Adsorption of Xyloglucan onto Cellulose Surfaces ... - ACS Publications

Jul 31, 2016 - Surface and Corrosion Science, KTH Royal Institute of Technology, 100 44 ... interaction is a major factor in the adsorption of XG on w...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/Biomac

Adsorption of Xyloglucan onto Cellulose Surfaces of Different Morphologies: An Entropy-Driven Process Tobias Benselfelt,*,† Emily D. Cranston,‡ Sedat Ondaral,§ Erik Johansson,∥ Harry Brumer,⊥ Mark W. Rutland,# and Lars Wågberg*,† †

Department of Fibre and Polymer Technology and Wallenberg Wood Science Center, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden ‡ Department of Chemical Engineering, McMaster University, Hamilton, Ontario L8S 4L7, Canada § Department of Pulp and Paper Technology, Karadeniz Technical University, 61080 Trabzon, Turkey ∥ Cellutech AB, 114 28 Stockholm, Sweden ⊥ The Michael Smith Laboratories and the Department of Chemistry, The University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada # Surface and Corrosion Science, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden S Supporting Information *

ABSTRACT: The temperature-dependence of xyloglucan (XG) adsorption onto smooth cellulose model films regenerated from N-methylmorpholine N-oxide (NMMO) was investigated using surface plasmon resonance spectroscopy, and it was found that the adsorbed amount increased with increasing temperature. This implies that the adsorption of XG to NMMO-regenerated cellulose is endothermic and supports the hypothesis that the adsorption of XG onto cellulose is an entropy-driven process. We suggest that XG adsorption is mainly driven by the release of water molecules from the highly hydrated cellulose surfaces and from the XG molecules, rather than through hydrogen bonding and van der Waals forces as previously suggested. To test this hypothesis, the adsorption of XG onto cellulose was studied using cellulose films with different morphologies prepared from cellulose nanocrystals (CNC), semicrystalline NMMO-regenerated cellulose, and amorphous cellulose regenerated from lithium chloride/dimethylacetamide. The total amount of high molecular weight xyloglucan (XGHMW) adsorbed was studied by quartz crystal microbalance and reflectometry measurements, and it was found that the adsorption was greatest on the amorphous cellulose followed by the CNC and NMMO-regenerated cellulose films. There was a significant correlation between the cellulose dry film thickness and the adsorbed XG amount, indicating that XG penetrated into the films. There was also a correlation between the swelling of the films and the adsorbed amounts and conformation of XG, which further strengthened the conclusion that the water content and the subsequent release of the water upon adsorption are important components of the adsorption process.



orientation and in the formation of the load-bearing network.3,4 In the primary plant cell wall, cellulose microfibrils are extensively coated with XG, preventing their assembly into larger fibril aggregates during cellulose synthesis.5 It has been

INTRODUCTION Xyloglucan (XG) is one of the main polysaccharides in the primary cell wall of dicotyledonous plants and it also accumulates in some seeds as nutrition storage. It has a linear β(1 → 4) glucan (cellulosic) backbone with D-xylo(α → 6) side groups which are further substituted with combinations of galactopyranose, fucopyranose, and arabinofuranose groups.1,2 It is suggested that XG performs a structural role in microfibril © XXXX American Chemical Society

Received: April 19, 2016 Revised: July 28, 2016

A

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

with high affinity that is rapidly saturated and a second binding site with less affinity but higher adsorption capacity. The ITC indicates an exothermic behavior for the first site which was assigned to hydrogen bonding. Adsorption to the second site was endothermic, indicating an entropy-driven interaction. However, the structures of these sites remain unknown. Zhang et al.18 used molecular dynamic simulations to predict the adsorption of XG model molecules onto hydrophilic cellulose surfaces and suggested that van der Waals interactions play a predominant role. Zhao et al.19 examined similar models for different crystal planes, a hydrophilic surface and hydrophobic (100) and (200) surfaces, and these simulations indicated that XG adsorbs more favorably to the hydrophobic surface and prefers a flat conformation. The rough hydrophilic surface induces a coiled formation in order to facilitate hydrogen bonding. These simulations support the explanation of the different adsorption sites suggested by Lopez et al.17 Considering the fundamentals behind polymer adsorption at a solid−liquid interphase,20 it is clear that specific interactions cannot per se be the driving force for polymer adsorption due to their extremely short range of action. There has to be a gain in enthalpy, that is, ΔH has to be negative or an increase in entropy of the system for spontaneous adsorption of the polymer to occur at the solid−liquid interphase. Specific interactions can, however, be involved in the adsorption process if, for example, the binding of the polymer segments to the surface is stronger than the binding of water molecules to the surface so that bound water molecules are released upon adsorption, which will give a large increase in the entropy of the system. This means, for instance, that even if hydrogen bonding is involved in an adsorption process, the release of water molecules can still be responsible for driving the adsorption of the polymer to the interphase. These different processes are often confused and this gives rise to difficulties in isolating the critical factors controlling the adsorption of xyloglucan to cellulose surfaces (among other polymers). One of the major purposes of the present work has thus been to determine whether the adsorption is driven by a gain in entropy or by a gain in enthalpy. To do this, the adsorption of XG onto smooth model cellulose films was studied at different temperatures using surface plasmon resonance spectroscopy (SPR) in order to investigate the endothermic behavior as suggested by Lopez et al.17 The model system approach is a complementary tool to investigate the driving force behind XG adsorption to cellulose, moving beyond the common perception that the adsorption is mediated by hydrogen bonds and van der Waals interactions. In addition, the adsorption of XG onto cellulose model films with different morphologies, different densities and different degrees of crystallinity was studied by quartz crystal microbalance with dissipation (QCM-D) and stagnation point adsorption reflectometry (SPAR). The novelty in our approach to understand XG adsorption lies in the comparison between well-defined model surfaces with different morphologies prepared from well-defined sources, and the use of high resolution measurement techniques. This approach makes it possible to investigate XG adsorption on different cellulose morphologies without the risk that the adsorption behavior is dominated by large-scale and unknown substrate variations as seen previously.10,14 Combinations of modified XG structures and cellulose morphologies also provide an insight into the adsorption behavior.

