Local force titration of wood surfaces by chemical force microscopy

Sapwood of spruce wood (Picea abies) from Switzerland was used. For wood functionalization, succinic anhydride and acetone were purchased from Sigma-Aldrich (St. Louis, MI, USA). Pyridine (anhydrous grade) was obtained from VWR (Radnor, PA, USA). Tip functionalization required 11-mercaptoundecanoic acid (HOOC(CH₂)₁₀SH, ≥ 95%, Sigma-Aldrich), 11-mercapto-1-undecanol (OH(CH₂)₁₁SH, ≥ 97%, Sigma-Aldrich) and ethanol (HPLC grade, ≥ 99.8%, Sigma-Aldrich). Buffer preparation in the pH range from pH 2 to pH 12 required water (HPLC plus grade, Sigma Aldrich), monobasic sodium phosphate (NaH₂PO₄ anhydrous, ≥ 99.0%, Sigma-Aldrich), phosphoric acid (H₃PO₄, 85wt%, Sigma-Aldrich), sodium phosphate dibasic dihydrate (Na₂HPO₄*2H2O, ≥ 99.5%, Riedel-de Haën, Seelze, Germany) and trisodium phosphate dodecahydrate (Na₃PO₄*12H₂O, ≥ 98%, Sigma-Aldrich).

Gold coated silicon nitride tips (NPG-10, Bruker, Santa Barbara, CA, USA) and small parts (~ 3 × 3 mm²) of a previously broken gold-coated silicon wafer (AUSW-51, Ø 2″, ~ 50 nm gold layer, NanoAndMore, Wetzlar, Germany) were cleaned for 20 min in an UV/Ozone-cleaner (UV/OZON ProCleaner, Bioforce Nanosciences, Salt Lake City, UT, USA) to remove organic contaminations. Directly after cleaning, tips and wafer were placed in a thiol solution overnight. For COOH-tips, the material was placed in a 1 mM solution of 11-mercaptoundecanoic acid in ethanol. For OH-tips, the material was placed in a 1 mM solution of 11-mercapto-1-undecanol in ethanol. After functionalization, the tips and wafers were removed from the functionalization solution and dipped in ethanol and water to remove excess functionalization chemicals. The success of the thiol functionalization was verified with contact angle measurements in preliminary tests on the thiol coated gold wafer.

Five wood blocks (1 × 1 × 0.5 cm³) were cut out of spruce wood and were chemically treated to increase the amount of carboxyl groups within the whole wood scaffold. A detailed publication can be found following a previously reported publication (Vitas et al. 2018). For this purpose, the wooden blocks were dried at 65 °C before they were immersed in the functionalization solution overnight in a flask equipped with a reflux condenser. The functionalization solution was prepared by adding succinic anhydride in 3 molar equivalent to wood glucopyranose unit (molecular weight = 162 g/mol) in 15 mL pyridine. The next day, the wood blocks were heated up in the solution at 65 °C for 2 h. Subsequently, the wood blocks were removed from the solution and submerged in acetone several times before they were dried again at 65 °C. This wood modification yielded a weight percentage gain of 15.8%, which is defined as the change in dry weight before and after the wood functionalization. The calculated degree of substitution was 0.17 per accessible OH-groups assuming a 50% cellulose content in spruce wood and that one hydroxyl group of the anhydroglucose unit reacts. The resulting material is referred to here as COOH-spruce. Wood samples which were not functionalized are referred to here as native spruce.

Polished surfaces were prepared with an ultramicrotome (Ultracut-R, Leica, Wetzlar, Germany). For this purpose, small pieces were removed from the wood blocks with a razor blade and glued onto metal disc holders. Afterwards, the wood sections were microtomed with three different knives successively, i.e., Trim 45, Histo and Ultra-AFM knife (DiATOME, Nidau, Switzerland). In this study, we polished and analyzed the radial wood surface (Fig. 1a).

Force-distance measurements were performed with an atomic force microscope (Dimension Icon, Nanoscope V, Bruker, Santa Barbara, CA, USA). An OH- or COOH-tip was fixed on the AFM scanner. We used cantilever A (triangular shaped, nominal spring constant = 0.35 N/m, nominal tip radius = 30 nm, resonance frequency 65 kHz) of the AFM probe for all measurements. The deflection sensitivity was calibrated on a sapphire sample and the spring constant was acquired by the built-in thermal tune method (Hutter and Bechhoefer 1993). After calibration, a freshly microtomed native wood sample was placed in the fluid chamber on the AFM sample stage and a plane wood cell wall was selected in the latewood area (Fig. 1b + c). The force-distance measurements were then performed in the center of the secondary (S2) cell wall (Fig. 1d). On a scan size of 1 µm², 64 force-distance measurements were performed with 512 samples per ramp. The ramp size was 400 nm or 1 µm for measurements in water or air, respectively, with a forward velocity of 1 µm/s. Measurements in water were performed on the same area as the measurements in ambient air, after the sample was submerged 20 min in the liquid. The acquired force curves were averaged and the adhesion forces were analyzed. The adhesion force is defined as the maximum force that is needed to pull the tip away from the surface.

