Bending Performance of Thermo-Hydro-Mechanically Densified Poplar Wood: Effects of Ultrasonic Pretreatment and Thermal Posttreatment at Different Compression Ratios

Abstract

Thermo-hydro-mechanical (THM) densification is an effective method for improving the mechanical performance of low-density, fast-growing hardwoods such as poplar. This study examined the bending performance of THM-densified poplar wood at different compression ratios (CR = 0%, 50%, 60%, and 65%), with emphasis on the effects of ultrasonic pretreatment (US) and thermal modification posttreatment (TM), applied individually and in combination. A paired sampling design was used to reduce material variability, and modulus of rupture (MOR) and modulus of elasticity (MOE) were evaluated using linear mixed-effects models (LMM). Bending tests were performed in accordance with EN 310:1993. Increasing the compression ratio led to substantial increases in MOR and MOE; compared with non-densified specimens, MOR increased by approximately 240% and MOE by about 140% at CR = 65%, confirming densification as the dominant factor controlling bending performance. US did not affect non-densified wood but significantly enhanced MOR and MOE after densification, particularly at CR = 50%. In contrast, TM consistently reduced MOR and, to a lesser extent, MOE across all compression ratios. The results demonstrate that the bending performance of densified poplar wood is governed by both compression ratio and compression-dependent treatment effects.

1. Introduction

Hybrid poplars are widely cultivated due to their rapid growth, good workability, and homogeneous structure. Globally, poplar plantations cover approximately 31.4 million hectares [1], while in Europe poplar-growing areas are estimated at about 1 million hectares [2]. In Serbia, plantations of various poplar hybrids cover an area of approximately 40,000 ha [3], making poplar wood an important raw material for veneer and plywood production, packaging, pulp and paper, and non-structural furniture components. Despite these advantages, poplar wood is characterized by low density, limited mechanical strength, and low natural durability, which restrict its use in load-bearing and high-wear applications [4]. Consequently, there is strong interest in modification routes that can upgrade the mechanical performance of poplar wood to enable higher-value and more demanding applications.

Among available wood modification techniques, THM densification has emerged as an effective and environmentally favorable approach for improving mechanical properties by increasing density and load-bearing capacity, as it relies on physical processing rather than chemical modification. By upgrading low-density, fast-growing hardwoods, it improves material efficiency and promotes more resource-efficient wood utilization. Thermo-hydro-mechanical (THM) densification, which combines heat, moisture, and pressure, enables controlled deformation of wood cells while minimizing cell-wall fracture, resulting in reduced porosity and enhanced stiffness and strength [5,6,7,8]. Several studies have demonstrated that increasing compression ratio (CR) generally leads to substantial increases in the modulus of rupture (MOR) and the modulus of elasticity (MOE) in fast growing hardwoods [9,10]. However, the effectiveness of densification depends strongly on wood species, anatomical structure, and processing conditions [5,11,12], and excessive compression may induce localized damage or non-uniform density profiles that limit further mechanical gains [7,9,13,14].

Although the influence of compression ratio on bending properties is well established, fewer studies have systematically examined how auxiliary treatments influence the development of MOR and MOE across different densification levels under a single, controlled THM schedule. In particular, it remains unclear whether such treatments improve properties beyond density-driven effects, and whether their influence depends on the degree of compression. Addressing this gap is essential for improving the efficiency and reliability of densification-based modification strategies for low-density hardwoods.

Physical pretreatments, such as ultrasonic (US) treatment have attracted attention as environmentally friendly methods for modifying wood structure and transport properties. US treatment in aqueous media has been shown to enhance permeability and moisture transport through pit disruption and formation of microchannels, thereby accelerating drying and improving fluid mobility [15,16,17]. In the context of THM densification, such effects are relevant because uniform heat and moisture distribution are critical for effective viscoelastic softening and controlled deformation during compression [6,7,8]. Previous studies have also suggested that US treatment may influence lignin structure and viscoelastic behavior, potentially lowering resistance to compression and contributing to more uniform deformation of the wood matrix [17]. Moreover, a positive effect of ultrasonic pretreatment on the bending performance of densified wood has been reported [18]. However, the extent to which these effects translate into measurable improvements in bending performance of densified wood remains insufficiently documented.

Thermal posttreatment (TM) is commonly applied after THM densification to stabilize compressed structure and reduce set-recovery and moisture sensitivity [19]. From a mechanical perspective, TM treatment alters the chemical composition and viscoelastic behavior of wood, primarily through degradation of hemicelluloses and modification of lignin [20]. These changes generally result in reduced bending strength and, to a lesser extent, stiffness, particularly at higher treatment intensities [21,22]. When combined with densification, TM may therefore modify the balance between density-induced mechanical gains and treatment-induced material degradation. The interaction between densification level and TM intensity; however, is not yet fully understood.

