratio, setup #15 produced the lowest average spring-back value of 5.3%, which was achieved with a low pre-steaming time and high levels of press temperature and pressing time. However, all the specimens showed a high standard deviation.

Based on the Pareto’s chart (Figure 12a,b), the pressing temperature was the most influential factor, followed by the pressing time, and the pre-steaming time was the third most influential factor. Since the Standardized Effect Estimate (SEE) values were negative for the press temperature and pressing time across both compression ratios, increasing these factor levels can effectively reduce the spring-back recovery.

According to Figure 12, only the pre-steaming factor changed depending on the compression ratio. With a 20% compression ratio, longer pre-steaming (60 min) slightly contributes to reducing the spring-back ratio. However, at the 50% compression ratio, the SEE value is positive, indicating that longer pre-steaming has a significant detrimental effect on the spring-back value. Figure 13a–c presents the response surface plots of the spring-back. Based on the response surface inclination direction and color zones we can optimize the factors’ level setup for a designed response.

3.3. Bending Strength—Modulus of Rupture (MOR)

Figure 14 presents the bending strength results of the 16 experimental setups and the control samples, measured according to the ASTM D143 [15] standard procedures.

Among all the experimental setups, setup #6 resulted in the lowest increase in the average MOR, showing only a 0.5% improvement compared with the control result of 90 MPa. The highest average MOR was obtained using the factor levels of setup #15, where the pre-steaming time was set low (30 min), the compression ratio was high (50%), the press temperature was high (150 °C), and the pressing time was also high (30 min). These experimental setups increased the average MOR value from the undensified 92.2 MPa to 147.9 MPa, representing a 60.4% increase. In addition to setup #15, three other setups (#7, #11, and #16) also yielded outstanding results, which did not differ significantly from each other, as determined by the Tukey HSD test. The common parameters across these setups were the high-level compression ratio and low-level pre-steaming time, except in the case of setup #16. The four significantly higher results closely align with the reported value (140 MPa) by Kutnar and Kamke [8], which was achieved when hybrid poplar wood was treated with 160 °C superheated steam for 3 min and pressed at a pressure of 5.5 MPa [8]. A similar result (MOR = 150 MPa) was reported by Bao et al. [19] when poplar wood was densified to a 75% densification ratio at a constant temperature of 160 °C for 20 min. Comparing the maximum value of 147.9 MPa with the published bending strength values of small defect-free samples in the Wood Handbook [23], it can be concluded that the densified wood considerably exceeds these values.

As shown in Figure 15, the Pareto chart indicates that the compression ratio emerged as the most important factor affecting bending strength, with an SEE value of 5.3.

The pre-steaming time was the second most significant factor, with an SEE value of −3.86. Despite the relatively low temperature (120–130 °C) of the steam, the detrimental effect of longer steaming lies in the hydrolytic decomposition of hemicelluloses caused by water molecules in the steam. Although the samples had an initial moisture content of 9%, the steaming process increased their moisture content. Inoue et al. [24] reported a similar negative relationship between the bending MOR and increasing steaming time, even when significantly shorter steaming durations (a maximum of 8 min) at elevated temperatures (180–220 °C) were used. In addition to the main effects of the factors, the ANOVA revealed a significant effect of the 2by3 and 2by4 factor interactions.

These interactions exhibit a spreading interaction type, meaning that the effect of the compression ratio is directly proportional to the levels of the pressing temperature and the pressing time. As shown by the response surfaces (Figure 16a,b), when both factors are at high levels, the positive effect is amplified.

The regressions model for the bending strength including the statistically significant main factors’ effect and two-way interactions:

$y_{MOR}=116.1-6.7x_1+9.2x_2+4.95x_3+4.67x_2{x}_3+3.78x_2{x}_{4 } R^2=0.35$

where x_i is the coded variable of the main factor effects (−1 ≤ x_i ≤ +1); x_ix_j are the coded variables of the factor’s two-way interactions, generated by multiplication of the main factor effects’ coded values; x₁ represents the pre-steaming time (factor A), x₂ represents the compression ratio (factor B), x₃ represents the press temperature (factor C), x₂x₃ represents the BC factor interaction, and x₂x₄ represents the BD factor interaction (factor D is the pressing time). Setting the factors at their optimum level (x₁ = −1; x₂ = +1; x₃ = +1) the model estimates the bending strength at the y_MOR = 145.4 MPa value, which is close to the measured average MOR value for the experimental setup #15, where the factors are also set at their optimum values. The difference can be explained by the lower coefficient of determination, i.e., the fitted regression model represents a lower proportion of variance.

