Soy protein isolate was supplied by Ruikang Biotechnology Co., Ltd. (Dezhou, China). Commercial glycerol diglycidyl ether (GDE, technical grade), hexamethylene diamine 98%, and dimethyl carbonate were bought from Sigma-Aldrich (Saint Louis, France).

The SPI-NIPU adhesive was prepared as follows: 80 g SPI is dissolved in 400 g of deionized water under magnetic stirring, then 54 g dimethyl carbonate is added and the mixture heated to 60°C for 120 min. Then 105 g hexamethylenediamine are then added and heated to 90°C for another 2 h, then cooled to room temperature. The mixture can be roto-evaporated under reduced pressure at 55°C to eliminate the excess of water and to obtain a final resin solids content of 45%. with a viscosity of 26.9 ± 1 Pa•s.

The resins were tested by thermomechanical analysis. The samples were prepared by applying each adhesive between two beech wood plies, with dimensions of 21 mm × 6 mm × 1.1 mm. These beech-resin-beech sandwiches were tested in non-isothermal mode between 25°C and 250°C at a heating rate of 10 °C/minute with a Mettler Toledo 40 TMA equipment (Mettler Toledo, Zurich, Switzerland). They were tested in three-point bending on a span of 18 mm exercising a force cycle of 0.1/0.5 N on the specimens, with each force cycle of 12 seconds (6 s/6 s). The classical mechanics relationship between force and deflection

E=[L3/(4bh3)][F/(fwood−fadhesive)]

where L = the length of the sample; b = its width; h = its thickness; F = the force applied f_wood the deflection of the wood and f_adhesive the deflection of the wood adhesive sandwich under test. This equation allows the calculation of the Young’s modulus E for each case tested. Such a measuring system has been introduced and is used to follow the progressive hardening of the adhesive with the increase of temperature and to indicate comparatively if an adhesive system is faster or slower hardening and if it gives stronger joints than another one.

All of the samples were analyzed with a PerkinElmer Frontier ATR-FT-MIR provided by an ATR Miracle diamond crystal. The powder and liquid samples were laid on the diamond eye (1.8 mm) of the ATR equipment and the contact for the sample was ensured by tightly screwing the clamp device. Each extract was scanned registering the spectrum with 32 scans with a resolution of 4 cm⁻¹ in the wave number range between 600 and 4000 cm⁻¹.

Samples for matrix assisted laser desorption ionization time-of-flight (MALDI-ToF) analysis were prepared by first dissolving 7.5 mg of sample powder in 1mL of a 50:50 v/v acetone/water solution. Then 10 mg of this solution was added to 10 µL of a 2,5-dihydroxy benzoic acid (DHB) matrix. The locations dedicated to the samples on the analysis plaque were first covered with 2 µL of a NaCl solution 0.1 M in 2:1 v/v methanol/water, and predried. Then 1.5 µL of the sample solution was placed on its dedicated location and the plaque was dried again. Red phosphorous was to standardize the MALDI equipment. MALDI-ToF spectra were obtained using an Axima-Performance mass spectrometer from Shimadzu Biotech (Kratos Analytical Shimadzu Europe Ltd., Manchester, UK) using a linear polarity-positive tuning mode. The measurements were carried out making 1000 profiles.

The performance of the SPI-NIPU adhesive was tested by preparing laboratory plywood panels and evaluating their shear strength dry, after 24 hours cold water soaking, and after 3 h in water at 63°C. The plywood panels of 400 mm × 200 mm × 5 mm were prepared for each adhesive using 2 mm poplar (Populus tremuloides) veneers. The glue spread used was of 300 g/m² double glue line, and hot-pressing time was of 6 minutes at 1.25 MPa pressure at a temperature of 200°C based on the consequence of the TMA trace. After hot pressing the plywood was stored under ambient conditions (20°C and 12% relative humidity) for 48 hours then it was cured and tested the dry shear strength, 24 h cold water soaking strength and for strength after 3 h at 63°C in water according to China National Standard GB/T 14074 (2006) [38], China National Standard GB/T17657 (1999) [39], and European Norm EN 636:2012 (2012) [40]. To reinforce the SPI-NIPU plywood also panels to which 10% or 15% by weight of a biosourced glycerol diglycidyl ether (GDE) were added to the SPI-NIPU based on solids were also pressed and tested under the same conditions.