Accordingly, this study investigates performance of THM-densified poplar wood across multiple compression ratios (CR = 0%, 50%, 60%, and 65%), with particular emphasis on the effects of ultrasonic (US) pretreatment and thermal posttreatment (TM), applied individually and in combination. While the influence of densification and TM on mechanical properties has been previously reported, the compression-dependent interaction between densification and auxiliary treatments—particularly US—remains insufficiently understood. The novelty of this study lies in the systematic evaluation of US and TM across several compression levels under a unified THM schedule, enabling differentiation between density-driven mechanical gains and treatment-specific effects on bending strength (MOR) and stiffness (MOE).

2. Materials and Methods

2.1. Specimen Description

Five poplar (Populus sp.) logs, aged 15–20 years, were obtained from a commercial plantation located in central-eastern Serbia. From each log, four tangentially sawn boards were produced, resulting in total of 20 boards. The boards were sawn such that growth rings were oriented parallel to the wide face (Figure 1). All boards were free of visible defects and had nominal dimensions of 2000 × 140 × 28 mm.

Prior to specimen preparation, the boards were kiln dried to a target moisture content of approximately 12% and conditioned to reach moisture equilibrium under laboratory conditions. The oven-dry density of the boards ranged from 376 to 550 kg·m⁻³, with a mean value of 434.1 kg·m⁻³ (SD = 43.8 kg·m⁻³; CV ≈ 10%).

To minimize the influence of natural material variability and enable reliable comparisons among treatments, a paired (matched position) sampling design was employed. Specimens assigned to different densification and treatments were cut-off from equivalent anatomical positions within each board, ensuring that each board contributed specimens to all experimental groups whenever possible. This approach reduced the influence of intra-board heterogeneity, such as density gradients, growth ring curvature, and earlywood–latewood ratio, allowing treatment effects to be evaluated independently of board-level variability.

From each board, four elements were prepared and assigned to one of the following compression-ratio groups: non-densified control (CR = 0%), and THM-densified specimens, with nominal compression ratios of 50%, 60% and 65% (the latter representing the upper compression limit of the press setup used in this study). From each element, specimens with initial dimensions 400 × 120 × 20 mm (length × width × thickness) were prepared for the densification process. One half of the specimens in each compression-ratio group was subjected to US prior to densification (Figure 1).

After densification and subsequent conditioning for two weeks at 20 ± 2 °C and 65 ± 5% relative humidity (RH), each specimen was longitudinally cut into two test specimens with final dimensions of 400 × 20 (length × width). One half of these test specimens were subsequently subjected to TM, while the remaining specimens were retained as non-thermally treated references. All specimens were conditioned again under the same climatic conditions prior to mechanical testing.

This experimental design resulted in four primary groups defined by CR (0% (non-densified), 50%, 60%, and 65%), each further subdivided into four treatment-based subgroups: specimens without pre- and posttreatment (Reference), ultrasonic pretreatment only (US), thermal posttreatment only (TM), and combined ultrasonic pretreatment and thermal posttreatment (US + TM) (

Table 1. Experimental groups, treatment combinations, group codes, and number of specimens (N) for each compression ratio (CR).

CR (%)	Ultrasonic Pretreatment	Thermal Posttreatment	Treatment Code	N
0 (control)	No	No	Reference	20
	Yes	No	US	20
	No	Yes	TM	20
	Yes	Yes	US + TM	20
50	No	No	Reference	20
	Yes	No	US	20
	No	Yes	TM	20
	Yes	Yes	US + TM	20
60	No	No	Reference	5
	Yes	No	US	5
	No	Yes	TM	5
	Yes	Yes	US + TM	5
65	No	No	Reference	10
	Yes	No	US	10
	No	Yes	TM	10
	Yes	Yes	US + TM	10

). Within each CR group, the untreated specimens served as reference samples, enabling direct assessment of the individual and combined effects of the applied treatments relative to densification alone. For the higher CR (60% and 65%), the number of specimens was reduced due to processing-related losses.

