Vibration welding of wood: x-ray tomography, additives, radical concentration.

Abstract

X-ray tomographic microscopy of wood fusion welded joints, without any adhesive, indicated that, in limited places, a reduction of intercellular material flow can occur in the bondline, and some breaks at the interface-fused composite/wood substrate can occur also. When proceeding

from the outer faces to the innermost layers of the bondline, one passes first undamaged wood cells, followed by interphases in which cells are mixed with long wood cells, and where wood cells are far predominating over the fused matrix, to finally reach an innermost zone where the composite occurs and where the proportion of the fused matrix appears to predominate in relation to the fibers. The use of some naturally derived additives such as tannins and furfural, by autcondensation and polymerization, afforded some improvements in cold water resistance of the welded bondline. Electron Spin Resonance spectroscopy showed an increase in radical concentration in friction-sanded wood surfaces, indicating that some of the reactions occurring in the holding phase after wood welding are likely to follow a radical reaction route.

**********

Mechanically induced friction welding techniques that are widely used in the plastic and car industries have recently been applied also to joining wood, without the use of any adhesive (Gfeller et al. 2003). These techniques work by melting some wood components and forming at the interface between the two wood surfaces to be joined a composite of entangled wood fibers drowned into a matrix of melted wood intercellular material, such as lignin. Linear mechanical friction vibration has been used to yield wood joints, satisfying the relevant requirements for structural applications by welding at a very rapid rate. Cross-linking chemical reactions also have been shown to occur by CP-MAS [.sup.13]C nuclear magnetic resonance (NMR). These reactions, however, are relatively minor contributors during the very short welding period proper but acquire more importance after the welding period proper (Gfeller et al. 2003).

X-ray tomographic microscopy (XTM) is a technique that allows nondestructive testing and observation of planes internal to the samples. It is ideal under these conditions to observe and evaluate the appearance of a glueline of a bonded joint without opening or testing the joint.

In the original wood welding study, it was indicated that what was obtained were only interior-grade bonded wood joints. Some additives that could be beneficial to improve this drawback should also be sought to expand the area of applicability of the wood-welding process.

Cross-linking chemical reactions also have been shown to occur, the most likely one of these identified by NMR appears to be a cross-linking reaction involving carbohydrate-derived furfural with itself and with lignin (Gfeller et al. 2003). This reaction, which is of relatively minor importance during the actual welding time, contributes to improved strength during the holding time after welding. Furfural and other furanic materials can be produced in wood by both ionic and radical reactions, and it is interesting to determine, under the experimental conditions used, if one or the other of these types of reactions occur.

This paper then deals with the observation of the welded wood bondline by non-destructive XTM, with the effect of some additives on welding. It also explores whether or not furfural and other cross-linking furan derivatives are obtained by radical reactions.

Experimental

The mechanical welding machine used was a Branson welding machine, type 2700, 100 Hz, normally used for the vibrational welding of engineering plastics and metals (Branson Ultrasonics Applied Technologies Group, Danbury, Connecticut).

Preparation of joints, with and without additives, and the test results

Specimens were composed of two pieces of beech (Fagus sylvatica), a hardwood, of Norway spruce (Picea abies), a softwood, of oak (Quercus robur), a higher density hardwood, and mixed joints of spruce/beech in which one surface was a hardwood and the other composed of a softwood. The dimensions of the each piece were 150 by 20 by 15 mm. The pieces were welded together to form a bonded joint 150 by 20 by 30 mm by a vibrational movement of one wood surface against another at a frequency of 100 Hz. When the fusion and bonding were achieved, the vibration process was stopped. The clamping pressure was then briefly maintained until the solidification of the bond. The welded samples were conditioned for 1 week in an environmental chamber (20[degrees]C and 65% relative humidity) before testing.

The tensile shear strength was measured according to European standard EN 204 (2004), method EN 205 (2003) for thermoplastic adhesives, which were chosen for the test rather than EN 301 (1993 and 301-1 (1993) for thermosetting adhesives because wood welding is exclusively used for interior furniture and joinery, hence as a substitute for PVAc, a thermoplastic adhesive. Saw cuts perpendicular to the specimen's wood grain, down to the bondline, were made from each one of the specimen's two surfaces. The distance between the two cuts was 2.5 cm. The specimens were then tested in tension on an Instron model 4467 universal testing machine at a rate of 2 s/mm.

