In situ identification of the molecular-scale interactions of phenol-formaldehyde resin and wood cell walls using infrared nanospectroscopy

Abstract

Atomic force microscope infrared spectroscopy (AFM-IR), contact resonance AFM (CR-AFM) measurement, and nanoindentation were combined to identify the interactions between wood cell wall and phenol-formaldehyde resin (PF) on the nanoscale. Significant differences in both chemical structure and mechanics were observed among the cell wall, resin, and interphase regions, indicating that PF resin had diffused to the cell wall effectively. In particular, the maximum penetration depth of the resin in the glueline reached approximately 3 μm, showing that PF resin was able to penetrate into the wood's secondary cell wall. The penetrating resin molecules not only dispersed within the cell wall but also reacted with cell-wall polymers, resulting in an increase in the elastic modulus and hardness of the wood cell wall. Nanoscale mechanical interlocks also formed between the resin and the wood cell wall in the interphase region, which may improve adhesion performance in wood-based composites.

---

Introduction

Adhesive bonding is a technology that has been extensively used in almost all fields of industry to assemble different substrates, for instance, engineering implants for enhanced osseointegration and bone repair, composite, natural-fiber reinforced polymer composites, and multilayer coatings.^1–3 A concept that has been gaining much support is that the overall performance of these multicomponent materials depends significantly on the quality of the “interphase” between the contacting solids.⁴ Therefore, it is clear that an accurate evaluation of the properties of the interphase region is essential for optimizing the design of these multicomponent materials. In biocomposites particularly, the geometry and performance of the interphase region, however, are prone to being influenced by the penetration and interactions of the low-molecule polymeric materials in biomaterials like natural fibres due to their porous structure and many inherited anatomical features.

Wood is a typically anisotropic cellular material with numerous levels of porous structure; wood has been widely used to develop high-quality and environmentally friendly materials to substitute for steel and concrete in larger structures, such as midrise buildings and bridges.⁵ In general, the penetration of polymeric materials in wood materials occurs on two or more levels of scale: gross penetration occurs through the lumens and interconnecting pits of the wood cells, filling the cell lumen; in addition, nano-penetration occurs within the wood cell wall.⁶ In the last few decades, research work has concentrated on observing the penetration phenomena at the micrometer, millimeter, or larger scale with optical microscopy and electron microscopy.^7–10 Recently, more attention has been paid to nano-penetration in cell walls because it is at the nanometer level that the resin molecules may interact with wood cell wall. Advanced tools, such as nanoindentation, contact-resonance force microscopy, atomic force microscopy (AFM), scanning thermal microscopy (SThM), and X-ray tomography (XT) have been employed to study resin penetration in cell walls.^11–15 However, those purely optical or electronic observational methods cannot provide exact chemical information at sub-micrometer resolutions. That is, the chemical characteristics at the interface between wood cell wall and resin have been difficult to determine. Nevertheless, understanding the chemical composition at the interface between wood materials and resin at submicron spatial resolution is critical for understanding how resins interact with the cell-wall polymer.

Infrared spectroscopy (IR) is perhaps the most widely used technique to gain insight into the chemical nature and molecular composition of materials. However, the spatial scale at which many important chemistries take place in wood cell wall is often well below the diffraction limit of conventional infrared spectroscopy (3–10 μm), so molecular-scale interactions cannot be easily studied in situ. In light of this, an improved technique, AFM-based infrared spectroscopy (AFM-IR), has been recently developed, enabling direct macromolecular characterization. AFM-IR leverages the advantages of both AFM and IR spectroscopy, rapidly acquiring nanoscale spatial resolutions (sub-100 nm) of chemical spectra to obtain both structural and chemical information about a sample.^16,17 Furthermore, high-resolution chemical images of the sample surface can also be collected by AFM-IR.^18,19 The technique has been successfully used to measure IR chemical absorption spectra and chemical imaging on microbiological,²⁰ biological,²¹ polymer, and polymer-based composites.²² In the present study, the specific molecular-scale interactions between phenol-formaldehyde resin (PF) and wood cell wall was assessed directly for the first time by using infrared nanospectroscopy (AFM-IR). Moreover, the mechanics of each phase in the wood–resin interphase region were also analyzed by nanoindentation to enhance understanding of the interaction mechanism.

