Variable-temperature solid-state NMR analysis of woody materials in the presence of small hydroxyl molecules

Abstract Interactions between the hierarchical structure of Japanese cypress and small hydroxylic solvents with high polarity were studied using variable-temperature 13C cross-polarization and magic-angle spinning nuclear magnetic resonance (CP/MAS) NMR spectroscopy, including low-temperature measurements. 13C CP/MAS NMR spectra were enhanced by dipolar interactions closely related to 1H–13C magnetization transfer, providing information about the interaction between woody materials and small hydroxyl molecules. The change in interaction was particularly near the melting point of the hydroxylic solvent, where a decrease in the intensity of the cypress carbohydrate signal and the appearance of the signal of solvent molecules were observed. The trend of signal intensity in the variable-temperature 13C CP/MAS NMR spectra of cypress and cellulose fibers in the presence of hydroxylic solvents indicated that the decrease in the carbohydrate signal intensity due to the transfer of magnetization near the melting point of the hydroxyl molecule is greater when the size of the hydroxyl molecule is smaller. Furthermore, molecular association in the nanopores of the hierarchical structure allowed the hydroxyl molecular signals of the impregnated cypress to appear above the melting point.


Introduction
Owing to the soaring cost of fossil fuels and concerns about carbon dioxide emissions, it is necessary to promote the use of biomass resources for a sound material-cycle society. Woody materials are typical biomass resources; however, in order to use them as industrial products, many aspects require improvement, such as moisture sensitivity, hardness, wear resistance, and low dimensional stability. Hence, wood-based materials are modified to mitigate these shortcomings, and the modification techniques include: (1) chemical treatment, (2) thermo-hydro and thermo-hydro-mechanical treatment, (3) treatment based on biological processes, (4) physical treatment using electromagnetic irradiation and plasma, [1] and (5) mechanical treatment with a grinder. [2] Raw materials with excellent properties can be produced via chemical modification of wood-based materials and their biomass constituents. Therefore, studies on the physical, mechanical, and biological properties of chemically modified wood are being actively conducted. [3] The chemical modification of wood-based materials is classified into non-reactive and reactive cell wall modifications. Conventional modification techniques include acetylation, furfuryl alcoholation, thermosetting resin modification, polymer monomer modification, and paraffin wax modification. [4] Unlike conventional industrial products, wood-based materials have a complex hierarchical structure and composition. At the microscale and nanoscale, wood-based materials consist of numerous cells with multilayered lignocellulosic cell walls that are divided into microfibrils, nanofibrils, and elementary fibrils. These multilayered lignocellulosic fibrils are composed of linear cellulose chains (40-50%), hemicellulose (10-30%), and lignin (20-30%) on the molecular scale. [5] The assembled linear cellulose chains are covered by hemicellulose to form cellulose-hemicellulose aggregates, which are then covered by a three-dimensional lignin network to form lignin-carbohydrate complexes. [6] The hierarchical structure of wood-based materials described above must be considered when performing chemical modifications. The hierarchical structure of wood-based materials is relatively well maintained when the chemical modification occurs inside the cell wall [4] as well as when in situ polymerization fills the lumen and nanoparticles fill the cell wall. [7] However, the hierarchical structure of wood-based materials is disrupted by the cleavage of glycosidic bonds in the cellulose chain via hydrolysis, oxidation, bacteria, and enzymes. [8,9] The behavior of small molecules in the nanopores of wood-based materials plays an important role in chemical modification, particularly when the hierarchical structure is maintained. Their behavior also affects the properties of the modified materials. Moisture in wood-based materials can be present in the cell wall materials (at the nanoscale and molecular scale) within the hygroscopic moisture range [10] and assume a capillary structure (at the microscale) in the over-hygroscopic moisture range. To understand moisture in wood, the equilibrium and sorption hysteresis of the moisture state and the expansion and contraction due to moisture have been investigated using unmodified and chemically modified wood. To understand molecular association between water and wood, micro-FTIR spectroscopy has been applied to investigate molecular structure of desorbed water in heat-treated wood during moisture desorption process. [11] Because