Disrupted cellulose aggregation leads to the reduced mechanical performance of wood–adhesive interphase during freeze–thaw cycles

The mechanical performance of engineered wooden products facing the freeze–thaw cycles (FTCs) arises as an attention-worthy issue since the application of timber architectures in cold climates spreads. Here, we reported an investigation to reveal the losses of the mechanical performance of the wood-phenol formaldehyde (PF) adhesive interphase after the FTCs. Results revealed that PF adhesive was barely affected by the FTCs due to the low moisture content and rigid networks, whereas the mechanical properties of the cell wall in wood-PF interphase reduced significantly (more than 30%) after 5 FTCs at – 40 °C. Cracks were observed in the cell wall and compound middle lamella after FTCs. Further investigation into the crystal structure of the cell wall in the wood-PF interphase demonstrated that the FTCs disrupt the aggregations of cellulose macromolecules. The stresses caused by the phase transition of free water and the external hydrogen bonds formed between water and cellulose disrupted hydrogen bond networks in the cell wall. A plausible mechanism for the FTCs reducing the mechanical properties of the wood-PF bonds can be concluded as the cracks and weakened cell walls crippled the structural integrity of the wood-PF interphase, making it a fragile and stress-concentrated site when subjected to load.


Introduction
The timber architecture is promising worldwide due to its remarkable sustainability and efficiency (Cornwall 2016). The outstanding strength-to-weight ratio compared with steel and concrete makes timber architectures a sustainable solution for high-rises and longspans (Zhang et al. 2018). Engineered wooden products (EWPs), including glue-laminated timber (GLT), cross-laminated timber (CLT), and parallel-strand lumber (PSL) et al., are commonly used as load-bearing elements in timber architectures. As the timber architectures are advanced, the stability and safety of long-term application of EWPs draw continuous attention, especially regarding variant environmental aggressors (i.e., moisture content, light, and ambient temperature).
The stability of EWPs facing moisture, light exposure and high temperature have been investigated, advancing the relevant fundamental understandings. Water acts as a plasticizer, softening and thickening of wooden cell walls. Thus, the hydrogen bonds between microfibrils in the cell wall are easier to disrupt, degrading the mechanics of EWPs (Meng et al. 2015). Furthermore, the mechano-sorptive creep of EWPs causes large deformation in long-term loadbearing (Rindler et al. 2019, Socha et al. 2021, Uwizeyimana et al. 2022. Photo-induced discoloration and surface roughing on the surface of wood and EWPs are also revealed and attributed to the photo-induced oxidation and degradation of lignocellulose (Evans 2012, Janesch et al. 2020, Bisht et al. 2021. Previous literatures also demonstrated the degradation in the mechanical performance of wood when exposed to high temperature (higher than 100 °C), which is attributed to the softening effects and the disrupted cell wall structure caused by the degradation of lignocellulose . The degradation of hemicellulose occurs at first as exposed to high temperature, followed by the one of cellulose and lignin. EWPs were normally utilized below 80 °C and the degradation of lignocellulose are minor under the circumstances, thus resulting in the neglectable losses in the mechanical performance (Gao et al. 2015, de Almeida et al. 2018). However, the mechanical performance of EWPs shows an evident reduction after freeze-thaw cycles (FTCs), as proved by recent research (Srubar 2015, Adibaskoro et al. 2021.
