Study on water resistance improvement of wood dowel rotation welding joints

Abstract Rotation welding of wood dowels has the advantages of high bonding strength, fast processing speed and green environmental protection, and is suitable for jointing nodes in furniture and wood products. However, most wood friction welding specimens have poor water resistance where the welded joints are more likely to be damaged in wet environments, which greatly limits their wider application. Previous studies focused on using natural and green materials or methods to enhance bonding strength and water resistance of friction-welded joints. This paper reveals an innovative chemical pretreatment method to improve the water resistance of rotary friction welded joints by treating the substrate blocks or dowels with reagents that successively oxidize and sulfonate the wood, and spraying the dowels with Zinc acetate alcohol solution that has a lubricant effect, followed by friction welding. The findings show that both the dry bonding strength of friction welded joints and the wet bonding strength after impregnation with cold, hot, and boiling water of dowels pretreated with oxidation and sulfonation reactions were higher than those without pretreatment and significantly superior to the traditional polyvinyl acetate (PVAc) adhesive bonding. Graphical Abstract

which are generally derived from petrochemicals. Among them, phenolic resin and urea-formaldehyde resin adhesives will release free formaldehyde in the process of using finished products, which represent a source of health concern and environmental contamination; [1][2][3] polyvinyl acetate adhesive is an environmentally friendly adhesive, but it does not have good water resistance, heat or creep resistance, these defects limit its wider use. [4,5] In addition, curing of polyvinyl acetate adhesive usually takes 4-24 h at room temperature, this disadvantage is a limiting factor in industrial production and significantly reduce productivity and yield. [6] Moreover, metal connectors consume substantial fossil fuels during production and use, resulting in carbon dioxide emissions. [7] Fortunately, recent years have seen the addition of mechanically induced friction welding techniques to these traditional joining methods, [8] where dowel-welded joints are easily applied to wood products and furniture manufacturing, [9] highlighting certain advantages compared to traditional petrochemical adhesive bonding technology and metal joining technology. This approach meets the requirements of low-carbon economy and circular economy and provides an effective research direction for achieving peak carbon emissions and carbon neutrality targets. [10] Wood friction welding technology is a glue-free bonding technology that enables fast, high-strength and efficient bonding between woods, [11][12][13][14] and can be divided into linear friction welding and rotary friction welding. [15,16] The technology involves the utilization of wood (mainly lignin and hemicellulose) to soften, melt, and flow the intercellular polymers without the aid of connectors or additives to form an entangled network between wood and wood through close friction. Lignin, which acts as an adhesive in natural wood and undergoes thermochemical changes during the welding process, is one of the most important chemical components of self-adhesive fibers without adhesives and water resistance. [17,18] When frictional movement stops, the molten polymer cools and solidifies rapidly. A good layer of the welded adhesive interface can be formed between wood and timber, [19,20] achieving an effective bond between timbers within ten seconds. [21] One advantage of this method is that it avoids the problem of corrosion caused by the use of metal connectors. [16] The adhesive-free bonding process is green and environmentally friendly since no adhesive is used, which minimizes petrochemical adhesives and shortens the time required for pressurized bonding.
Although wood welding technology can efficiently glue timber or wood materials, wood welded joints are more prone to damage in humid environments. The dry shrinkage and wet rise properties of wood, [8] the inhomogeneity of the weld interface and the open, penetrable structure of the fibers and melted matrix composite lead to the failure of welded specimens easily in humid environments. [22][23][24] Thus, welded products are limited in their application potential due to their poor water resistance. Numerous attempts have recently been made to enhance the water resistance of welded wood products via the modification of the wood surface before friction welding. [8] This includes (1) optimization of welding parameters, parameters such as insertion speed, welding pressure and welding time; [19,25,26] (2) parameter optimization for chemical composition and structure of the wood, for example, the increased content of lignin and hemicellulose in wood will improve the water resistance of welded nodes; [13,27] (3) heat treatment, but the resistance of heat-treated welded wood joints to water remarkably decreased; [28,29] [4) treatment with natural additives such as acetylated lignin, [30] vegetable oils, [31] tannin aqueous solution, furfural, acetylene, [32] natural rosin, oils, [25,30,33,34] citric acid [35] and Padauk extracts; [8] and (5)chemical treatments, such as by immersion in copper chloride aqueous solution [36] and by acetylation and furfurylation. [30] Various approaches have been explored to enhance water resistance of the welded wood. However, a way to oxidize and sulfonate the substrate block and dowel before welding and add lubricant has not been explored. The acidic oxidizing reagents Ammonium persulphate ((NH 4 ) 2 S 2 O 8 ), sulfonating agents Ferrous sulfate (FeSO 4 ·7H 2 O) with Anhydrous sodium sulfate (Na 2 SO 3 ) and lubricant Zinc acetate alcohol solution ((CH 3 COO) 2 Zn•2H 2 O, CH 3 CH 2 OH) are used to oxidize and sulfonate the wood. It is hypothesized that this approach would open up the surface structure of the wood, which in turn facilitates the reaction when the friction welding is taking place. Aside from that, since lignin is the main adhesive for friction welding of wood, oxidation can lead to cracking or demethoxylation of the aromatic ring of lignin in wood, improving its reactivity. [37] In addition, the sulfonation reaction also introduces sulfonic acid groups into lignin, resulting in functionalizing and enhancing its activity, [38] which subsequently increases the likelihood that condensation reactions will occur when frictional temperatures are high. [39] As a result, promoting cross-linking of lignin would significantly improve the bonding strength of friction-welded interfaces. Additionally, the impregnation of wood with acidic solutions such as ammonium sulfate leads to oxidative degradation of hemicellulose and cellulose amorphous zones by converting the hydroxyl groups to carboxyl groups. [40] Therefore, increasing the proportion of cellulose and lignin in the weld interface, in combination with increasing the crystallinity of cellulose, will eventually enhance the mechanical strength and reduce the hygroscopic properties of the interface.
Therefore, based on wood rotational friction welding technology, this paper tries to find a process that can balance the environmental performance with waterproofing performance and mechanical properties. The study aims to evaluate the effect of impregnating or spraying wood dowels and substrate holes with an acidic oxidizing agent (aqueous ammonium persulphate) and an aqueous mixture of ferrous sulfate and sodium sulfite (sulfonating agent) on the water resistance and strength of rotary friction welded joints. Specifically, an in-depth evaluation was performed to study the dry bonding strength of welded nodes, the wet bonding strength after impregnation with cold, hot or boiling water, and the jointing strength after impregnation and drying. Then, the friction-welded interface of pretreated wood was examined on its microscopic morphology, interfacial density, changes in molecular clusters and physical phases and the crystallinity of cellulose via X-ray micro densitometry analysis, scanning electron microscopy (SEM), X-ray diffraction analysis (XRD), and Fourier transform infrared (FT-IR) spectroscopy.