Experimental

Materials

Loblolly pine (Pinus taeda L.) wood was collected from Crossett, Arkansas, USA; the wood samples were cut parallel to the direction of the grain and sawn into blocks of 5 × 5 × 10 mm³. The commercial PF resin used in this study was obtained from Arclin Corporation (Mississauga, Canada). The PF resin had a viscosity of 150 mPa s at 25 °C, a solid content of 45%, a density of 1.025 g mL⁻¹, and a pH value of 10.5.

Preparation of wood–resin bonds

To obtain the interphase between wood and resin, wood–resin bonds were prepared by using wood blocks and PF resin. The surfaces of the wood blocks in both the radial and tangential directions were smoothed by microtome before gluing. The tangential surface of one wood block and the radial surface of another block were spread with 75 g m⁻² PF resin and were then glued together at a consistent temperature of 140 °C according to the manufacturer's recommendations at a specific pressure of 0.8 MPa and a curing time of 30 min.

AFM-IR analysis

Basic principles of AFM-IR. As shown in Fig. 1a, a pulsed, tunable IR laser source illuminates a thin sample from the top side. When the IR laser is tuned to a wavelength corresponding to absorption by the sample, the absorbed radiation induces a rapid thermal expansion of the absorbing region of the sample. The rapid thermal expansion pulse then excites resonant oscillations in the AFM cantilever through contact with the AFM tip. Hence, the resonant oscillation of the cantilever is directly proportional to the absorption coefficient of the sample. The instantaneous position of the cantilever is then determined by reflecting a visible laser off the top of the cantilever. The AFM tip can sense and map variations in thermal expansion from IR absorption to spatial resolutions greater than 100 nm. The high-resolution chemical images can be collected by tuning the IR laser source to a single wavelength matching the frequency of a specific functional group absorption band across the sample surface.

Fig. 1 AFM-IR experimental setup: (a) scheme of the AFM-IR measurement; (b) the sample obtained from wood–resin bonds; (c) schematic representation of the interphase region between wood cell wall and resin.

Sample preparation. A gently sloping (30°) apex was created using a microtome on the transverse surface of the wood–resin bond. The bonded sample was mounted into a holder designed for ultramicrotomy in order to obtain an ultra-thin slice of 200 nm thickness with a diamond knife (Fig. 1b). The slice was transferred onto a flat silicon substrate with dimensions of 5 mm × 5 mm.

AFM-IR analysis. A nanoIR2™ AFM-IR instrument (Anasys Instruments Corp., Santa Barbara, USA) was used to collect the spatially resolved IR spectra. The flat silicon substrate with the ultra-thin sample was mounted on the sample stage of the AFM-IR instrument. Using an incident light micrograph as a guide, locations of the interphase region between wood and resin could be spectroscopically examined. As shown in Fig. 1b and c, two typical types of interphase regions on wood–resin bond were selected for AFM-IR analysis: type I is located at the glueline, i.e., PF resin has penetrated from the outside directly into the wood cell wall; type II is located away from the glueline, i.e., the PF resin filling the cell lumen through the porous structure of wood has penetrated from the inside into the wood cell wall. All AFM-IR data were obtained in contact mode with an access-C cantilever (AppNano, Mountain View, USA). The tunable IR laser produced laser pulses of 10 ns duration at a repetition rate of 1 kHz. The AFM-IR spectra were collected from 1000 to 1800 cm⁻¹, with a data point spacing of 4 cm⁻¹. One hundred twenty-eight cantilever ringdowns were covered for each data point. AFM-IR images were then collected in contact mode at a scan rate of 0.1 Hz using a gold-coated silicon nitride probe. The mechanical properties of the same locations were measured using a different AFM cantilever in Lorentz contact resonance (LCR) mode.²³

Nanoindentation

The residual sample from which the 200 nm-thick AFM-IR section was cut was glued on a metal disk for the nanoindentation test. A Hysitron TriboIndenter system (Hysitron Inc., USA) equipped with scanning probe microscopy (SPM) and Berkovich indenter was used. Using the scanning probe micrographs as a guide, the same locations that were used for the AFM-IR experiments were precisely chosen to investigate the nanoindentation elastic modulus and hardness of the control cell wall, resin, and cell wall penetrated with resin in the interphase regions (Fig. 2b–d). At least 20 valid indentations were performed on the cell wall and resin in load-controlled mode using a three-segment load ramp: load application within 5 s, hold time 5 s, and unload time 5 s. The peak load was 400 μN for all indents in the experiment. The reduced elastic modulus (E_r) and hardness (H) were calculated according to the method of Oliver and Pharr (1992),²⁴ as follows:where P_max is the peak load, and A is the projected contact area at peak load.where E_r is reduced modulus (which accounts for the fact that elastic deformation occurs in both the sample and indenter); S is initial unloading stiffness; and β is a correction factor correlated to the indenter's geometry (β = 1.034 for a Berkovich indenter).