The phase transition of free water caused by FTCs may significantly impact the mechanical performance of the EWPs. Despite that, few efforts are applied to the specific mechanism, hindering the development of reliable EWPs in cold climates. One common-supported concept is that the mechanical performance of the EWPs is highly contingent on the wood-adhesive interphase (Herzele et al. 2018). Revealing the losses in the mechanical performance of wood-adhesive interphase during the FTCs is thus essential, considering the critical role of wood-adhesive interphase in the stress transferring of EWPs when subjected to load (Gindl et al. 2005, Jakes et al. 2015. However, the relevant investigation is limited. Generally, the losses in the mechanical performance of raw wood in the FTCs are related to the phase transition of free water and as-caused damage in microfibrils bundles (Szmutku et al. 2013, Lamason et al. 2014, Hunt et al. 2018, Zhao et al. 2020. The FTCs also play significant roles in the rapture of the crosslinked network of biopolymer-based adhesive, which may lead to losses in the mechanical performance of wood-adhesive interphase (Wang et al. 2013, Tran et al. 2020. Here, nanoindentation (NI), accompanied by an investigation of the chemical structures and morphologies of wood-adhesive interphase was applied. The objective of this work is to reveal the losses in the mechanical performance of wood-adhesive interphase and illustrate the origination of the damage caused by the FTCs (freeze temperature covers from 0 to -40 °C and cycles for 2-20 times). We believe that such investigation can carve a way to fundamentally understand the mechanism underlying the losses in the mechanical performance of wood-adhesive interphase caused by the FTCs. Thus, it can provide an opportunity to improve the frost resistance of EWPs.

Materials
Wood blocks (10 × 3 × 15 cm 3 ) were cut from the same defect-free region of Douglas fir (Pseudotsuga menziesii), and the moisture content was adjusted to 6% before further procedures. Phenol-formaldehyde (PF) adhesive was applied in this work. The viscosity and solid content of PF adhesive were measured as 153.2 mPa·s and 48.7%, respectively. Wood-PF adhesive bonds were prepared via a hot-press procedure (140 °C and 0.8 MPa for 40 min). The corresponding consumption of PF adhesive was set as 150 g m − 2 . As-prepared wood-PF adhesive bonds were then kept at constant temperature and relative humidity (25 °C and 55%) for 48 h to minimize the inner stress induced by the curing of PF adhesive. The moisture content of all prepared samples was 10.9% before the FTCs.

3
Vol.: (0123456789) The FTCs of wood-PF bonds According to previous literature ) and a standing standard (ASTM D6662-13), the FTCs were performed in a CDLJ-2 W ultra-low temperature freezer (Wuxi Guanya, China), and the desired freezing temperature was set as 0, -10, -20, and − 40 °C, respectively. Each FTC consists of the following three steps: (i) The samples were placed in distilled water (pH = 7.1) at 20 °C for 12 h; (ii) The samples were kept at the freezing temperatures for 12 h; and (iii) The samples were placed in the environment of 20 °C and 55% (relative humidity) for 12 h. As displayed in Fig. 1a, infrared thermography (H10, Hikmicrotech, China) was utilized to verify that the temperatures of samples meet the set. According to Fig. 1b and c, the temperature reduction rates of wood-PF bonds were determined as 17.4 °C h − 1 , 30.8 °C h − 1 , 34.6 ℃ h − 1 and 40.1 ℃ h − 1 , corresponding to different freeze temperatures. After FTCs, the samples were placed in a constant temperature and humidity chamber (LRH-70 F, Shanghai Yiheng scientific instruments, Co., Ltd., China) to adjust the moisture content to 10%, thus minimizing the undesired turbulence induced by varied moisture content on the mechanical performance of wood cell wall.

Mechanical performance of wood-PF interphase
A universal mechanical testing machine (Instron5960, USA) was utilized to characterize the shear strength of wood-PF bonds according to the Chinese National Standard (GB/T 9846-2015). The loading speed of all tests was set as 2 mm s − 1 . A schematic diagram of shearing strength analysis was also provided (Fig. 1a) for further clarifying the morphology and dimensions of the samples. An ultra-field depth optical microscope (VHX-500, Keyence, China) and a scanning electronic microscopy (SEM, SU8010, Hitachi, Japan) were utilized to evaluate the fracture sites of wood-PF bonds.