Materials
Fast-growing poplar (Populus × euramericana 'San Martino' I-72) was selected as the substrate, produced in Huai'an, Jiangsu Province, with an average air-dry density of about 0.45 g/cm 3 and a plain surface. A substrate block with dimensions of 50*50*28 mm was dried to 8-10% moisture content, in which the center of the surface of the substrate block needed to be pre-drilled in advance. The depth of the pre-drilled hole was equal to the thickness of the substrate. For the diameters of both pre-drilled holes, the former for friction welding was 9 mm while the latter for gluing was 12 mm. Schima Superba was chosen as the material for the round dowel with an air-dry density of about 0.68 g/cm 3 , and the moisture content was dried to 3-5%. The surface of the dowel was smooth with 45° chamfering at both ends. The average size of the dowel after drying was 11.3 mm × 100 mm, and the dowel diameter was around 1.26 times that of the pre-drilled hole diameter. All the experiments were conducted with the same batch of wood.

Treatment
The surfaces of the substrate holes or dowels were modified by conducting oxidation and sulfonation reactions followed by the treatment of zinc acetate coating. About 3-5% moisture content was achieved for all welding materials (wood dowels and wood blocks) after drying. Then, the substrate holes and the dowels are pretreated separately according to the following steps.

Pretreatment method of the substrate hole.
A moisture content of between 3% and 5% was achieved for the poplar substrate blocks after drying, and then their pre-drilled holes were treated with oxidizing and sulfonating agents. Firstly oxidation treatment of the pre-drilled holes of poplar substrate blocks: spraying with aqueous ammonium persulfate ((NH 4 ) 2 S 2 O 8 ) at a mass concentration of 4% (oxidizing agent), spraying amount of 1 g more or less, after 2 h reaction at room temperature, the mixture was placed at 50 °C to continue the chemical reaction and dried to the original weight. Next, sulfonation treatment of the holes of substrate blocks: spraying with a mixture of ferrous sulfate (FeSO 4 ) and sodium sulfite (Na 2 SO 3 ) at a pH of 10.5 (sulfonating agent) and a mass ratio of 1:5, spraying volume about 1 g, after 2 h reaction in constant temperature and humidity chamber at 65 °C, [3,41] the mixture was dried at 50 °C to the water content of 8-10%.

Pretreatment method of the dowel.
The dowels were dried to 3-5% moisture. Then, their surfaces were treated with oxidation and sulfonation reactions followed by zinc acetate coating. Firstly, the lower half of the dowel was impregnated with aqueous ammonium persulfate solution with a mass concentration of 4% (oxidizing agent) for 12 h, after that the depth of soaking wood was 2-3mm and the water absorption rate of wood was about 20%, and then dried at 50 °C to the original weight. Then, the mixture was impregnated with ferrous sulfate (FeSO 4 ) and sodium sulfite (Na 2 SO 3 ) at a pH of 10.5 (sulfonating agent) in constant temperature and humidity chamber at 65 °C for 12 h, with a mass ratio of 1:5. After that, the penetration depth of the wood was more than 3-4mm, and then the resulting mixture was dried at 50 °C to a 3-5% moisture. Lastly, zinc acetate (CH 3 COO) 2 Zn•2H 2 O) and ethanol (75% solution) mixed solution was sprayed onto the dowels surface and dried with a hair dryer, repeated this operation 10 times. Zinc acetate coating aims to eliminate hydroxide radicals and provide lubrication.

Water-resistant treatment and test methods
After soaking the rotating friction welded specimens in cold, hot, and boiling water, their jointing strength was measured. Then, they were dried for another jointing strength test to eliminate the effect of tensile strength on the welded specimens due to the dimensional expansion of the dowels.