Fig. 2 Microscopic images showing the samples and positioning of indents: (a) incident light micrographs of the wood–resin bond from which the AFM-IR section was cut; (b–d) SPM images of test locations and the indents after characterization by nanoindentation.

Results and discussion

AFM-IR spectra and chemical imaging

Obtaining chemical information about the interphase region is essential to understand how resins interact with wood cell-wall polymers. Fig. 3 illustrates the AFM topography images and a series of AFM-IR spectra collected along a line with a length of 4 μm from cell wall to resin. The black IR spectra were recorded in the cell wall region, the blue spectra in the resin region, and the red spectra in the interphase region (Fig. 3b and d). The black and blue IR spectra suggested that the chemical nature of the cell wall is clearly different from the resin. As is known, wood cell wall is mainly composed of cellulose, hemicelluloses, and lignin. The prominent band at 1040 cm⁻¹ and 1263 cm⁻¹ on the black IR spectra (Fig. 3b) arise from the C–O stretching vibration in cellulose and hemicellulose, and the absorption at 1724 cm⁻¹ is indicative of the C=O stretching vibration in acetyl groups in the hemicellulose.²⁵ Phenol-formaldehyde resins (PF) are synthetic polymers obtained by the additive reaction and polymerization of phenol with formaldehyde (Fig. 4). The intensive band at 1722 cm⁻¹ originates from the C=O stretching vibration in formaldehyde, while the absorptions at 1603 cm⁻¹ and 1470 cm⁻¹ are attributable to the C=C stretching vibration in the aromatic ring of the phenol.²⁶ It is intriguing that a gradual change in the intensity of those characteristic peaks as a function of distance from the resin region to cell wall region was observed in the AFM-IR spectra. As shown in Fig. 3b, spectra 5 and 9 (close to the boundary) are similar to spectra 1 and 13 (farther from the boundary), respectively, while a number of spectral modifications appeared on the spectra of 6, 7, and 8 even though the general aspect of the spectra remained unchanged. The intensity of the C=O stretching at 1724 cm⁻¹ decreased significantly in the interphase region. The corresponding changes in the C=O stretching of cell wall are due to the reaction of the acetyl groups (–CHO) in the hemicellulose with phenolic resin.²⁷ On the other hand, the reaction of the cellulose monomer units and the formaldehyde in the PF resin may be another influential factor, but a negative one, on the intensity of the peak at 1724 cm⁻¹ (Fig. 4b). Besides the changes of absorption intensity, the stretching absorption of carbonyl and carboxyl groups at 1724 cm⁻¹ of wood cell wall gradually moves to a lower wavenumber at 1722 cm⁻¹ when mixed with PF resin, indicating that PF resin accelerates the oxidation of cell-wall materials.²⁸ The decrease of the intensity of C=C stretching vibration at 1603 and 1470 cm⁻¹ can be attributed to the addition of wood materials with fewer C=C bands. Nevertheless, a relatively broader band between 1150 and 1000 cm⁻¹ appeared in all three spectra (6, 7, and 8), corresponding to the asymmetric stretching vibration of C–O–C aliphatic ether in the interphase region. In agreement with the literature, the cross-linking reactions of –OH groups of wood polymer and phenol formaldehyde resin occurred in the composite system at the cell wall level (Fig. 4c).

Fig. 3 AFM height images and AFM-IR spectra: (a) and (c) AFM height images of the two types of test locations in the bond (the positions of the spectral measurements with 100 nm spacing); (b) and (d) representative AFM-IR spectra recorded from the correspondingly colored locations on the AFM image.

Fig. 4 Schematic chemical reactions of cell-wall polymers and phenol formaldehyde resin: (a) synthesis of phenol formaldehyde resin; (b) reaction of cellulose and residual formaldehyde; (c) reaction of cellulose and resin oligomer.