2.2. Wood Modification

2.2.1. Ultrasonic Pretreatment

US was performed using a stainless steel ultrasonic bath (ASonic Pro300, Ljubljana, Slovenia) operating at nominal frequency of 28 kHz and at output power of 600 W, corresponding to an estimated nominal power density of approximately 0.40 W/cm², assuming uniform energy distribution over the bath bottom area. Specimens were fully immersed in distilled water in a 28 L ultrasonic bath and treated for 45 min. The initial water temperature was 25 °C and increased to about 35 °C during sonification as a result of ultrasonic energy input. This temperature rise was consistent across treatment batches and was not externally controlled.

To ensure uniform exposure to the ultrasonic field, specimens were placed in custom holders that maintained spacing between individual samples and prevented direct contact with the bath walls (Figure 2). Specimens were weighted to remain fully submerged throughout the treatment. US treatment was conducted in batches of six specimens, and fresh distilled water used for each batch to avoid the cumulative effects of dissolved extractives.

The selected ultrasonic frequency, power, and treatment duration were chosen to be comparable with conditions commonly reported in previous studies on ultrasound-assisted wood modification and permeability enhancement [23]. Following US, all specimens were conditioned at 20 ± 2 °C and 65 ± 5% RH for two months prior to the subsequent densification process.

2.2.2. Thermo-Hydro-Mechanical (THM) Densification

THM densification was carried out using an open-system hydraulic hot-press (Langzauner “Perfect” LZT-UK-30-L, Lambrechten, Austria) equipped with a water-cooling system. Both US pretreated and non-pretreated specimens were densified under identical processing conditions.

Specimens were compressed in the radial direction from an initial thickness of 20 mm to thickness of 10 mm, 8 mm and 7 mm, corresponding to nominal compression ratios (CR) for 50%, 60% and 65% respectively. Metal stops were used to define the final platen position. Based on thickness measurements before and after densification, the actual compression ratio (CR) was calculated as (Equation (1)):

$$CR=\left({\displaystyle \frac{h_0-h_d}{h_0}}\right)\times 100 \left[\%\right]$$

(1)

where $h_0$ is the initial specimen thickness and $h_d$ is the thickness after densification.

The THM cycle (Figure 3) consisted of the following stages: (a) pre-heating of the upper and lower platens to 170 °C; (b) loading and closing at constant platen speed of 3 mm s⁻¹ until the metal stops were reached, corresponding pressure to a nominal pressure of 4 MPa applied in the radial direction; (c) holding phase of 3 min under constant compression; (d) a high-temperature phase in which platen temperature was increased to 200 °C and maintained for 2 min; and (e) cooling under load to 60 °C prior to unloading, as in [24]. The THM schedule was kept constant for all specimens in order to isolate the effects of compression ratio and auxiliary treatments rather than to optimize densification conditions for individual groups.

2.2.3. Thermal Posttreatment

TM was applied to selected specimens following densification and conditioning. Prior to TM treatment, all specimens were conditioned for two weeks at 20 ± 2 °C and 65 ± 5% relative humidity to ensure comparable initial moisture states.

TM treatment was conducted in a laboratory kiln (Kambič VS-258 ML, Semič, Slovenia) under normal atmospheric pressure at a target temperature of 210 °C. The temperature was increased from ambient conditions at an average heating rate of approximately 13.5 °C/h. The total treatment duration was 22 h, of which the specimens were maintained at the maximum temperature for 7 h. The total treatment time exceeded the initially planned schedule due to delayed attainment of the target temperature; however, the applied time-temperature combination remained within the range commonly reported for industrial thermal modification of hardwoods [25,26].

2.3. Physical Characterization and Mechanical Testing

2.3.1. Determination of Moisture Content and Density

Following densification and, where applicable, TM, all specimens were conditioned for two months at 20 ± 2 °C and 65 ± 5% relative humidity prior to physical and mechanical testing. This conditioning period was selected to ensure that specimens reached moisture equilibrium under identical climatic conditions, allowing meaningful comparison among compression ratios and treatment groups.

From each test specimen, a subsample was extracted for determination of moisture content and density at testing conditions as well as oven-dry density. Moisture content was determined using the oven-dry method in accordance with EN 13183-1:2002 [27]. Density at testing conditions and oven-dry density were determined following ISO 13061-2:2014 [28], based on measured mass and dimensions at the corresponding moisture states.

2.3.2. Bending Tests

Bending tests were performed in accordance with EN 310:1993 [29] using two universal testing machines (ZwickRoell Z005 and Z100, Ulm, Germany). Due to different final thicknesses resulting from THM densification at different compression ratios, the support span and loading rate were adjusted to maintain a constant span-to-thickness ratio (L/h = 15) across all specimen groups, ensuring mechanical comparability and compliance with standard. The selected support spans and corresponding loading rates for each compression-ratio group are summarized in

Table 2. Support span and specimen thickness for bending tests.

