Anisotropic Diffusion and Phase Behavior of Cellulose Nanocrystal

Jan 23, 2019 - ICS-3, Forschungszentrum Jülich , D-52424 Jülich , Germany. Langmuir , Article ASAP. DOI: 10.1021/acs.langmuir.8b03792. Publication Dat...
0 downloads 0 Views 4MB Size
Article Cite This: Langmuir XXXX, XXX, XXX−XXX

pubs.acs.org/Langmuir

Anisotropic Diffusion and Phase Behavior of Cellulose Nanocrystal Suspensions Jonas Van Rie,† Christina Schütz,‡ Alican Gençer,† Salvatore Lombardo,† Urs Gasser,§ Sugam Kumar,∥ Germań Salazar-Alvarez,∥ Kyongok Kang,⊥ and Wim Thielemans*,†

Downloaded via UNIV OF GOTHENBURG on January 23, 2019 at 15:47:14 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



Renewable Materials and Nanotechnology Research Group, Department of Chemical Engineering, KU Leuven, Campus Kortrijk, Etienne Sabbelaan 53, B-8500 Kortrijk, Belgium ‡ Physics and Materials Science, Campus Limpertsberg, Université du Luxembourg, 162 A avenue de la Faïencerie, L-1511 Luxembourg, Luxembourg § Laboratory for Neutron Scattering and Imaging, Paul Scherrer Institute, CH-5232 Villigen, Switzerland ∥ Department of Materials and Environmental Chemistry, Stockholm University, SE-10691 Stockholm, Sweden ⊥ ICS-3, Forschungszentrum Jülich, D-52424 Jülich, Germany S Supporting Information *

ABSTRACT: In this paper, we use dynamic light scattering in polarized and depolarized modes to determine the translational and rotational diffusion coefficients of concentrated rodlike cellulose nanocrystals in aqueous suspension. Within the range of studied concentrations (1−5 wt %), the suspension starts a phase transition from an isotropic to an anisotropic state as shown by polarized light microscopy and viscosity measurements. Small-angle neutron scattering measurements also confirmed the start of cellulose nanocrystal alignment and a decreasing distance between the cellulose nanocrystals with increasing concentration. As expected, rotational and translational diffusion coefficients generally decreased with increasing concentration. However, the translational parallel diffusion coefficient was found to show a local maximum at the onset of the isotropic-to-nematic phase transition. This is attributed to the increased available space for rods to move along their longitudinal axis upon alignment. This increased parallel diffusion coefficient thus confirms the general idea that rodlike particles gain translational entropy upon alignment while paying the price for losing rotational degrees of freedom. Once the concentration increases further, diffusion becomes more hindered even in the aligned regions due to a reduction in the rod separation distance. This leads once again to a decrease in translational diffusion coefficients. Furthermore, the relaxation rate for fast mode translational diffusion (parallel to the long particle axis) exhibited two regimes of relaxation behavior at concentrations where significant alignment of the rods is measured. We attribute this unusual dispersive behavior to two length scales: one linked to the particle length (at large wavevector q) and the other to a twist fluctuation correlation length (at low wavevector q) along the cellulose nanocrystal rods that is of a larger length when compared to the actual length of rods and could be linked to the size of aligned domains.



INTRODUCTION Liquid crystals (LCs) are materials that have no positional order in molecular arrangement but maintain orientational order.1 Owing to this anisotropy, the molecules or particles exhibit an anisotropic macroscopic physical response to, for example, light, flow, or electric or magnetic fields. LCs are applied in devices we use every day including in phones, computers, and television displays.2,3 One can distinguish two main classes of liquid crystals: thermotropics, which form in response to a temperature change, and lyotropics, which form with addition of a solvent and respond to changes in concentration. Typical lyotropic LCs are formed by amphiphiles, compounds possessing both hydrophilic and lipophilic properties, but anisotropic particles also belong to the class of lyotropic LCs. LCs can form different phases with characteristic orientations of molecules/particles with respect to each other, © XXXX American Chemical Society

all characterized by a specific orientational order (indicated by the director, which represents the overall direction of the molecules/particles) and positional order. The different types, based on the way molecules arrange, are nematic, smectic, columnar, and cubic.4,5 Nematic liquid crystals are the most common, and despite the random positioning of the particles in this phase, they are directionally correlated and aligned in a general direction. A special case of a nematic phase is the chiralnematic phase, where the director rotates around a helical axis. Cellulose nanocrystals (CNCs) belong to the class of lyotropics and form a chiral-nematic LC phase. They are a renewable biomaterial mainly found in plants, providing Received: November 11, 2018 Revised: December 23, 2018

A

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. Pictures of cellulose nanocrystal suspensions between crossed polarizers. Birefringence is clearly present after 3 wt % and lasts for a couple of seconds after shaking, indicating alignment of the CNCs. Below 3 wt %, no birefringence is observed, as the green color disappears instantly after shaking.

In general, rod-shaped particles in suspension undergo fast and slow translational diffusion parallel and perpendicular to their main axis.32 The measurement of the corresponding diffusion coefficients is normally carried out under dilute conditions to allow total freedom of movement to the nanoparticles and to suppress secondary effects. In line with this, several studies on rotational and translational dynamics have already been carried out on dilute cellulose nanocrystal suspensions in aqueous suspension.32,41−44 Diffusion coefficients of cellulose nanocrystals have also been studied in dilute suspensions, often to calculate the dimensions of the rodlike particles.32,41,42 One of the interesting findings by De Souza Lima et al. was that rotational diffusion coefficients for their CNC suspensions (conc. 0 can be neglected when the rods are small compared to the wavevector, that is, qL < 5. In this case, the EACF in the VV mode becomes47 gE,VV ̂ (q , t ) = exp{−q2Dt t} for qL < 5

In this wavevector range, the only contribution to the decay of the IACF is due to translational diffusion. For such small values of qL, the rotation of rods does not lead to a significant change in the intensity in the VV mode. The EACF in the VH mode, on the contrary, does contain a contribution due to rotational diffusion and is equal to

Figure 5. Overview of the viscosities of 1−5 wt % cellulose nanocrystal suspensions at shear rate 100 s−1.