Force titration measurements were performed on COOH-functionalized gold wafers, native spruce and COOH-spruce specimens. In the force titration set-up, adhesion forces were acquired in buffer solutions at different pH values with an atomic force microscope (Dimension Icon, Nanoscope V). Therefore, phosphate buffers with a concentration of 0.01 mol/L were prepared ranging from pH 2 to pH 12. At first, the sample was placed in the middle of the AFM fluid chamber (Fig. 2), which was floated with a phosphate buffer of pH 2. The ambient temperature was 23 °C. Wood samples were allowed to stabilize for 20 min. Subsequently, peak force tapping scans of 4 × 512 pixels (line scans) were performed with COOH-tips mapping the cell wall from one adjacent middle lamella to the other middle lamella (Fig. 1d).

The same scanning settings were used for force titrations on COOH-functionalized gold wafers. Line scanning was done in peak force tapping mode (PeakForce Quantitative Nanoscale Mechanical Mapping, Bruker, Santa Barbara, CA, USA) with the following measurement settings: peak force frequency = 1 kHz, peak force set-point = 5nN, disabled y-scan axis, scan rate = 0.1 Hz, peak force amplitude ~ 100 nm, gain ~ 3 and lift height ~ 25 nm. After the line scan at pH 2 was finished, phosphate buffer of pH 4 was added to the AFM fluid chamber with a peristaltic pump (Ismatec Reglo ICC, Cole-Parmer, Wertheim, Germany) using chemical resistant tubes (Tygon LMT-55, Ø 1.42 mm, Cole-Parmer, Wertheim, Germany). The pH change in the fluid chamber was monitored with a pH-electrode (InLab Nano, Mettler-Toledo, Greifensee, Switzerland), which was placed next to the analyzed surface as close as possible. In this way, the pH was increased step by step by adding buffer with a higher pH value until the final pH of 12. Pumping was stopped when the targeted pH value was stable and adhesion forces were acquired. Adhesion forces of different cell wall areas, i.e., inner lumen surface (ILS) and the adjacent secondary (S2) cell wall layers, were extracted by a mask tool in the scanning probe microscopy analysis software Gwyddion (Nečas and Klapetek 2012). Finally, the average adhesion forces together with the mean absolute deviation were plotted against pH and a sigmoidal curve was fitted by a Boltzmann function in OriginPro software (coefficient of determination is displayed as R²). We interpreted the inflection point of the sigmoid curve as the pK_a value of the analyzed sample. Additionally, the swelling behavior from the corresponding topographical lines were analyzed. For this purpose, one single line of the cell wall topography image was extracted and plotted in OriginPro. The vertical height distance between the maximum Z value in the secondary cell wall and the minimum Z value in the adjacent middle lamella was calculated from each line scan per pH value. In this way, a maximum vertical height distance was acquired for each titration cycle which is indicated as “max. swelling” in Fig. 6b. The ratio of the vertical height distance at a certain pH and the maximum vertical height distance is defined as the swelling value and was calculated for each line scan per pH value.

Before performing force titration measurements, force-distance curves were analyzed in preliminary tests to study adhesion phenomena on wood depending on the tip functionality. Therefore, adhesion forces were acquired with OH-tips and COOH-tips on the secondary wood cell wall as indicated in Fig. 1d by 64 simple force-distance measurements in an area of 1 × 1 µm².

For this purpose, the radial section of native spruce wood was ultramicrotomed and tested with different tips in ambient air or aqueous surrounding. Varying adhesion phenomena were occurring on wood substrates depending on the applied tip functionality and the selected surrounding (Fig. 3). We could observe that native spruce developed large adhesion forces towards OH-tips of about 100 nN in ambient air (Fig. 3c). In these force curves, a clear pull-off force can be identified in the retraction movement of the tip. Subsequently, the wood sample was submerged in water for 20 min and force-distance measurements were performed on the same position. In this configuration, the adhesion forces get drastically lower and the force profile on retraction changed behavior as indicated by the change in shape of the retraction curve. Several rupture events influenced the pull-off force (Fig. 3d). COOH-tips developed adhesion forces of approximately 17 nN towards native spruce wood in ambient air (Fig. 3a). When submerged in water, low or hardly any adhesion forces were measured with COOH-tips on the same position and there were no rupture events as seen with OH-tips (Fig. 3b).