The mechanical performance of the cell walls and PF adhesive in the wood-PF interphase was evaluated by NI, and the samples for NI analysis (5 × 5 × 8 Fig. 1 a Schematic diagram of the FTCs of wood-PF bonds and the corresponding thermal images, which reflects the distribution of radiant temperature on the surface of wood-PF bonds. b The temperature curves at different freeze temperatures in a single FTC. c The cooling rate of wood-PF bonds at different freezing temperatures mm 3 ) were cut from as-prepared wood-PF bonds. An ultra-microtome (Leica MZ6, Germany) equipped with a diamond knife was applied to prepare smooth surfaces on the cross-section of wood-adhesive bonds according to previous literature (Jakes et al. 2008). The samples went through no embedding procedures. An iMicro Nanoindenter (Nanomechanics, USA) equipped with a Berkovich indenter was utilized to reveal the mechanical performance of cell wall and PF adhesive. According to previous literature, a NI Mapping methodology (NanoBlitz 3D) was applied in an area of 40 μm × 40 μm in wood-PF interphase . 40 × 40 indents were performed with a uniform indent depth (200 nm) in an array. The reduced elastic modulus (Er) and hardness (H) of each indent can be obtained according to the method as follows (Oliver and Pharr 1992): where S is the initial unloading stiffness, β is 1.034 for a Berkovich indenter, and A corresponds to the projected contact area at peak load. P max in Eq. 2 represents the peak load of indent.

Chemical structures and morphologies of wood-PF interphase
Fourier transform infrared spectroscopy (FT-IR) and X-Ray Diffraction (XRD) were performed to investigate the chemical and crystal structures of wood and PF adhesive after the FTCs, respectively. FT-IR analyses were performed to evaluate the FTCs-caused potential changes in the chemical structures of wood and the PF adhesive, respectively, and all spectra were recorded using a Nexus870 (Nicolet, USA) FT-IR spectroscopy. For each sample, sixteen scans were conducted from 4000 to 500 cm − 1 at a resolution of 4 cm − 1 .
A smartlab9 X-ray diffractometer (Rigaku, Japan) equipped with the Ni-filtered Cu Kα1 radiation (λ = 1.542 Å) was applied to reveal the potential effects of FTCs on the crystalline structures of cured PF adhesive and wood, operating at 40 kV, 40 mA and 0.02° s − 1 (2θ value ranging from 3 to 60°). The XRD analysis was conducted in a reflection mode. All samples were powder and contained in the holder of 8 mm diameter and 1 mm thickness. PF adhesive was cured at 140 °C for 30 min in an oven (PH-010 A, Shanghai Yiheng scientific instruments, Co., Ltd., China) and then milled to 60 mesh in a micro plant grinding machine (FT102, Hangzhou Aipu scientific instruments, Co., Ltd., China). Wood was conditioned to a moisture content of 10% for 7 d using a constant temperature and humidity chamber (LRH-70 F, Shanghai Yiheng scientific instruments, Co., Ltd., China), and then was also milled to 60 mesh. The methodology described by Segal et al. (1959) was applied to calculate the crystallinity index (CrI) of cell wall as shown in Eq. 3, which requires the subtraction of background signal. Here, a blank XRD test with the empty sample holder was performed prior to the formal tests with samples. Data obtained from the blank test was seen as the background signal which is subtracted from the XRD signal with samples using JADE 6.5 software.
where I 200 represents the height ratio of the intensity of 22.8° (corresponding to 200 crystal plane of cellulose) (Hoyos et al. 2020). It should be noted that the intensity of the amorphous region of cellulose (I Am ) was obtained from the minimum intensity between the overlapped peaks of (110) and (200) (French  2020).
The crosslinked structure of PF adhesive was also revealed by a thermogravimetric analysis (TGA) using a 209 F3 Tarsus thermal gravimetric analyzer (Netzsch, Germany) from 30-800 °C at the heating rate of 10 °C min − 1 and nitrogen flow rate of 30 ml min − 1 . A SU8010 cold field emission electron microscopy (Hitachi, Japan) was applied to reveal the morphology of the wood-PF interphase after the FTCs.

