Transport Properties of Commercial Cellulose Nanocrystals in Aqueous Suspension Prepared from Chemical Pulp via Sulfuric Acid Hydrolysis.

A cellulose nanocrystal (CNC) sample prepared from chemical pulp via sulfuric acid hydrolysis procedures has been supplied by InnoTech Alberta Inc. in the shape of white dry powder as a prototype product. Some transport coefficients were precisely investigated for the CNC sample in aqueous suspensions at the room temperature of 25 °C such as the average rotational and translational diffusion coefficients (D r and D t) and viscoelastic relaxation times (τv) at dilute conditions. The determined values, D r ≈ 2.3 × 103 s-1 and D t ≈ 1.0 × 10-11 m2 s-1, using depolarized and usual dynamic light scattering (DLS) techniques, respectively, proposed the consistent length and width of L ≈ 170 nm and W ≈ 7.6 nm via a theoretical model for monodisperse rigid rods dispersed in pure water. The viscoelastic behavior for aqueous CNC suspensions containing spherical probe particles was examined using DLS rheological techniques. The obtained value of τv = 1.0 × 10-4 s fairly agrees with that of (6D r)-1 ≈ 7.4 × 10-5 s. Because the theoretical model for monodisperse rods denotes the relationship τv = (6D r)-1, this observation strongly confirms that the CNC sample behaves as approximately monodisperse rigid rodlike particles. Transmission electron microscopy (TEM) images clearly demonstrated a bimodal distribution in rod length with major and small minor peaks at ca. 150 and 240 nm, respectively. Then, the reason for the observed disagreement between the L values resulted from the transport coefficients and the major peak in TEM images is the presence of the small minor component with L ≈ 240 nm. Consequently, individual nanosize rodlike crystalline particles in the CNC sample well disperse without forming large aggregations because of strong interactions and behave as isolated individual rods in dilute aqueous suspensions.

Introduction

Cellulose nanocrystals (CNCs) are nanosize crystalline rodlike particles made from cellulose via acid hydrolysis procedures by the use of several types of acidic agents.[1−4] CNCs are highly potential nature base organic polymeric materials for a wide range of applications such as a high-performance reinforcement agent in common plastics because of their low density, ca. 1.6 g cm–3; high strength, ca 10 GPa;[5] and high modulus, ca 150 GPa.[2,3,6,7] CNCs are also highly potential materials for optical applications because produced composites containing CNC particles can be transparent because of the size of CNC particles less enough than the wavelength of visual light.[8,9] Moreover, it has been well-known that many kinds of CNCs have the ability to make a liquid crystalline phase in aqueous suspensions at concentrations higher than critical values depending on the CNC species.[10,11] This liquid-crystalline-forming feature also fascinates many researchers who work in the field of basic science and also industrial practical applications.[10,11] Moreover, CNCs have a highly promising future to be used in the fields of food and pharmaceutical industries.[12−15] The fact that cellulose is the most abundant organic materials, that is, biomass, on the globe as the main component of the plant cell walls, which are generated repeatedly every year by a vast kind of plants infinitely, should be an important point to be deeply considered to develop a well-established sophisticated sustainable society on our globe.[2,3,7]

CNCs possessing different dimensions and morphologies are able to be prepared by choosing the source of cellulose samples, the kind of hydrolysis acidic agents and conditions.[2−4] If one subjects higher plant cellulose samples originated from, for example, cotton and ordinary wood pulp to hydrolysis reactions, needle- or spindle-like CNC particles possessing lengths shorter than a few hundred nanometers and widths of a few nanometers are prepared.[2−4] Enormously long whiskerlike CNC particles on the order of micrometers in length are obtained from tunicate and also green algae cellulose by hydrolysis reactions.[2]