On a freshly cut wood surface, primarily OH-groups are arising from mainly cellulose, hemicellulose and lignin polymers, which bind water molecules from the ambient atmosphere (Berthold et al. 1996). Therefore, large adhesion forces in ambient air can be explained by high capillary forces between hydrophilic tips and a layer of water molecules bound on the wood surface (Sedin and Rowlen 2000; Noy et al. 1997). The retraction movement of the tip is clearly depending on the tip functionality, when the measurements were performed in water (Fig. 3b + d). OH-tips exhibited saw-tooth like retraction behavior on native spruce wood with rupture distances of about 100–150 nm. In fibril structures, this behavior can be explained by the AFM tip that picks up molecules from the surface of the sample, which are gradually pulled away and disentangled from the substrate until they rupture from the tip or eventually from the sample (Li et al. 2016; Ahola et al. 2008). In terms of wood fibrils, the AFM tip shows multiple rupture events in the retraction due to microfibrils that bind to the tip when they come in contact and are peeled from the surface (Yan and Li 2013; Furuta and Gray Derek 1998). We assume that the microfibrils in swollen state are better accessible to the AFM tip to pick up and peel the fibrils from the surface. Unlike OH-tips, COOH-tips showed hardly any adhesion forces in water. First, capillary forces are eliminated compared to the measurements in ambient air (Weisenhorn et al. 1989; Butt et al. 2005). Second, the measurements were performed at pH 7, which leads to deprotonation and negative charge of the carboxyl groups of the tip and wood cell wall components (Bastidas et al. 2005; Vezenov et al. 1997). This behavior leads to low adhesion forces because of electrostatic repulsion between the tip and the wood surface.

Since OH-tips showed varying rupture events in the force-distance measurements and CH₃-tips did not exhibit pH dependent adhesion forces (Bastidas et al. 2005), we performed force titration measurements applying COOH-tips.

Adhesion forces between COOH-tips and COOH-functionalized gold wafers were studied in varying pH buffers to assure the quality of the functionalized tips and to verify the capability of peak force tapping for performing force titration measurements. A continuous increase of solvent pH was enabled by gradually adding and removing phosphate buffers with a peristaltic pump (Fig. 2). In this set-up, a functionalized AFM tip was recording adhesion forces while tapping an area of 4 × 512 pixels on the sample, referred to here as line scan. After the first line scan at pH 2 was finished, pH 4 was added to the fluid chamber and a further line scan was recorded on the same position. The adjusted pH value of the buffer was determined with a nano-electrode. In this way, line scans were acquired until approximately pH 12 while increasing the pH buffer in small steps. Average adhesion forces were extracted of each scanned area, which showed pH dependency (Fig. 4). At pH 2.4, the average adhesion force was found to be 174 pN between the COOH-tip and COOH-functionalized gold wafer. The highest adhesion force (448 pN) was measured at pH 3.9. Between pH 4 and pH 6 the adhesion force dropped and leveled at around pH 8 to 0 pN. The adhesion force data was treated with a sigmoidal fit, of which the inflection point was located at pH 5.2.

The pH dependent adhesion changes between carboxylated surfaces originate from the protonation state of the COOH groups (Vezenov et al. 1997; Vancso et al. 2005; van der Vegte and Hadziioannou 1997; Bastidas et al. 2005). Low adhesion forces at high pH can be explained by electrostatic repulsion between the negatively charged deprotonated COO^–-tip and the deprotonated COO^–-wafer. At low pH, the COOH groups mainly appear uncharged and adhesion forces are existing due to hydrogen bonding. Low adhesion forces around pH 2 might be caused by strong, ionic out-of-plane hydrogen bonding between neutral and ionized functional groups arising from the tip and sample, which could be prevented in the case of high electrolyte concentration buffer due to the formation of an electric double layer (Smith et al. 2000).