As expected, similar changes are observable on the AFM-IR spectra in Fig. 3d. However, the characteristic bands in the overall spectra of the phenol-formaldehyde resin in the cell lumen were weaker than those of the resin in the glueline as well as those of the bands between 1150 and 1000 cm⁻¹ in the spectra of interphase region, indicating that the PF resin in the cell lumen had a less efficient interaction with the cell-wall polymer. On the other hand, the C=C stretching vibration at 1603 cm⁻¹ and 1470 cm⁻¹ in the aromatic ring of the PF resin in the cell lumen appeared band-shifts. As discussed in our previous work, organic extractives composed of fats and waxes in loblolly pine wood are able to retard the chemical reaction of PF molecules with the cell wall.²⁹ In this experiment, the PF resin in the cell lumen contacted the wood extractives adequately when it flowed through the porous structure of the wood (Fig. 1c), i.e., the chemical nature of the resin in the cell lumen was modified. Furthermore, wood extractives may negatively affect the infiltration capacity of the PF resin. The width of the type I interphase region reached approximately 1.2 μm, according to the distances from spectra 5 to 9, while the width of the type II interphase region was about 0.5 μm.

To track the diffusion of PF resin in the cell wall precisely, the wavenumber of the IR laser source from the AFM-IR instrument was fixed at the wavenumber of the characteristic peak of PF resin to scan and collect the chemical images of the same areas on the AFM height images (Fig. 3a and c). As shown in Fig. 5a and b, the AFM-IR chemical images of the test locations were obtained when the wavenumber was fixed at 1603 cm⁻¹, which corresponds to an absorption peak that arises from the C=C stretching vibration in the aromatic ring of PF resin. The AFM cantilever scanned over the specimen's surface and measured absorption at each scan point. In each absorption image, the red color indicates regions of stronger IR absorption at this wavelength and hence serves to localize the penetration of PF resin into the cell wall. It is obvious that the PF resin has penetrated into the cell wall from both the glueline and cell lumen. In Fig. 5a, most of the resin molecules interpenetrated in the cell wall had a strong IR absorption at 1603 cm⁻¹, implying that these interpenetrating molecules occupied the free volume within the cell wall first and then self-polymerized after curing at high temperature. A mechanical interlocking effect as “fingers” of the cured resin extended from the glueline into the cell wall, which has a beneficial effect, generating intrinsic adhesion forces across the interphase. Some areas in the interphase region had a weak IR absorption at 1603 cm⁻¹ (yellow domains), which can be attributed to the chemical cross-linking reactions between the resin molecules and the cell-wall polymeric components. As the curing treatment progressed, the self-polymerization of resin and the chemical interactions between resin and cell wall appear to have restricted the flow of resin and finally formed a stable interphase region. However, it can be observed from Fig. 5b that the resin in cell lumen barely penetrated into the cell wall. Most resin molecules were concentrated near the boundary between the cell wall and the resin.

Fig. 5 Corresponding IR amplitude images: (a) corresponding chemical mapping of the C=C bond (1603 cm⁻¹) on the type I; (b) corresponding chemical mapping of the C=C bond (1603 cm⁻¹) on type II locations.

The average penetration depth was measured according to the high-resolution chemical images in this experiment. Image processing software was applied to extract the image's edge feature points using the SOBEL operator, and then a segmented image was formed. On Fig. 6a and b, 20 data were collected along the length of the interphase to calculate the average penetration depth of the PF resin. The average penetration depths of the resin from the glueline and cell lumen were about 1.3 μm and 500 nm, respectively. In general, wood cell walls are composed of layers of different thickness, including the thin primary wall and the thick secondary wall, which has three layers (S₁, S₂, and S₃). The S₂ layer generally occupies about 80% of the cell wall (in terms of thickness) and it dominates the natural properties of the cell wall.³⁰ As shown in Fig. 2, the thickness of the wood cell wall seen in this study was approximately 5 μm. The maximum penetration depth of the resin from the glueline and the cell lumen reached approximately 3 μm and 1 μm, respectively, indicating that PF resin had penetrated into the S₂ layer.

Fig. 6 (a and b) Segmentation images of type I and type II interphase region; (c) measurement of penetration depth of resin into cell wall. SD is standard deviation.