Cold water strength test method.
Referring to GB/ T9846-2015 "Plywood for general use" in the class II glue standard, 20 rotary friction welded specimens were placed in a constant temperature soaking tank with pH 7 ± 1 and temperature (20 ± 3) °C for 24 h. Then, 10 specimens were removed and dried of excess water prior to tensile strength testing, the result obtained was referred to as cold water wet strength. The other 10 welded specimens were put into a drying oven at (63 ± 3) °C until 95-100% of the original mass prior to the tensile strength testing, the result obtained was referred to as cold water drying strength.
Hot water strength test method. Referring to GB/ T9846-2015 "Plywood for general use" in the Class III gluing standard, 20 rotary friction welded specimens were placed in a constant temperature soaking tank at temperature (63 ± 3) °C for 3 h, then dried the surface water of 10 specimens after they were taken out and cooled at room temperature for 10 min. The result obtained was referred to as cold water wet strength. The remaining 10 welded specimens were put into a drying oven at (63 ± 3) °C and dried until 95-100% of the original mass prior to the tensile strength measurement, the result obtained was referred as hot water drying strength.
Boiling water strength test method. Referring to GB/ T17657-2013 "Test methods of evaluating the properties of wood-based panels and surface decorated wood-based panels" in the determination of moisture resistance -boiling water test treatment method, 20 specimens were placed in the (90 ± 10) °C water tank heated to the boiling point. After boiling for 2 h, the specimens were put into (20 ± 3) °C cold water for 1 h. Then, 10 specimens were dried and tested. The measured result was referred to as boiling wet water strength. Another 10 specimens were put into the (70 ± 2) °C drying oven until 95-100% of the original mass was obtained to determine its tensile strength. This result obtained was referred to as boiling water drying strength.

Fabrication of specimens
Fabrication of dowel-welded specimens. A dowel is inserted into the substrate block at both welding and rotating speed of 400 mm/min and 1500 rpm, respectively, under temperature range of 20-25 °C and relative humidity of 45-65%. When the dowel completely passes through the substrate and surpasses the substrate bottom by approximately 2 ~ 3 mm, the rotation is ceased immediately and held at the pressure for 5 ~ 10 s. After that, all welded specimens are placed in a humidity chamber with relative humidity of 50-70% at 20-25 °C for 7 days to modulate before being tested for jointing strength.
Fabrication of glued specimens. Firstly, a hole of φ12mm is pre-drilled in the center of the wooden substrate, and then the surface of the dowel and the inner wall of the pre-drilled hole are coated with glue on both sides, with the total amount of glue applied being 200-250 g/m 2 . The glue type is polyvinyl acetate emulsion (PVAc), pH 6.7, solid content 48.2%, and viscosity 0.63 Pa-s. Then, after 15-20 min, the dowel of diameter φ11.5 mm is inserted into the pre-drilled hole until the bottom of the dowel exceeds the bottom of the substrate by about 3~5mm, and then the small amount of extruded glue is wiped. Finally, it is placed in an indoor environment with an ambient temperature of 20 ~ 25 °C and relative humidity of 45 ~ 65% for 7t days until the glue layer is completely cured.

Tensile mechanical test
The universal testing machine (UTM5105) tests the bonding strength of all-welded and glued specimens. The tensile mechanical test method referred to GB/ T14018-2009 "Method of testing nail holding power of wood ", conducted at a relative humidity of 45-65% at 20-25 °C. The test was performed with the dowel fixed and the wooden substrate was pulled upward at a loading speed of 2 mm per minute until the substrate was entirely separated from the dowel and stopped. Then, the maximum pull-out force was measured with 10 replicates.
Scanning electron microscope analysis method SEM scanned samples were made of wood dowel welding area of 10*2*2mm samples, and the samples of welding interface were conditioned to oven dry. Scanning electron microscopy (SEM) micrographs of the surfaces were obtained after metallizing with goldpalladium by a Hitachi S-4800 (Tokyo, Japan) sputter coater at an accelerated voltage of 5.0 kV.

X-ray micro densitometry analysis method
X-ray micro densitometry analysis is a nondestructive testing technique that effectively detects the density distribution of materials. It has proven to be a valuable technical means for evaluating the wood welding quality. Since friction welding has already demonstrated a considerable densification of the welded wood interface using a similar technique, such a technique is most appropriate for proving the densification of wood interfaces. [42] The GreCon DAX-5000 (Germany) X-ray micro densitometer was used with a scanning step of 0.02 mm and a test accuracy of ±1%. The weld zone of the welded specimen was scanned along the thickness direction of the substrate block(28 mm thickness), and the testing environment of the X-ray micro densitometer was maintained at a humidity of 55-65% and temperature of 20-25 °C.

The Fourier transform-infrared spectroscopy technology
A Thermo Scientific iN10 FT-IR spectrometer was used to perform FT-IR analysis in the mid-infrared region (400-4000 cm −1 ) by KBr compression method. Firstly, the wood particles were scraped from the surface of untreated dowel or treated dowel for sampling, and the molten sample particles were scraped from the untreated welded-dowel or treated welded-dowel, and each of them was ground by ball milling method, sieved with 0.05 mm mesh and dried at 70 °C until completely dried. Then, in a dry environment, approximately 1 mg of sample and 300 mg of dry potassium bromide powder (KBr) were added to a mortar and were ground thoroughly for several times, and then pressed on a tablet press machine (into transparent sheets), the background and then the FT-IR spectrum of the sample were collected. Each sample was scanned 32 times at a resolution of 4 cm −1 and the data obtained were processed using Origin Pro 10.1 software. [43] The baseline corrected and normalized IR spectrogram eliminates the effect of sample volume on the test, allowing comparison of the relative intensity changes of different absorption peaks and semi-quantitative analysis of the composition. And usually, the strongest peak of cellulose is used for normalization by default.