suggested that the interaction between XG and cellulose microfibrils is driven by co-operative hydrogen bonding and van der Waals interactions,6 but model experiments are still needed to clarify the major factors that control the adsorption of XG to cellulose. The strong affinity of XG to cellulose has attracted researchers for its application in cellulose-based materials. One of the major uses of XG is as an additive in the papermaking process, which results in an increased adhesion and lowered friction between cellulosic fiber surfaces, leading to a better formation and greater strength of the paper.7−9 It has been reported that the molecular weight, sugar composition and side-group structure of XG affect the interactions between XG and cellulose.10−12 It has also been shown that the chemical composition and physical properties of cellulose substrates change the adsorption tendencies of XG.5,10−12 Zhou et al.13 studied XG adsorption on different pulps and found that a lower surface coverage of lignin and extractives resulted in an increase in XG adsorption due to the greater accessibility of cellulose in these fibers, indicating that the XG−cellulose interaction is a major factor in the adsorption of XG on woodbased fibers. Chambat et al.10 focused on the influence of crystallinity on the XG adsorption profile, and they found that a higher saturation adsorption was related to a lower degree of cellulose crystallinity as well as to a larger specific surface area. Gu and Catchmark14 investigated the adsorption of XG on cellulose nanocrystals (CNC) and on amorphous model samples prepared from the same material. They actually showed trends opposite to those reported by Chambat et al., with a higher degree of adsorption on the highly crystalline CNC than on the amorphous model samples, and they stated that this was observed due to differences in surface area and porosity of the cellulose samples. More recently, CNC films and dispersions were used by Cathala and co-workers to study the kinetic aspects of xyloglucan adsorption,15 the structure of the adsorbed xyloglucan at different concentrations, and its effect on the colloidal stability of the CNC dispersion.16 A Langmuir adsorption isotherm analysis of the adsorption data suggested that there were two relaxation processes: k1, the diffusionlimited adsorption to an uncovered surface; and k2, the rearrangement at the surface. At low concentration (99.9%, SigmaAldrich, Germany) by a method adapted from Berthold et al.27 Substrates were coated following the method of Eriksson et al.28 Dissolving-grade pulp (0.5 g) was dispersed in Milli-Q water and solvent exchanged to methanol followed by DMAc by immersion for 3 × 30 min in each solvent. LiCl (1.5 g) was added to 18 mL of DMAc, which was heated to 110 °C. When a clear solution was obtained, the solution was allowed to cool. The solvent-exchanged pulp was then added to the solution of LiCl in DMAc in small portions. The pulp was allowed to dissolve at room temperature for 24 h, and the solution was subsequently diluted with 80 mL of DMAc to a final cellulose concentration of approximately 0.5 wt % and heated to 110 °C before spin-coating onto silica substrates pretreated with PVAm. These cellulose surfaces were then placed in Milli-Q water to remove excess LiCl and solvent before being blown dry with N2. Preparation of Cellulose Model Surfaces for SPR Measurements. The SPR substrates consisted of Schott glass D263 (thickness 0.5 mm) sputter-coated with Au and a Cr adhesive layer (total thickness 52 nm) which were purchased from BioNavis and cleaned in chromosulfuric acid solution (Merck Chemicals), rinsed with Milli-Q water and ethanol, and dried under N2. NMMO-regenerated cellulose surfaces were prepared on the goldcoated Schott D263 glass substrates without a precursor layer. The cellulose solution was spin-coated for 15 s at 1500 rpm followed by 30 s at 2500 rpm. To precipitate the cellulose, a drop of Milli-Q water was left to stand on the film for 10 s before spinning dry. The cellulosecoated gold substrate was immersed in Milli-Q water for 2 × 20 min to regenerate the cellulose and remove residual solvent. Atomic Force Microscopy (AFM). The different cellulose model films on SiO2 surfaces were imaged by atomic force microscopy (AFM) run in the tapping mode (Nanoscope III, Multimode SPM, Veeco Inc., U.S.A.). All the experiments, in which standard rectangular noncontact silicon cantilevers (RTESP, Veeco Instruments Inc., U.S.A.) were used, were conducted under ambient conditions (23 °C and 50% RH). Experiments using the Quartz Crystal Microbalance with Dissipation (QCM-D). The adsorbed XG amount, the associated water, and the viscoelastic properties of the adsorbed XG layers on the cellulose model films were determined using a quartz crystal microbalance with dissipation (QCM-D, E4 model, Q-Sense Ab,

EXPERIMENTAL SECTION

Preparation of XG. XG with different molecular weights (high XGHMW and low XGLMW) were prepared by enzyme hydrolysis of native XG from tamarind seed (Innovassynth Technologies Ltd., India). Before enzyme treatment, XG (5 g/L) was dissolved in water at 65 °C under stirring for 2 h, centrifuged at 4000 rpm at 4 °C for 1 h, and filtered through glass filter paper (Whatman GFA) to remove impurities, followed by a subsequent freeze-drying. XG (5 g/L) was hydrolyzed by cellulase (1 mU/mg XG) from Trichoderma reesei (Sigma-Aldrich) at a constant mixing rate at 27 °C in a buffer solution of 100 mM NaOAc (Sigma-Aldrich) giving a pH of 4.75. The hydrolysis was terminated after 240 min by increasing the pH to 9.5 with 25% ammonia (Sigma-Aldrich). XG solutions were purified from cellulase by filtration through a Q Sepharose Fast Flow column (GE Healthcare Life Science) pretreated with 5% ammonia at a speed of 10 mL/min, followed by subsequent freeze-drying. The molecular weights of the XGs were determined by gel permeation chromatography (GPC) using a Waters 616 HPLC system equipped with TSK gel G5000HHR (7.8 mm × 30 cm, pore size 1 × 105 Å, particle size 5 μm) and G3000HHR (7.8 mm × 30 cm, pore size 1500 Å, particle size 5 μm) columns connected in series. HPLC-grade dimethyl sulfoxide (DMSO, Sigma-Aldrich), in which XG is completely soluble, was used as the eluent (flow rate 1 mL/min, column temperature 60 °C), with evaporative light-scattering detection (Polymer Laboratories PL-ELS 1000).13 Pullulan standards (Polymer Laboratories) were used as calibration over the range of 180 to 1660000 Da. Calibration and GPC data are shown in the Supporting Information, Figure S2. The GPC analyses of the XG fractions are given in Table 1 and show the relative Mw values.

Table 1. GPC Results of XG Hydrolysatesa XG

Mn

Mw

PDI

XGHMW XGLMW

449000 54600

2060000 93100

4.7 1.7

a

Number-average molar mass (Mn), mass average molar mass (Mw), and polydispersity index (PDI) of the xyloglucan specimens.