A positive correlation between the MOR and compression rate was previously reported by Bao et al. [19]. However, the relationships with other manufacturing parameters have not been reported in previous studies. Although the pressing time was not significant as a main effect, its interaction with the compression ratio was found to be slightly significant (p = 0.03).

3.4. Compression Strength Parallel to Grain

Figure 17 presents the compression strength results for the 16 experimental setups and the control samples, measured in accordance with the ASTM D143 (Section 9) [15] procedures.

Regardless of the densification parameters, the average compression strength of all the treated samples greatly exceeds that of the control samples, which is 38 MPa. Among the treated samples, only the average strength of 45 MPa in setup #2 did not differ significantly from the control samples at a 5% significance level. No significant difference was found between the lower average values of setups #1, #2, #9, and #10, as determined by the Tukey honestly significant difference (HSD) test. In all these cases, the compression ratio and pressing temperature were set at a low level. In contrast, when these factor levels were set at the high level, as in the case of setups #7, #15, and #16, the highest strength values were obtained. With a low steaming time, high pressing rate, high pressing temperature, and low pressing time, setup #7 achieved an average compression strength of 82.3 MPa, which is 116.6% greater than that of the 38 MPa control samples.

The key role of the compression ratio and press temperature was confirmed via ANOVA and Pareto analyses. Figure 18 shows the Pareto chart with the Standard Effect Estimate (SEE) values for the four main effects and the two-level interactions.

The steaming time also significantly affected the compressive strength values, whereas the pressing time did not significantly affect the compressive strength values. Based on the Pareto analyses, the high level of setup parameters for both the compression ratio (50%) and the press temperature (150 °C) was beneficial; however, the longer pre-steaming time (60 min) was detrimental to the strength performance. The strongest interaction can be observed between the compression ratio and the pressing temperature.

The positive “standardized effect estimate” (SEE) of these interactions indicates that increasing both factors simultaneously leads to higher strength values, as also demonstrated by the response surface (Figure 19a).

A significant crossover interaction between the pre-steaming time and compression ratio factors is evident on the response surface (Figure 19b).

Since this interaction is significant, a low level of pre-steaming enhances the effect of a higher compression ratio on MOR performance. Inoue et al. [24] reported a similar detrimental effect on the mechanical properties due to high pre-steaming conditions, specifically when the steam temperature exceeded 180 °C for 10 min. The authors attributed the observed reduction in mechanical performance to the degradation of the microfibril framework and the viscous flow of matrix components, likely resulting from the breakdown of cross-linked structures within the wood.

The regressions model for the compression strength including the statistically significant main factors’ effect and two-way interactions is as follows:

$y_{{\sigma }_c}=59.2-2.5x_1+7.4x_2+5.5x_3-1.2x_1{x}_2+2.0x_2{x}_3, R^2=0.62$

where x_i is the coded variable of the main factor effects (−1 ≤ x_i ≤ +1); x_ix_j are the coded variables of the factor’s two-way interactions; x₁ represents the pre-steaming time (factor A), x₂ represents the compression ratio (factor B), x₃ represents the press temperature (factor C), x₁x₂ represents the AB factor interaction, and x₂x₃ represents the BC factor interaction.

The model approximates the dependent variable (compression strength) at y_σC = 77.8 MPa at the factors’ optimum level, which is comparable with the average of the measured data at experimental setup #15. The higher coefficient of determination indicates a better fit of the model to the observed data.

3.5. Hardness

Figure 20 shows that the average hardness values of all densified samples exceeded the control value of 1384 N measured on the tangential surface following the ASTM D 143 standard (Section 13) [15]. However, for setups #1, #2, #5, #6, and #9, the Tukey HSD test did not reveal significant differences compared with the control value.