The location of the inflection point of the sigmoidal fit can be interpreted as an approximation of the surface pK_a value (Vezenov et al. 2008), as seen in Fig. 4: inflection point at pH 5.2 ≈ pK_a. This finding accords with previous estimations found in research. The surface pK_a value between COOH-tips and COOH-terminated gold surfaces was previously estimated to be 5.5 (Vezenov et al. 1997) or 4.8 (van der Vegte and Hadziioannou 1997). In contrast to these publications, we observed a rather peak-shaped titration behavior than a S-shaped pattern. As described previously, this property could be caused by strong hydrogen bonding between neutral and ionized functional groups between the tip and surface.

After validating the capability of peak force tapping for force titration, the same experimental set-up was applied on native and chemically modified spruce wood. Line scans of approximately 0.08 × 45 µm were performed on the radial section of an ultramicrotomed cell wall area. In this way, we were able to measure a flat wood area while being able to differentiate several wood tissues, i.e., S2 secondary cell wall, inner lumen surface (ILS) and adjacent middle lamella (ML), as seen in Fig. 5a. Since the wood structure is swelling and the cell wall is expanding several micrometers, the wood samples were submerged in the first phosphate buffer (pH 2) for 20 min. Subsequently, adhesion and topography information were acquired simultaneously while continuously increasing the pH of the buffer solution. Therefore, the local distribution of adhesion was given at different pH levels. We extracted the adhesion forces at varying pH levels and differentiated between three different cell wall areas: inner lumen surface ILS, and the left and right adjacent secondary S2 cell wall layers (the positions are indicated with colored rectangles in Fig. 5a). Native and COOH-spruce wood exhibited pH dependent adhesion properties (Fig. 5b + c). It has to be noted that the two measurement series in Fig. 5b + c were obtained with two different COOH-tips. The highest average adhesion force of 868 pN was measured at pH 2 at the left S2 cell wall of native spruce wood.

A drop of adhesion forces was visible starting at pH 5 and low adhesion force of about 280 pN were exhibited after pH 8. The inner lumen surface showed similar adhesion changes but the scattering of the measured forces was higher. Additionally, the right neighboring S2 cell wall showed also similar adhesion changes but the general magnitude of adhesion forces was lower on this site. Data points presented as filled squares in Fig. 5b are control values, which were recorded after the pH series was finished, to verify the reversibility of the measurements. We assume that the slightly lower adhesion values of the control measurements could originate from structural changes, which are analyzed in Fig. 6. Moreover, the long measurement duration of up to 1.5 h for one force titration cycle might harm the tip or alter the wood surface. The left and right S2 secondary cell wall of COOH-spruce showed similar trends but the general magnitude of adhesion forces was lower to native spruce. The adhesion forces measured in the inner lumen surface showed high scattering in this configuration. The data shown in Fig. 5b + c was modelled with a sigmoidal fit to estimate the pK_a value of the substrate. The position of the inflection point of the fitted line can be used to estimate the pK_a and is found to be similar in all cell wall layers of native spruce (pK_a ≈ 5.3–5.4).

The estimated pK_a was higher in COOH-spruce compared to native wood: we calculated a pK_a value of 7.7 at the inner lumen surface and the right S2 cell wall area. The left S2 cell wall area exhibited a slightly lower pK_a value of 7.3. Differences in the left and right cell wall areas could originate from the cutting direction in the “twisted” fibril alignment of the wood cell wall. The varying microfibril angle, i.e. the angle between the microfibrils and the cell axis, is between 0° and 30° in the secondary cell, also visualized by AFM (Casdorff et al. 2018), and could influence the roughness, which further influences the adhesion properties.

It has to be considered, that the magnitudes of adhesion forces are depending on the quality and dimensional properties of the functionalized AFM tip, which slightly differ from tip to tip. Therefore, comparison of the magnitude of adhesion forces from two different measurement series using two different tips needs to be done carefully, since the contact area and the amount of interaction molecules between tip and surface varies (van der Vegte and Hadziioannou 1997). Comparing Fig. 5b + c, slightly overall lower adhesion forces of COOH-spruce than native wood are visible. This effect could be caused by a slightly different tip property. Furthermore, wood materials provide an inhomogeneous fibrous structure, where surface morphology could differ from measurement position to measurement position. It was found that the wood roughness and chemical variances influence the adhesion behavior in atomic force microscopy (Jin and Kasal 2016). This is why, truly reproducible adhesion forces of wood substrates by force titration are difficult, due to the inherent heterogeneous chemical and structural differences and the sensitive force titration measurements.