CR [%]	Specimen Thickness h (mm)	Support Span L (mm)	L/h	Loading Rate (mm/min)
0	20	300	15	7
50	10	150	15	5
60	8	120	15	4.5
65	7	105	15	4

. Loading rates were chosen such, that maximum load was reached within 60 ± 30 s, as specified in EN 310:1993.

Specimens were loaded in three-point bending with the load applied perpendicular to the wide face of the specimen. Modulus of rupture (MOR) and modulus of elasticity (MOE) were calculated according to EN 310:1993 as (Equations (2) and (3)):

$$MOR={\displaystyle \frac{3F_{max}l}{2b{h}^2}}$$

(2)

$$MOE={\displaystyle \frac{l^3∆F}{4b{h}^3∆y}}$$

(3)

where $F_{max}$ is the maximum applied load (N), $l$ is the support span (mm), $b$ is the specimen width (mm), $h$ is the specimen thickness (mm), ∆F is the load increment within the linear elastic region (N), and ∆y is the corresponding mid-span deflection (mm).

2.3.3. Statistical Analysis

Statistical analysis was conducted using the R statistical environment (R 4.5.2, R Core Team, Vienna, Austria). Due to the paired (matched position) sampling design, in which multiple specimens originated from the same board, LMM were applied to account for the grouping of specimens within boards [30]. Board was included as a random intercept to account for intra-board variability and non-independence of observations.

For each mechanical property (MOR and MOE), the primary model was defined as:

Property ~ CR × Treatment + (1|Board)

where compression ratio (CR), treatment (Reference, US, TM, US + TM), and their interaction were treated as fixed effects. Type III analysis of variance was performed using the Kenward–Roger approximation for denominator degrees of freedom. Post hoc comparisons were conducted on estimated marginal means (EMM) using Tukey-adjusted pairwise contrasts [31].

For higher compression ratios (CR = 60% and 65%) the number of specimens was reduced due to processing-related losses, resulting in lower statistical power for detecting treatment effects at these levels. Consequently, non-significant differences at high compression ratios were interpreted conservatively.

In addition to the primary analysis, secondary models were evaluated for specimens densified at a CR = 50%, in which density at testing conditions was included as a covariate. These models were used to examine whether treatment effects on MOR and MOE persisted at comparable density levels.

3. Results

3.1. Compression Ratio

Based on post-densification measurements, the mean achieved compression ratios were 50.6%, 60.1%, and 65.8% for the nominal CR 50%, 60%, and 65% groups, respectively. The mean thickness immediately after densification decreased from approximately 20 mm in the non-densified specimens to about 9.9 mm (CR 50%), 8.0 mm (CR 60%), and 6.9 mm (CR 65%).

After conditioning at 20 ± 2 °C and 65 ± 5% relative humidity, a slight increase in thickness was observed for all densified groups, indicating partial set-recovery. The mean conditioned thicknesses were approximately 10.0 mm, 8.1 mm, and 7.1 mm for CR 50%, CR 60%, and CR 65%, respectively, while non-densified specimens retained a thickness close to the initial value.

3.2. Moisture Content and Density

Following two months of conditioning under identical conditions prior to mechanical testing, specimens exhibited different MC depending on the applied treatments (

Table 3. Density and moisture content at testing conditions, together with bending properties (MOR and MOE), presented as mean values with standard deviations (SD) for each experimental group.