Among the 16 experiments, five setups (#7, #8, #11, #15, and #16) yielded significantly higher hardness values, with setup #11 achieving the highest average hardness of 3519 N. When a low steaming time (30 min), high compression ratio (50%), low pressing temperature (110 °C), and high pressing time (30 min) were used, the hardness increased by 154% compared with that of the control samples. This increase aligns well with the findings of Peliet and Yorulmaz [25], who reported that 40% densification of poplar wood led to a 138% increase in tangential hardness. Budakçı et al. [26] reported a 147% increase in hardness for black poplar when a pressing temperature of 140 °C was applied alongside a 50% compression ratio. When the maximum value of 3519 N is compared with those of solid red oak or white oak, which are widely used wood species for flooring, it can be concluded that this value is considerably lower than the average values of 6300 N or 6600 N published by Uzcategui et al. [27].

The results of the Pareto analysis (Figure 21) support the fact that the compression ratio has an exceptionally strong effect on surface hardness.

The SEE value of the compression ratio is 2.6 times greater than that of the press temperature, which is the second most significant factor.

The remarkable effect of the compression ratio on hardness has already been confirmed by numerous previous studies [2]. The third significant factor is the pressing time, with an SEE value of 2.68. Laine et al. [18] reported the beneficial effect of a longer pressing time at higher temperatures on surface hardness. The pre-steaming time did not significantly affect the hardness properties; however, setting this factor to a low level (30 min) was beneficial. The strongest interaction was observed between the compression ratio and the pressing temperature. Owing to the spreading interaction type, setting both factors to high levels significantly contributed to the increase in hardness (Figure 22). Budakçı et al. [26] reported the same phenomenon during the densification of black poplar and linden species when different pressing temperatures and compression ratios were combined.

The regressions model for the hardness, including the statistically significant main factors’ effect and two-way interactions, is as follows:

$y_{MOR}=2492.2-147.3x_1+627.1x_2+321.9x_3+229.1x_2{x}_3 R^2=0.55$

where x₁ represents the pre-steaming time (factor A), x₂ represents the compression ratio (factor B), x₃ represents the press temperature (factor C), and x₂x₃ represents the BC factor interaction.

The model approximates the dependent variable (hardness) at 3917.6 MPa value at the factors’ optimum level, which is comparable with the average of the measured data at experimental setup #15. The modest coefficient of determination indicates a moderate fit of the model to the observed data.

4. Conclusions

The comprehensive factorial analysis demonstrates that the compression ratio is the most influential factor affecting the mechanical and physical properties of yellow poplar (Liriodendron tulipifera) densified wood, significantly enhancing compression strength, bending strength, and hardness. However, a higher compression ratio (50%) also resulted in a significant disadvantage, leading to greater spring-back compared to the 20% compression ratio.

Pressing temperature is another critical factor, with higher levels (150 °C) boosting strength, hardness, and spring-back performance.

The analysis revealed the negative effect of longer pre-steaming treatment on the examined mechanical properties. Pressing time, though less significant as an independent factor, positively interacts with high compression ratios and pressing temperatures to improve performance.

Notable interactions between factors were observed. High compression ratios combined with elevated pressing temperatures significantly improved strength and hardness. The densification process applied in this study increased compression strength by 117%, bending strength by 60%, and hardness by 154% compared to control samples.

Lower coefficients of determination values of the regression models, fitted on the observed data, indicated that the high variability of the individual measurements mask the involved factors’ main effect.

The enhanced properties of densified yellow poplar (Liriodendron tulipifera) suggest its potential for high-performance structural applications, reducing reliance on exotic hardwood species. However, further research should evaluate the long-term performance of densified yellow poplar (Liriodendron tulipifera) under diverse environmental conditions to assess its durability and suitability for outdoor applications.

These findings highlight the need for precise optimization of process parameters to balance mechanical performance and dimensional stability, particularly at higher compression ratios where variability challenges may arise.

Author Contributions

Conceptualization, G.S.O. and B.B.; methodology, G.S.O. and B.B.; software, B.B. and L.D.; validation, B.B. and L.C.; formal analysis, B.B.; investigation, G.S.O.; resources, J.M.; original draft preparation, G.S.O.; writing—review and editing, B.B., G.S.O. and L.D.; supervision, G.S.O. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The research data supporting the findings of this study will be made available upon reasonable request.

Acknowledgments

Big thanks to Carol Dawson and Nathan Pulcifer from Quintek Measurement Systems for performing our density profile measurements.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. During the hot press stage with a 20% compression ratio. Metal spacers are shown in the picture.

Figure 2. Density distribution profile of a densified specimen (example).

Figure 3. Static bending test setup.

Figure 4. Compression test parallel to the grain.