A less branched structure (XGG) was also produced by partially removing galactose side groups from the native Tamarind seed XG using a β-galactosidase enzyme treatment. The complete βgalactosidase enzyme treatment procedure can be found elsewhere21 and resulted in a galactose removal of approximately 40%. The modified XG was dissolved (50 mg/L) at 65 °C for 1 h without visible aggregates. Preparation of Cellulose Model Surfaces for QCM-D and SPAR Studies. Cellulose model surfaces were prepared on two different types of SiO2 surfaces used for the two adsorption measurement techniques: Si wafers (boron-doped, p-type) used in SPAR experiments were purchased from MEMC Electronic Materials SpA (Novara, Italy), and QCM-D crystals coated with SiO2 (QSX 303/50) were supplied by Q-Sense AB (Gothenburg, Sweden). Si wafers were oxidized at 1000 °C for 3 h followed by consecutive rinsing with Milli-Q water (18.2 MΩ·cm, Millipore Milli-Q Purification System), ethanol, and Milli-Q water and dried with N2. The oxide layer thickness was measured by ellipsometry (43702− 200E, Rudolph Research, Flanders NJ, U.S.A.) to be approximately 90 nm. The oxidized wafers were cut into pieces approximately 10 × 50 mm in size to be used for SPAR measurements. Both oxidized wafers and QCM-D crystals were rendered hydrophilic by treatment for 2 min in a plasma cleaner at a reduced pressure (PCD 002, Harrick Scientific Corp., Ossining, NY, U.S.A.) at 30 W before the cellulose films were prepared. After plasma treatment, the surfaces were soaked in a solution containing 0.1 g/L polyvinylamine (PVAm, BASF, Sweden) for 15 min to create an adhering precursor layer for cellulose adsorption. Before use, the PVAm was dialyzed against Milli-Q water and freeze-dried. The PVAm layer was rinsed with deionized water and dried with N2. Three kinds of model surfaces were prepared: C

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

proportional to reflectivity ratio of the surface and therefore changes with the adsorption of XG. The adsorbed amount (Γ) can be calculated as

Gothenburg, Sweden). Crystals coated with different cellulose model films were mounted in the QCM-D chamber and exposed to a buffer solution to create a stable baseline before XG injection. All the QCMD experiments were conducted at a constant temperature of 24.0 °C and pH 7 with a flow rate of 0.1 mL/min and a XG concentration of 50 mg/L. The adsorbed amount was calculated using the Sauerbrey equation with sensitivity constant C = −0.177 mg m−2 Hz1− and overtone number n = 3.29 The Sauerbrey equation is a model created for rigid films, but studies have shown that the there is only a small difference between Sauerbrey and more advanced viscoelastic models even in systems with high dissipation.30−32 Surface Plasmon Resonance Spectroscopy (SPR). A multiparameter SPR spectrometer (MP-SPR Navi from BioNavis) was used in the “full angular scan” mode to monitor the adsorption of XG onto the model cellulose surfaces as a function of temperature. The instrument is equipped with a laser of wavelength 670 nm (and the laser and detector move on a double goniometer) with a BK7 half cylinder prism equipped with a refractive-index-matching polymer to ensure optical coupling between the prism and the glass side of the substrate in the Kretschmann configuration. For more information about the optical and flow systems of the instrument as well as the sample holder and channel configuration, see the publication by Liang et al.33 The SPR method is highly sensitive to changes in refractive index near the gold (SPR active) sensor surface. At a specific angle of incidence, the laser energy is coupled into the sensor and used to excite surface plasmons which leads to a large dip in the reflected laser intensity (termed the SPR peak). As such, full scans of reflected intensity vs angle were collected and the dip position was fitted. The experiment and fitting is done in three steps: first, the bare sensor layer parameters (thickness and refractive index) in water are determined; next the sensor is removed from the instrument and coated with cellulose, placed back in the SPR and after equilibrating in water for 10 min a new scan is collected and fit; finally the XG adsorption is performed in situ followed by rinsing and this final scan is fit to determine the thickness and refractive index of the adsorbed XG. Sample data illustrating these three steps is presented in the Supporting Information, Figure S3. The final adsorbed XG amount was determined in water at 24, 32, and 40 °C by fitting the SPR data to the Fresnel equations using the software Winspall 3.02, 2009 (freeware from J. Worm, MPI fur Polymerforschung). The “best fit” came from an iterative least-squares fitting routine that fits the SPR peak minimum in the x-axis (i.e., the peak angle) and converges at 0.01%, which gives peak angles in the simulated curve and experimental data that match to the third decimal place. The intensity (y-axis) at the SPR angle varied due to surface roughness and minor film/substrate uniformity. The refractive index of water at different temperatures was taken from the Web site http:// www.luxpop.com/.34 The layer model used in the Winspall program is described in the Supporting Information, Table S1, and best fits are provided in Table S2 and shown graphically in Figures S4, S5, and S6. XGHMW was used for the SPR measurements and was prepared in a solution of 50 mg/L in Milli-Q water. Solutions were pH adjusted using NaOH (Sigma-Aldrich) to pH 7. XG solutions were heated to 10 °C above their desired measurement temperature, degassed and sonicated (Aries 101, Farfield) for 2 h under vacuum before reequilibration at the desired temperature. Baselines and adsorption measurements were made in Milli-Q water and XG solutions at 24, 32, and 40 °C with a flow speed of 0.1 mL/min. XG was introduced to the SPR flow cell for 200 min. Stagnation Point Adsorption Reflectometry (SPAR). The adsorbed XG amount was also measured by of stagnation point adsorption reflectometry (SPAR; Laboratory of Physical Chemistry and Colloidal Science, Wageningen University). Detailed information about the SPAR technique has been given by Dijt et al.35 Briefly, a beam of linearly polarized light is focused on the stagnation point of flow over the silica surface and reflected toward the detector. The parallel and perpendicular polarized light components, Ip and Is, are separated by a beam splitter, and the intensity of each component is recorded as a voltage by two photodiodes. The ratio S = Ip/Is is

Γ = A s−1·

ΔS S0

(1)

where S0 is the initial reflectometer reading and As−1 is the sensitivity factor that is proportional to the refractivity index increment with respect to concentration (c), that is, dn/dc of the adsorbent. The sensitivity factors used in the experiments were determined by Prof. Huygens software (Dullware, The Netherlands). The dn/dc of XG (0.13 mL/g) was determined using an Abbe refractometer (Carl Zeiss, Germany). SPAR experiments were conducted under ambient conditions at pH 7, with a flow speed of 1 mL/min and a XG concentration of 50 mg/L. Error Calculations. All error intervals (±Δx) are at least 95% confidence intervals calculated from the standard deviation (Sx) of repeat measurements (N), that is, Δx = Sx × t-value/(N)1/2, where the t-value is obtained from Student’s t-distribution at a confidence level of 95% for N − 1 degrees of freedom. In some cases, the standard deviation (Sx) is presented instead of an error interval, because only two measurements were made; these numbers are then presented after the average value in parentheses.