polymer components (cellulose, hemicellulose, and lignin) behave like typical solid polymers, a phenomenological polymer science approach is useful for understanding the effects of moisture on diffusion. This indicates that the high swelling pressures in the cell walls are due to the hierarchical structure of wood. [12] In addition to the phenomenological approach, instrumental analytical methods have been used to evaluate the molecular behavior of the cell walls with hierarchical structures. Raman imaging has been used to monitor the distribution of biodegradable polymers grafted onto the cell wall after chemical modification [13] and the reduction in lignin due to delignification. [14] X-ray and neutron scattering have been used to study the effects of moisture changes on the nanostructure of wood, showing the dependence of the nanostructure on the type of wood. [15] In recent years, solid-state nuclear magnetic resonance (NMR) spectroscopy has expanded its application range to complex biological systems through the use of ultrafast magic angle spinning (MAS) [16] and 2D dipolar-assisted rotational resonance (DARR) techniques. [17] Solid-state NMR is also a powerful technique for investigating sustainable and renewable cellulose-based materials [18] and alkyl-O-aryl bond determination in lignin preparations. [19] The reproducibility of the raw material composition is crucial for the industrial use of wood. Solid-state NMR and infrared characterizations show that the molecular mobility and homogeneity of the constituent polymers in wood samples of the same species from different provenances led to differences in the chemical and physical properties. [20] To evaluate the moisture content in the wood, one-and two-dimensional 1 H NMR relaxometry techniques have been used to determine the dry mass of the wood. [21] Recently, a combination of multidimensional 13 C MAS NMR and molecular dynamics modeling has revealed the role of water in the molecular architecture of softwood secondary cell walls. [22] The multidimensional solid-state NMR was effective in analyzing the polymer interactions in cell walls of the softwood to elucidate the molecular architecture derived from the different cellulose environments. [23] Furthermore, T 1ρ H relaxation time and two-dimensional 13 C-1 H correlation (FSLG CP HETCOR) NMR experiments have revealed specific interactions between the aromatic part of the resin and the lignin, as well as a network of hydrogen bonds involving all components of the system. [24] We previously reported the chemical changes and physical disruption of vascular bundle units in steam-heated bamboo using a new multi-scale instrumental analysis technique that combined several solid-state NMR methods. [25] Regarding the hierarchical structure, multi-scale instrumental analysis using solid-state NMR methods revealed interactions between water molecules and the nanostructure of the delignified and hemicellulose-free Japanese cypress. [26] We also reported that the impregnation of Japanese cedar with phenol formaldehyde resin and moisture affects the solid-state NMR spectra, usually 13 C cross-polarization and magic-angle spinning NMR ( 13 C CP/MAS NMR), and the relaxation times typically in the 13 C CP/MAS NMR spectra. [27] Furthermore, in Japanese cypresses impregnated with polyethylene glycols (PEG) of various sizes, the hierarchical structure brought significant spectral changes in solid-state NMR spectra around the melting point of PEG due to interactions within the nanopores of the cell wall and lumen. [28] Based on the results of our previous studies described above, this study investigated the behavior of small hydroxylic solvents with high polarity in cellulose isolated from Japanese cypress using variable-temperature 13 C CP/MAS NMR spectroscopy. As reported previously, the rheological tests showed that the wood impregnated with various organic solvents, such as ethylene glycol (EG), dimethyl formamide, and dimethyl sulfoxide, have unique solvent relaxation processes. [29] Based on our previous rheological results, this study used low-temperature 13 C CP/MAS NMR spectroscopy to study woody materials impregnated with small hydroxylic solvents. The 13 C CP/MAS NMR method reflects dipolar interactions closely related to 1 H-13 C magnetization transfer, which are difficult to observe with other spectroscopic methods. In particular, the dipolar interaction between woody materials and small hydroxyl molecules is greater when their interaction is greater, thereby enhancing the signal intensities. Using the 13 C CP/ MAS NMR method to reflect the interaction of woody materials, we analyzed the effects of the hierarchical structure of woody materials and the molecular size of small hydroxylic solvents around their melting points. We also investigated the possibility of using polar solvents as molecular probes for 13 C CP/MAS NMR spectroscopy.