The parameters that were used were adapted from those optimized in the initial study (Gfeller et al. 2003): a welding time (WT) of 3 seconds; a contact holding time (HT) maintained after the welding vibration had stopped of 5 seconds; a welding pressure (WP) exerted on the surfaces of 1.3 MPa; a holding pressure (HP) exerted on the surfaces after the welding vibration had stopped of 2.0 MPa; and the amplitude (A) of the shift imparted to one surface relative to the other during vibrational welding of 3 mm. The frequency of welding was maintained at 100 Hz. The equilibrium moisture content of the samples was 12 percent.

In the case of the trials with additives, sunflower oil, a water solution of a polyflavonoid tannin (pine tannin extract, Pinus radiata. ex Diteco, Chile) and furfural were added during the welding process. The conditions to obtain welding had to be changed according to which of these additives were added.

X-ray tomographic microscopy

Two types of samples were prepared: one sample was 3.5 by 3.5 by 3.5 mm and one was 1.4 by 1.4 by 1.4 mm. These dimensions were chosen because the equipment used can only take images in width or in diagonal of 1.4 mm or of 3.5 mm. The samples were fixed onto the sample holder using wax and examined on Sohne & Berger x-ray tomography equipment. The 3.5-mm visual field results were magnified 8 times, while the 1.4-mm visual field results were magnified 20 times. Scanning of the samples was done every 0.01 mm.

Electron Spin Resonance (ESR) of wood surfaces

ESR spectra of three basswood (Tilia americana) surfaces at 12 percent equilibrium moisture content were done before and after mechanical sanding with 60 mesh sand paper to reproduce the action of welding. The basswood specimens used were 45 by 2.8 by 2.5 mm. The spectra were done at 293 K on a JEOL ESR spectrometer. Fine structures of the phenoxyl radical signals in the range 3345[+ or -]50 Gauss were studied with field modulation intensity of 3.2 G at 100K and sweep times of 4 minutes, amplitude 6.3 X 100, response 0.1 seconds.

Results and discussion

Figure 1 shows the XTM image of the radial-axial section of the bondline of the whole specimen, still bonded (thus showing the whole plane of the bondline). This confirms directly in situ, in still-bonded specimens, what was observed by scanning electron microscopy on tested samples (Gfeller et al. 2003). The bondline welded area has the appearance of an amorphous, fused mass. One can also observe that while the area of the bondline is covered to a great extent by the fused material film, this is not nonetheless continuous, and some gaps can be observed where lower flow or no flow of the intercellular material has occurred. Figure 1 shows the areas where the lack of flow has been more marked. One finding of interest is that a homogeneous mass of fused wood constituents appears in Figure 1. If a very marked, predominant, chemical reaction had occurred, this would have been noted through the intense color variations that are characteristic of materials analyzed by XTM. XTM is quite sensitive to these color changes (Abela, unpublished results). This is not the case here, which confirms that fusion of the intercellular material is indeed the predominant mechanism in wood welding.

Figure 2 shows the tangential section of the bondline region (top) and the tangential-radial section of the bondline region (bottom) of a bonded joint. The top image shows wood cells (wood fibers) that have been detached from the cellular wood structure during welding, entangled and immersed in the matrix of fused intercellular material. This same image shows also that the movement of the wood surfaces during welding may cause, in some places, a break at the interface-fused composite/wood, which is circled in the figure. This implies that even if the joint strength is above standard specification requirements, considerable room for improvement exists, as results are likely to improve further once the extent of these microcracks is markedly reduced by further parameter optimization. The bottom image of the tangential-radial section of the wood joints in Figure 2 shows the following when proceeding from the outermost to the innermost layers: 1) untouched, undamaged wood cells (the small beehive appearance zones); 2) two interphases in which cells are mixed with long wood cells, and where wood cells are far predominating over the fused matrix; these zones are quite probably those in which the fibers expelled from the bondline during welding are likely to have been generated; 3) an innermost zone where the composite occurs and where the proportion of the fused matrix appears to predominate in relation to the fibers. This image shows that the thickness of this zone, under the welding conditions used, is from slightly thinner than 300[micro] to about double this thickness. This measure confirms wood welding to be a predominantly surface-affecting process. It means that for each wood surface, the effect to any extent is limited to the first 150 [micro] to 450 [micro] in depth of the surface layer.