In this study, we pay attention to a white dry CNC sample obtained from wood pulp. InnoTech Alberta Inc. (Edmonton) has started to operate a CNC pilot plant which can produce CNC samples as prototype products targeting commercially available products from any fibrous materials containing a high α-cellulose contentlike chemical pulp. They say that their CNC samples contain nanosize spindlelike crystalline particles with a relatively monodisperse length distribution that maybe based on transmission electron microscopic (TEM) images.[16] However, the dispersibility and stability of CNC samples in aqueous suspensions essentially important for practical industrial applications are highly dependent on the surface chemistry of CNC particles controlled by hydrolysis procedures and cannot be simply rated from only TEM images in general. Here, the distribution of length (L) and width (W) or diameter for nanosize spindlelike CNC particles contained in a CNC sample that was kindly supplied by InnoTech Alberta Inc. were analyzed based on carefully taken TEM images, and the average, LTEM and WTEM, values and their standard deviations were also evaluated. Then, transport coefficients for the commercial prototype CNC sample in dilute aqueous suspensions, such as rotational and translational diffusion coefficients (Dr and Dt) and viscoelastic relaxation time (τv), were determined to rate the dispersibility into aqueous media highly related to the efficiency of the CNC samples for industrial applications.

The determined values of Dr, Dt, and τv were used to evaluate the consistent length, Ltr, and width, Wtr, of the nanosize spindlelike particles contained in the CNC sample as rodlike ones using a theoretical model for monodisperse rodlike particles dispersed into viscous fluid.[17] The average values of the length and diameter of rodlike particles, LTEM and WTEM, of the CNC sample were compared with those of Ltr and Wtr. Recently, theoretical models[17] to discuss precisely the shape and dimension of dispersed particles at extremely dilute conditions have been established. However, some reported experimental transport coefficient data obtained at dilute conditions sometimes contain serious problems because of difficulties of experiments. Correct transport coefficients evaluated from carefully carried experiments are necessary to determine Ltr and Wtr precisely. Moreover, experimental studies to determine correct transport coefficients at dilute conditions are essential for especially newly obtained particles such as the CNC sample examined in this study for the investigation of transport and rheological features of moderate to concentrated suspensions.

Very recently, Mao et al.[18] investigated the size and morphology of the particles contained in the same CNC sample as we examined. They used several kinds of scattering techniques, TEM observation, and also dynamic light scattering (DLS) experiments to obtain precise information on the morphology of the CNC particles and concluded that the length of the CNC particles in the sample derived from the Dr and Dt data did not coincide with each other. The values of consistent Ltr and Wtr determined in this study will be compared with their values and discuss the validity of our Ltr and Wtr.

Results and Discussion

TEM Images

Figure shows a typical TEM image of CNC particles of the tested white dry CNC sample, which was highly diluted in water suspension and stained by the negative staining method. Many highly extended needle or spindle shape objects, which seem to be individual primary CNC particles and possess the length, L, and width, W, less than 300 and 20 nm, respectively, are observed in Figure . The distribution of the L and W was quantified in the form of histograms, as seen in Figure . The observed L value shows the presence of a bimodal distribution: a major peak found at 150 nm and a small minor one at 240 nm. Similar bimodal L distributions in CNC samples were also clearly recognized in the previous systematic study.[4] On the other hand, the W value has only one sharp peak at 10 nm. Then, the TEM image shown in Figure looks similar to that previously published as CNC particles made from chemical pulp via sulfuric acid hydrolysis techniques including the bimodal distribution of L seen in Figure .[4,18,19]

Figure 1

TEM image of negatively stained CNC particles contained in the examined white dry powder sample resulted from the ordinary chemical pulp via a hydrolysis procedure with sulfuric acid. A black scale bar means 200 nm.

Figure 2

Normalized distribution functions for the length, L, (a) and width, W, (b) of primary particles in the CNC sample evaluated from the TEM image, as shown in Figure . The LTEM and WTEM values represent the average L and W, respectively. Moreover, the Ltr and Wtr represent the consistent values estimated from the translational and rotational diffusion constants, which will be described later in detail.