CR [%]	Treatment	N	Oven-Dry Density [Kg/m3]	Density [Kg/m3]	MC [%]	MOR [MPa]	MOE [GPa]
			(SD)	(SD)	(SD)	(SD)	(SD)
0	Ref.	20	434.1 (43.8) a	469.6 (36.4) a	10.9 (0.3) a	74.4 (8.9) a	9.65 (1.35) a
	US	20	442.0 (52.2) a	463.3 (33.4) a	11.1 (0.2) a	72.0 (9.5) a	9.25 (1.06) a
	TM	20	417.4 (35.3) a	433.0 (33.6) b	5.1 (0.2) b	53.5 (12.7) b	10.01 (1.31) a
	US + TM	20	418.8 (51.1) a	427.9 (38.3) b	5.1 (0.2) b	47.2 (14.6) b	9.49 (1.24) a
50	Ref.	20	826.1 (70.0) a	864.7 (52.6) a	7.9 (0.5) a	147.5 (22.4) a	12.57 (1.51) a
	US	20	867.3 (69.3) b	905.4 (72.2) b	7.4 (0.8) a	174.2 (22.1) b	16.20 (2.18) b
	TM	20	762.4 (107.8) c	792.4 (71.5) c	3.7 (0.4) b	78.8 (16.7) c	11.30 (1.99) c
	US + TM	20	850.2 (88.7) a	864.5 (78.0) a	3.6 (0.5) b	119.4 (37.3) d	16.45 (2.80) b
60	Ref.	5	1048.6 (61.9) a	1123.4 (57.3) a	8.1 (0.3) a	242.6 (20.9) a	22.48 (2.38) a
	US	5	1058.6 (52.1) a	1121.2 (76.5) a	8.4 (0.2) a	235.6 (26.0) a	22.36 (2.86) a
	TM	5	958.6 (63.1) b	1004.2 (64.9) b	3.4 (0.5) b	134.4 (22.0) b	18.56 (1.70) b
	US + TM	5	1014.8 (65.4) ab	1050.4 (55.0) c	3.4 (0.4) b	150.4 (39.3) b	20.36 (3.02) ab
65	Ref.	10	1159.6 (83.7) a	1183.4 (81.3) a	5.3 (0.7) a	254.3 (55.3) a	23.48 (4.35) a
	US	10	1171.8 (80.6) a	1190.4 (92.5) a	5.4 (0.6) a	269.4 (57.8) a	24.60 (4.98) a
	TM	10	1030.4 (74.6) b	1089.9 (44.1) b	2.7 (0.1) b	164.7 (29.1) b	20.43 (2.13) b
	US + TM	10	1011.7 (91.6) b	1089.0 (62.4) b	3.1 (0.5) b	160.9 (32.7) b	20.54 (2.80) b

^a,b,c,d Different superscript letters within the same compression ratio indicate statistically significant differences between treatment groups (p < 0.05), based on Tukey-adjusted pairwise comparisons of EMM from the LMM.

). TM consistently resulted in lower MC values compared with non-thermally treated specimens within each compression-ratio group. In contrast, US alone did not result in a systematic change in MC across compression ratios.

Increasing compression ratio was associated with a general decrease in MC. For non-thermally treated specimens, MC ranged from approximately 11% in non-densified wood to about 8% at CR = 50% and 60%, and further to approximately 5% at CR = 65%. TM post-treated specimens exhibited even lower MC values, reaching below 3% at CR = 65%. These trends confirm the combined influence of densification and TM modification on hygroscopic behavior.

Density increased markedly with increasing compression ratio across all treatment groups. The mean density at testing conditions increased from approximately 450 kg·m⁻³ in non-densified specimens to values approaching 800 kg·m⁻³ at CR = 50% and exceeding 1100 kg·m⁻³ at CR = 65% (Table 3). This increase was observed consistently for all treatments, confirming compression ratio as the dominant factor governing density.

TM resulted in a reduction in density at both testing conditions and in the oven-dry state within each compression-ratio group, whereas US alone did not lead to systematic density changes relative to reference specimens. Oven-dry density exhibited a clear and nearly monotonic increase with compression ratio, accompanied by a reduction in relative variability. Treatment-related differences in oven-dry density were negligible at CR = 0% but became increasingly pronounced at higher compression ratios.

Inferential analysis using LMM confirmed that compression ratio had a highly significant effect on both density at testing conditions and oven-dry density. Treatment effects on density were statistically significant only in combination with densification, indicating that auxiliary treatments modulated density outcomes without overriding the dominant influence of compression ratio.

3.3. Bending Strength (MOR)

Bending strength increased markedly with increasing compression ratio for all treatment groups (Table 3). Mean MOR values rose from approximately 70–75 MPa in non-densified specimens to values exceeding 250 MPa at the highest densification level, confirming the dominant influence of compression ratio on bending strength. This trend was observed consistently across all treatments, indicating that densification was the primary factor governing MOR (Figure 4). Linear mixed-effects analysis confirmed a highly significant effect of compression ratio on MOR (p < 0.001), as well as a highly significant CR × treatment interaction (p < 0.001), indicating that treatment effects depended on the level of densification. Compared with non-densified specimens (CR = 0%), MOR increased by approximately 98% at CR = 50%, by about 226% at CR = 60%, and by roughly 242% at CR = 65%, demonstrating a strong but gradually diminishing incremental gain at higher compression levels.