Figure 5. Hardness test setup.

Figure 6. Density profile of (a) two samples of undensified samples, (b) two samples from setup #1, (c) two samples from setup #7, and (d) two samples from setup #16. Each line corresponds to an individual sample.

Figure 7. SEM images of setup #6 with 20% densification. (a) Magnification: 25×; (b) magnification: 250×.

Figure 8. Relative differences between various types of average densities categorized by compression ratio.

Figure 9. Average maximum density for samples from the two compression ratios and the control group.

Figure 10. Average distance between the specimens’ surface and the average maximum density, categorized based on the compression ratio.

Figure 11. Average spring-back of the 16 experimental setups.

Figure 12. Pareto chart of the effects on the spring-back. (a) 20% compression ratio, (b) 50% compression ratio.

Figure 13. Response surface plots of the spring-back related to specific variables. (a) Interaction between pressing time and pre-steaming time at a 20% compression ratio; (b) interaction between pressing time and press temperature at a 20% compression ratio; (c) interaction between pressing time and press temperature at a 50% compression ratio.

Figure 14. Average bending strength of the 16 experimental setups and control samples.

Figure 15. Pareto chart of the effects on the bending strength.

Figure 16. Response surface plots of the bending strength as response variable. (a) MOR in function of press temperature and compression ratio; (b) MOR in function of pressing time and compression ratio.

Figure 17. Average compression strength of the 16 experimental setups and control samples.

Figure 18. Pareto chart of the effects on the compression strength.

Figure 19. Response surface of the factor interactions with compression strength as the response variable. (a) Interaction between press temperature and compression ratio; (b) interaction between pre-steaming time and compression ratio.

Figure 20. Average hardness of the 16 experimental setups and control samples.

Figure 21. Pareto chart of the effects on the hardness.

Figure 22. Response surface plots of the hardness as response variable.

Table 1

The four controlled factors and their levels.

Level	Factors
	Pre-Steam Time (min)(A)	Compression Ratio (%)(B)	Press Temperature (°C)(C)	Pressing Time (min)(D)
Low (−1)	30	20	110	15
High (+1)	60	50	150	30

Table 2

Experiment design.

Experiment Setups	Pre-Steaming Time (min)(A)	Compression Ratio (%)(B)	Press Temperature (°C)(C)	Pressing Time (min)(D)
1	30	20	110	15
2	60	20	110	15
3	30	50	110	15
4	60	50	110	15
5	30	20	150	15
6	60	20	150	15
7	60	50	150	15
8	30	50	150	15
9	30	20	110	30
10	60	20	110	30
11	30	50	110	30
12	60	50	110	30
13	30	20	150	30
14	60	20	150	30
15	30	50	150	30
16	60	50	150	30

Table 3

The average density and thickness values across all 16 experimental setups.

Experiment Setups	Initial Density (kg/m3)	Final Density (kg/m3)	Density Increase (%)	Initial Thickness (mm)	Target Thickness (mm)	Final Thickness (mm)
1	502.9 (37.9)	564.4 (50.4)	12.2	19.1 (0.1)	15.2	15.5 (0.2)
2	455.9 (39.1)	519.7 (54.4)	14.0	19.1 (0.2)	15.2	15.1 (0.1)
3	457.4 (27.3)	674.5 (31.9)	47.5	19.1 (0.2)	9.5	12.3 (0.9)
4	495.8 (46.1)	667.9 (58.4)	34.7	19.5 (0.2)	9.5	13.9 (0.7)
5	465.1 (38.6)	529.7 (43.8)	13.9	19.0 (0.2)	15.2	15.1 (0.1)
6	440.6 (41.1)	518.6 (52.9)	17.7	18.9 (0.1)	15.2	15.1 (0.2)
7	494.4 (33.7)	770.4 (94.7)	55.8	19.3 (0.2)	9.5	10.6 (1.5)
8	493.4 (26.7)	747.5 (61.5)	51.5	19.4 (0.2)	9.5	11.4 (0.8)
9	450.6 (49.5)	515.6 (54.4)	14.4	19.0 (0.2)	15.2	15.1 (0.1)
10	471.4 (43.6)	554.1 (49.4)	17.5	19.0 (0.1)	15.2	15.3 (0.2)
11	488.1 (41.2)	836.1 (88.6)	71.3	19.4 (0.4)	9.5	10.8 (0.1)
12	468.7 (40.0)	713.6 (57.4)	52.3	19.4 (0.2)	9.5	11.0 (0.5)
13	461.4 (47.1)	526.5 (63.3)	14.1	19.0 (0.2)	15.2	15.0 (0.1)
14	462.1 (47.4)	491.0 (39.9)	6.3	18.9 (0.1)	15.2	14.9 (0.1)
15	496.6 (32.6)	835.7 (65.1)	68.3	19.5 (0.3)	9.5	10.0 (0.9)
16	482.9 (21.6)	769.0 (73.5)	59.3	19.4 (36.2)	9.5	10.7 (1.3)