Aggregation of cellulose macromolecules
Small-angle X-ray scattering (SAXS) was also utilized to further illustrate the cellulose aggregation in the cell walls of wood-PF interphase after the FTCs. Figure 2 represents the sample preparation for SAXS analysis.
As displayed, the original thickness of wood-PF bonds was 5 mm with the bond line locating in the middle of the sample. A wire-electrode cutting (DK7625, Sodick Inc., Japan) was used to reduce the thickness of wood-PF bonds layer by layer (a step of 100 μm) from the both sides till the desired thickness (~ 200 μm). Aforementioned preparation aims to minimize the undesired wood substrate and to obtain the information of wood-PF interphase which locates in the middle of the prepared sample. SAXS patterns were obtained using a D8 Advance X-ray diffractometer (Bruker, Germany) with a scattering vector (q) of 0.04-5 nm − 1 . The distance between samples and detector was maintained at 576 mm. FIT2D v.12.077 (Hammersley 1998) was used to calibrate and obtain the background-subtracted 1D scattering profiles, i.e., the scattering intensity I(q) vs. q. The scattering vector q is related to the scattering angle 2θ and the wavelength λ (for a Cu Kα1 radiation, λ = 1.542 Å) as determined by the Bragg-equation (q = 4π·sinθ/λ) (Jakob et al. 1996). The Guinier approximation at low q range allows one to determine the radius of gyration (R g ) of the nanocellulose crystal from the slope of a linear fit as shown in Eq. 4 (Guinier et al. 1964, Jungnikl et al. 2008. It should be noted that the slope fitting reveals well in the range of 0 < q 2 < 0.003 Å −2 , thus the same fitting range was used in this work. In the smaller q ranges (less than 0.035 Å −1 ), I(q) follows a Power law with a formula of I(q) = P·q −α , where P is the Porod constant and the exponent α is related to the surface roughness of cellulose fibrils (Rennhofer et al. 2021). Lower α indicates a rougher surface, with values towards 3 corresponding to a surface fractal and values between 2 and 3 corresponding to a mass fractal (Jakob et al. 1994). By using an adapted Porod evaluation, I(q) = P·q − 4 +C·q − 2 , the Porod constant P can be determined as the intercept of the linear fitting of q 4 ·I(q) = C·q 2 + P (Kohnke et al. 2016, Jafta et al. 2017).
The scattering invariant Q can be calculated according to Eq. 5, and it can be seen as the area below the socalled Kratky plot, i.e., q 2 ·I(q) vs. q (Lichtenegger et al. 1998). The distance between cellulose macromolecules (d) can be obtained from Eq. 6, and the q max represents the peak location in the Kratky plot using a parabola fit on the top of the peak (Virtanen et al. 2015).

Results and discussion
The FTCs-caused losses in the mechanical performance of wood-PF interphase Figure 3 displays the shear strength of wood-PF bonds after the FTCs. The shear strength of wood-PF bonds after the FTCs displays a more evident reduction ratio (more than 35%). This finding is in accordance with Fig. 2 The sample preparation procedures for SAXS analysis. The coordinate axis represents the longitudinal (L), tangential (T) and radial (R) directions of wood grain, respectively previous literature which also revealed the losses in the mechanical performance of glulam after the FTCs . In this work, the freezing temperature exhibits a more significant impact on the mechanical performance of wood-PF bonds than the cycle times. As displayed in Fig. 3c, an evident reduction can be found in the shear strength of wood-PF bonds after 20 FTCs at − 20 °C, whereas it only takes 5 FTCs at -40 °C to induce more significant losses in those mechanical performances. It can be attributed to the difference in the phase transition of free water and the formation of ice crystals at different freezing temperatures. Figure 1c represents the colling rate of wood-PF bonds at different freezing temperatures. The slow freezing of free water at 0 and − 20 °C leads to smaller ice crystals than at -40 °C, resulting in minor damage in the wood-PF interphase (Lamason et al. 2014, Zhao et al. 2021. Microscopy analysis also confirmed the evident fracture of wood-PF interphase after shearing ( Fig. 2d and e), yielding the wood-PF interphase fractures prior to the bulk wood when subjected to tensile. Previous research also