Within individual compression-ratio groups, statistically significant differences among treatments were observed (Table 3). At CR = 0%, no significant difference in MOR was detected between control specimens and US pretreated specimens, whereas TM post-treated specimens exhibited significantly lower MOR values compared with non-thermally treated groups. This indicates that US alone did not affect bending strength of non-densified wood, while TM reduced MOR even in the absence of densification.

For densified specimens, treatment effects were compression-dependent. At CR = 50%, US significantly increased MOR, whereas TM caused a pronounced reduction. The combined US and TM treatments resulted in intermediate MOR values. At CR = 60% and 65%, US did not result in significant differences relative to the reference group, while TM consistently reduced MOR. However, the reduced number of specimens at these compression ratios limited statistical power, and these results should therefore be interpreted cautiously.

LMM confirmed that bending strength was significantly influenced by compression ratio and treatment, as well as by their interaction. The significant CR × treatment interaction indicates that the effect of US and TM treatments on MOR depended on the level of densification. Model-based EMM illustrates distinct MOR development trends for different treatments across the compression range (Figure 5), with the strongest relative treatment effects observed at moderate densification.

To evaluate whether observed treatment effects could be attributed solely to differences in density, a secondary analysis was conducted for specimens densified at CR = 50%, where sufficient density variability was present. As shown in Figure 6, MOR increased with increasing density at testing conditions for all treatments, confirming density as a major determinant of bending strength. However, at comparable density levels, US pretreated specimens exhibited higher MOR values than reference specimens, while TM post-treated specimens consistently exhibited lower MOR values.

When density at testing conditions was included as a covariate in the LMM, density had a strong positive effect on MOR. Importantly, treatment remained a significant factor after density adjustment, and a significant interaction between density and treatment was detected. These results indicate that treatment-related differences in bending strength cannot be explained by density changes alone, and that US and TM treatments modify the relationship between density and MOR.

Overall, the MOR results demonstrate that while compression ratio governs the absolute strength level of THM-densified poplar, auxiliary treatments significantly influence strength development in a compression-dependent manner. US enhanced MOR at moderate densification levels, whereas TM consistently reduced bending strength across all compression ratios.

3.4. Modulus of Elasticity (MOE)

The modulus of elasticity increased systematically with increasing compression ratio for all treatment groups (Table 3). Mean MOE values increased from approximately 9–10 GPa in non-densified specimens to values exceeding 22 GPa at the highest densification levels, confirming compression ratio as the dominant factor governing elastic stiffness. Compression ratio had a highly significant effect on MOE (p < 0.001), while treatment and the CR × treatment interaction were also statistically significant (p < 0.001). Relative to non-densified specimens, MOE increased by approximately 30% at CR = 50%, by about 133% at CR = 60%, and by roughly 143% at CR = 65%, confirming compression ratio as the dominant factor governing elastic stiffness. Compared with bending strength, MOE exhibited lower relative variability within compression-ratio groups, reflecting the more uniform response of elastic stiffness to densification.

Within individual compression-ratio groups, treatment-related differences in MOE were observed (Figure 7), although these differences were generally less pronounced than those observed for MOR. At CR = 0%, no statistically significant differences in MOE were detected among treatment groups, indicating that neither US nor TM affected elastic stiffness of non-densified wood.

For densified specimens, treatment effects on MOE were compression-dependent. At CR = 50%, US significantly increased MOE, whereas TM significantly reduced stiffness, with the combined treatment resulting in intermediate values. At CR = 60%, no significant differences were detected between the reference, US, and combined treatments, while TM specimens exhibited lower MOE values. At CR = 65%, TM again resulted in significantly lower MOE values, whereas differences between reference and US treatments were not statistically significant.

LMM confirmed that MOE was significantly influenced by compression ratio and treatment, as well as by their interaction. The significant CR × treatment interaction indicates that treatment effects on elastic stiffness depended on the level of densification. EMM derived from the LMM illustrates distinct stiffness development trends among treatments across the compression range (Figure 8), with the most pronounced relative treatment effects occurring at moderate densification levels.

To assess the influence of density on elastic stiffness, a secondary analysis was conducted for specimens densified at CR = 50%. As shown in Figure 9, MOE increased with increasing density at testing conditions for all treatments, confirming density as the primary determinant of elastic stiffness. Compared with MOR, the MOE–density relationship exhibited a clearer and less scattered trend.

When density at testing conditions was included as a covariate in the LMM, density had a strong positive effect on MOE. After density adjustment, treatment remained a significant factor, although its contribution was smaller than that of density. These results indicate that while MOE is predominantly controlled by density increase associated with densification, auxiliary treatments introduce secondary but statistically detectable modifications to stiffness development.