We then conducted depolarized and polarized dynamic light scattering experiments to determine the diffusion coefficients of the CNC suspensions for these different concentrations. With this technique, the intensity fluctuations of scattered light are recorded in terms of intensity autocorrelation functions. The decay rate of the autocorrelation functions can be determined and used to calculate the diffusion coefficients. Two different scattering modes were used: vertical−vertical (VV) and vertical−horizontal (VH) modes. The difference between both modes is the polarization direction of the incident and the detected light. In the VV mode, both polarization directions are perpendicular to the plane spanned by the propagation direction of the incident light and the direction in which scattered light is detected. In the VH mode, the polarization direction of the scattered light is within that plane. The VV mode is used to derive translational diffusion coefficients, whereas the VH mode can be used to determine the rotational diffusion coefficients. In our system of interacting rodlike cellulose nanocrystals, the effects of rotational and translational coupling have to be considered. First of all, the autocorrelation function of the scattered intensity is measured as a function of the scattering angle. The normalized intensity autocorrelation function ĝI(q, t) (IACF) is related to the normalized field autocorrelation function ĝE(q, t) (EACF) through the Siegert relation gÎ (q , t ) = 1 + |gÊ (q , t )|2

(7)

gE,VH(q , t ) ∝ exp[−(q2Dt + 6Dr )t ]

(8)

These expressions for the EACFs hold for monodisperse systems of noninteracting rods. Our system is however quite polydisperse (as concluded from AFM data, S1), which can be accounted for by introducing stretched exponentials. Furthermore, we sometimes find two modes in the EACFs, which is attributed to rod−rod interactions, as will be discussed later. The experimental EACFs are thus fitted to gE(t ) = A1 exp{−(Γ1t )} + A 2 exp{−(Γ2t )β } + B

(9)

where A1 = 0.1, A2 = 0.9, Γ1 = 10, Γ2 = 1, B = 0, and β = 0.5 (the so-called stretching exponent) were taken as typical starting values for the fitting procedure. The diffusion coefficients corresponding to the two modes were obtained from the resulting Γ1 and Γ2 by plotting them as a function of q2 and using that (see eqs 7 and 8) ΓVV = q2Dt

(10)

ΓVH = q2Dt + 6Dr

(11)

The value for the rotational diffusion coefficient can be obtained from the intercept of the linear fit of the data in a Γ

(4) E

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 6. Normalized intensity autocorrelation functions. (Left column) CNC suspensions measured in the VH mode and (right column) CNC suspensions measured in the VV mode with their corresponding exponential fits according to eq 9. The residuals for the fits can be found in the Supporting Information in Figure S5.1 (VH mode) and Figure S5.2 (VV mode). F

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 7. (a) Average intensities as a function of q for 1−5 wt % CNC suspensions and (b) the stretch exponent β (eq 7) as a function of q for 1−5 wt % CNC suspensions. The evolution of amplitude in slow (c) and fast (d) modes as a function of q for 1−5 wt % CNC suspensions (eq 9).

Table 1. Theoretically Calculated and Experimentally Determined Rotational Diffusion Coefficients Dr (s−1) cellulose wt %

VH

infinite dilution 1 2 3 4 5

− 310 130 140 20 0

noninteracting particle approximationa 17 000 8500 5300 4200 3000 2000

Debye approximationb − 1700 1500 1400 1100 800

Tirado and Garcı ́a de la Torre, 1979, 1980 and Broersma, 1960. bUsing Debye length instead of the average length determined by AFM.

a

vs q2 plot in VH geometry, and the translational diffusion coefficient is obtained from the slope in VV and VH geometries. Typical normalized intensity autocorrelation functions and their corresponding fits are shown in Figure 6. The left column displays the autocorrelation functions for scattering angles ranging from 40° to 110° in the VH mode. The right column displays the autocorrelation functions resulting from DLS experiments in the VV mode with scattering angles from 40° to 110°. Two relaxation modes can be observed in the autocorrelation functions, where the scattering intensity is dominated by the slow mode In our experiments, the empirical function with the best fit consisted of two modes; one is a fast mode (at shorter timeframe) that is assumed to be homogenous, whereas the other is a slow mode (at longer times), which is stretched. The stretching exponent power of the slow mode, β, was found to be roughly 0.5 (see Figure 7b), obtained from the fits of the

dispersion relations (in Figure 6). This stretching is generally taken to be due to the existing polydispersity of the CNCs. Here, we considered that due to the anisotropic shape of the cellulose nanocrystals that in our fit of the scattered electricfield autocorrelation functions of the translational motion of the nanocrystals, the fast mode would be parallel diffusion, i.e. diffusion along the longitudinal axis of the nanoparticles, whereas the slow mode would be perpendicular diffusion, i.e. diffusion perpendicular to the nanoparticle longitudinal axis. This is reasonable in the sense that the timeframe for motion of anisotropic particles is expected to occur faster in the parallel direction, rather than in the perpendicular direction, in averaged collective motions of slowly varying bulk phase behavior in “crowded” suspensions.48 These experimentally determined fast- and slow-mode movements enable us to determine the parallel and perpendicular translational as well as rotational diffusion G

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Table 2. Theoretically Calculated and Experimentally Determined Translational Diffusion Coefficientsa D⊥t (μm2/s)

D∥t (μm2/s) b

c

cellulose wt %

VH

VV

app.

Noninteracting particle appr.

Debye appr.

VH

VV

App.