demonstrated that the stress concentration occurs in the wood-adhesive interphase when subjected to tensile, leading to fracture of cell wall and the adhesive (Cao et al. 2021. Therefore, the reduction in the shear strength of wood-PF bonds originated from the crippled mechanical performance of the wood-PF interphase. SEM and NI were applied to evaluate the morphologies and the mechanical performance (Er and H) of the wood-PF interphase as displayed in Fig. 4. The raw wood-PF interphase exhibits smooth and intact morphology (Fig. 4a1), whereas cracks can be found in the compound middle lamella (CML) at the wood-PF interphases after 20 FTCs at -20 °C (Fig. 4a2). SEM analysis demonstrated the location of cracks verified in a corresponding manner of wood microstructure. Most cracks were observed in the CML, which can be attributed to the difference in the hydrophilicity of polysaccharide and lignin. Cellulose and hemicellulose exhibit notable hydrophilicity, thus stabilizing the surround water from frozen by forming hydrogen bonds (Barnat-Hunek et al. 2019). In contrast, the hydrophobicity nature of lignin makes itself as an active ice nucleation, accelerating the formation of ice crystal surrounding lignin macromolecules. It cripples the plasticity at low temperature and leads to the generation of cracks in the lignin-rich CML (Bogler andBorduas-Dedekind 2020, Zeng et al. 2022). Cracks weaken the structural integrity of the wood-PF interphase, leading to stress concentration and the fracture of wood-PF bonds when subjected to load. As shown in Fig. 4a3, 5 FTCs at -40 °C cause the prolongation of cracks in the wood-PF interphase, further weakening its structural integrity. Thus, wood-PF bonds possess the lowest macro-mechanical performance after 5 FTCs at -40 °C.
The investigation into the mechanical performance of wood-PF interphase revealed more information about the damage induced by the FTCs. Figure 4b1 and c1 illustrate that the Er and H values of the cell wall contacts with PF adhesive in the raw wood-PF interphase are 21.1 GPa and 902.8 MPa, respectively. The corresponding mechanical performance reduced rapidly to 17.2 GPa (Er) and 725.7 MPa (H) after 20 FTCs at -20 °C. The mechanical performance of the cell wall with PF adhesive infiltration are 16.7 GPa (Er) and 689.2 MPa (H) after 5 FTCs at -40 °C, respectively ( Fig. 4b3 and c3). This finding demonstrates that the rapid freezing rate leads to severe losses in the mechanical performance of the wood-PF interphase, turning the wood-PF interphase into a fragile and easy-fractured site. It has been also proved by former microscopy analysis that wood-PF interphase fractured prior to the bulk wood.
An interesting finding is that the FTCs exhibit minor impact on the mechanical performance of PF adhesive. It can be attributed to the lower moisture content in PF adhesive (0.7%) than in wood (10.7%) before the FTCs, limiting the phase transition of free water. The 3D structure in PF adhesive serves as the unactive ice nucleation, restricting the formation of ice crystal. Furthermore, the 3D rigid structure in PF adhesive also imparts itself the resistance to the shrinkage/expanding stress caused by the phase transition of free water (Xing et al. 2021). Figure 5 also reveals that the FTCs lead to the severe losses in the mechanical performance of the cell walls without PF adhesive infiltration. The corresponding Er and H are 15.4 GPa and 586.1 MPa, whereas 20 FTCs at -20 °C reduced those mechanical performances to 13.3 GPa and 461.5 MPa. Compared to the cell walls with PF adhesive infiltration, the FTCs weaken the mechanical performance of the wood cell wall more significantly. The interpenetrating polymer networks (IPNs) formed in the cell walls with PF adhesive infiltration prevent the contact of free water and elevate 1 3 Vol.: (0123456789) Fig. 3 (a) The schematic diagram of the shear strength analysis. The shear strength of wood-PF bonds after FTCs at different freeze temperatures (b) and after 2-20 FTCs (c). The optical and SEM images of the fracture of wood-PF interphase before (d) and after FTCs (e) the resistance to frost . The failure of the wood-PF interphase after FTCs originates from the rapid disruption of the wood-PF interphase, followed by the deconstruction of wood.