Materials
Disks with a diameter of 3 mm were prepared from 1 mm-thick sapwood slices of Japanese cypress (Hinoki, Chamaecyparis obtusa) from Gifu Prefecture, Japan, using a perforator-hole puncher to obtain wood samples. The wood disk samples were heated in hot water at 80 °C for 1 h, dried at ambient temperature for one week, and then dried in a vacuum dryer at 80 °C for 24 h before impregnation with solvent molecules to remove growth stress. Cellulose fibers were obtained by pulping a mixture of 50% lodgepole pine (Pinus contorta), 40% white spruce (Pinus glauca), and 10% Douglas fir (Pinus menziesii). Small hydroxylic solvents (EG, diethylene glycol (DEG), and methanol) were purchased from Wako Pure Chemical Co. Ltd. (Osaka, Japan) and used without further purification. The dried wood disks and cellulose fibers were wrapped in non-woven filters and introduced into the solvent. The impregnated samples were depressurized with a vacuum pump and left undisturbed for one week.

Solid-state NMR spectrum measurements
MAS NMR spectra were obtained using a Varian 400 MHz NMR spectrometer (Palo Alto, USA) with a Varian T3 double-resonance 4 mm solid-state probe. The samples were loaded into a 4-mm liquid ZrO 2 rotor, except for dry cypress and cellulose that were loaded into a 4-mm regular ZrO 2 rotor. The sample rotor was spun at 10 kHz; the moisture-absorbing cypress was spun at 5 kHz, while the dry cypress and cellulose were spun at 15 kHz. The temperature of the sample rotor was manually controlled using the variable-temperature unit of the NMR instrument. The variable temperature measurement began 15 min after the temperature became constant but was limited to 128 transient delays owing to instrument constraints in low-temperature measurements. The 13 C MAS NMR spectra were acquired for the 13 C nuclei with 2.6 µs π/2 pulse at 100.56 MHz and a 40 ms acquisition period over a 30.7 kHz spectral width. Proton decoupling was performed with an RF field strength of 86 kHz and a small phase incremental alteration decoupling pulse sequence. [30] CP/MAS NMR spectra were measured with a 5.0 s recycle delay using a ramped-amplitude pulse sequence [31] with a 2 ms contact time and a 2.5 µs π/2 pulse for the 1 H nuclei. The amplitude of the 1 H nuclei was linearly increased from 90.5% of its final value during the cross-polarization contact time. The contact time for CP/MAS was optimized by measuring several spectra with various contact times (100-8000 μs). All data acquisition and processing were performed using the software provided with the NMR instrument, and all spectra were obtained by applying a 30 Hz line-broadening to the original data. All comparisons between the spectra were performed visually, without peak deconvolution.

Results and discussion
In our previous study, we reported the changes in the signal intensity in the variable-temperature 13 C CP/ MAS NMR spectra of Japanese cypress plates. [28] In this study, the signal intensities of the pretreated cypress plates decreased with increasing temperature, whereas those of the PEG 200-impregnated plates showed a small change. A decrease in the signal intensity due to increasing temperature was also observed in the 13 C CP/MAS NMR spectrum of dried cypress, as shown in Figure S1 (Supporting Information). The signal changes in 13 C CP/MAS NMR spectra of cypress disks, which were sufficiently dried to exclude the effects of moisture, observed in this study were similar to those observed when the pretreated cypress disks were cooled after increasing the temperature in the previous study. However, the untreated and chemically treated Japanese cypress disks under humid conditions showed stronger signals in 13 C CP/MAS NMR spectra than those that were air-dried and heat-dried. [26] The results shown in Figure S1 were in good agreement with our previous findings, indicating that the water molecules significantly influence the interactions of cypress constituent polymers during CP/MAS NMR spectroscopy. To investigate the state and environmental impact of the water molecules on the interactions in the constituent polymers, we examined the variable-temperature 13 C CP/MAS NMR spectra of Japanese cypress in the presence of water molecules using a liquid rotor. Figure 1 shows the variable-temperature 13 C CP/ MAS NMR spectra of cypress in the presence of water molecules between −50 and 80 °C. High-temperature (25-80 °C) measurements were performed by increasing the temperature from 25 °C, while low-temperature (−50 to 0 °C) measurements were conducted by increasing the temperature from −50 °C. Each signal was assigned based on the results of our previous studies, [25,26] and the assignments are shown for each spectrum in Figure 1.