RESULTS Characterization of Cellulose Model Films. The cellulose films used in this study were prepared by previously established methods26 to give model surfaces with different morphologies. The AFM height and phase images of CNC, NMMO-regenerated, and amorphous cellulose films are shown in Figure 1. Figure 1 shows that the CNC film has a rod-like structure consistent with the dimensions of the CNCs and also that there is a slight orientation due to the spin coating procedure. A nonfibrillar structure is dominant on the surface of both NMMO-regenerated and amorphous cellulose films showing randomly oriented cellulose aggregates. The film thickness, determined by scratch-height tests in AFM, and the RMS surface roughness for the cellulose surfaces are presented in Table 2. The CNC film is the smoothest with a thickness similar to that of the NMMO-regenerated cellulose film that shows a higher roughness. The amorphous cellulose film was twice as thick and showed the highest roughness of the films prepared. To evaluate the adsorption capacity of the cellulose films, the adsorbed amount of water (Γwater) was calculated using QCMD data. The total change in the frequency after water injection is due to both the bulk effect of the silica crystal (≈−386 Hz) and the swelling of the cellulose film. Γwater was calculated according to the Sauerbrey equation and is given in Table 2. When the Γwater is normalized with respect to the dry film thickness, it is evident that the adsorbed water is of the same order of magnitude in both the CNC film and NMMOregenerated cellulose films and significantly lower than that of the amorphous cellulose film. Cellulose (without a precursor layer) was successfully regenerated on gold surfaces from the NMMO solution for the SPR experiments. The surface coverage of cellulose on the Au appeared to be complete, although the surface roughness and morphology differed slightly from that of NMMO-cellulose prepared on silicon wafers. The RMS roughness was 10 nm on Au compared to 4.4 nm on the silicon wafers, probably due to the inherent roughness of the gold SPR sensors. The AFM tapping mode image is shown in the Supporting Information, D

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Table 3. XG Adsorption on NMMO-Regenerated Cellulose at Various Temperatures Measured by SPRa T (°C)

XG film thickness (nm) in water

24 32 40

1.4 (0.0) 2.1 (0.6) 2.6 (0.1)

Film thickness is derived from fitting of full-angle SPR data to Fresnel equations; the same exact thickness was extracted from repeat measurements at 24 °C, hence, the standard deviation for this value is 0.

a

XG Adsorption onto Cellulose Films Using QCM-D and SPAR. Two complementary methods were used to investigate the adsorption of xyloglucan onto the cellulose films. QCM-D measures the adsorbed amount including associated water while SPAR measures only the adsorbed amount of XG. The different methods make it possible to draw conclusions regarding the morphology of the adsorbed XG at the interface, including any relaxation of the cellulose film during the adsorption process. The XGHMW adsorbed mass and the dissipation shift due to the adsorption, measured by QCM-D, are shown in Figure 2a and b, respectively. The amount of XGHMW adsorbed onto the amorphous cellulose film is higher than the amount on either the CNC film or the NMMO-regenerated cellulose film. As can be seen in Figure 2a, the adsorption kinetics are similar for all the cellulose films: a fast initial adsorption followed by a slower adsorption region before saturation is reached. The dissipation value, representing the viscoelastic nature of the adsorbed XG layer, increased with time and on almost the same scale for all the cellulose films (Figure 2b). The SPAR data in Figure 3 show that the amount of XGHMW adsorbed on the amorphous cellulose film was significantly higher than the amount on the other cellulose films. Moreover, as with the QCM-D data, the adsorption on the CNC film was greater than that on the NMMO-regenerated cellulose surface. Effect of XG Molecular Weight and Molecular Structure on the Adsorption. XGLMW, prepared by enzymatic hydrolysis of XGHMW using cellulase, and the less branched XGG, prepared by a β-galactosidase enzymatic treatment, were adsorbed onto the cellulose films in order to investigate the effects of both molecular weight and molecular structure of XG on the adsorption. Figure 4a,b shows the adsorbed XG amounts and dissipation from QCM-D. Overall, the adsorbed amount and the dissipation related to XGG were higher than for the branched high and low molecular weight XG, XGHMW and XGLMW. The greater dissipation is due to the larger amount of adsorbed XGG, as demonstrated in the SPAR data in Figure 4c. As with XGHMW, more XGLMW was adsorbed onto the CNC film than onto the NMMO-regenerated cellulose surface in both the QCM-D (Figure 4a) and SPAR measurements (Figure 4c). After XGLMW adsorption, the cellulose film was rinsed with Milli-Q water in order to remove weakly bound XG chains. The fact that rinsing changed the

Figure 1. AFM height (left) and phase (right) tapping mode images of smooth cellulose films on SiO2 surfaces. Image size is 1 × 1 μm.

Figure S7. The cellulose film thickness for SPR measurements was 14.9 ± 0.7 nm calculated from the SPR data. Temperature Dependence of XG Adsorption. SPR was used to measure the adsorption of XGHMW onto NMMOregenerated cellulose at different temperatures (24, 32, and 40 °C) and adsorbed film thicknesses for wet films were calculated, as shown in Table 3. The SPR measurements are internally consistent in their experimental protocol and data modeling are intended as a comparison of adsorbed XG amount at the three temperatures; however, SPR “optical” thicknesses cannot be directly compared to the QCM-D and SPAR measurements. The thickness of the adsorbed XG film increased with increasing temperature. The increase in thickness with increasing temperature indicates an endothermic behavior, similar to that shown by Lopez et al.17 and is elaborated further in the Discussion. Table 2. Properties of Model Cellulose Films on SiO2 Substrates

a

film type

dry thicknessa (nm)

roughnessa RMS (1 μm2; nm)

crystallinityb (%)

Γwater (mg/m2)

Γwater/dry thickness (kg/m3)

CNC NMMO-regenerated Amorphous

20 ± 3 15 ± 4 40 ± 5

2.27 4.41 6.59

85.1 60.0 14.8

19 13 62

950 870 1550

Determined by AFM and scratch-height analysis. bFrom Aulin et al.26 E

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 2. QCM-D data showing (a) the adsorption of XGHMW on the films of CNC (red), NMMO-regenerated cellulose (black) and amorphous cellulose (blue), and (b) the dissipation change due to the adsorbed layers. The amount of XG adsorbed was estimated using the Sauerbrey equation. The adsorption was measured at 24 °C, pH 7, a flow rate 0.1 mL/min, and a concentration of 50 mg/L.

Figure 4. QCM-D and SPAR data showing adsorption of XGLMW and XGG on the cellulose films: (a) Adsorbed XGG on CNC films (green) and XGLMW on CNC films (red) and NMMO-regenerated cellulose films (black), (b) dissipation values from the same measurements, and (c) SPAR data for the same combinations. The adsorption was measured at 24 °C, pH 7, a flow rate 0.1 mL/min for QCM, a flow rate 1.2 mL/min for SPAR, and a concentration of 50 mg/L for both techniques.