Chemical and crystal structures of PF adhesive and cell wall after the FTCs
The former analysis demonstrates a slight impact of the FTCs on the mechanical performance of PF adhesive. To further illustrate the mechanism, the chemical structures of PF adhesive were investigated (Fig. 6). As displayed in Fig. 6a, the vibration peaks of C = C covalent bonds in phenol rings, PhOH, and aliphatic C-O covalent bonds can be found at 1610 − 1510 cm − 1 , 1270 − 1130 cm − 1 , and 1005 cm − 1 of FT-IR spectra, respectively . Few changes in the intensity of these vibration peaks can be observed, indicating the limited structural deformation of PF adhesive after the FTCs. TGA curves of PF adhesive can also supplement this finding (Fig. 6b). The carbon residues corresponding to PF adhesive after 20 FTCs at -20 °C and 5 FTCs at -40 °C are 61.2% and 60.9%, respectively, showing a slight reduction from the raw PF adhesive (63.6%). XRD curves in Fig. 6c reveal that PF adhesive holds an amorphous physical conformation during the FTCs. The results demonstrate that PF adhesive exhibits excellent resistance to the phase transition of free water, and its mechanical performance can be maintained after the FTCs.
FT-IR and XRD were also utilized to provide insights into the damage of the FTCs to the chemical and crystal structures of cell walls. As displayed Fig. 4 The SEM images, Er mapping, and H mapping corresponding to the raw wood-PF interphase (a1-c1), wood-PF interphase after 20 FTCs at -20 °C (a2-c2), and wood-PF interphase after 5 FTCs at -40 °C (a3-c3) respectively in Fig. 7a, the vibration peaks located at 3393 and 2902 cm − 1 are associated with the hydroxyl groups and C-H covalent bonds, whereas the characterized vibration peaks corresponding to lignin and polysaccharide (cellulose and hemicellulose) can be observed at 1639, 1510, 1454, 1269 and 1168 cm − 1 (Schnabel et al. 2015, Esfandiar et al. 2020. The comparison between FT-IR spectra of wood after the FTCs reveals no significant variant, indicating the chemical structures of the cell wall remain unaffected. However, the XRD patterns of wood after the FTCs (Fig. 7b) suggest a reduced relative crystallinity during the FTCs. The peaks located at 14.9°, 22.8°, and 33.4° correspond to the crystal planes of (110), (200), and (004) in cellulose macromolecules, respectively (Wen et al. 2017). The CrI of wood cell wall reduces from 62.8% (raw) to 52.6% after 20 FTCs at -20 °C, and the CrI corresponds to wood cell wall after 5 FTCs at -40 °C is 44.7% ( Fig. 7b and d). This finding indicates that the FTCs exhibit significant effects on the crystal structures of the cell wall, further resulting in the reduction in the mechanical performance. One plausible trigger for the FTCs reducing the relative crystallinity of cell walls is that the phase transition of free water causes stress in the amorphous region of cellulose macromolecules in the elementary fibril bundles, disrupting the original hydrogen bond networks (Sun et al. 2022). External hydrogen bonds and van der Waals interactions between free water and cellulose also lead to the destruction of hydrogen bond networks inside cellulose macromolecules during the FTCs (Li et al. 2020). Dispersion of cellulose macromolecules after the FTCs SAXS was utilized to provide further information about the cellulose aggregation in cell walls after the FTCs, and the results are displayed in Fig. 8. The SAXS patterns (Fig. 8a1-a3) are alike, indicating the orientation of cellulose macromolecule remains unaffected after the FTCs. The shoulder peak located at q = 0.1 Å −1 in Fig. 8b1 is associated with the aggregation structures of cellulose macromolecules, and a reduction in the corresponding SAXS intensities  et al. 2021). Further calculating the slopes of SAXS intensity-q curves yields the radius of gyration (R g ) and the grain size (L) of the nanocellulose crystal in the cell wall (Penttila et al. 2013). It should be noted that the size of the crystal region in the cellulose macromolecules is much larger than the detection range of SAXS. Thus, the obtained grain size reflects the thickness of nanocellulose crystal in elementary fibril bundles in the cell wall (Virtanen et al. 2015). As displayed in Table 1, the grain size of nanocellulose crystal in the raw wood-PF interphase is 179.4 Å, and it reduces to 145.1 and 135.8 Å after 20 FTCs at -20 °C and 5 FTCs at -40℃, respectively. This finding suggests that the FTCs disperse the cellulose aggregation in the wood-PF interphase, thus reducing the cell wall's relative crystallinity and mechanical performance.