Lignin signals were observed at 56 ppm (OCH 3 ) and 110-160 ppm (aromatic and olefinic carbons), respectively, from the carbohydrate signals (60-110 ppm). The cellulose and hemicellulose signals overlapped, except for C4 (80-90 ppm). The crystalline (c) and amorphous (a) signals were assigned to cellulose C4 and C6, respectively. Unlike the dried cypress disks where the signal intensities decreased with increasing temperature (Figure S1), the signal intensities of the water-impregnated cypress disks increased from 25 °C to 60 °C and then remained constant until 80 °C (Figure 1). The signal intensities of the water-impregnated cypress at low (−50 °C) and high (60 °C) temperatures were similar. The intensities of the signals at −50 °C rapidly decreased with increasing temperature up to 0 °C and then increased at 25 °C; in particular, the signal intensity was the weakest at 0 °C in the variable-temperature measurement. The ratio of the crystalline (c) to amorphous (a) signals for cellulose C4 and C6 in water-impregnated and dried cypress disks monotonically decreased with increasing temperature. [28] Variable-temperature 13 C CP/MAS NMR spectra of Japanese cypress in the presence of EG molecules were obtained using a liquid rotor to determine the effect of solvent size on the signal intensity. Figure 2 shows the variable-temperature 13 C CP/MAS NMR spectra of Japanese cypress in the presence of EG. As in the case of water, high-temperature (25-80 °C) and low-temperature (−50 to 0 °C) measurements were individually performed. In Japanese cypress, the carbohydrate signals were the weakest at −20 °C near the  melting point of EG (-13 °C). Above −20 °C, the carbohydrate signal intensities increased with increasing temperature up to 60 °C and then decreased at 80 °C. Simultaneously, although the lignin portion in the dry cypress disk were observed at all measured temperatures ( Figure S1), the lignin portion of the EG-impregnated cypress disk exhibited very weak signals only at −50 °C. The EG signal was evident at −20 °C; the signal intensity increased to 25 °C and then decreased when the temperature exceeded 40 °C. Because of the cross-polarization effect of 13 C CP/ MAS NMR spectroscopy, molecules in the liquid state cannot be detected on their own but only when they interact with solid molecules. [28] The signal changes observed in this study for EG-impregnated cypress in variable-temperature 13 C CP/MAS NMR were in good agreement with our previous results. 13 C CP/MAS NMR spectra of cellulose fibers (no hierarchical structure) and mesoporous silica (periodically repeated nanostructure) in the presence of EG were obtained at low temperatures near the melting point of EG to investigate the effects of the cypress hierarchical structure on EG interaction. Unlike dried cypress, where the signal intensities decreased with increasing temperature, the cellulose fiber signals gradually increased with increasing temperature during 13 C CP/MAS NMR spectroscopy, as shown in Figure S2 (Supporting Information). As shown in Figure 3, variable-temperature 13 C CP/MAS NMR spectra of cellulose fibers were measured only at low temperatures, where remarkable spectral changes were observed in the dried cypress. Despite the different trends of dried cellulose fibers and dried cypress, the signal intensities of 13 C CP/MAS NMR spectra of cellulose fibers were also the weakest at −20 °C. However, the decrease in the signal intensity of cellulose at −20 °C was smaller than that of the carbohydrates in cypress. Furthermore, the CH 2 signal of EG appeared at −50 °C and decreased monotonically with increasing temperature. Thus, CP/MAS spectroscopy revealed that cellulose had a weaker dipolar interaction with EG than with cypress, particularly at ambient temperatures where EG was liquefied.