Figure 3. SPAR data showing the adsorption of XGHMW on the films of CNC (red), NMMO-regenerated cellulose (black) and amorphous cellulose (blue). The adsorption was measured at 24 °C, pH 7, a flow rate 1.2 mL/min and a concentration of 50 mg/L.

ment of XGLMW by XGHMW on the NMMO-regenerated cellulose surface.



adsorbed amount indicates that not all XG molecules were firmly bound to the surface. When XGHMW was introduced to the surface, an increase in the adsorbed mass of XG and increased dissipation were observed, indicating some displace-

DISCUSSION Xyloglucan Adsorption at Different Temperatures. A Van’t Hoff-type plot is shown in Figure 5 indicating that XG

F

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

though crystal regions do not adsorb water, is due to the high porosity in the dry films, which is estimated to be 21% by volume for spin-coated films from a 1 wt % CNC suspension.39 When CNC films are immersed in water, their pores fill with water and the space between the CNCs is increased (by 4−6 molecular layers of water) due to favorable water-cellulose interactions which screen CNC−CNC interactions. Williams and Philipse calculated the packing of particles of different aspect ratios40 and showed that a maximum packing of 70% by volume was achieved for particles with a low aspect ratio. However, an increased polydispersity would enable a packing above 70% by volume, which is consistent with the results for the CNC films. We suggest that the semicrystalline NMMOregenerated films have the lowest swelling because the combination of less ordered (flexible) and crystalline (dense but rigid) cellulose allows a denser packing in the initial dry film. Neutron reflectivity has shown that amorphous cellulose films have a cellulose volume fraction of 0.8−0.9.36 Similar measurements have shown a volume fraction of 0.4−0.6 for CNC films,41,42 supporting the statement that the porosity of CNC films is not negligible. Kontturi et al.36 also state that the amorphous cellulose is indeed highly dense with a substantial supramolecular structure, and a density of 1490 kg/m3 has previously been used for amorphous cellulose.43 This makes the density 1500 kg/m3 a sufficient approximation for the density of the amorphous and the NMMO-regenerated cellulose films. Table 2 shows the amount of water adsorbed normalized with respect to the dry film thickness. In order to explain the swelling behavior, the contribution of water released from crystalline and amorphous regions can be roughly estimated to highlight the different morphology of the CNC films. Assuming low porosity of the NMMO-regenerated and amorphous cellulose films compared to the CNC film, Table 2 leads to the expression:

Figure 5. Van’t Hoff-type plot showing the natural logarithm of the XG film thickness in water as a function of the reciprocal of the temperature. Linear regression analysis for the three data points is merely a guide for the eye and an approximation at best. Nonetheless, the trend indicates an endothermic reaction (negative slope which means positive ΔH, and a positive intercept which means positive ΔS). The standard deviation is shown as error bars.

adsorption is associated with an entropy gain due to the increasing mass adsorbed with increasing temperature. The XG film thicknesses in Table 3 were used as a measure of the equilibrium constant (K) assuming that the reaction “XG unbound ⇌ XG bound” is in fact an equilibrium reaction and that the reaction is at equilibrium after 200 min of adsorption, that is, ln K = −ΔH /RT + ΔS /R

(2)

This adsorption behavior, in combination with the simulations made by Zhang et al.18 and the calorimetric titration made by Lopez et al.,17 supports the hypothesis that XG adsorption to cellulose is endothermic. An elevated temperature would reduce the efficiency of the hydrogen bonding and van der Waals interactions making them unlikely driving forces for XG adsorption. The entropy loss when a polymer is adsorbed to a surface must be compensated for by a gain in entropy of another part of the system. In this case, the release of water molecules that are unfavorably arranged on a surface provides that entropy gain when the total surface area, cellulose, and XG, on which water can arrange is reduced. This entropy-driven process would explain the increasing adsorption with increasing temperature. Water Content in Cellulose Films. The calculated swelling values (wt %) of the model films used in this work are 79, 58, and 103%, for CNC films, NMMO-regenerated films, and amorphous films, respectively. These values are calculated using the dry-thickness and the adsorbed water, Γwater, from Table 2, a density of 1200 kg/m3 for CNC films and a density of 1500 kg/m3 for NMMO-regenerated and amorphous films. The lower density chosen for the CNC films is due to their porosity, as discussed below. While cellulose film swelling data are frequently available in the literature, different film preparation methods, different characterization techniques and inconsistencies in the definition of “dry” cellulose, make a comparison with our results difficult.26,36−38 However, Aulin et al.26 used similar preparation and characterization methods and their swelling data (26, 22, and 48%, for CNC films, NMMOregenerated films, and amorphous films, respectively) shows the same trend. The relatively large swelling of the CNC films, even

Γwater /dry thickness = ϕcrystalline ·C + ϕamorphous ·A

(3)

where ϕ is the composition fractions shown in Table 2, C is the water content of the crystalline regions (kg/m3), and A is the water content of the amorphous regions (kg/m3), and this leads to 1550 kg/m 3 = 0.148 ·C + 0.852 ·A (Amorph)

(4)

870 kg/m 3 = 0.6·C + 0.4·A (NMMO)

with the solution A = 1773 kg/m3 and C = 268 kg/m3. Note that these values are based on the dry thickness and are therefore higher than would be obtained if the water content were based on the wet thickness. Using these values, the total calculate water content of the CNC films, based on the same model, is 495 kg/m3 in contrast to the measured value of 950 kg/m3, which indicates that almost 50% of the water is located in pores. This illustrative swelling model agrees well with the previously mentioned data from neutron reflectivity.41,42 Given the literature values, the swelling model, and the swelling values in Table 3, the specific surface area of the model films decreases in the order: amorphous cellulose > CNC > NMMOregenerated cellulose, which is reflected in the amount of XG adsorbed according to the SPAR data in Figure 3. It is here suggested that the best representation of the films is a density of 1500 kg/m3 for the NMMO-regenerated and the amorphous cellulose films, and a density of 1200 kg/m3 for CNC films due to the porosity. The pores in the CNC films increase the G