The dispersion in the cellulose macromolecules caused by the FTCs can be further supported by the Porod law (Fig. 8b2). Calculated fractal dimensions corresponding to all wood-PF interphase are less than 3.0 (Table 1), indicating mass fractal for the dispersion of cellulose macromolecules (Penttila et al. 2013). The FTCs lead to an evident elevation in the fractal dimension, suggesting the dispersed cellulose macromolecules in the wood-PF interphase, especially after severe conditions (5 FTCs at -40 °C). A peak located at 0.15 Å −1 can be observed in the Kratky plot (Fig. 8b3), and it relates to the distance between cellulose macromolecules (d). As listed in  SAXS analysis confirms the finding obtained from XRD analysis, i.e., the FTCs lead to the reduced relative crystallinity in cell walls. A plausible explanation for that can also be concluded based on SAXS analysis. As displayed in Fig. 8c, the phase transition of free water can result in stress in the cellulose macromolecules within the elementary fibril bundles. Meanwhile, external hydrogen bonds and van der Waals interactions between water and cellulose further accelerate the dispersion of cellulose macromolecules, displaying elevation in the fractal dimensions and intermolecular distance of cellulose. Dispersed cellulose macromolecules lead to the disrupted elementary fibril bundles in the cell wall and reduced  Table 1 Radius of gyration (R g ), the grain size of the nanocellulose crystal (L), fractal dimension, and distances between cellulose macromolecules (d) of the cell wall in wood-PF interphase samples a The letter m in the brackets represents the fractal behavior of the cellulose macromolecules corresponding to mass fractal mechanical performance of the cell wall and wood-PF interface after the FTCs.

Conclusion
In this work, the losses in the mechanical performance of wood-PF interphase caused by the FTCs were revealed, and the specific mechanism was also illustrated. The FTCs exhibited slight effects on the structures and mechanical properties of PF adhesive, whereas they reduced the mechanical properties of the cell walls in the wood-PF interphase. It is the primary reason underlying the evident losses in the shear strength of wood-PF bonds (more than 30% after 5 FTCs at -40 °C). Cracks were also observed in the wood-PF interphase after the FTCs, weakening the structural integrity. Both XRD and SAXS analyses revealed that the FTCs dispersed the cellulose aggregations. The stress in cellulose crystals caused by the phase transition of free water and the external hydrogen bonds between free water and cellulose led to disruption of the original hydrogen bond networks, thus reducing the mechanical properties of the cell wall. Freezing temperature exhibited a more significant effect on the mechanical properties of wood-PF interphase than cycle times. This work demonstrated the relationship between the FTCs and reduced mechanical properties of wood-PF interphase. A reliable explanation was that the FTCs disrupted the cellulose aggregations in the cell wall, making wood-PF interphase fragile facing stress concentration and fracture prior to the bulk wood.