X-ray diffraction analysis method
X-ray diffraction (XRD) analysis is an analytical method to analyze the composition of the physical phase, molecular or atomic structure of the sample. By this method, it is possible to obtain precise information on the changes in phase and crystallinity of the friction welding interface, [44] and to understand further what causes the bonding strength to increase and improvement in water resistance. Morphologically, cellulose can be divided into crystalline and amorphous regions. [45] The percentage of the crystalline zone in the supramolecular structure of cellulose is called cellulose crystallinity, which reflects the degree of crystallization of cellulose when it aggregates. And the crystallinity of wood has an important effect on the physical, mechanical, and chemical properties of wood-based materials. [46,47] The crystallinity of cellulose in dowels was analyzed by XRD. The molten material in the non-welded and welded areas of the dowels was prepared as a powder and placed in an Ultima IV X-ray diffractometer (Rigaku, Japan) for continuous scanning at 2θ from 5-90° in steps of 0.02° at a scanning speed of 2° min −1 . The crystallinity was calculated using the peak splitting method, i.e., the diffraction curve of the XRD pattern was split using the Lorentzian function while the crystallinity was calculated as follows: where CrI is the crystallinity index, IA is the integral area of the non-crystalline area, and SP is the total integrated area of crystallization peaks in the crystallization area. Figure 1 shows the wet tensile strength and dry tensile strength for each group. Group A represents specimens in which dowels and substrates were treated, whereas group B represents specimens in which only the dowels were treated. Group C represents untreated specimens and group D represents conventional coated specimens (PVAc). The dry and wet tensile strengths of all groups were, in descending order, group B > group A ≈ group C > group D. The wet jointing strengths of all specimens meet the requirements of Class II and Class III plywood (i.e., ≥ 0.7 MPa). The highest dry tensile strength was achieved in group B (6.9 MPa), which was 16.7% higher than the dry bond strength of the untreated material friction specimen in group C (5.9 MPa) and 67.2% higher than the dry strength of the glued specimen in group D (4.1 MPa). For group B, the welding specimens that were impregnated with cold water, hot water, and boiling water show wet strength of 5.1, 4.6, and 4.5 MPa, respectively, which are higher than the specimens in group C (53.0%, 28.3%, and 47.1%, respectively) and group D (304.7%, 166.3%, and 198.0%, respectively) that impregnated in the same conditions. It can be seen that friction specimens, which were only pretreated with oxidation and sulphonation reactions on the dowels, exhibited high bonding strength and excellent water resistance. The group A specimens were pretreated with substrate holes and dowels to increase their jointing strength and water resistance. However, the group A specimens did not exhibit a significant advantage over the untreated specimens (group C). An increase in the diameter of the predrilled hole in group A was observed due to the absorption and desorption of moisture by the wood surrounding the hole walls after pretreatment. In turn, the diameter of the dowel decreased in proportion to the hole diameter. This is a critical factor in the effectiveness of rotary friction welding that subsequently results in bonding strength reduction. The differences in the bonding strength of each group of specimens can be determined by observing the spilt melt during welding, the profile after welding, and the failure form during mechanical tensile testing.

Dry and wet bonding strength of specimens
As shown in Figure 2a-c, the molten material spewed from the group C welded specimens without any pretreatment was light brown and relatively small in amount. In contrast, the pretreated group A and B welded specimens were found to extrude an enormous quantity of black molten material from the gap between the dowel and the hole during friction welding. Specifically, the molten material spilt from the group B specimens was almost continuous in the form of flakes, which indicates that the molten material ground out from the group B friction specimens has higher adhesion and ductility, which is conducive to the formation of high-strength gluing interfaces. Three groups of welded specimens of the section are displayed in Figures 3a-c. When the wood dowel was subjected to high-speed rotational friction, molten material was extruded into the substrate hole wall wood, resulting in a similar glue nail appearance. This glue nail-like structure was conducive to improving the actual gluing area of the welded joint. In particular, the number of glue pegs in the welded joints of friction specimens of group B was higher than in groups A and C. This resulted in a larger glue area, which contributed to the improvement in joint strength. Three groups of friction welding specimens have been shown in Figure 4a-c, in the form of tensile testing failures. The group B welding specimens demonstrate strong bonding between the welding  interfaces when wood dowel tear or direct fracture is applied to the tensile test failures. In addition, many poplar substrate fibers were bonded to the welded areas of the dowels in the friction specimens of groups A and B after tensile damage. In contrast, the fibers were barely visible in the welded zone of the dowels in group C specimens, implying that the groups A and B friction specimens possessed higher bond strength at the welded interface.