The MALDI ToF analysis shows that SPI-NIPU urethane linkages do form with at least some of the soy protein amino acids. However, the analysis also shows that for SPI it is not a straightforward reaction to NIPUs as described for other bio-sourced materials (Figs. 1a–1d). The assignments for the relevant peaks of the spectra are shown in Tab. 1. Thus, the peaks in Fig. 1a at 189 Da and 215 Da can be assigned to the reaction product of leucine with DMC without and with Na⁺ being present, respectively, hence to a structure of Type (I):

The same structure is present for 284 Da and 311 Da which can be assigned to triptofan-DMC without or with Na⁺. At 271 Da the structure can be attributed to serine-DMC-hexamethylene diamine +Na⁺. However, at 271 Da and 295 Da the structure that can be deduced is leucine-DMC-hexamethylene diamine without or with Na⁺ of Structure (II)

It must be clearly pointed out that while a urethane linkage, namely R-NH-COO-R is formed, the structure so formed is also a lactam, as indicated by the -CO-O-CO- structure. The same type of urethane/lactam linkage is found also at 339 Da and 368 Da these being respectively arginine-DMC-diamine + Na⁺ and tryptophan-DMC-diamine + Na⁺. The small peak at 453 Da indicates the formation of a different oligomer, namely: leucine-DMC-diamine-DMC-leucine +Na⁺, hence a more complex structure of Type (III):

The same type of structure but involving different amino acids is present at 495 Da, 539 Da and 602 Da, these oligomers being respectively leucine-DMC-amine-DMC-arginine + Na⁺, arginine-DMC-amine-DMC-arginine + Na⁺ and tryptophan-DMC-amine-DMC-tryptophan + Na⁺.

It must be pointed out that:

This reaction, as seen, leads to urethanes that eventually share a lactam structure. To avoid the presence of urethane/lactam type structure then a longer carbonate, cyclic or not should be used.

2. That after urethane dimers are formed such as those of the peaks at 453 Da, 495 Da, 539 Da and 602 Da the reaction can further proceed either (i) By reactions forming amides as the NH₂ usable for reaction with DMC do not lead to urethanes, or (ii) By further reaction of the residual NH groups in the urethane linkages with DMC-aminoacid. In effect, there is a peak at 610 Da (with Na⁺) assigned to a compound formed in this latter manner by reaction of a third leucine of structure (IV)

And even the peak at 746 Da (without Na⁺) of assigned structure (V):

Similar peaks for other soy protein amino acids are also present.

There are then some urethane linkages formed, more than enough to cross-link the protein, even if these are coupled with lactam-like structures. Thus, once the two NH₂ of the diamine are saturated then any further reaction can proceed either (i) Through the formation of amides, hence polyamides [41], or (ii) By a chain reaction with any residual -NH- group that is obtained when forming a urethane linkage, or (iii) By using a polyamine rather than a diamine. This means that in any case the hardened network formed is most likely to be cross-linked by a mix of urethane bridges and amides ones.

Parasite side reactions also appear to occur as indicated by the peaks at 256 Da (VI) and 398 Da (VII) assigned to structures derived by the reaction of DMC with the diamine.

And

Although these reactions form also compounds that are platforms for further urethane linkages to be formed either among themselves or with the SPI amino acids.

In summary rather complex tridimensional structures can results from these combinations.

The reaction of formation of soy protein urethanes, when using DMC as the carbonate, leads to urethanes that eventually share a lactam structure. The use of a longer carbonate, preferably cyclic such as propylene carbonate, does not eliminate the presence of such urethane/lactam type structures.