Noninteracting particle appr.b

Debye appr.c

infinite dilution 1 2 3 4 5

− 1.2 0.7 0.6 0.3 0.2

− 1.4 0.7 0.7 0.3 −

− 1.3 0.7 0.7 0.3 0.2

5.8 2.9 1.8 1.4 1 0.7

− 2.3 1.5 1.2 0.9 0.6

− 5.9 4.3 4.1 1.7 1.8

− 7 6 7.4 2.1 −

− 6.5 5.2 5.7 1.9 1.8

8.7 4.4 2.7 2.1 1.6 1

− 3.2 2.2 1.8 1.3 0.9

a

Translational diffusion coefficients were determined in VV and VH modes. The apparent translational diffusion coefficient is the average value of both values. bTirado and Garciá de la Torre, 1979, 1980 and Broersma, 1960. cUsing Debye length instead of the average length determined by AFM.

coefficients, reported in Tables 1 and 2. Table 1 gives the rotational diffusion Dr, whereas the sidewise (Dt⊥) and lengthwise (Dt∥) translational diffusivities are reported in Table 2 (with standard deviations given in Table S6 in the Supporting Information). The plots of the frequency Γ as a function of q2 for the VV and VH measurements are shown in Figures 8 and 9. As one can see from the dispersion relation in Figure 8 for the fast mode in VH geometry, as a function of the cellulose concentration, there is a nonzero intercept value at q = 0. The slow mode on the other hand does reach a value of 0 at q = 0. The intercept is directly related to the rotational diffusion coefficient of CNCs, being equal to 6Dr. Figure 8 and Table 1 clearly show a decrease in the rotational diffusion coefficient with increasing concentration (in the fast mode), which is due to interparticle interactions hindering the rotational motion of the CNCs. The slopes of the two modes in Figures 8 and 9 are proportional to the translational diffusion coefficient for parallel and perpendicular motions of the cellulose crystals along their long axis and perpendicular to it, respectively. Since motion of a rod parallel to its orientation is significantly less affected by interparticle interactions as compared to its motion in the perpendicular (sidewise) direction, the fast mode is reported as D∥t and the slow mode as D⊥t . The dispersion relations from light scattering experiments in the VV mode show a zero intercept at q = 0 as expected (Figure 9). However, CNCs show a nonlinear increase in relaxation rate as a function of q2, that is, there is a linear dependence at low q, which transitions to a linear dependence at higher q with a larger slope. This may have two origins: it may result either from “longer” range twist fluctuations in the existing chiralnematic or from scattering of both translational and rotational couplings in nematic phases dispersed within the isotropic phase before macroscopic phase separation. This agrees with our results of neutron scattering experiments and the crossedpolarizer images (Figures 1−3), and similar observations of an early stage of crystalline ordering in small domains have been reported by Uhlig et al. using SANS measurements to detect their formation at concentrations below macroscopically visible nematic structure formation.53 It is worthwhile to compare these experimentally determined diffusion coefficients to theoretical diffusion coefficients for noninteracting rods characterized by viscous drag. The translational diffusion coefficients related to these two displacements as well as the rotational diffusion coefficient for noninteracting rods can be calculated according to the following equations where the ribbonlike CNC particles were modeled hydrodynamically as cylinders49

Dt⊥ =

1 kBT (ln p + γ⊥) πηL 4

(12)

Dt =

1 kBT (ln p + γ ) πηL 2

(13)

Dr =

kBTp2 A 0πηL3(1 + δ)

(14)

with p = L/d being the aspect ratio, d being the diameter of the cylinder, and L being the length of the particles. The assumptions of cylindrical shape, monodispersity, and noninteracting rod deviate quite a lot from the reality of polydisperse ribbon-shaped CNCs stabilized in suspension by electrostatic repulsions. As can be seen from these expressions, for noninteracting rods, the translational diffusion coefficients for parallel and perpendicular motions are estimated to be approximately a factor of two different (comparing eqs 12 and 13). In eqs 12−14, γ⊥, γ∥, δ⊥, and δ∥ are end-effect terms, obtained as numerical constants using the shell-model methodology49−52 γ⊥ = 0.839 +

0.185 0.233 + p p2

γ = −0.207 + δ=

0.980 0.133 − p p2

0.677 0.183 − p p2

(15)

(16)

(17)

Depending on the aspect ratio of the particles, different theoretical approaches exist to calculate the end-effect corrections, and therefore slightly different values may be obtained. The values resulting from the theoretical calculations and experiments (vide infra) are shown in Tables 1 and 2. As these expressions are only valid for dilute suspensions and our experiments were carried out in concentrated systems, the viscosity of the suspension at 100 s−1 (Figure 5) was used to estimate the diffusion coefficients rather than the pure solvent viscosity. In both tables, a similar trend is observed. Generally speaking, as the suspension becomes more concentrated, the diffusion coefficients of the rods decrease. The rods are thus more hindered in their movement in the more concentrated suspensions, as could be expected. Furthermore, the experimental values for translational diffusion coefficients are of the same order of magnitude as the theoretical values, indicating that the theoretical calculations give a good prediction for the H

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 8. Dispersion relations for the fast- and slow-mode relaxations of CNCs in depolarized light scattering in vertical−horizontal (VH) polarization geometry.

translational diffusivity in this colloidal system when using the mean-field approximation for the solution viscosity. The experimentally determined rotational diffusion coefficients, on the contrary, are an order of magnitude off when compared to the theoretical values. This is most likely a direct result of the noninteracting rigid rod assumption of the model. Indeed, large deviations between the theoretical and experimental values can be expected as the charged surfaces of the rods result in rod− rod repulsion. Owing to these intrinsic charges of the rods, their

interacting radius is much larger compared to the actual one. This mainly reflects on the theoretical values for rotational diffusion as the magnitude is several times larger when compared to translational diffusion. To show the effect of the surface charges, the theoretical values were recalculated using the same eqs 12−14, correcting the length and diameter with the Debye length (eq 1 and Figure S3). As can be seen in Tables 1 and 2, the values are closer to the experimentally obtained values, as the calculated numbers for the rotational diffusion I

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 9. Dispersion relations for the fast- and slow-mode relaxations of CNCs in depolarized light scattering in vertical−vertical (VV) polarization geometry.