Bonding strength of wet specimens after drying
All specimens met the gluing strength requirements for Class II and Class III plywood (≥ 0.7 MPa), as shown in Figure 5. The glue strength of groups A and B welding specimens is slightly higher than group C welding specimens and significantly higher than group D PVAc gluing specimens. Group B friction specimens show the highest tensile strength and the best water resistance after cold (3.5 MPa), hot (4.4 MPa) and boiled water (3.8 MPa) impregnation and then drying. In addition, comparing Figures 1 and 5, the bonding strength of dry specimens after immersion is generally lower than the strength of wet specimens. This is due to the expansion of the wood under wet conditions, [8] which makes the welding nodes tighter, and the fact that the specimen cannot return to its presoaking state after drying, the wood voids become larger, as well as the wood bonding interface has a non-uniformity.
Overall, the findings revealed that the water resistance of the friction-welded specimens with pretreated dowels was significantly better than that of the untreated welded specimens and the glued specimens. The strength after redrying of group B (only the dowels were treated) is 50.6% of the dry strength, and that of group C (untreated specimens) is 38.4%. In contrast to the study of A. Pizzi, they performed welding with beech (Fagus sylvatica), the dowel diameter was 10 mm, the substrate size was 50*50*28 mm, the pre-drilled hole was 8 mm, the drilling speed was 1600 rpm, the welding depth was 24 mm, and the   welding time was 1.5-2 s. When rosin was used as an additive, after 24 h cold water soaking and drying, the redrying bonding strength of the Group A(the dowel dipped in molten rosin) was 74.5% of its dry strength, while 48.7% for Group B(both substrate and dowel are treated with 25% Rosin solution in ethanol). [34] And when lignin was used as an additive, after 24 h cold water soaking and 72h air drying, the redrying bonding strength of the no-treatment group was 46.2% of its dry strength, and 59.9% for group with non-acetylated wheat lignin; 62.8% for group with wheat straw acetylated lignin; 74.2% for group with depolymerized acetylated wood lignin. [30] Since the water resistance of welded wood joints depends on the material properties, welding parameters and additives, the effect of these treatments cannot be compared quantitatively. However, all of the above methods are effective in improving the water resistance of rotary friction welded wood products. Figure 6a shows that the surface of the welded dowel is bonded with many poplar fibers wrapped transversely around the dowel surface. Figure 6b demonstrates the dowel underwent continuous rotation and friction, the longitudinal base fibers of the dowel (indicated by the arrows in Figure 6(b)) were covered with a composite interfacial layer consisting of a mixture of wood fibers and molten intercellular material. Figure 6c displays a flat and continuous composite phase consisting of wood fibers and molten material in a tightly bound layer formed on the dowel surface after welding. The high frictional temperature and pressure promote a homogeneous encapsulation of the wood fibers by the molten intercellular material. Figure 6d shows a strong, continuous, highly densified gelatinous welded interface where the wood cell structure has been completely lost on the welded interface layer. It can be seen that the welding interfacial layer has a highly densified structure, which indicates a low porosity of the weld interface, thus reducing the permeability of water.

X-ray micro densitometry analysis
The X-ray microdensity profiles of the untreated and treated welded specimens scanned along the thickness direction of the substrate blocks (28 mm) are shown in Figure 7. The density of group B welded specimens (only the dowels were treated) was about 750-800 kg/ m 3 , while that of group C welded specimens (untreated specimens) was about 650-700 kg/m 3 . It was found that when the density of the friction welding interface is higher, the mechanical properties of the welded joint are better, suggesting that the bonding strength of group B friction specimens is stronger than group C.

FT-IR spectroscopy analysis
Fourier transform-infrared (FT-IR) spectroscopy had been proved to be useful for analysis and identification on functional groups of polysaccharides in plant materials. [48] FT-IR spectra could be used to analyze the changes in the molecular groups of wooden dowels before and after the treatment and the changes in the chemical functional groups of untreated and treated wooden dowels before and after friction welding. The changes in cellulose, hemicellulose and lignin content in the welded interface were also analyzed under different treatments (i.e., dowel and hole treatment and dowel-only treatment) to reveal the reasons for the increased adhesive strength of the specimens and the increased water resistance of the welded interface after the treatment. Figure 8a, b shows the FT-IR spectrum of untreated dowels and pretreated dowels welded or not based on the characteristic absorption peaks of the infrared spectra of the wood, while their corresponding attributions are shown in Table 1. It is clear that all four specimens contain cellulose, hemicellulose and lignin. The infrared spectra showed no new absorption peaks in the samples tested. The main difference between them was the variation in the intensity of the original absorption peaks. Cellulose contains different types such as crystalline cellulose and non-crystalline cellulose. The chemical environment of the C-O-C bond in different types of cellulose varies depending on the hydrogen bonding, cellulose type, or space junctions. The FT-IR spectra detected at 1425, 1159 , 1116, and 1057 cm −1 are corresponded to the existence of cellulose, [49][50][51] while the bands from 1617 to 1238 cm −1 correspond to the lignin. [52] The relative intensity of the absorption peaks at 1160 to 1032 cm −1 mainly reflected variation in the relative amounts of different cellulose types or their microstructures. [51] The baseline corrected and normalized IR spectrogram eliminates the effect of sample amount on the test, allowing comparison of the relative intensity changes of different absorption peaks and semi-quantitative analysis of the composition. The absorption peaks of cellulose C─O─C stretching vibration at 1057 cm −1 were selected as the highest point for normalization. It was found that there were inconsistent changes in the absorption peaks of cellulose stretching vibration at 1159, 1116, and 1032 cm −1 , and the absorption peaks of lignin stretching vibration at 1617-1238 cm −1 . Figure 8 shows that the treated dowels and welded area of the treated dowels had significantly higher relative cellulose contents than the other two untreated specimens. The lignin content of the melt at the welded interface of the pretreated dowels was much higher than that of the other three. Following is a comparison of the cellulose and lignin content of the dowels before and after chemical pretreatment and friction welding. Figure 9a, b, the main difference between the IR spectra of the treated and untreated dowels is that the treated one has significantly enhanced absorption peaks at 1159, 1116, and 1735cm −1 , of which 1159 cm −1 and 1116 cm −1 correspond to the absorption peaks of cellulose. This indicates that the treated dowels have higher cellulose content than the untreated dowels or changes in the molecular structure of cellulose. This may be due to the treatment applied (i.e., oxidation and sulphonation) on dowel surfaces which alters the molecular structure of hemicellulose, cellulose and lignin on the wood surface and remove abundant hydroxyl groups (─OH) from its molecular structure. [37,38] On top of that, oxidation and sulphonation reactions can result in a greater degradation degree of hemicellulose in wood, causing to an increment in the relative content of cellulose and lignin. [37,38] Since the amorphous zone of cellulose is more reactive than the crystalline zone, this can lead to chemical reaction on the amorphous zone of  cellulose by peroxydisulphate, sulfate radicals, hydrogen peroxide and hydrogen ions to degrade the noncrystalline cellulose. [40] As a result, the crystalline regions that are structurally compact and chemically stable remain, resulting in a rearrangement of the apparent cellulose structure and increment in the relative cellulose crystallinity.