Standard deviation is shown in parentheses; density increase = (final density − initial density)/initial density × 100.

References

1. Ali, M.R.; Abdullah, U.H.; Ashaari, Z.; Hamid, N.H.; Hua, L.S. Hydrothermal Modification of Wood: A Review. Polymers; 2021; 13, 2612. [DOI: https://dx.doi.org/10.3390/polym13162612] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34451152]

2. Cabral, J.P.; Kafle, B.; Subhani, M.; Reiner, J.; Ashraf, M. Densification of timber: A review on the process, material properties, and application. J. Wood Sci.; 2022; 68, 20. [DOI: https://dx.doi.org/10.1186/s10086-022-02028-3]

3. Hajihassani, R.; Ghahri, S.; Zamani, S.M.; Nourbakhsh, A. Performance of Densified Wood Glulam as Building Bio-Material. J. Renew. Mater.; 2022; 10, pp. 511-526. [DOI: https://dx.doi.org/10.32604/jrm.2022.017781]

4. Sandberg, D.; Haller, P.; Navi, P. Thermo-hydro and thermo-hydro-mechanical wood processing: An opportunity for future environmentally friendly wood products. Wood Mater. Sci. Eng.; 2013; 8, pp. 64-88. [DOI: https://dx.doi.org/10.1080/17480272.2012.751935]

5. Wang, X.; Tu, D.; Chen, C.; Zhou, Q.; Huang, H.; Zheng, Z.; Zhu, Z. A thermal modification technique combining bulk densification and heat treatment for poplar wood with low moisture content. Constr. Build. Mater.; 2021; 291, 123395. [DOI: https://dx.doi.org/10.1016/j.conbuildmat.2021.123395]

6. Yin, Y.; Berglund, L.; Salmén, L. Effect of Steam Treatment on the Properties of Wood Cell Walls. Biomacromolecules; 2011; 12, pp. 194-202. [DOI: https://dx.doi.org/10.1021/bm101144m] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21133402]

7. Kutnar, A.; Kamke, F.A. Influence of temperature and steam environment on set recovery of compressive deformation of wood. Wood Sci. Technol.; 2012; 46, pp. 953-964. [DOI: https://dx.doi.org/10.1007/s00226-011-0456-5]

8. Kutnar, A.; Kamke, F.A. Compression of wood under saturated steam, superheated steam, and transient conditions at 150 °C, 160 °C, and 170 °C. Wood Sci. Technol.; 2012; 46, pp. 73-88. [DOI: https://dx.doi.org/10.1007/s00226-010-0380-0]

9. Zhou, Q.; Chen, C.; Tu, D.; Zhu, Z.; Li, K. Surface densification of poplar solid wood: Effects of the process parameters on the density profile and hardness. BioResources; 2019; 14, pp. 4814-4831. [DOI: https://dx.doi.org/10.15376/biores.14.2.4814-4831]

10. Xiang, E.; Li, J.; Huang, R.; Gao, Z.; Yang, S. Effect of superheated steam pressure on the physical and mechanical properties of sandwich-densified wood. Wood Sci. Technol.; 2022; 56, pp. 899-919. [DOI: https://dx.doi.org/10.1007/s00226-022-01383-0]

11. Shao, Y.; Li, L.; Chen, Z.; Wang, S.; Wang, X. Effects of thermo-hydro-mechanical treatments on various physical and mechanical properties of poplar (Populus) wood. BioResources; 2020; 15, pp. 9596-9610. [DOI: https://dx.doi.org/10.15376/biores.15.4.9596-9610]

12. Rahimi, S.; Singh, K.; DeVallance, D. Effect of Different Hydrothermal Treatments (Steam and Hot Compressed Water) on Physical Properties and Drying Behavior of Yellow-Poplar (Liriodendron tulipifera). For. Prod. J.; 2019; 69, pp. 42-52. [DOI: https://dx.doi.org/10.13073/FPJ-D-18-00028]

13. ASTM D4442-20 Standard Test Methods for Direct Moisture Content Measurement of Wood and Wood-Based Materials; ASTM International: West Conshohocken, PA, USA, 2020.