fractions in the biphasic region.54−56 This is similar to a work by Lettinga et al. in their investigation of the diffusion coefficients of rodlike fd in the isotropic and nematic phases.45 The wavevectors where the two linear slopes in the relaxation rates intersect (see Figure 7, starting from 2 wt %) are plotted in Figure 11 (right, red lines), together with the relaxation rates for both the fast and the slow mode. Similar to the trend in the diffusion coefficients, a peak in the corresponding relaxation rates is observed at an intermediate concentration of around 3 wt %. This is the concentration where the isotropic phase starts forming a nematic phase by local alignment of the rods in suspension (confirming what was already observed through SANS experiments (vide supra) and

coefficients reduced with almost one order of magnitude. However, for the rotational diffusion, it is still a poor estimate. The experimental results suggest that the rotational diffusion coefficient tends to zero at higher concentrations, showing that rotational degree of freedom is lost, which is expected when the particles align to form the nematic state. The concentration dependences of the diffusion coefficients and their standard deviations are shown in Figure 10. The perpendicular translational and rotational diffusion coefficients are seen to go through a plateau around 2−3 wt %, and the parallel diffusion coefficient even appears to go through a local maximum. This can be explained by the fractionating effect of polydisperse rodlike colloids, resulting in variable volume J

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 10. Concentration dependencies of (left) translational and (right) rotational diffusion coefficients.

Figure 11. (a) Sketch of the twisted orientation of rods in the chiral (twist)-nematic state, within the isotropic-nematic two-phase region. Twist fluctuations are found in the translational parallel diffusion, indicated as a “twist-correlation” length. (b) Wavevector where the linear branches in the dispersion relation intersect for the fast modes.

here indicated in Figures 10 and 11 by the gray area). Within the isotropic phase, the available volume for the rods to move diminishes with increasing concentration. With local alignment of the rodlike particles, the excluded volume taken up per rod decreases and more free volume becomes available than when the rods are arranged in a completely isotropic phase of the same concentration. The increase in translational parallel diffusion freedom (increased translational entropy along the length axis of the particles) thus seems to drive the CNC alignment where rotational freedom (rotational entropy) is sacrificed, and this can directly be measured using these light scattering experiments. The small local increases in the rotational and translational perpendicular diffusion coefficients can also be explained by the appearance of the aligned domains: locally aligned rods will reduce the concentration in the isotropic domains, giving a small increase in diffusional freedom, which is of course offset immediately with further increases in concentration. On further increasing the concentration when the particles are aligned, the available free space again diminishes as the CNCs stack closer together as is also seen in the rotational diffusion coefficient tending to zero. This explains the maximum observed in Figure 10 (left) for the translational diffusion coefficient in the parallel and perpendicular directions. Please note that the wavelength 2π/q for the wavevector q where the linear branches in the dispersion relations in Figure 9 intersect is a few times the linear dimension of the rods, which hints at the possibility that twist fluctuations are responsible for the nonlinear dispersive

behavior. The cartoon in Figure 11 is a sketch of the twisted orientation in a chiral-nematic, where the length of a full 2π twist can thermally fluctuate and is a few times larger than the bare length of a CNC rod. Interestingly, this is observed in the translational parallel component as a fast mode of VV-scattering geometry, indicating that a majority of CNC rods are cooperatively “twisted” along each other in the transition from the isotropic to the nematic phase.



CONCLUSIONS The relation between the diffusivity of CNC rods and their phase behavior has been studied. SANS showed that alignment starts to take place at 2 wt % already, but apparently domains are too little in number or size to have an effect on average parameters as rheological measurements indicated alignment only from 3 wt %. Translational and rotational diffusion coefficients in concentrated suspensions were experimentally obtained using dynamic light scattering in VV and VH geometries. Both translational and rotational diffusion coefficients were found to decrease with increasing concentrations. In VV geometry, probing translational diffusion, two relaxation modes were found, which are attributed to motion parallel and perpendicular to the long axis of rods. The concentration dependences of the diffusion coefficients for parallel and perpendicular motions are semiquantitatively described by using the suspensions’ viscosity in expressions that apply for diffusion of noninteracting rods parallel and perpendicular to their long axis. The fast translational diffusion K