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules specific surface area in the film as well as the amount of surfacebound water, and the adsorbed amount is hence higher than that expected for an intractable flat surface. XG Adsorption onto Cellulose Films. QCM-D and SPAR results in Figures 2a and 3 show that more XGHMW is adsorbed onto the amorphous cellulose film than onto the more crystalline surfaces of CNC and NMMO-regenerated cellulose. The dissipation change, seen in Figure 2b, is related to the adsorbed mass of the polymer for a specific conformation. The dissipation for XG adsorption onto the different cellulose morphologies do not follow the adsorbed amount, indicating that the adsorbed XG binds less water when it is adsorbed onto the amorphous film or that a large amount of water is released from the film during the adsorption process. The most reasonable explanation is that XG penetrates into the amorphous film and is adsorbed in a three-dimensional configuration while expelling water from the film, resulting in a collapse of the highly swollen amorphous cellulose structure. Notley44 and Enarsson et al.45 have reported a similar behavior in the case of polyelectrolyte adsorption onto charged NMMOregenerated cellulose surfaces. QCM-D measurements showed that the dissipation change was negative and the frequency shift positive after the adsorption due to release of counterions followed by extensive deswelling. It is reasonable to suggest that XG can induce a similar relaxation of the cellulose film which will affect the dissipation of the system. The data were further evaluated by comparing the adsorbed amount of XG to the thickness of the different cellulose films. Figure 6a indicates that there is a strong correlation between the adsorbed amount of XG and the dry cellulose film thickness, which suggests that XG can penetrate into the cellulose films and adsorb inside pores and amorphous regions. However, when the data is normalized with respect to the film thickness in Figure 6b, it is found that the XG adsorption decreases in the order: amorphous cellulose > CNC > NMMOregenerated cellulose, which emphasizes the previously discussed correlation with specific surface area and swelling. Given that XG adsorption is driven by the release of water, we suggest that the swelling of the films is related to the normalized amount of adsorbed XG in a similar way. Figure 7 shows the correlation between the normalized adsorbed amount and the swelling calculated from the film thickness and the assumed densities. The water content in the amorphous regions or pores can be related to the greater adsorption due to a larger specific surface area. The swelling of the cellulose films, when the water located in pores in the CNC films is subtracted, will thus be proportional to the amount of available adsorption sites from which water can be released to gain entropy in the system. Effect of Xyloglucan Molecular Weight and Molecular Structure on Adsorption. The molecular weight and the PDI were large which is not unreasonable for a native XG from Tamarind seeds and are in accordance with earlier reports,21,46 but will, of course, depend on the extraction procedure. Picout et al.46 investigated the properties of XG extracted from Tamarind seeds in aqueous solutions, and they observed a radius of gyration (Rg) of around 100 nm, a persistence length of 4−6 nm, a second virial coefficient of roughly 10 × 10−4, and a Flory exponent of 0.51. This means that water is close to a theta-solvent for XG from tamarind seeds, but it is commonly observed that XG samples aggregate over time. It can be speculated that this is due to the presence of a less soluble XG fraction in the sample, or an elevated local concentration at the

Figure 6. Relationship between the adsorbed amount of XG (SPAR data at 90 min) and the film thickness (Table 2) presented as (a) adsorbed XG as a function of cellulose film thickness for the different model surfaces, and (b) adsorbed XG normalized with respect to the cellulose film thickness of the different surfaces. A linear relation is shown (a) with the coefficient of determination R2 = 0.99. There is a significant difference in the normalized adsorption in (b) calculated using Student’s t test at the significance level (*) α = 0.01 and (¤) α = 0.05. The error bars are 95% confidence intervals based on the deviation in film thickness which is assumed to be the main contribution to the variation.

Figure 7. Relationship between the normalized adsorption and the swelling of the cellulose films calculated using a cellulose density of 1500 kg/m3 for the amorphous and NMMO-regenerated cellulose film, and a density of 1200 kg/m3 for the CNC film. The line is a linear fit with the coefficient of determination R2 = 0.94. The error bars are 95% confidence intervals based on the deviation in film thickness which is assumed to be the main contribution to the variation.

H

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules air−water interphase which favors aggregation. The limited water solubility is something all adsorption studies of XG have to take into account, but it is hard to predict the effect it will have on the adsorption behavior. The XGHMW sample might be affected by a slight aggregation and this will influence the adsorption behavior. XG is fully soluble in DMSO used for the GPC analysis and aggregation should not affect the Mw and PDI. It can be hypothesized that the extensive time needed to reach equilibrium is a result of aggregation, where small polymer chains reach the surface first due to faster diffusion and are later exchanged for larger chains or aggregates, which is a slow process. XGLMW adsorption was comparable to that of XGHMW on both CNC and NMMO-regenerated cellulose surfaces (Figure 4). When XGHMW was introduced to a surface previously covered with XGLMW a replacement of XGLMW with XGHMW should have taken place. The replacement mechanism is thermodynamically favorable because of the increase in entropy of the system as shown in earlier publications,20 but in this case, the replacement was not significant. Higher molecular weights provide a more extended conformation of polymers out from the surface in loops and tails, which leads to an increase in the amount of water in the polymer layer, seen as a greater increase in adsorbed mass in QCM-D than in the SPAR measurements. On the other hand, the adsorption of XGHMW on the CNC surface in the QCM-D data appear to be slower than for the lower molecular weight XGLMW, which can be seen in the slope of the adsorption curve in Figure 4a. This behavior would be expected for a porous film where a lower molecular weight can facilitate penetration and reconformation at the surfaces inside the porous film. The results in Figure 4 also show that there is a slightly greater adsorption of the modified XGG than of the native XGHMW onto the CNC film (Figure 4a and Figure 4c). On βgalactosidase enzyme treatment, XG turns into a less branched structure (XGG) due to the removal of galactose side groups. The porosities of the swollen cellulose films suggest that the less branched XGG can more easily penetrate/diffuse into the film, resulting in a larger amount adsorbed. To evaluate the layer structure of XGs on the cellulose films, the effective hydrodynamic thickness (deff) of the adsorbed layer can be considered. This thickness, a measure of the layer extension that in turn can be interpreted in terms of layer structure, can be calculated from the adsorbed masses (ΓQCM and ΓSPAR) and the bulk density of the polymer (ρXG = 1500 kg/m3) according to the following equation:47

Table 4. Effective Layer Thickness (deff) of XG on the Cellulose Films combination

ΓQCM (mg/m2)

ΓSPAR (mg/m2)

ΔD × 106

ρeff (g/cm3)

deff (nm)

XGHMW on CNC XGHMW on NMMO XGHMW on Amorph XGLMW on CNC XGLMW on NMMO XGG on CNC

3.29 2.37 3.67 3.17 1.50 3.85

0.919 0.473 2.22 0.883 0.468 1.03

1.5 1.5 1.6 1.7 0.9 2.1

1.14 1.10 1.30 1.14 1.16 1.13

2.89 2.15 2.82 2.78 1.29 3.41

plotted in Figure 8 (red/patterned) showing specific volumes of 3 mm3/mg for most of the combinations. However, there are

Figure 8. XG properties on the different surfaces showing (red/ patterned) the specific volume of XG, calculated by normalization of the effective thickness deff with the SPAR data, and (blue) dissipation from QCM-D normalized with the SPAR data. (1) and (2) are deviating combinations: (1) XGHMW on NMMO-regenerated cellulose and (2) XGHMW on amorphous cellulose.