The drop of adhesion forces around pH 5.4 of native spruce is according to force titration characterizations of cellulosic fibers, which exhibited a sharp change in adhesion forces between pH 4 and 6 (Bastidas et al. 2005). Considering a freshly cut wood cell wall exhibiting OH-groups to a large extend, the pH dependent adhesion behavior can be explained by attractive forces at low pH towards protonated COOH-tips and repulsive forces towards the ionized COO^–-tips at high pH on the wood surface. The pK_a value of COOH-modified fir wood was reported to be 5.75 (Balaba and Subramanian 1984), which is lower than the pK_a of the COOH-spruce in our study. It was reported that the pK_a value of a dispersion of microfibrillated cellulose rods was depending on the ionic strength of the dispersion (the pK_a value shifted from 7.4 at 1 mM to 5.3 at 300 mM) assuming a typical pK_a value of 4.8 for carboxyl groups (Wågberg et al. 2008). In our experiments, the adhesion forces were acquired with an ionic strength of 0.01 mM, which could explain the higher observed pK_a value. Furthermore, the differences can be caused by the varying surface morphology of fir and spruce wood in combination with the differently applied methodological approaches.

To investigate topographical differences, one cross-sectional profile per pH value was extracted from the height image, which was simultaneously recorded with adhesion mapping (Fig. 6a). In both substrates, i.e. native and COOH-spruce, the position of the middle lamella was fixed, on which all pH height profiles were aligned. The samples showed different pH dependent swelling behavior in the secondary S2 cell wall area, whereas the middle lamella (ML) showed hardly any pH dependent height changes in both samples (Fig. 6a + b). The force titration measurements were performed in pH 2 to pH 12 in ascending order, and subsequently, a measurement at pH 2 (e.g. 2.8 for spruce and 2.4 for COOH-spruce) was repeated after the force titration series was finished, referred to as control measurement (dashed line in Fig. 6a + b). The insert in Fig. 6a depicts the dry ultramicrotomed cell wall structure when scanned in ambient air and shows the topographical differences between dry and swollen cell walls.

It can be seen, that native wood showed general low height expansion and the control line at pH 2.8 showed slightly higher height changes compared to the other cross-sections (Fig. 6a). Interestingly, the topographical changes were much higher in COOH-spruce compared to native spruce and the control measurement at pH 2.4 was located between the cross-sections of pH 7.3–7.6. A swelling value was defined to model the pH dependent behavior (explanation can be found in the experimental section). Native spruce expressed a linear increase in swelling (Fig. 6c), whereas COOH-spruce showed a sigmoidal behavior (Fig. 6d). Therefore, a Boltzmann fit was applied, of which the inflection point was located at pH 7.9.

The secondary wood cell wall is composed out of cellulose fibril aggregates, which are embedded in a lignin and hemicellulose matrix (Salmén 2015). The middle lamella is mainly composed out of lignin, which is “gluing” the cells together. Therefore, the high structural changes in the secondary cell wall can be explained by the liquid uptake in the fibrous wood scaffold. When the aqueous solution diffuses in the secondary S2 cell wall, H-bonding with hydrophilic wood polymers takes place leading to an expansion in the fibrillary matrix (Bossu et al. 2018; Barbetta et al. 2017). On the other hand, the middle lamella binds less water to the structure compared to cellulose fibrils due to the comparably hydrophobic and complex structural character of lignin (Rowell et al. 2012). The inner lumen surface appeared also rougher with greater height changes, which alters the contact area between tip and substrate. This condition could also explain the high scattering of adhesion forces in this area (Fig. 5b + c). Regarding pH dependent structural changes, two assumptions can be made. First, the esterification of the wood cell wall for COOH-spruce leads to decreased mechanical properties (Gusenbauer et al. 2020). Therefore, the expansion of COOH-spruce is facilitated due to the weakened wood matrix. Second, the carboxyl content in these wood scaffolds is increased and at high pH, the deprotonated COO^–-form dominates. This ionic form leads to higher electrostatic repulsion and the carboxylated wood fibrils are able to swell more. The assumption is supported by the control measurement (dashed line in Fig. 5b), which shows that the structural expansion decreased after the solvent pH is changed from pH 11 to 4. Similar pH dependent swelling effects were observed at wood nanocellulose (Chinga-Carrasco and Syverud 2014). By introducing COOH-groups into the wood scaffold, the surface charge is changed, which is influencing the surface pK_a value and additionally, it was shown that an increased ionic strength leads to decreased electrostatic surface potential, which changed the pK_a value (Wågberg et al. 2008). The structural changes of about 2 µm in COOH-spruce might change the accessibility of functional groups available close to the wood surface, which could also explain a different tip-sample interaction. This combination of increased surface charge in COOH-spruce and high structural changes could explain why the estimated pKa value was higher than expected.