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Throughout the experiment, the scattering angle can be changed in steps of 1/16° with the ALV/LSE 5001/II Controller. A SpectraPhysics/Newport “Excelsior 532 Single Mode” laser with 150 mW maximum power is used. A reflective neutral density filter reduces the applied power to a value below 10 mW. The beam used has a diameter of only 300 μm, resulting in a relatively large divergence. A lens with local length of 400 mm is used to compensate for this divergence, it collimates the beam to a diameter less than 1 mm. Another lens (with focal length of 250 mm) is used to focus the beam to a diameter of about 0.2 mm in the measurement volume. A Glan prism polarizer (B. Halle, extinction ratio 105) is used to polarize the laser beam vertically with respect to the scattering plane. The scattered light passes the Glan-Thompson prism analyzer (B. Halle, extinction ratio 10−6), which is also vertically or horizontally adjusted with respect to the scattering plane. A photomultiplier pair detector (ALV/SO-SIPD) is used together with a single mode fiber (OZ) and a focusing collimator (LINOS MB 02). Finally, the photon counts are fed into a correlator (ALV 5000/E Multiple Tau Correlator with Fast option) with 12.5 ns shortest sample time. To avoid uncorrelated after-pulsing effects, the autocorrelation function is calculated in pseudo-cross-correlation mode. All measurements were carried out with an angular step of 5° between 30 and 120°. AFM. On a freshly scotch-tape-cleaved mica surface (NanoAndMore GMBH), 20 μL of poly-L-lysine solution (Sigma-Aldrich, 0.1% w/v in water) was deposited for 3 min, subsequently rinsed with deionized water, and dried with compressed air. Following, 20 μL of a CNC dispersion (0.001 wt %) was placed on the treated surface for 3 min, then rinsed with deionized water and dried with compressed air. The samples were left in a vacuum oven at 40 °C overnight. A multimode V AFM (Digital instruments Nanoscope Veeco) was used in tapping mode, and AFM probes from Budget Sensors (Tap300 AlG, resonance frequency 300 kHz and force constant 40 N/m) were used to image the samples. The length of 300 particles was measured manually using imageJ,61 and the mean length and standard deviation as well as a log−normal distribution were determined. The width was extracted as the height for at least 800 points on representative particles. Elemental Analysis. A Thermo Scientific Flash 2000 Elemental Analyser was used for the analyses of N and S. Enriched phenanthrene (0.49% N; 0.52% S) standard was used for the linear calibration. Vanadium(V) oxide along with 1−2 mg of the sample was weighed in a tin capsule. The empirical formula of cellulose anhydroglucose units containing sulfate half ester groups was used to determine the degree of substitution. Nitrogen and sulfur contents were reported as mass percentage of the total sample. Thermogravimetric Analysis. A Netzsch F3 Tarsus thermogravimetric analyzer was used to determine the water content of dried samples (for correction of elemental analysis data) and suspensions (concentration determination) under nitrogen atmosphere. The method involved heating of the sample during two segments: the first segment included heating to 85 °C at a rate of 10 °C/min, followed by an isothermal period of 30 min. During the second segment, heating was continued to 150 °C at a rate of 10 °C/min. The water content was determined in the linear region of the resulting TGA curve, as the mass loss at 85 °C. Acid−Base Titrations. An SI Analytics Titroline 6000 autotitrator was used for the acid−base titrations. Sodium hydroxide, first standardized with oxalic acid, was used to titrate the sulfate groups present on the CNCs. Then, 2 mL of 26.5 mg/mL CNCs was titrated with a 0.01084 M NaOH solution. The inflection point was determined, and the molar amounts of sulfate groups were derived. The experiment was repeated three times. Small-Angle Neutron Scattering. SANS measurements were performed on SANS-II at SINQ (Paul Scherrer Institute, Villigen, Switzerland).62 The samples were measured at a 6 m sample−detector distance, lambda 0.73 nm. The setup has a 3He detector with 128 × 128 pixels. One millimeter path length quartz glass capillaries were used for all measurements; the five CNC samples (1−5 wt %, all in 100% D2O) were loaded in a sample holder and measured in the same run. On the basis of standard procedures, all obtained data were

mode, obtained in VV geometry, exhibited a two-step diffusion behavior where particle alignment became prominent around 3 wt %. This unusual dispersive behavior can be understood to correspond to two different length scales: one related to the particle length (at large q) and the other to a twist fluctuation correlation length (at slow q) along the CNC rods, which is larger in length in comparison with the bare length of rods. The translational parallel diffusion coefficient was further found to increase with the formation of aligned CNC domains from an isotropic phase, attributed to the increased translational entropy for rods upon the isotropic-to-nematic phase transition. It would be interesting to employ the cellulose system in future experiments to probe the scattering properties at very small wavevectors in order to distinguish the translational diffusion of cellulose rods in parallel and perpendicular directions, independently. Scattering at such very small wave vectors probe displacements of the rods over larger length scales, where differences between the two diffusivities become the most pronounced. Comparable research has been performed by one of the present authors (K.K.) with charged fibrous-virus suspensions.59,60 In addition, it would be valuable to study the dynamics and phase behavior of the cellulose system under external electric fields, with possible technological applications such as spin-coating of cellulose suspensions in the presence of an electric field.



EXPERIMENTAL SECTION

Materials. Cotton wool was purchased from Carl Roth GmbH & Co. KG. Sulfuric acid (95%, RECTAPUR) was purchased from VWR international. Amberlite MB-6113 (for ion chromatography, mixed resin) was obtained from Acros Organics. Deuterium oxide (99 atom % D) was purchased from Aldrich. Spectra/Por type 3 dialysis tubing (MWCO 3.5 kD) was obtained from Spectrumlabs. Preparation of Cellulose Nanocrystal Suspensions. Sulfuric acid (10.06 M) hydrolysis of cotton wool was employed to obtain a cellulose nanocrystal suspension according to standard literature procedures.11 The excess non-H+ and OH− ions were removed through ion exchange with the Amberlite MB6113 mixed resin, which was afterward removed by filtration. A rotary evaporator was used to concentrate the initial aqueous suspension (around 1 wt %) to the desired concentrations. Full characterization of the obtained nanocrystals is provided in the Supporting Information. Dialysis against deuterated water was performed to prepare samples for neutron scattering measurements: dialysis tubes were filled with 1 mL of each CNC sample (1−5 wt %) and left to dialyze in 10 mL of deuterium oxide. After 2 days, the deuterated water was replaced with 10 mL of fresh deuterated water, and dialysis was continued for another 2 days to complete the solvent exchange. Rheological Measurements. An Ar-G2 stress-controlled rotational rheometer of TA Instruments was used to analyze the aqueous CNC suspensions. Shear rates between 10 and 100 s−1 were employed to obtain steady shear viscosity versus shear rate curves. A transient test was used to determine the time required to reach steady state. The strain for frequency sweep experiments was determined by amplitude sweep experiments at a frequency of 10 rad/s. Double-wall concentric cylinder geometry was used with gap sizes of 380 and 420 μm. Water evaporation was avoided by using a solvent trap. All experiments were conducted at 20 °C. Polarized and Depolarized Dynamic Light Scattering Measurements. An ALV SLS/DLS setup was used for the depolarized dynamic light scattering measurements. The goniometer with the sample holder forms the central part of the setup. The sample cell and bath cuvette can be positioned with the goniometer axis. Heat conduction occurs through the toluene bath, and reflexes from the outer cell surface can be reduced through refractive index matching with the glass. A Julabo F12 circulation thermostat is used to control the bath temperature. L