Mechanical response

The mechanical properties of the same locations displayed in Fig. 3a and c were measured in Lorentz contact resonance (LCR) mode. The individual mechanical response spectra were collected along the line with 150 nm spacing. According to the mechanical spectra, the contact resonance peak frequency was plotted as a function of distance from the boundary to the resin and cell wall regions. It can be observed clearly from Fig. 7a and b that the cell wall had a higher peak contact resonance frequency, corresponding to higher stiffness as compared to the PF resin, and also, there was a gradient of mechanical stiffness in the interphase region. This result is consistent with the AFM-IR spectral results shown in Fig. 3, where the IR spectra in the three regions are also different. In Fig. 7a, the measured stiffness of the location within 300 nm of the boundary in the interphase region approximated that of the resin region, resulting from the accumulation of amounts of resin molecules as shown in Fig. 5a. In other words, both the chemical and mechanical properties of this area were similar to those of the PF resin. However, the contact resonance peak frequency increased significantly as the distance increased from 300 nm to 1 μm, which further confirmed the existence of nanoscale mechanical interlocks or an interpenetrating polymer network (IPN) between the resin molecules and the wood polymer. Accordingly, the mechanical properties became more and more similar to those of the cell wall due to the lesser penetration of the resin molecules with the increase of distance to the boundary, especially when the distance exceeded 1.5 μm. Similarly, there was an obvious increase in peak frequency at the location close to the boundary in the type II interphase region (Fig. 7b). Nevertheless, the area enhanced by resin in the type II interphase region was much smaller as compared to the area in the type I interphase region. This is in agreement with the results of the AFM-IR spectra and chemical images, showing that PF resin from the glueline had better penetration and distribution in the wood cell wall.

Fig. 7 Plots of contact resonance peak frequency corresponding to the distance from the boundary between the wood cell wall and resin: (a) type I location; (b) type II location.

Nanoindentation

Fig. 8 illustrates the average elastic modulus (E_r) and hardness (H) values of the S₂ layer of the control cell wall and the resin-penetrated cell wall (types I and II), and the resins in glueline and cell lumen, with error bars representing the standard deviations. The E_r and H values obtained from the control cell wall and PF resin were consistent with the data in the literature.^31,32 It was observed that the E_r values of the cell walls (type I and type II) (Fig. 2c and d) increased by 7.9% and 3.1%, respectively, and that the hardness increased by 24.9% and 11.5%, respectively. The results presented here support the findings of AFM-IR measurement, in which a great number of resin molecules have diffused into the S₂ layer. The interpenetrating polymer network (IPN) between resin molecules and the cell-wall polymer had a positive effect on the mechanical properties of wood cell wall. On the other hand, the PF resin with a higher density of 2.28 g cm⁻³ (calculated according to the solid content and density of liquid resin) can increase the density of the wood cell wall in the interphase region significantly after curing, resulting in the higher mechanical properties. In particular, the cell wall in the type I region revealed significantly improved mechanics compared to that of the cell wall in the type II region, resulting from more resin molecules directly penetrating into cell wall from the glueline. Furthermore, the hardness increased significantly more than the elastic modulus, which can be ascribed to the lower E_r and higher H of the resin polymer itself.

Fig. 8 Nanoindentation mechanics of PF resin and wood cell wall: (a) reduced elastic modulus; (b) hardness.

Conclusions

AFM-IR and nanoindentation were applied to accurately characterize the chemical and mechanical properties of the interphase region between wood cell wall and PF resin at nanometer spatial resolutions. PF resin penetrated into the S₂ layer of wood cell wall; in particular, the resin diffused into the cell wall from the glueline effectively. The resin molecules not only occupied the free volume within the cell wall but also formed chemical interactions between the resin and the cell-wall polymers. The effective penetration of resin contributed greatly to the mechanical properties of wood cell wall. The nanoscale interpenetrating polymer network (IPN) between resin molecules and the cell-wall polymers expanded as “fingers” of the cured resin extended from both the glueline and cell lumens into the cell wall, which was beneficial to the adhesive forces across the interphase.

Acknowledgements

The authors gratefully acknowledge the financial support of the project funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions, UTIA 2015 Innovation Grant, Foundation of Zhejiang Key Level 1 Discipline of Forestry Engineering (2014lygcz005), Elite Program of South Lake Taihu in Huzhou, and the Natural Science Foundation of China (No. 31570552).