MCM-8M, hydroxyl-covered mesoporous silica with a pore size of approximately 8 nm, was used in this study. Figure S3 illustrates the low-temperature 13 C CP/MAS NMR spectra of MCM-8M in the presence of EG (Supporting Information). Although only trace EG signals were observed at −30 °C, the EG signal in mesoporous silica became evident at −20 °C and then slightly decreased at 25 °C. In the case of EG-impregnated Japanese cypress, although the EG signal appeared at low temperatures (−50 °C), an EG signal was also observed above the melting point. In the case of cellulose fiber, a small signal was observed near the melting point (−20 °C). Because cypress and mesoporous silica have nanopores with hydrogen groups, the EG molecules bonded with each other in their nanopores. Dipolar interactions between the bound molecules can cross-polarize 1 H nuclei to 13 C nuclei, similar to the solid phase, resulting in 13 C CP/ MAS signals even above the melting point. Because the cellulose fibers had incomplete nanopores due to pulping, EG bonding was weak, and strong EG signals were observed only below the melting point.
We then studied DEG, which has a longer molecular backbone than that of EG. Figure 4 shows the low-temperature 13 C CP/MAS NMR spectra of Japanese cypress in the presence of the DEG. The DEG signal of the impregnated cypress was evident at 0 °C, and this signal of the small hydroxyl molecules near the melting point (DEG: −7 °C) was also observed in 13 C CP/MAS NMR spectra of the EG-impregnated cypress, as shown previously. Furthermore, our previous study showed that PEG signals appear above their melting points in 13 C CP/ MAS NMR spectra of the PEG-impregnated cypress plates. [28] The intensities of the carbohydrate signals in the DEG-impregnated cypress were weakest at −20 °C, which increased with increasing temperature. However, the decrease in the carbohydrate signal intensity near the melting point of the DEG was smaller than that of water and EG. In our previous study on longer glycol molecules, [28] the carbohydrate signal intensity decreased slightly with increasing temperature, which was independent of the melting point of impregnated PEG, unlike EG or DEG, as shown above. These results indicate that the intensity of the carbohydrate signal in the cypress at the melting point of glycol is greater when the molecular size of glycol is smaller.
We then investigated the effects of methanol, which is smaller than glycol compounds. Figure 5 shows the variable-temperature 13 C CP/MAS NMR spectra of cypress in the presence of methanol. Because of instrumental limitations, the low-temperature 13 C CP/ MAS NMR spectrum was measured above −50 °C instead of near the melting point of methanol (-98 °C). Nevertheless, the carbohydrates in the methanolimpregnated cypress appeared as relatively weak signals at −50 °C, and their intensities increased with increasing temperature up to 0 °C. Figure 6 shows the variable-temperature 13 C CP/MAS NMR spectra of the cellulose fibers in methanol. The carbohydrate signals exhibited low intensities at −50 °C, which then increased with increasing temperature, similar to those of the impregnated cypress. A decrease in the carbohydrate signal intensity due to methanol addition was also observed for EG addition, although the cellulose fibers had no hierarchical structure. In particular, the intensities of the carbohydrate signals decreased for cypress and cellulose fibers with the addition of small hydroxyl molecules. The height of the methanol CH 3 signal in the 13 C CP/MAS spectrum of the methanol-impregnated cypress was almost identical to that of the cellulose C4 at −50 °C; the methanol signal decreased with increasing temperature and almost disappeared at 0 °C ( Figure 5). However, the methanol CH 3 in the cellulose fibers only appeared as a trace signal ( Figure 6). The low signal intensity of the hydroxylic solvent was similar to that of the EG-impregnated cellulose. To summarize the variable-temperature 13 C CP/MAS NMR spectra of   the small hydroxyl-molecule-impregnated materials in the present study, the aliphatic group signals did not appear in the cellulose fibers but appeared in cypress, which had nanopores due to the hierarchical structure of wood, and mesoporous silica.