FT-IR analysis of untreated dowels compared to untreated wood welded interfaces.
As can be seen from Figure 10a, b, the infrared spectra of treated dowels compared to that of untreated dowels at the welded interface melt show less variation. The infrared spectra of the two tested specimens at 2000-800 cm −1 are highly overlapping, indicating that the changes in lignin, hemicellulose, and cellulose in poplar wood after the friction welding treatment are smaller than the chemical pretreatment method.

Comparative infrared spectroscopy of treated dowels versus treated dowel welded interfaces.
As can be seen from Figure 11(a) and (b), the IR spectra of the molten material at the welded interface of the treated dowels (representing the welded zone of Group B dowels) and the non-welded zone is significantly higher in the relative intensity of the absorption peaks at 1617, 1514, 1461, 1425, 1238, 1159, and 1116 cm −1 . Of these, the adsorption peak at 1617, 1514, 1461, and 1238 cm −1 correspond to the absorption peaks of lignin. Pretreated dowels with welded interfaces display significantly higher lignin absorption peaks than those with non-welded zones, suggesting that the treated wood can significantly increase the relative lignin content by friction welding at high temperatures. This may be because the wood is subjected to oxidation reactions that can crack or demethoxylate the aromatic rings of the lignin and sulphonation reactions that can introduce the sulfonic acid groups (-SO 3 H) into the lignin, which thereby makes it functional and active. [37,38] As a result, high frictional temperatures make the wood more susceptible to condensation reactions. The absorption peaks at 1425, 1159, and 1116 cm −1 correspond to cellulose. The relative intensity of the absorption peaks of the pretreated dowels at these three locations is higher than that of their welded interface melt, which indicates that the relative cellulose content in the pretreated dowels has decreased after the high-temperature friction welding, further suggesting that cellulose degradation has occurred. In addition, 1735 cm −1 represents the C═O stretching vibration on the carboxyl group of zinc acetate, where the pretreated dowel has a significant absorption peak, while its welded interface melt has no absorption peak at this wavenumber, indicating complete consumption of the zinc acetate after friction welding.

Comparative infrared spectroscopy of welded interfaces between treated and untreated dowels.
As can be seen from Figure 12a, b, the infrared spectra of the pretreated wood dowel welding interface melt show a higher relative intensity than the untreated one for the absorption peaks at 1617, 1514, 1461, 1425, 1238, 1159, and 1116 cm −1 . Among them, 1617, 1514, 1461, and 1238 cm −1 correspond to the lignin characteristic absorption peaks. Pretreated wood dowel welding interfaces exhibit a significant increase in the relative intensity of lignin absorption peaks, indicating that the wood can significantly increase the relative content of lignin after chemical treatment and frictional high-temperature reaction. The significant   increase in the relative intensity of the absorption peaks assigning to cellulose at 1425, 1159, and 1116 cm −1 infers that the high-temperature reaction of the pretreated wood by friction causes an increment in the relative cellulose content or its microstructure changes.
In summary, it is evident the untreated wood (i.e., logs) showed a small change in cellulose and a small increment in the relative lignin content before and after friction welding. In contrast, wood that had been oxidized, sulfonated and chemically treated with zinc acetate and then treated with high-temperature friction showed a significant increase in cellulose and the relative lignin content, as well as substantial changes in its microstructure. Overall, the greatest impact on the relative cellulose content and microstructure of the wood is observed in the chemical pretreatment, followed by the high-temperature friction welding treatment. Then, the greatest impact on the relative lignin content of the wood was found in the high-temperature friction welding treatment, followed by the chemical pretreatment in which the former had a much greater impact than the latter. The effects on cellulose content and its microstructure were, treated dowel > treated dowel melt ≥ untreated dowel ≈ untreated dowel welded interface melt. For the effects on lignin content were, treated dowel weld interface melt > treated dowel > untreated dowel weld interface melt > unwelded dowel.
It was noted from the findings that the high-temperature friction welding treatment has a negligible effect on the lignin and cellulose content of the wood, with a slight increase in the relative lignin content and a reduction in the relative cellulose content. In contrast, the oxidation, sulphonation and zinc acetate coating followed by high-temperature friction welding of the dowels shows a noticeable increase in the relative lignin content and obvious changes in the relative cellulose content and its microstructure. In addition to chemical pretreatment, high-temperature friction welding may cause a higher degree of degradation of cellulose (primarily in the amorphous zone) and hemicellulose. As a result, the wood significantly reduces amorphous cellulose and hemicellulose content and increases cellulose (particularly crystalline cellulose) and lignin content. It can also be seen that the chemical pretreatment of dowels followed by high-temperature friction welding increases the crystalline cellulose and lignin at the friction welding interface.