Overall, the MOE results demonstrate that elastic stiffness of THM-densified poplar wood is governed primarily by compression ratio and associated density increase, with US enhancing stiffness development at moderate compression levels and TM consistently reducing MOE across all compression ratios. Compared with bending strength, treatment-related effects on MOE were more modest, reflecting the fundamentally different mechanical nature of elastic stiffness.

4. Discussion

4.1. Effect of Compression Ratio on Bending Performance

The results clearly demonstrate that compression ratio is the dominant factor governing both bending strength (MOR) and elastic stiffness (MOE) of THM-densified poplar wood. Increasing compression ratio resulted in substantial and nearly monotonic increases in both properties, consistent with previous studies on densified hardwoods [7,9,10]. This behavior reflects the increasing effective load-bearing fraction of the material due to lumen collapse and cell-wall compaction, leading to higher density and stiffness. To verify that these trends were not merely a consequence of treatment-induced differences in moisture content, exploratory LMM including moisture content as an additional covariate were evaluated. The analysis confirmed that the effects of compression ratio and treatment on both MOR and MOE remained statistically significant when moisture content was accounted for, indicating that the observed mechanical enhancements are primarily driven by densification-related structural modifications rather than moisture effects alone.

At higher compression ratios, however, the rate of mechanical improvement tended to diminish, suggesting a gradual approach toward structural saturation, where further densification yields smaller incremental gains. Similar trends have been reported for poplar and other low-density species, where excessive compression may introduce localized damage or non-uniform density profiles that limit further improvements in mechanical performance [13,14].

4.2. Effect of Ultrasonic Pretreatment on Bending Performance

US did not significantly affect bending properties of non-densified poplar wood, indicating that US alone is insufficient to enhance MOR or MOE under the applied conditions. In contrast, US significantly enhanced bending strength and stiffness after THM densification, as reported in [18], particularly at a moderate compression ratio (CR = 50%). Importantly, density-adjusted analyses confirmed that these improvements persisted at comparable density levels, indicating that the observed effects cannot be explained by density differences alone.

These findings suggest that US influences the densification process itself rather than directly strengthening the wood material. Previous studies have shown that US treatment can increase wood permeability and moisture mobility through pit disruption and microchannel formation [15,17], which may promote more uniform heat and moisture distribution during THM processing. More specifically, ultrasonic waves in an aqueous medium induce cavitation, a phenomenon where the implosive collapse of micro-bubbles generates localized shockwaves and micro-jets [15]. These mechanical stresses can lead to physical damage on wood surfaces, pit membrane disruption, and the formation of microchannels [32]. Scanning electron microscopy has confirmed that ultrasonic treatment can damage the thin-walled tissues of wood cells and open internal pores, significantly enhancing fluid migration [18]. Such modification of the cellular structure increases the transverse permeability of wood, as demonstrated in studies on Douglas-fir, where multiple increases in permeability coefficients were observed with extended ultrasonic treatment times [17]. Some authors suggest that lateral chains of lignin are removed, indicating the physico-chemical effects of ultrasound and cavitation at the cellular level [33].

Improved permeability and more homogeneous moisture distribution are critical for controlled viscoelastic softening and uniform cell-wall deformation during THM processing [6,8,34]. While microstructural or chemical characterization was beyond the scope of the present study, the compression-dependent nature of the observed effects supports the interpretation that US enhances the efficiency of densification rather than altering intrinsic material properties.

At higher compression ratios (CR = 60% and 65%), the beneficial influence of US became less pronounced. This trend likely reflects the increasing dominance of density and extensive cell-wall compaction in controlling the mechanical response at high densification levels, thereby reducing the relative contribution of auxiliary treatments.

4.3. Effect of Thermal Posttreatment on Bending Performance

TM consistently reduced bending strength and, to a lesser extent, bending stiffness across all compression ratios. These reductions remained after adjusting for density differences, indicating that TM modification reduces the mechanical performance of densified wood beyond mass loss or density reduction alone. Similar strength losses after TM treatment are widely reported and are commonly attributed to degradation of hemicelluloses and associated changes in the load-bearing capacity of the wood cell wall [20,21,35]. This degradation is a key factor, as hemicelluloses are the least thermally stable wood component, breaking down at temperatures above 150 °C and leading to a significant reduction in strength [8]. Studies have correlated these losses in bending properties specifically to the degradation of hemicelluloses, while the modulus of elasticity losses is often linked to cellulose decomposition, particularly at higher treatment severities [9]. While lignin is generally more resistant to thermal degradation compared to hemicelluloses, heat treatment can still induce significant changes in its structure [36]. This chemical alteration contributes to a shift in the mode of failure, with heat-treated wood often exhibiting a more brittle response in mechanical tests [8].