[FIGURE 1 OMITTED]

It can be noted that the addition of a traditional thermosetting adhesive to a friction-welding bondline works well (Properzi, unpublished results not reported here), but this is of no great interest. In short, the heat generated by the friction cures the thermosetting adhesive. In this context, there are much more effective and already industrially established methods other than vibration welding to obtain the same result (e.g., radio frequency curing etc.). However, a certain number of additives were also added to the bondline to test if wood welding was rendered easier or not: 1) sunflower oil to test if a more water-resistant bondline was generated; 2) a water solution of polyflavonoid pine tannin extract to see if by tannin autocondensation some improvement in wet strength of the bondline could be obtained; 3) a compound, polybutylene adipate, reputed as capable of decreasing the glass transition temperature of lignin, hence of rendering easier its flow and used in pulp technology (Li and Sarkanen 2002); and 4) furfural, a compound obtained from agricultural waste and capable of both resinifying by autocondensation to thermosetting, water-resistant resins as well as reacting with lignin. Addition of these additives was done just to see if they had some potential and not to optimize their use. The results obtained are shown in Table 1.

[FIGURE 2 OMITTED]

The addition of sunflower oil rather than improving the situation made it much worse. In short, the presence of the oil on the wood surface functions as a lubricant and as a consequence welding time had to be lengthened just to bond to a much lower strength than the control. There was no noted resistance to water. The polybutylene adipate water solution did not work at all: the water evaporated almost instantly as welding started, depositing dry salt at the interface that impeded bonding. Pine tannin and furfural performed better. Pine tannin dry results were comparable to the control but the joints still showed some resistance (too weak to measure) and they did not fall apart after 4 hours in cold water. Furfural, in small amounts, also gave results comparable to the control, but yielded joints that were still holding up rather well after 4 hours in cold water, although still far from an exterior-type performance. Furfural autocondensation to furanic resins during the considerable increase in temperature of the bondline due to vibration welding (Gfeller et al. 2003) is the cause of this behavior. Self-condensation of the traces of furfural and other furanics produced from the carbohydrates by the temperature increase during welding has already been shown to contribute to the strength of the joint (Gfeller et al. 2003). A combination of pine tannin and furfural, which would be capable of giving better results, was not considered because this mix is a recognized type of thermosetting adhesive (Pizzi 1983, 1994) and hence, even though all compounds are of natural origin, it would be disqualified by the same argument advanced for synthetic thermosetting adhesives outlined at the beginning of the last paragraph. It must be pointed out that in the case of the more successful additives, the results reported are notoptimized ones and that further optimization of the relevant welding parameters are likely, in some cases, to further improve the results.

[FIGURE 3 OMITTED]

Chemical cross-linking reactions tied to the presence of traces of furfural and other furanic compounds from carbohydrate reactions produced by the temperature increase during welding contribute to the strength of the joint (Gfeller et al. 2003). During welding, the temperature of the joint ranges from 170[degrees]C to as high as 210[degrees] to 220[degrees]C (Gfeller et al. 2003, Leban et al. 2004, Pizzi et al. 2004, Properzi et al. 2004). Furfural, methylfurfural, and furanic derivatives can be produced in wood through both ionic and radical reactions (Fengel and Wegener 1989). Since water evaporates rapidly from the bondline during welding due to the high temperature reached (210[degrees] to 220[degrees]C), the possibility of ionic reactions in solution could be rather limited. The main route for the formation of furfural and other furanic compounds is then likely to be by radical decomposition of wood carbohydrates. The thermal treatment of xylans and polyoses in general, at temperatures higher than 150[degrees]C, proceeds by a radical route to produce furfural and hydroxymethylfurfural (Fengel and Wegener 1989). The same route yields levulinic acid from both these two compounds. CP-MAS [.sup.13]C NMR of welded wood bondline indicated that welding produced furanic compounds, probably furfural. Traces of levulinic acid were also observed in the same spectra. Equally, lignin is subject to radical degradation reactions at elevated temperatures (Fengel and Wegener 1989). Its phenoxyl radicals are very stable and as a consequence can be observed with relative ease by ESR testing. Their presence would indicate that radical reaction routes are of importance in friction vibration wood welding. To check the extent and influence of the radical route to furanic compounds and in lignin, ESR testing of wood surfaces, which were sanded to increase the surface temperature as for welding, was carried out. The radical concentration of wood surfaces before and after sanding is shown in Figure 3. Replicate radical concentration curves of wood surfaces before and after testing are superimposed in Figure 3. In all cases, a friction action has clearly increased radical concentration in all the replicate tests shown. The curves indicate that the radicals present are phenoxyl radicals, as they are centered in the range 3345[+ or -]50 Gauss exclusively characteristic of phenoxyl radicals (Bielski and Gebicki 1989). They are, hence, phenoxyl radicals belonging to lignin. This indicates that radical reactions occur and improve the joint strength after welding (Gfeller et al 2003).