References

1. R. Agarwal and A. J. García, Adv. Drug Delivery Rev., 2015, 94, 53–62.
2. F. Ansari, M. Skrifvars and L. Berglund, Compos. Sci. Technol., 2015, 117, 298–306.
3. J. Karger-Kocsis, H. Mahmood and A. Pegoretti, Prog. Mater. Sci., 2015, 73, 1–43.
4. A. Baldan, Int. J. Adhes. Adhes., 2012, 38, 95–116.
5. A. Özçifçi, Build. Environ., 2007, 42, 1510–1514.
6. F. A. Kamke and J. N. Lee, Wood Fiber Sci., 2007, 39, 205–220.
7. T. Furuno and H. Saiki, Mokuzai Gakkaishi, 1988, 34, 409–416.
8. A. J. Bolton, J. M. Dinwoodie and D. A. Davies, Wood Sci. Technol., 1988, 22, 345–356.
9. A. P. Singh, B. Dawson, C. Rickard, J. Bond and A. Singh, Microsc. Anal., 2008, 22, 5–8.
10. C. Xing, B. Riedl, A. Cloutier and S. M. Shaler, Wood Sci. Technol., 2005, 39, 374–384.
11. S. H. Lee, S. Wang, G. M. Pharr and H. Xu, Composites, Part A, 2007, 38, 1517–1524.
12. S. S. Nair, S. Wang and D. C. Hurley, Composites, Part A, 2010, 41, 624–631.
13. S. Wang, S. D. Nair, D. Hurley and S. H. Lee, Adv. Mater. Res., 2010, 123–125, 403–406.
14. J. Konnerth, D. Harper, S. H. Lee, T. G. Rials and W. Gindl, Holzforschung, 2008, 62, 91–98.
15. J. L. Paris, F. A. Kamke, R. Mbachu and S. K. Gibson, J. Mater. Sci., 2014, 49, 580–591.
16. K. Kjoller, J. R. Felts, D. Cook, C. B. Prater and W. P. King, Nanotechnology, 2010, 21, 185705.
17. H. Cho, J. R. Felts, M. F. Yu, L. A. Bergman, A. F. Vakakis and W. P. King, Nanotechnology, 2013, 24, 444007.
18. A. Dazzi, R. Prazeres, F. Glotin and J. M. Ortega, Ultramicroscopy, 2007, 107, 1194–1200.
19. F. De Angelis, G. Das, P. Candeloro, M. Patrini, M. Galli, A. Bek, M. Lazzarino, I. Maksymov, C. Liberale, L. C. Andreani and E. Di Fabrizio, Nat. Nanotechnol., 2010, 5, 67–72.
20. C. Mayet, A. Dazzi, R. Prazeres, J. M. Ortega and D. Jaillard, Analyst, 2010, 135, 2540–2545.
21. E. Kennedy, R. Al-Majmaie, M. Al-Rubeai, D. Zerulla and J. H. Rice, RSC Adv., 2013, 3, 13789–13795.
22. H. S. Purohit and L. S. Taylor, Mol. Pharmaceutics, 2015, 12, 4542–4553.
23. C. Marcott, M. Lo, Q. Hu, E. Dillon, K. Kjoller and C. B. Prater, Am. Lab., 2014, 46, 23–25.
24. W. C. Oliver and G. M. Pharr, J. Mater. Res., 1992, 7, 1564–1583.
25. D. De, B. Adhikari and D. De, Polym. Adv. Technol., 2007, 18, 72–81.
26. I. Poljanšek and M. Krajnc, Acta Chim. Slov., 2005, 52, 238–244.
27. P. Ghosh, S. Biswas and C. Datta, J. Mater. Res., 1989, 24, 205–212.
28. G. He and B. Riedl, Wood Sci. Technol., 2004, 38, 69–81.
29. X. Wang, Y. Li, S. Wang, W. Yu and Y. Deng, J. Adhes., 2016 DOI:10.1080/00218464.2015.1130629.
30. A. C. Wiedenhoeft and R. B. Miller, Handbook of wood chemistry and wood composites, Florida, 2005.
31. Y. Meng, Y. Xia, M. Y. Timothy, Z. Cai and S. Wang, RSC Adv., 2015, 5, 47538–47547.
32. F. Stoeckel, J. Konnerth and W. Gindl, Int. J. Adhes. Adhes., 2013, 45, 32–41.

---