The FTIR spectra of SPI and of the SPI-NIPU resin are shown in Fig. 2. The peaks at 1703 cm⁻¹, 1543 cm⁻¹ and 1255 cm⁻¹ are characteristic of the asymmetric stretching of urethane linkages, with the 1703 cm⁻¹ for the SPI NIPU spectrum belonging the C=O of the urethane being quite intense and being absent in the spectrum of SPI alone. This conclusion of the presence of urethane linkages is confirmed by the presence of other peaks. Thus, the appearance of the relatively small peak at 1766 cm⁻¹ in the SPI NIPU spectrum is characteristic the R-COO-R of esters derived from the lactam structures, these being esters of the carbamic acid (-NH-COO-) forming the urethanes, confirming the structures assigned by the MALDI-ToF analysis. The big peak at 1639 cm⁻¹ is characteristic of amide groups, the peak being quite marked already for SPI alone, thus describing the amide (peptide) linkages holding together the protein skeleton. This same peak becomes even more dominant in the SPI NIPU spectrum where the amides derived from the reactions of the diamine with DMC and the proteins amino acids generate a great number of amides in the urethane structure superposed to the ones already present in the protein skeleton. The presence of the amide groups is supported by the peak at 1370 cm⁻¹ (thus of-CO-NH-C-) and the wide peak at 3200 cm⁻¹ that mask a number of different group effects. The peak at 1441 cm⁻¹ represents the asymmetric stretch of -CH₂- groups of the alkane chains of the diamine included in the finished network. Lastly the peak at 1390–1400 cm⁻¹ belongs to the protonated N-containing groups, such as -NH₃⁺ on the protein aminoacids. The two sharp peaks at 2900 cm⁻¹ and 2800 cm⁻¹ are assigned to the asymmetric stretching of both the –CH₃ groups of DMC and the asymmetric stretching of the R-CH₂-R chains from the amine, both before and after reaction. They support such assignments in relation to the peak at 1441 cm⁻¹ for the –CH₂- groups. It must be also pointed out that these peaks are particularly intense as they are the superposition of the asymmetric stretching band of both –CH₃ and –CH₂- groups.

The thermomechanical analysis traces of SPI-NIPU adhesive is shown in Fig. 3. The curve of the increase of Young’s modulus as a function of temperature show two upsurges and two peaks of MOE, namely at 130°C and at 220°C, this latter starting at around 185°C. The curves decrease in MOE value between the two peaks. This means that the peak at 130°C indicates the formation of an unstable network that is degraded and transformed if the temperature is increased causing a marked internal rearrangement. The second peak at higher temperature indicates that curing of the SPI-NIPU resin occurs at a high temperature as already reported in previous literature for similar types of adhesives [34,35]. The increase of Young’s modulus has been shown to correlate with the bonding strength of the adhesive [42–48].

It is interesting to speculate what is the cause of the marked decrease between the two MOE peaks in Fig. 3. While it could be an unstable structure, a more likely reason could derive from the formation of linear polymers arranging themselves in a physically entangled network, rather than a chemically cross-linked one. As the temperature increases further, the Brownian movements would make the viscosity of these physically entangled chains markedly decrease and extensively disentangle, the MOE again starting to increase at the much higher temperature of the second peak once chemical cross-linking does finally occur.

Before the TMA curves show a significant increase in the modulus value, the overall trend of the modulus values in the TMA curves of the NIPU adhesive decreases between 50°C and 75°C. This is mainly due to the decrease in viscosity of the adhesive as the temperature increases, resulting in a lower viscosity and bonding strength. Conversely, the adhesive penetrates into the wood, softening or degrading it during heating thus reducing its strength. It is evident from the TMA trace that the curing temperature to be used for such adhesives is relatively higher than what common to-day for equivalent panels bonded with traditional adhesives. After 200°C degradation of the wood substrate is known to start [45–48], thus the second peak is rather high when considering that wood degradation has already started to occur at the same time. The decrease of the curve afterward is the reflexion of the continuing degradation of the wood substrate.

The results of the 3-plys plywood panels prepared are presented in Tab. 2. These indicate that the SPI NPU adhesive by itself just passes at 0.74 MPa the interior grade limit of the relevant China national standards [38–40] that requires a shear strength equal to, or higher than 0.7 MPa. By adding 10% and 15% biosourced GDE the dry shear strength results improved markedly in both cases passing to 1.34 MPa and 1.73 MPa respectively. This indicates that also the dry strength requirements for interior grade plywood of European Norm EN 636:2012 (2012) are satisfied [40] for interior grade plywood. Such a marked shear strength increase indicates that even the addition of smaller percentages of GDE, around 5% by weight, would be sufficient to comfortably pass the requirements for interior grade plywood panels under industrial conditions. The addition of GDE improves the performance due to additional cross-linking. The 24 hours cold water soaking results and the 3 hours at 63°C tests did not pass the requirements for exterior or semiexterior grade plywood as shown in Tab. 2 which are 0.7 MPa and 1.0 MPa for both the China National Standard and in the European Norm, respectively. This is consistent with the results of the TMA reported in Fig. 3, where it is indicated that these adhesives, in the reaction proportions used in this work, need a higher curing temperature to develop all their strength and achieve good results in the wet-type tests. It means also that such an adhesive under the conditions used is possibly better suited to its use in particleboard where higher pressing temperatures are used. An approach to explore to further decrease the high energy of activation of the curing reaction and thus achieve better wet bonding results would be to use small percentages of ionic liquids as already experienced for other adhesives [49].