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Langmuir



calibrated for detector nonlinearity using the incoherent scattering from 1 mm-thick D2O samples. BerSANS software was used to subtract dark background due to electronic noise/stray neutrons and the background due to the sample environment to obtain the scattering signal of the sample. BerSANS was furthermore used to obtain values for the transmissions. Further data reduction was done using GRASP software to obtain the final one-dimensional scattering patterns. The structure factor (S(q)) was calculated according to the following equation

S(q) =

I(q) F(q)

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.8b03792. Cellulose nanocrystals’ length and width distribution data determined by AFM, acid−base titration data for charge determination on cellulose nanocrystal surface, Debye length calculation as a function of cellulose nanocrystal concentration, rheological data on cellulose nanocrystal suspensions, residuals for VH and VV autocorrelation function fits, experimentally determined diffusion coefficients with standard deviations (PDF)



REFERENCES

(1) Axenov, K. V.; Laschat, S. Thermotropic Ionic Liquid Crystals. Materials 2011, 4, 206−259. (2) Kelly, S. M.; O’Neill, M. Liquid Crystals for Electro-Optic Applications. In Handbook of Advanced Electronic and Photonic Materials and Devices; Academic Press: Burlington, 2001; pp 1−66. (3) Giavazzi, F.; Crotti, S.; Speciale, A.; Serra, F.; Zanchetta, G.; Trappe, V.; Buscaglia, M.; Bellini, T.; Cerbino, R. Viscoelasticity of Nematic Liquid Crystals at a Glance. Soft Matter 2014, 10, 3938− 3949. (4) Bolhuis, P.; Frenkel, D. Tracing the Phase Boundaries of Hard Spherocylinders. J. Chem. Phys. 1997, 106, 666−687. (5) Bates, M. A.; Frenkel, D. Influence of Polydispersity on the Phase Behavior of Colloidal Liquid Crystals: A Monte Carlo Simulation Study. J. Chem. Phys. 1998, 109, 6193−6199. (6) Boluk, Y.; Lahiji, R.; Zhao, L.; McDermott, M. T. Suspension Viscosities and Shape Parameter of Cellulose Nanocrystals (CNC). Colloids Surf., A 2011, 377, 297−303. (7) Thielemans, W.; Warbey, C. R.; Walsh, D. A. Permselective Nanostructured Membranes Based on Cellulose Nanowhiskers. Green Chem. 2009, 11, 531−537. (8) Morandi, G.; Heath, L.; Thielemans, W. Cellulose Nanocrystals Grafted with Polystyrene Chains through Surface-Initiated Atom Transfer Radical Polymerization (SI-ATRP). Langmuir 2009, 25, 8280−8286. (9) Yue, Y.; Zhou, C.; French, A. D.; Xia, G.; Han, G.; Wang, Q.; Wu, Q. Comparative Properties of Cellulose Nano-Crystals from Native and Mercerized Cotton Fibers. Cellulose 2012, 19, 1173−1187. (10) Beck-Candanedo, S.; Roman, M.; Gray, D. G. Effect of Reaction Conditions on the Properties and Behavior of Wood Cellulose Nanocrystal Suspensions. Biomacromolecules 2005, 6, 1048−1054. (11) Jasmani, L.; Eyley, S.; Wallbridge, R.; Thielemans, W. A Facile One-Pot Route to Cationic Cellulose Nanocrystals. Nanoscale 2013, 5, 10207−10211. (12) Eyley, S.; Thielemans, W. Surface Modification of Cellulose Nanocrystals. Nanoscale 2014, 6, 7764−7779. (13) Habibi, Y.; Lucia, L. A.; Rojas, O. J. Cellulose Nanocrystals: Chemistry, Self-Assembly, and Applications. Chem. Rev. 2010, 110, 3479−3500. (14) Chu, G.; Wang, X.; Chen, T.; Xu, W.; Wang, Y.; Song, H.; Xu, Y. Chiral Electronic Transitions of YVO 4:Eu 3+ Nanoparticles in Cellulose Based Photonic Materials with Circularly Polarized Excitation. J. Mater. Chem. C 2015, 3, 3384−3390. (15) Zubik, K.; Singhsa, P.; Wang, Y.; Manuspiya, H.; Narain, R. Thermo-Responsive Poly(N-Isopropylacrylamide)-Cellulose Nanocrystals Hybrid Hydrogels for Wound Dressing. Polymers 2017, 119. (16) Sampath, U. G. T. M.; Ching, Y. C.; Chuah, C. H.; Singh, R.; Lin, P. C. Preparation and Characterization of Nanocellulose Reinforced Semi-Interpenetrating Polymer Network of Chitosan Hydrogel. Cellulose 2017, 24, 2215−2228. (17) Zhao, W.; Li, X.; Gao, S.; Feng, Y.; Huang, J. Understanding Mechanical Characteristics of Cellulose Nanocrystals Reinforced PHEMA Nanocomposite Hydrogel: In Aqueous Cyclic Test. Cellulose 2017, 24, 2095−2110. (18) Tang, J.; Sisler, J.; Grishkewich, N.; Tam, K. C. Functionalization of Cellulose Nanocrystals for Advanced Applications. J. Colloid Interface Sci. 2017, 494, 397−409. (19) Wu, Y.; Wang, L.; Qing, Y.; Yan, N.; Tian, C.; Huang, Y. A Green Route to Prepare Fluorescent and Absorbent Nano-Hybrid Hydrogel for Water Detection. Sci. Rep. 2017, 7, No. 4380. (20) Shao, C.; Wang, M.; Chang, H.; Xu, F.; Yang, J. A Self-Healing Cellulose Nanocrystal-Poly(ethylene Glycol) Nanocomposite Hydrogel via Diels-Alder Click Reaction. ACS Sustainable Chem. Eng. 2017, 5, 6167−6174. (21) Hou, K.; Li, Y.; Liu, Y.; Zhang, R.; Hsiao, B. S.; Zhu, M. Continuous Fabrication of Cellulose Nanocrystal/poly(ethylene Glycol) Diacrylate Hydrogel Fiber from Nanocomposite Dispersion: Rheology, Preparation and Characterization. Polymer 2017, 123, 55− 64.