Table 1. -- Influence of additives on mechanical performance of
vibration-welded wood. Ten welded specimens for each case. Welding for 3
seconds except where otherwise indicated.

                          Avg. tensile strength  SD (a)  SD percentage
                                  (MPa)                        (%)

Control                           8.72           1.1           12.5
Sunflower oil (b)                 5.28           0.54          10.3
Polybutylene adipate (c)          --             --            --
Pine tannin (d)                   7.95           1.63          20.4
Furfural (e)                      8.81           1.04          11.8

                          Standard error  Max. tensile
                                              (MPa)

Control                        0.42           11.22
Sunflower oil (b)              0.45            6.25
Polybutylene adipate (c)       --             --
Pine tannin (d)                1.27           10.91
Furfural (e)                   0.41           11.11

(a) SD = standard deviation.
(b) Welding had to be increased to 5.5 seconds instead of 3 seconds,
falling apart in cold water after 1 hour.
(c) No bonding due to dry out of solution due to heat rise on bondline.
(d) Weak but still holding in water after 4 hours.
(e) Holding strongly in water after 4 hours.

Literature cited

Bielski, B.H.J. and J.M. Gebicki, 1989. Atlas of Electron Spin Resonance Spectra. Academic Press, NY. pp. 349,463,541.

European Standard. 1993. EN 301: Classification for thermosetting adhesives for wood: phenolics and aminoplastics.

__________. 2004. EN 204: Classification for thermoplastic adhesives.

__________. 2003. Method EN 205: Determination of the tensile strength.

__________. 1993. EN 301-1: Tensile strength.

Fengel, D. and G. Wegener. 1989. Wood; Chemistry, Ultrastructure, Reactions. Walther de Gruyter, Berlin, Germany.

Gfeller, B., M. Lehmann, M. Properzi, F. Pichelin, M. Zanetti, A. Pizzi, and L. Delmotte. 2004. Interior wood joints by mechanical friction welding of wood surfaces. Forest Prod. J. 54(7/8):72-79.

Leban, J.-M., A. Pizzi, S. Wieland, M. Zanetti, M. Properzi, and F. Pichelin. 2004. X-ray microdensitometry analysis of vibration-welded wood. J. Adhesion Sci. Technol. 18 (6):673-675.

Li, Y. and S. Sarkanen. 2002. Alkylated kraft lignin-based thermoplastic blends with aliphatic polyesters. Macromolecules 35(26): 9707-9715.

Pizzi, A. 1983. Wood Adhesives Chemistry and Technology. Marcel Dekker Inc., NY.

__________. 1994. Advanced Wood Adhesives Technology. Marcel Dekker Inc., NY.

__________. J.-M. Leban, H. Kanazawa, M. Properzi, and F. Pichelin. 2004. Wood dowels bonding by high speed rotation welding. J. Adhesion Sci. Technol. 18(11):1263-1278.

Properzi, M., J.-M. Leban, A. Pizzi, S. Wieland, F. Pichelin, and M. Lehmann. 2004. Influence of grain direction in vibrational wood welding. Holzforschung 59(1).

S. Wieland

Bozhang Shi

A. Pizzi*

M. Properzi

M. Stampanoni

R. Abela

Xiaoning Lu

F. Pichelin

The authors are, respectively, Project Leader, HSB, Hochschule fur Architektur, Bau und Holz, Univ. of Applied Science, 102 Solothurnstrasse, CH-2504 Biel, Switzerland; Laboratory Head, ESR Analysis, Chemistry Dept., Nanjing Forestry Univ., Nanjing, P.R. China; Professor, ENSTIB-LERMAB, Univ. of Nancy 1, 27 Rue du Merle Blanc, BP 1041, F-88051 Epinal, France; Project Leader, HSB, Hochschule fur Architektur, Bau und Holz, Univ. of Applied Science; Scientist and Scientist, Synchrotron Radiation Research Dept., Paul Scherrer Inst., 5232 Villigen PSI, Switzerland; Professor, College of Wood Science and Technology, Nanjing Forestry Univ., and Head of Panels R & D Dept., HSB, Hochschule fur Architektur, Bau und Holz, Univ. of Applied Science. This paper was received for publication in October 2003. Article No. 9767.

*Forest Products Society Member.