14. ASTM D2395-17 Standard Test Methods for Density and Specific Gravity (Relative Density) of Wood and Wood-Based Materials; ASTM International: West Conshohocken, PA, USA, 2022.

15. ASTM D143-22 Standard Test Methods for Small Clear Specimens of Timber; ASTM International: West Conshohocken, PA, USA, 2022.

16. Kutnar, A.; Kamke, F.A.; Sernek, M. Density profile and morphology of viscoelastic thermal compressed wood. Wood Sci. Technol.; 2009; 43, pp. 57-68. [DOI: https://dx.doi.org/10.1007/s00226-008-0198-1]

17. Rautkari, L.; Laine, K.; Kutnar, A.; Medved, S.; Hughes, M. Hardness and density profile of surface densified and thermally modified Scots pine in relation to degree of densification. J. Mater. Sci.; 2013; 48, pp. 2370-2375. [DOI: https://dx.doi.org/10.1007/s10853-012-7019-5]

18. Laine, K.; Rautkari, L.; Hughes, M. The effect of process parameters on the hardness of surface densified Scots pine solid wood. Eur. J. Wood Prod.; 2013; 71, pp. 13-16. [DOI: https://dx.doi.org/10.1007/s00107-012-0649-0]

19. Bao, M.; Huang, X.; Jiang, M.; Yu, W.; Yu, Y. Effect of thermo-hydro-mechanical densification on microstructure and properties of poplar wood (Populus tomentosa). J. Wood Sci.; 2017; 63, pp. 591-605. [DOI: https://dx.doi.org/10.1007/s10086-017-1661-0]

20. Balasso, M.; Kutnar, A.; Niemelä, E.P.; Mikuljan, M.; Nolan, G.; Kotlarewski, N.; Hunt, M.; Jacobs, A.; O’reilly-Wapstra, J. Wood Properties Characterisation of Thermo-Hydro Mechanical Treated Plantation and Native Tasmanian Timber Species. Forests; 2020; 11, 1189. [DOI: https://dx.doi.org/10.3390/f11111189]

21. Laskowska, A. The Influence of Process Parameters on the Density Profile and Hardness of Surface-densified Birch Wood (Betula pendula Roth). BioResources; 2017; 12, pp. 6011-6023. [DOI: https://dx.doi.org/10.15376/biores.12.3.6011-6023]

22. Xu, X.; Tang, Z. Vertical compression rate profile and dimensional stability of surface-densified plantation poplar wood. Lignocellulose; 2012; 1, pp. 45-54.

23. Forest Products Laboratory. Wood Handbook: Wood as an Engineering Material; U.S. Department of Agriculture, Forest Service, Forest Products Laboratory: Madison, WI, USA, 1999; [DOI: https://dx.doi.org/10.2737/FPL-GTR-113]

24. Inoue, M.; Norimoto, M.; Tanahashi, M.; Rowell, R.M. Steam or heat fixation of compressed wood. Wood Fib. Sci.; 1993; 25, pp. 224-235.

25. Pelit, H.; Yorulmaz, R. Influence of densification on mechanical properties of thermally pretreated spruce and poplar wood. BioResources; 2019; 14, pp. 9739-9754. [DOI: https://dx.doi.org/10.15376/biores.14.4.9739-9754]

26. Budakçı, M.; Pelit, H.; Sönmez, A.; Korkmaz, M. The Effects of Densification and Heat Post-Treatment on Hardness and Morphological Properties of Wood Materials. BioResources; 2016; 11, pp. 7822-7838. [DOI: https://dx.doi.org/10.15376/biores.11.3.7822-7838]

27. Carmona Uzcategui, M.G.; Seale, R.D.; França, F.J.N. Physical and mechanical properties of clear wood from red oak and white oak. BioResources; 2020; 15, pp. 4960-4971. [DOI: https://dx.doi.org/10.15376/biores.15.3.4960-4971]