The CNC 1 wt % background-corrected intensity data (I(q)) were used to determine the form factor (F(q)) to obtain the structure factors shown in Figure 2.



Article

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]. ORCID

Jonas Van Rie: 0000-0002-2658-3561 Christina Schütz: 0000-0003-0238-1639 Alican Gençer: 0000-0001-5225-7121 Wim Thielemans: 0000-0003-4451-1964 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors want to thank Hartmut Kriegs for helping with the setup of the DLS experiments. Dr Urs Gasser and Dr Joachim Kohlbrecher are acknowledged for helping in setting up the SANS experiments. Furthermore, professors Jan Dhont and Pavlik Lettinga are acknowledged for the interesting discussions and theoretical input. The authors would like to thank Research FoundationFlanders (FWO) for funding under the Odysseus grant (G.0C60.13N), KU Leuven for grant OT/14/072, and the EU Interreg Vlaanderen-Nederland program for the Accelerate3 equipment investment grant. W.T. also thanks the Provincie West Vlaanderen for his Chair in Advanced Materials. C.S. thanks the Alexander von Humboldt Foundation for financial support via a Feodor Lynen scholarship. This work uses SANS results from experiments performed at the SANS-II instrument at the Swiss spallation neutron source SINQ, Paul Scherrer Institute, Villigen, Switzerland (proposal 20180361) for which we are also grateful. M

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir (22) Martin, C.; Barker, R.; Watkins, E. B.; Dubreuil, F.; Cranston, E. D.; Heux, L.; Jean, B. Structural Variations in Hybrid All-Nanoparticle Gibbsite Nanoplatelet/Cellulose Nanocrystal Multilayered Films. Langmuir 2017, 33, 7896−7907. (23) Goikuria, U.; Larrañaga, A.; Vilas, J. L.; Lizundia, E. Thermal Stability Increase in Metallic Nanoparticles-Loaded Cellulose Nanocrystal Nanocomposites. Carbohydr. Polym. 2017, 171, 193−201. (24) Kaboorani, A.; Auclair, N.; Riedl, B.; Hosseinaei, O.; Wang, S. Cellulose Nanocrystal (CNC)-Based Nanocomposites for UV Curable High-Solid Coating Systems. J. Coat. Technol. Res. 2017, 14, 1137− 1145. (25) Arredondo, J.; Jessop, P. G.; Champagne, P.; Bouchard, J.; Cunningham, M. F. Synthesis of CO2-Responsive Cellulose Nanocrystals by Surface-Initiated Cu(0)-Mediated Polymerisation. Green Chem. 2017, 19, 4141−4152. (26) Ladhar, A.; Ben Mabrouk, A.; Arous, M.; Boufi, S.; Kallel, A. Dielectric Properties of Nanocomposites Based on Cellulose Nanocrystals (CNCs) and Poly(styrene-Co-2-Ethyl Hexylacrylate) Copolymer. Polymer 2017, 125, 76−89. (27) Popescu, M.-C.; Dogaru, B.-I.; Popescu, C.-M. The Influence of Cellulose Nanocrystals Content on the Water Sorption Properties of Bio-Based Composite Films. Mater. Des. 2017, 132, 170−177. (28) Dai, S.; Prempeh, N.; Liu, D.; Fan, Y.; Gu, M.; Chang, Y. Cholesteric Film of Cu(II)-Doped Cellulose Nanocrystals for Colorimetric Sensing of Ammonia Gas. Carbohydr. Polym. 2017, 174, 531−539. (29) Van Rie, J.; Thielemans, W. Cellulose-Gold Nanoparticle Hybrid Materials. Nanoscale 2017, 9, 8525−8554. (30) Onsager, L. The Effects of Shape on the Interactions of Colloidal Particles. Ann. N. Y. Acad. Sci. 1949, 51, 627−659. (31) Lagerwall, J. P. F.; Schütz, C.; Salajkova, M.; Noh, J.; Hyun Park, J.; Scalia, G.; Bergström, L. Cellulose Nanocrystal-Based Materials: From Liquid Crystal Self-Assembly and Glass Formation to Multifunctional Thin Films. NPG Asia Mater. 2014, 6, No. e80. (32) Boluk, Y.; Danumah, C. Analysis of Cellulose Nanocrystal Rod Lengths by Dynamic Light Scattering and Electron Microscopy. J. Nanopart. Res. 2014, 16, No. 2174. (33) Hirai, A.; Inui, O.; Horii, F.; Tsuji, M. Phase Separation Behavior in Aqueous Suspensions of Bacterial Cellulose Nanocrystals Prepared by Sulfuric Acid Treatment. Langmuir 2009, 25, 497−502. (34) Araki, J.; Kuga, S. Effect of Trace Electrolyte on Liquid Crystal Type of Cellulose Microcrystals. Langmuir 2001, 17, 4493−4496. (35) Salajková, M.; Berglund, L. A.; Zhou, Q. Hydrophobic Cellulose Nanocrystals Modified with Quaternary Ammonium Salts. J. Mater. Chem. 2012, 22, 19798−19805. (36) Gebauer, D.; Oliynyk, V.; Salajkova, M.; Sort, J.; Zhou, Q.; Bergstrom, L.; Salazar-Alvarez, G. A Transparent Hybrid of Nanocrystalline Cellulose and Amorphous Calcium Carbonate Nanoparticles. Nanoscale 2011, 3, 3563−3566. (37) Dong, X. M.; Revol, J.-F.; Gray, D. G. Effect of Microcrystallite Preparation Conditions on the Formation of Colloid Crystals of Cellulose. Cellulose 1998, 5, 19−32. (38) Dong, X. M.; Kimura, T.; Revol, J.-F.; Gray, D. G. Effects of Ionic Strength on the Isotropic−Chiral Nematic Phase Transition of Suspensions of Cellulose Crystallites. Langmuir 1996, 12, 2076−2082. (39) Yoshida, H.; Kobashi, J. Flat Optics with Cholesteric and Blue Phase Liquid Crystals. Liq. Cryst. 2016, 43, 1909−1919. (40) Liu, Y.; Wu, Y.; Chen, C.-W.; Zhou, J.; Lin, T.-H.; Khoo, I. C. Ultrafast Pulse Compression, Stretching-and-Recompression Using Cholesteric Liquid Crystals. Opt. Express 2016, 24, 10458−10465. (41) De Souza Lima, M. M.; Wong, J. T.; Paillet, M.; Borsali, R.; Pecora, R. Translational and Rotational Dynamics of Rodlike Cellulose Whiskers. Langmuir 2003, 19, 24−29. (42) Mao, Y.; Liu, K.; Zhan, C.; Geng, L.; Chu, B.; Hsiao, B. S. Characterization of Nanocellulose Using Small-Angle Neutron, X-Ray, and Dynamic Light Scattering Techniques. J. Phys. Chem. B 2017, 121, 1340−1351. (43) Tsvetkov, N. V.; Lebedeva, E. V.; Lezov, A. A.; Perevyazko, I.; Petrov, M. P.; Mikhailova, M. E.; Lezova, A. A.; Torlopov, M. A.;