Effect of various pretreatment approaches on the welding interface composition
As seen from Figure 13 and Table 1, all three groups of friction specimens (i.e., A, B and C) contained cellulose, hemicellulose and lignin in the molten material at the welded interface. As seen from the infrared spectrum, no new absorption peak formed, the only difference being the intensity change of the existing absorption peak. After baseline correction and normalization, the infrared spectra shows that the molten material at the welding interface of the A and B friction specimens showed a significant increase in the intensity of the C-O-C stretching vibration absorption peaks at 1159 and 1112 cm −1 . This again indicates that the relative cellulose content of the wood increases significantly after chemical pretreatment and high-temperature friction welding. The cellulose molecular structure is rearranged and associated with the content changes of cellulose and non-crystalline celluloses. The wavenumbers at 1616 and 1514 cm −1 correspond to the characteristic peaks of lignin in different woods. Depending on the relative intensities of the absorption peaks, the relative intensity of the lignin in the melt of the welding interface of group A friction specimens at 1616 and 1514 cm −1 are significantly lower than group B, indicating the lignin content in the group A is significantly lower. As a result, group A friction specimens have a lower welded bond strength than group B friction specimens. These thermochemical modifications following the welding process are in accordance with the findings of previous studies. [53,54] Figure 14a shows the XRD pattern of the dowel without any chemical pretreatment. The analysis of the plots and control standard cards by Jade 6.5 software shows that the main phase component of the dowel is cellulose. Figure 14b shows the XRD pattern of the molten material at the friction welding interface between the untreated wood dowel and the poplar substrate. As shown by the analysis of the plots and control standard cards, the physical phase at the welded interface of the friction specimen remains unchanged. It remains cellulose, regardless of whether the dowel or the base material has been pretreated. After friction-welding and melting the wood at high temperatures, the cellulose at the welded interface appears to be more ordered and narrower than when the wood is not welded, indicating a change in its cellulose structure and an increase in crystallization.

Analysis of the physical composition and crystallinity of untreated material before and after friction welding
By splitting the XRD diffraction curves using the Lorentz function (Figures 14(a) and (b)), the relative crystallinity of cellulose in the untreated wood dowel and the welded interface of their friction specimens were 17.09% and 35.05%, respectively. The cellulose crystallinity at the welded interface compared to the wood in the unwelded area was 105.09% higher. This is due to the poor stability of the hemicellulose during friction at high temperatures, which can cause thermal degradation of the hemicellulose. It is also associated with the amorphous cellulose degradation, the rearrangement of the molecular structure, and the crystallization of cellulose, which increases the relative crystallinity of cellulose. Figure 15a shows the composition of the dowel following oxidation and sulphonation reactions and zinc acetate coating. The dowel contained two different structures of Na 2 SO 4 (space groups Fddd and Pbnn) and Na 2 Fe(SO 4 ) 2 -4H 2 O, in addition to cellulose in its physical phase composition. Based on the physical phase changes of the wood before (Figure 14(a)) and after pretreatment (Figure 14(b)), the wood contained abundant amounts of Na 2 SO 4 and a small amount of Na 2 Fe(SO 4 ) 2 -4H 2 O on its surface. This is due to the low melting point of Na 2 SO 3 , which can melt under high-temperature conditions and react with oxygen (O 2 ) to produce Na 2 SO 4 with higher melting points and stability. FeSO 4 is oxidized to basic iron sulfate  on the surface under humid conditions and becomes the tetrahydrate (FeSO 4 -4H 2 O) at 50-60 °C and combines with sodium sulfate to form Na 2 Fe(SO 4 ) 2 ·4H 2 O.

Analysis of the physical composition and crystallinity of wooden dowels before and after chemical pretreatment
In Figure 15a, the Lorentz function was applied to split the XRD diffraction curves, which resulted in a crystallinity of 54.06% for cellulose, an increase of 216.33% relative to that of untreated lignocellulose (17.09%). It has been demonstrated that the oxidation and sulphonation reactions increase the crystallinity of cellulose in the wood. Consequently, the FT-IR analysis suggests that the molecular structure of cellulose has been rearranged in the pretreated dowels. The finding concurs with the increased cellulose content in the pretreated dowels compared to the other three materials. In light of these results, pretreatment with oxidation and sulphonation reactions likely results in the degradation of hemicellulose and cellulose amorphous regions of the wood, which results in a significant increase in the crystalline cellulose content.

Analysis of the physical composition and crystallinity of the experimental group with both dowels and substrates pretreated before and after friction welding
XRD patterns of molten material at the weld interface of the group A friction specimen are shown in Figure  15b. Based on the result of Jade 6.5 software analysis of the plots and control standard cards, it was discovered that when the substrate hole and the dowel were chemically treated with oxidation and sulphonation, and friction welding, the main components of the molten material at the welding interface were cellulose, Na 2 SO 4 , NaHSO 4 , (Na 0.90 Zn 0.05 ) 2 SO 4 and Na 2 Fe(SO 4 ) 2 ·4H 2 O.
From group A of pretreated material before and after welding, the physical phase changes ( Figures  15(a) and (b)), and the material phase composition of the welded interface have increased by two substances after high-temperature friction welding, namely: NaHSO 4 and (Na 0.90 Zn 0.05 ) 2 SO 4 . This is because the aqueous sodium sulfite solution contains sodium hydroxide (NaOH) and sulfite (H 2 SO 3 ), which is very unstable and easily oxidized and pyrolyzed to produce sulfuric acid in high-temperature environments. As a result, both sodium hydroxide and sulfuric acid react to produce sodium hydrosulfate (NaHSO 4 ) and water during high-temperature friction welding. Zinc acetate hydrate (Zn(CH 3 COO) 2 ·2H 2 O) can be stripped of two water molecules from crystallization above 100 °C to form Zn(CH 3 COO) 2 , which melts at around 237 °C and decomposes completely to ZnO when the temperature rises to 370 °C. However, by monitoring the temperature at the welded interface of the friction specimens, it was found that the friction temperature did not exceed 350 °C. Hence, the zinc acetate only melted and did not decompose into ZnO. Thus, zinc acetate (Zn(CH 3 COO) 2 ) reacted with sulfate at high temperatures to produce zinc sulfate (Zn 2 SO 4 ) and combined with sodium sulfate (Na 2 SO 4 ) to form (Na 0.90 Zn 0.05 ) 2 SO 4 as evident from the diffraction peak near 2θ = 25,34,47 of the XRD pattern. As shown in Figure 15(b), by splitting the peaks of the X-ray diffraction pattern of the wood following welding using the Lorentz function, it was determined that the cellulose crystallinity of the group was 51.33%, a slight decrease from 54.06% before welding. This was primarily due to a small high-temperature degradation of the crystalline cellulose at the welding interface of the pretreated material.