two combinations that differ from the others: (1) XGHMW on the NMMO surface (4.5 mm3/mg) and (2) XGHMW on the amorphous surface (1.3 mm3/mg). The same behavior can be seen when the dissipation from QCM-D is normalized with respect to the adsorbed amount from SPAR shown in Figure 8 (blue). The normalization shows how each adsorbed mg of XG affected the viscoelastic properties of the films, and this information indicates the conformation of XG in the adsorbed layer in a similar way as for the effective thickness. The first case (1) can be explained by a limited penetration of XGHMW due to the high molecular weight and high density of the NMMOregenerated films, leading to an extended adsorption on the surface of the cellulose film. The second case (2) can be explained by integration of XGHMW into the amorphous cellulose network, where it can be adsorbed in a threedimensional configuration and exclude water, as previously discussed. This hypothesis is supported by the adsorption behavior in Figures 2a and 4a, which show the equilibrium kinetics after the fast initial adsorption. The slope after 30 min is quite steep for the amorphous cellulose, indicating that the equilibrium was slowly reached and that it would be reasonable to assume extensive penetration into the cellulose film. In the NMMO case, equilibrium is reached more rapidly with the conclusion that the surface is mainly covered with XGHMW. On the other hand, XGLMW seems to penetrate more with a specific volume of 2.75 mm3/mg. There is also a distinct increase in

deff = ΓQCM /ρeff = ΓQCM /[(ρXG ·ΓSPAR /ΓQCM) + ρwater (1 − ΓSPAR /ΓQCM)]

(5)

where ρeff is the effective density of the adsorbed layer and ρwater = 1000 kg/m3. This equation is based on a shear model that assumes that the material in the range of 0 to deff oscillates with the crystal and thus contributes fully to the adsorbed mass obtained by QCM-D, whereas material located further from the surface does not oscillate with the crystal and does not contribute to the adsorbed mass.48 Since the adsorbed mass increased with time and since equilibrium could not be achieved, the adsorbed masses from QCM-D and SPAR at 90 min were used to calculate the deff shown in Table 4. In order to make the values comparable, the deff values were normalized with respect to the adsorbed amount calculated from the SPAR data, giving the specific volume of adsorbed XG. The data, for each surface and XG combination, are I

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules dissipation when XGHMW is adsorbed onto an already covered surface in Figure 4b, indicating a structure that extends into the solvent. According to the specific volume calculations, there is only a slight difference between XGG and XGHMW on CNC surfaces, and the conclusion is that there is no significant conformational change at the cellulose surface due to the removal of side groups. The specific volume of XGG compared to that of XGHMW also shows that the increase in adsorbed mass of XGG is an effect of different adsorption/penetration rates, as previously mentioned, which is most probably a result of a different conformation in solution. Cosgrove has, in recent reviews,49−51 introduced the “biomechanical hotspot” model that describes the interactions between cellulose and XG in the primary cell wall. The model is based on data which shows that the cellulose-XG interactions are rather scarce. Cosgrove suggests that cellulose microfibrils connect in junctions with a layer of tightly and flatly bound XG segments as the load bearing link, while the rest of the XG molecule maintain a random coil shape as in solution. This model would explain the difference in the bound specific volume for the different XG structures: XGHMW, XGLMW, and XGG, when steric hindrance and the size of the random coil structure are the important factors. A random coil in an amorphous cellulose network is more likely to have several parts of the XG chain attached to cellulose while only a few connection points would be found for a random coil in a CNC film with larger pores. Alteration of the molecular weight would lead to a higher accessibility and a denser packing, which is consistent with a smaller specific volume for XGLMW than for XGHMW on NMMO-regenerated cellulose.

on amorphous cellulose. The former showed a greater specific volume and hence a less dense adsorption on the surface, and the latter showed a very dense adsorption into the amorphous cellulose network. A less branched xyloglucan, XGG, showed a slightly greater adsorption compared to branched xyloglucan onto CNC films, but without distinct conformational differences. The hypothesis is that the greater adsorption is a result of kinetic differences during adsorption onto or penetrating into the cellulose network due to a different conformation in the aqueous phase. The results and hypotheses presented can be used to understand XG adsorption to cellulose during cell-wall formation, knowledge that can be further utilized to create new biobased nanomaterials with CNC/CNF and XG as the main components.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biomac.6b00561. An illustration of the nomenclature for the conformation of polymers at interfaces, the pullulan calibration curve and the GPC data, images showing the SPR fitting procedure, an AFM image of the NNMO-regenerated cellulose on the gold SPR sensor, tables containing the parameters for the SPR model, and a table containing the result from the SPR model is included (PDF).





AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected].

CONCLUSION The adsorption properties of xyloglucan on cellulose model films with different morphologies have been investigated. The amount of native xyloglucan, XGHMW, adsorbed onto NMMOregenerated cellulose films was found to increase with increasing temperature, consistent with an endothermic process. The most probable explanation of such behavior is a system where the adsorption is driven by the entropy gain of released water molecules ordered around cellulose and xyloglucan. The amount of native xyloglucan adsorbed on films with different morphologies was greatest on amorphous cellulose films followed by CNC films and NMMO-regenerated cellulose films. The adsorbed amount was correlated to the film thickness, which would be expected for a system where the adsorbent can penetrate into the film. However, there was a slight difference in the normalized adsorbed amounts between the different films, owing to structural differences such as porosity and the volume fraction of crystalline regions. There was also a correlation between the swelling of the films and the normalized adsorbed amounts, governed by the specific surface area on which water molecules are confined and subsequently exchanged for xyloglucan. The higher swelling and the higher XG adsorption onto the cellulose nanocrystal films compared to NMMO-regenerated films was attributed to a larger specific surface area due to the packing of charged rod-shaped particles in the wet layers, which leads to a more porous network. There were two combinations of cellulose surface and xyloglucan sample that showed results significantly different from the others regarding the properties of the wet adsorbed layers: XGHMW on NMMO-regenerated cellulose and XGHMW

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We acknowledge the Swedish Center of Biomimetic Fiber Engineering (Biomime), the Wallenberg Wood Science Center (WWSC), and the Knut and Alice Wallenberg Foundation for financial support. We also thank Dr. Marcus Ruda for assistance with the enzymatic hydrolysis of xyloglucan, and Dr. Joby Kochumalayil Jose for providing β-galactosidase-treated xyloglucan.



REFERENCES

(1) Hayashi, T. Annu. Rev. Plant Physiol. Plant Mol. Biol. 1989, 40, 139−168. (2) Chanliaud, E.; De Silva, J.; Strongitharm, B.; Jeronimidis, G.; Gidley, M. J. Plant J. 2004, 38, 27−37. (3) Zhou, Q.; Rutland, M.; Teeri, T.; Brumer, H. Cellulose 2007, 14, 625−641. (4) Fry, S. C.; York, W. S.; Albersheim, P.; Darvill, A.; Hayashi, T.; Joseleau, J.-P.; Kato, Y.; Lorences, E. P.; Maclachlan, G. A.; McNeil, M.; Mort, A. J.; Grant Reid, J. S.; Seitz, H. U.; Selvendran, R. R.; Voragen, A. G. J.; White, A. R. Physiol. Plant. 1993, 89, 1−3. (5) Vincken, J. P.; de Keizer, A.; Beldman, G.; Voragen, A. Plant Physiol. 1995, 108, 1579−1585. (6) Hanus, J.; Mazeau, K. Biopolymers 2006, 82, 59−73. (7) Lima, D. U.; Oliveira, R. C.; Buckeridge, M. S. Carbohydr. Polym. 2003, 52, 367−373. (8) Stiernstedt, J.; Brumer, H.; Zhou, Q.; Teeri, T. T.; Rutland, M. W. Biomacromolecules 2006, 7, 2147−2153.