Krivoshapkin, P. V. Hydrodynamic and Optical Characteristics of Hydrosols of Cellulose Nanocrystals. Colloid Polym. Sci. 2017, 295, 13−24. (44) Wilcoxon, J.; Schurr, J. M. Dynamic Light Scattering from Thin Rigid Rods: Anisotropy of Translational Diffusion of Tobacco Mosaic Virus. Biopolymers 1983, 22, 849−867. (45) Lettinga, M. P.; Barry, E.; Dogic, Z. Self-Diffusion of Rod-like Viruses in the Nematic Phase. Europhys. Lett. 2005, 71, 692−698. (46) May, K.; Stannarius, R.; Kang, K.; Challa, P. K.; Sprunt, S.; Jákli, A.; Klein, S.; Eremin, A. Collective Dynamics in Dispersions of Anisometric Pigment Particles. J. Mol. Liq. 2018, 322−329. (47) Dhont, J. K. G. An Introduction to Dynamics of Colloids; Elsevier: Amsterdam, 1996. (48) Kang, K.; Wilk, A.; Buitenhuis, J.; Patkowski, A.; Dhont, J. K. G. Diffusion of Spheres in Isotropic and Nematic Suspensions of Rods. J. Chem. Phys. 2006, 124, No. 044907. (49) de la Torre, J. G.; Bloomfield, V. A. Hydrodynamic Properties of Complex, Rigid, Biological Macromolecules: Theory and Applications. Q. Rev. Biophys. 1981, 14, 81−139. (50) Tirado, M. M.; Martínez, C. L.; de la Torre, J. G. Comparison of Theories for the Translational and Rotational Diffusion Coefficients of Rod-like Macromolecules. Application to Short DNA Fragments. J. Chem. Phys. 1984, 81, 2047−2052. (51) Filson, D. P.; Bloomfield, V. A. Shell Model Calculations of Rotational Diffusion Coefficients. Biochemistry 1967, 6, 1650−1658. (52) Bloomfield, V.; Dalton, W. O.; van Holde, K. E. Frictional Coeffiecients of Multisubunit Structures. I. Theory. Biopolymers 1967, 5, 135−148. (53) Uhlig, M.; Fall, A.; Wellert, S.; Lehmann, M.; Prévost, S.; Wågberg, L.; von Klitzing, R.; Nyströ m, G. Two-Dimensional Aggregation and Semidilute Ordering in Cellulose Nanocrystals. Langmuir 2016, 32, 442−450. (54) van Bruggen, M. P. B.; Lekkerkerker, H. N. W.; Dhont, J. K. G. Long-Time Translational Self-Diffusion in Isotropic Dispersions of Colloidal Rods. Phys. Rev. E 1997, 56, 4394−4405. (55) Vroege, G. J.; Lekkerkerker, H. N. W. Theory of the IsotropicNematic-Nematic Phase Separation for a Solution of Bidisperse Rodlike Particles. J. Phys. Chem. 1993, 97, 3601−3605. (56) Buining, P. A.; Lekkerkerker, H. N. W. Isotropic-Nematic Phase Separation of a Dispersion of Organophilic Boehmite Rods. J. Phys. Chem. 1993, 97, 11510−11516. (57) Schütz, C.; Agthe, M.; Fall, A. B.; Gordeyeva, K.; Guccini, V.; Salajková, M.; Plivelic, T. S.; Lagerwall, J. P. F.; Salazar-Alvarez, G.; Bergström, L. Rod Packing in Chiral Nematic Cellulose Nanocrystal Dispersions Studied by Small-Angle X-Ray Scattering and Laser Diffraction. Langmuir 2015, 31, 6507−6513. (58) Schütz, C.; Van Rie, J.; Eyley, S.; Gençer, A.; van Gorp, H.; Rosenfeldt, S.; Kang, K.; Thielemans, W. Effect of Source on the Properties and Behavior of Cellulose Nanocrystal Suspensions. ACS Sustainable Chem. Eng. 2018, 8317−8324. (59) Kang, K. Soft-Mode of Charged Chiral Fibrous Viruses (Fd). Soft Matter 2016, 12, 6385−6399. (60) Kang, K. Diffusivity in an Electric-Field−induced Homeotropic Phase of Charged Colloidal Rods. EPL 2010, 92, No. 18002. (61) Schneider, C. A.; Rasband, W. S.; Eliceiri, K. W. NIH Image to ImageJ: 25 Years of Image Analysis. Nat. Methods 2012, 9, 671−675. (62) Strunz, P.; Mortensen, K.; Janssen, S. SANS-II at SINQ: Installation of the Former Risø-SANS Facility. Phys. B 2004, 350, E783−E786.

N

DOI: 10.1021/acs.langmuir.8b03792 Langmuir XXXX, XXX, XXX−XXX