Analysis of the physical composition and crystallinity of the experimental group with only dowels pretreated before and after friction welding
The X-ray diffraction pattern of the melt at the welded interface of the friction specimen from group B is shown in Figure 16. The phases at the weld interface of this friction specimen are mainly: cellulose, Na 2 SO 4 (space group Fddd and space group Pbnn), NaHSO 4 (sodium bisulphate) and (Na 0.90 Zn 0.05 ) 2 SO 4 . The function peak splitting calculations show that the cellulose crystallinity of this group of specimens was: 50.92%. From the comparison of the crystallinity of the melt at the friction interface of groups A, B and C, it was found that the cellulose crystallinity of groups A (46.45%) and B (39.57%) are higher than group C, suggesting that the chemical pretreatment method in groups A and B can improve the crystallinity of the wood welding interface.
As seen above, wood pretreated by chemical reactions or friction welding at high temperatures can change the cellulose content and microstructure of the wood, with a relative increase in the crystalline cellulose content. The cellulose crystallinity of untreated wood increased from 17.09% to 35.05% after high-temperature friction welding. After pretreatment with oxidation and sulphonation, the cellulose crystallinity increased from 17.09% to 54.06%. This suggests that chemical pretreatment contributes more to the cellulose crystallinity of the wood than high-temperature friction welding, which is consistent with the results of the FT-IR analysis. Physico-mechanical properties of wood are strongly influenced by the crystallinity of cellulose. As the crystallinity of the wood and the crystalline material increases, the dimensional stability, tensile strength, hardness, and density of the wood increase while the moisture absorption, swelling, and chemical reactivity of the wood fibers decrease.
In general, the hygroscopicity and desorption of cellulose occur mainly on the linear molecular chains in the amorphous region with many free hydroxyl groups and the surface of the crystalline region. [52] A higher crystallinity indicates a larger crystalline area, a tighter bond between fibers, and a reduced ability of water molecules to access cellulose molecules. The relative reduction in the amorphous zone reduces the number of hydroxyl groups that can bind water molecules and reduces the moisture absorption capacity of the cellulose. In contrast, the larger crystalline zone binds the amorphous zone more, resulting in less dry shrinkage and more dimensional stability. Therefore, according to the previous glue strength test results, the reason for the better wet water resistance at the welded interface in both groups A and B than in group C (untreated group) specimens is due to the significantly higher cellulose crystallinity at the welded interface and the increased relative lignin content.

Conclusion
The results show that both the mechanical performance and the water resistance of the pretreated welded specimens are significantly better than that of the untreated welded and glued specimens. In terms of mechanical performance, the dry bonding strength of the treated group (group B) was 16.7% higher than that of the untreated group (group C); in terms of water resistance, the wet bonding strength of the specimens in the treated group (Group B) was 53.0%, 28.3% and 47.1% higher than that of the untreated group (Group C) after impregnation with cold, hot and boiling water, respectively. And the wet strength of the specimens in the treated group (Group B) after impregnation with cold water was 74.3% of their unimpregnated (i.e., original dry strength), which implies a decrease of only 25.7%, while the wet strength of the specimens in the untreated group (Group C) after impregnation with cold water decreased by 43.3% compared to their original dry strength. Furthermore, comparing the results of this test with the experimental results of Pizzi's acetylated lignin additive treatment method, [30] the pretreatment method of this paper was found to be superior in terms of water resistance, and was comparable to the natural 25% ethanol rosin solution impregnation treatment method. [34] At the same time, the addition of lignin [30] or impregnation with citric acid [35] resulted in a slight decrease in the dry bonding strength of the samples. In comparison, the reagents chosen in this paper were more readily available and improved the dry bonding strength of the friction welded specimens, while also offering comparable water resistance.
The mechanism for enhancing the water resistance of the pretreated dowel welded joints was studied by some methods. The SEM analysis shows the welding interfacial layer has a highly densified structure, which indicates a low porosity of the weld interface and reduces water permeability. The X-ray micro densitometry analysis shows that the density of the welded interface of pretreated dowels was higher than that of the untreated dowels. The higher the density of the weld interface, the greater the bonding strength. The FT-IR spectroscopy analysis shows that pretreatment by sulfonation and oxidation can significantly increase the relative content of lignin and cellulose in wood and rearrange the molecular structure of cellulose, while subsequent high temperature friction welding can further increase the relative content of lignin. The XRD analysis shows that chemical pretreatment can substantially improve the crystallinity of cellulose. Therefore, higher cellulose crystallinity and relative lignin content both contribute to the friction welding strength and water resistance. Overall, this research reveals a new method for improving the water resistance of wood dowel rotation welding joints.

Disclosure statement
No potential conflict of interest was reported by the author(s).