The reduction in mechanical properties observed in this study is not solely due to mass loss or density reduction but rather reflects a fundamental alteration of the cell wall matrix, making the wood more fragile [37].

The lower MC observed in TM treated specimens indicates reduced hygroscopicity, consistent with previous studies reporting reduced hygroscopicity after TM [20,38]. From an application perspective, this represents a trade-off between improved dimensional stability and reduced mechanical performance, which is consistent with mechanisms described for heat/steam-treated lignocellulosic materials [39]. The present results indicate that this trade-off persists even in highly densified wood, underscoring the importance of carefully balancing densification and TM treatment intensity depending on the intended end use.

4.4. Density- and Treatment-Specific Contributions to Bending Performance

The joint interpretation of MOR and MOE highlights important differences in the sensitivity of strength and stiffness to treatment-induced modifications. While both properties increased strongly with density, MOR exhibited greater relative sensitivity to auxiliary treatments than MOE. This distinction reflects the fundamentally different mechanical nature of the two properties: MOE primarily reflects the elastic response governed by the continuity and stiffness of the compacted cell-wall network, whereas MOR is controlled by damage initiation and propagation processes that are more sensitive to treatment-induced changes in cell-wall chemistry and microstructure [9,40].

The density-adjusted analyses provide important evidence that mechanical performance of densified wood cannot be fully explained by density differences alone. Instead, auxiliary treatments modify the efficiency with which density is translated into mechanical performance, particularly for bending strength at moderate compression ratios.

4.5. Limitations and Practical Implications

The reduced number of specimens at higher compression ratios (CR = 60% and CR = 65%) limited the statistical power for detecting treatment effects, at, and these results should therefore be interpreted cautiously. In addition, the US applied in this study was conducted under laboratory conditions using batch processing, which may pose challenges for industrial-scale implementation. From an industrial perspective, THM densification is technically feasible, but achieving uniform densification and limiting material loss at higher compression ratios remain key challenges. Large-scale implementation would require balancing mechanical gains with process efficiency and the trade-off introduced by thermal posttreatment.

Despite these limitations, the results provide valuable insights into the compression-dependent role of auxiliary treatments in THM densification. The pronounced benefits of US at moderate densification levels suggest a potential pathway for improving the mechanical efficiency of densification processes without increasing compression severity. Conversely, the consistent strength reduction after TM highlights the need for application-specific optimization when combining densification with stabilization-oriented thermal modification.

5. Conclusions

Thermo-hydro-mechanical (THM) densification substantially improved the bending performance of poplar wood, with compression ratio (CR) identified as the dominant factor controlling both modulus of rupture (MOR) and modulus of elasticity (MOE). Increasing CR resulted in pronounced and almost steady increases in strength and stiffness, confirming the effectiveness of densification as a strategy for upgrading low-density, fast-growing hardwoods. Densification led to substantial mechanical enhancement, with MOR increasing by approximately 98% at CR = 50%, 226% at CR = 60%, and 242% at CR = 65%, while MOE increased by about 30%, 133%, and 143%, respectively, compared with non-densified specimens.

US did not affect bending properties of non-densified wood but significantly enhanced MOR and MOE after densification, particularly at CR = 50%, where MOR increased by approximately 18% and MOE by about 29% relative to the corresponding reference group. Importantly, these improvements remained evident after adjusting for density, indicating that US improved the densification effect on strength and stiffness rather than just increasing density.

TM consistently reduced bending strength and, to a lesser extent, bending stiffness across all compression ratios. These reductions remained after adjusting for density, demonstrating that thermal modification alters the mechanical potential of densified wood despite its beneficial effect on reducing equilibrium moisture content. The combined US and TM treatment resulted in intermediate mechanical performance, suggesting that US can partly offset the strength loss caused by TM treatment.

Overall, the bending performance of THM-densified poplar wood is controlled not only by compression ratio and density but also by the interaction between densification and auxiliary treatments. US offers a promising route for improving densification efficiency at moderate compression levels, whereas TM creates a trade-off between mechanical performance and moisture-related stability. Higher compression ratios (CR = 60%–65%) provide the greatest absolute increases in strength and stiffness but are associated with greater material compaction and yield loss. Consequently, the optimal treatment strategy depends on balancing maximum mechanical performance with material efficiency and application-specific requirements.