J

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

(44) Notley, S. M. Phys. Chem. Chem. Phys. 2008, 10, 1819−1825. (45) Enarsson, L.-E.; Wågberg, L. Biomacromolecules 2009, 10, 134− 141. (46) Picout, D. R.; Ross-Murphy, S. B.; Errington, N.; Harding, S. E. Biomacromolecules 2003, 4, 799−807. (47) Höök, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73, 5796−5804. (48) Olanya, G.; Iruthayaraj, J.; Poptoshev, E.; Makuska, R.; Vareikis, A.; Claesson, P. M. Langmuir 2008, 24, 5341−5349. (49) Cosgrove, D. J. Curr. Opin. Plant Biol. 2014, 22, 122−131. (50) Park, Y. B.; Cosgrove, D. J. Plant Cell Physiol. 2015, 56, 180− 194. (51) Cosgrove, D. J. J. Exp. Bot. 2016, 67, 463−476.

(9) Christiernin, M.; Henriksson, G.; Lindström, M.; Brumer, H.; Teeri, T. T.; Lindström, T.; Laine, J. Nord. Pulp Pap. Res. J. 2003, 18, 182−187. (10) Chambat, G.; Karmous, M.; Costes, M.; Picard, M.; Joseleau, J.P. Cellulose 2005, 12, 117−125. (11) Levy, S.; Maclachlan, G.; Staehelin, L. A. Plant J. 1997, 11, 373− 386. (12) Lima, D. U.; Loh, W.; Buckeridge, M. S. Plant Physiol. Biochem. 2004, 42, 389−394. (13) Zhou, Q.; Baumann, M. J.; Brumer, H.; Teeri, T. T. Carbohydr. Polym. 2006, 63, 449−458. (14) Gu, J.; Catchmark, J. M. Cellulose 2013, 20, 1613−1627. (15) Villares, A.; Moreau, C.; Dammak, A.; Capron, I.; Cathala, B. Soft Matter 2015, 11, 6472−6481. (16) Dammak, A.; Quémener, B.; Bonnin, E.; Alvarado, C.; Bouchet, B.; Villares, A.; Moreau, C.; Cathala, B. Biomacromolecules 2015, 16, 589−596. (17) Lopez, M.; Bizot, H.; Chambat, G.; Marais, M.-F.; Zykwinska, A.; Ralet, M.-C.; Driguez, H.; Buléon, A. Biomacromolecules 2010, 11, 1417−1428. (18) Zhang, Q.; Brumer, H.; Ågren, H.; Tu, Y. Carbohydr. Res. 2011, 346, 2595−2602. (19) Zhao, Z.; Crespi, V.; Kubicki, J.; Cosgrove, D.; Zhong, L. Cellulose 2014, 21, 1025−1039. (20) Fleer, G. J.; Stuart, M. A. C.; Scheutjens, J. M. H. M.; Cosgrove, T.; Vincent, B. Polymers at Interfaces; Chapman & Hall: London, 1993. (21) Kochumalayil, J. J.; Berglund, L. A. Green Chem. 2014, 16, 1904−1910. (22) Edgar, C.; Gray, D. Cellulose 2003, 10, 299−306. (23) Dong, X. M.; Revol, J.-f.; Gray, D. g. Cellulose 1998, 5, 19−32. (24) Eriksson, M.; Notley, S. M.; Wågberg, L. Biomacromolecules 2007, 8, 912−919. (25) Gunnars, S.; Wågberg, L.; Cohen Stuart, M. A. Cellulose 2002, 9, 239−249. (26) Aulin, C.; Ahola, S.; Josefsson, P.; Nishino, T.; Hirose, Y.; Ö sterberg, M.; Wågberg, L. Langmuir 2009, 25, 7675−7685. (27) Berthold, F.; Gustafsson, K.; Berggren, R.; Sjöholm, E.; Lindström, M. J. Appl. Polym. Sci. 2004, 94, 424−431. (28) Eriksson, J.; Malmsten, M.; Tiberg, F.; Callisen, T. H.; Damhus, T.; Johansen, K. S. J. Colloid Interface Sci. 2005, 284, 99−106. (29) Sauerbrey, G. Eur. Phys. J. A 1959, 155, 206−222. (30) Karabulut, E.; Pettersson, T.; Ankerfors, M.; Wågberg, L. ACS Nano 2012, 6, 4731−4739. (31) Krivosheeva, O.; Sababi, M.; Dedinaite, A.; Claesson, P. M. Langmuir 2013, 29, 9551−9561. (32) Aulin, C.; Varga, I.; Claesson, P. M.; Wågberg, L.; Lindström, T. Langmuir 2008, 24, 2509−2518. (33) Liang, H.; Miranto, H.; Granqvist, N.; Sadowski, J. W.; Viitala, T.; Wang, B.; Yliperttula, M. Sens. Actuators, B 2010, 149, 212−220. (34) Boisset, G. Luxpop, For index of refraction values and other photonics calculations; www.luxpop.com. (35) Dijt, J. C.; Stuart, M. A. C.; Fleer, G. J. Adv. Colloid Interface Sci. 1994, 50, 79−101. (36) Kontturi, E.; Suchy, M.; Penttilä, P.; Jean, B.; Pirkkalainen, K.; Torkkeli, M.; Serimaa, R. Biomacromolecules 2011, 12, 770−777. (37) Kittle, J. D.; Du, X.; Jiang, F.; Qian, C.; Heinze, T.; Roman, M.; Esker, A. R. Biomacromolecules 2011, 12, 2881−2887. (38) Jiang, F.; Kittle, J. D.; Tan, X.; Esker, A. R.; Roman, M. Langmuir 2013, 29, 3280−3291. (39) Niinivaara, E.; Faustini, M.; Tammelin, T.; Kontturi, E. Langmuir 2015, 31, 12170−12176. (40) Williams, S. R.; Philipse, A. P. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2003, 67, 051301. (41) Jean, B.; Heux, L.; Dubreuil, F.; Chambat, G.; Cousin, F. Langmuir 2009, 25, 3920−3923. (42) Cerclier, C.; Cousin, F.; Bizot, H.; Moreau, C.; Cathala, B. Langmuir 2010, 26, 17248−17255. (43) Krässig, H. A. Cellulose: Structure, Accessibility, And Reactivity; Gordon and Breach Science: Yverdon, Switzerland, 1993. K

DOI: 10.1021/acs.biomac.6b00561 Biomacromolecules XXXX, XXX, XXX−XXX