As shown, the decrease in the signal intensity of the bulk constituents near the melting point of the solvent is greater when the molecular size of the hydroxylic solvent is smaller. This phenomenon can be explained by the relationship between the melting energy of the hydroxyl molecule and the excitation energy of the bulk constituent irradiated by radiofrequency pulses during 13 C CP/MAS NMR spectroscopy, as shown in Figure 7. As reported in our previous study, [28] for large molecules relative to the hydroxyl groups, such as PEG, the correlation time of the mobility is long. Because the magnetization of the bulk constituents is not transferred to the melting energy of larger hydroxyl molecules, a slight decrease is observed in the 13 C CP/MAS signal near the melting point of the compound ( Figure 7A). However, the hydroxylic solvents investigated in this study had short molecular correlation times and strong hydrogen bonds. Therefore, magnetization transfer from the bulk constituents to the melting energy of smaller hydroxyl molecules is efficient, significantly decreasing the signal near the melting points ( Figure 7B).
Based on the above results and discussion and publications on the hierarchical structure of woody materials, [5,6] we propose the possible magnetic behaviors of the small hydroxyl molecules in woody materials ( Figure 8). Because cellulose fibers do not have nanopores ( Figure 8A), the small hydroxyl molecules in the cellulose fibers are less associative than molecules in confined spaces. Lower molecular association allows CP/MAS spectroscopy to detect signals in the solid state with low mobility; however, it is difficult to detect signals in the liquid state with high mobility. Therefore, the CP/MAS signals of the small hydroxyl molecules in cellulose increased in intensity at low temperatures and disappeared above their melting points. In materials with hierarchical structures ( Figure  8B), the small hydroxyl molecules were trapped in the nanopores and behaved like solids, even above the melting point; the CP/MAS signals are detected because of the higher molecular association in the nanopores. However, as the temperature increased, the mobility increased, the material behaved like a liquid even in the nanopores, and the CP/MAS signal disappeared. Below the melting point, the small hydroxyl molecules interacted closely with the constituents in the nanopores and behaved magnetically as a part of the constituents. Therefore, cross-polarization does not occur in the small hydroxyl molecules, but in the bulk constituents, and the CP/MAS signals of the small hydroxyl molecules can hardly be detected.  As mentioned above, our new approach to low-temperature 13 C CP/MAS NMR using small hydroxylic solvents with high polarity can be used to study the hierarchical structures of woody materials. In particular, this approach is useful for evaluating the nanostructural changes caused by the chemical modification of wood and wood-based materials using small hydroxylic solvents as molecular probes in 13 C CP/MAS NMR spectroscopy. In the future, we aim to investigate manufacturing processes that utilize wood as a natural material to reveal the hierarchical structure and nanostructure of each chemical and mechanical process.

Conclusions
In this study, impregnated Japanese cypress was analyzed at low temperatures near the melting points of the hydroxyl molecules using variable-temperature 13 C CP/MAS NMR spectroscopy to investigate the behavior of the small hydroxyl molecules in woody materials. In the variable-temperature 13 C CP/MAS NMR spectra of impregnated cypresses, the carbohydrate signal intensity was the weakest at the melting point of the added hydroxy molecules. In the case of cellulose fibers with no hierarchical structure, although the carbohydrate signals were the weakest at the melting point, the intensity of the hydroxyl molecules monotonously decreased with increasing temperature. For mesoporous silica with a periodically repeating nanostructure, the hydroxyl molecular signal appeared above the melting point, which is similar to cypress with a hierarchical structure. 13 C CPMAS signal intensity is enhanced because of the magnetization transfer from protons to carbon-13 via 1 H-13 C dipolar interaction. Therefore, impregnation with small hydroxylic solvents with high polarity reduced the carbohydrate signal intensity owing to the transfer of magnetization to hydroxyl molecules. Magnetization transfer is more efficient near the melting point and in small hydroxyl molecules with short correlation time motion. Furthermore, impregnation caused the signals of hydroxyl molecules to appear above their melting point owing to the molecular association in the hierarchical structure of cypress. This is because the small hydroxyl molecules are trapped in the nanopores and behave like solids, even above the melting point. These spectral changes related to the correlation time and molecular association can be extended to various nanostructures and impregnated chemicals. Future work would benefit from recording complementary direct polarization 13 C spectra recorded for short recycle delays that would reveal mobile components that are not observed in a 13 C CP MAS spectrum. We aim to analyze a wide range of chemical modifications and manufacturing processes from now on.