In the case of dried out cellulose fibril samples made from kinds of higher plant resources such as softwood and hardwood pulp, ramie, cotton, and so on, cellulose microfibrils (or microfibril aggregates) are constructed by numbers of (Iα and Iβ type) crystallites connected sequentially by amorphous (or disordered) cellulose chain regions along the elongated long-axis direction.[2−4,19,20] In the short-axis (or a transverse sectional) direction, microfibrils (or microfibril aggregates) are constructed by one (or a few) microfibril(s) formed by, for example, 36–68 β-1,4-glucan chains for higher plants[20,21] (or more β-1,4-glucan chains depending on the species of source plants[21]) and held together by intra- and intermolecular hydrogen bonding, whereas β-1,4 glycoside linkages sustain the rod- or threadlike structure of cellulose microfibrils in the elongated direction. Then, it has been widely accepted that most disordered regions existing every ca 150 nm (approximately) periodically along cellulose microfibrils are effectively decomposed, and then isolated crystallites (CNC particles) possessing their length close to 150 nm are generated in the acidic hydrolysis procedures.[20] Moreover, it has been well-known that a small amount of sulfate (−OSO3–) groups are introduced on the surface of CNC particles in the procedure of hydrolysis reaction using sulfuric acid, and the obtained CNC particles demonstrate rather higher dispersibility in aqueous suspension because of electrostatic repulsion between negatively charged sulfate groups than CNC particles without any ionic groups obtained by hydrochloric acids.[1−5] Consequently, one might expect high dispersibility for the examined CNC sample in aqueous suspension as in the form of isolated individual CNC primary particles, as seen in Figure .

The size of the transverse section of a microfibril formed by 36–68 β-1,4-glucan chains has been discussed by many researchers.[2−5,21,22] Ding et al.[22] recently estimated the size to be 3.2 nm × 5.3 nm. The observed W value for CNC particles in the sample markedly larger than the size of the microfibril, as seen in Figure b, clearly reveals that a couple of microfibrils are still bundled tightly by strong hydrogen bonding even after the hydrolysis procedure.

There are two possible explanations following the bimodal distribution of the L values observed in Figure a. The first explanation is that the amorphous or disordered regions exist perfectly in order with the constant periodicity of ca 150 nm along cellulose microfibrils, and the degradation process of β-1,4 glycoside linkages by hydrolysis reaction is not perfect. The second explanation is that the periodicity of disordered regions is not perfect, but the degradation process for glycoside linkages is perfect. Agarwal et al.[23] recently discovered that crystalline domains are absent in the never-dried native-state wood cellulose samples based on the Raman spectroscopic data. Moreover, Horikawa et al.[19] very recently claimed an idea that the disordered regions along dried out cellulose microfibrils are preferentially resulted from the drying out process of cellulose samples and are not there in the raw (or wet) native-state cellulose fibrils. Then, this idea[19] is strongly supported by the discovery by Agarwal et al.[23] Then, the latter explanation for the bimodal L distribution seems plausible.

The average LTEM and WTEM values were evaluated to be 152 and 8.84 nm, respectively, from the distribution functions, as shown in Figure . The standard deviations of the L and W were also calculated to be 34.0 and 2.37 nm, and these values mean the distribution of L and W rather sharp. However, it should be noted that TEM images obtained by negative-staining methods sometimes provide slight overestimation in object sizes, especially in the nanometer range similar to the W values, because of artifacts due to an essential problem for broad adsorption sites for heavy metal cations such as uranyl ions to objects in staining processes. Coagulative aggregation of primary CNC particles in the drying process during specimen preparation is of course another serious problem.[21]

Rotational and Translational Diffusion

If suspended particles into viscous medium fluids have anisotropic structure such as spindle, rodlike, or rotational ellipsoidal shapes, the rotational diffusion constants of the particles can be determined by the use of depolarized DLS measurements.[18,24−28] The incident light source is polarized in the vertical, v, direction. Then, the depolarized, vh, condition is able to be set by placing a polarizing optical device in the horizontal, h, direction in front of a photodetector. According to Berne and Pecora,[28] the intercept of a plot of the first cumulant (Γ1vh), which is calculated from the autocorrelation function of scattered light electric field obtained under the depolarized condition [g1vh(t)] converted from the autocorrelation function of scattered light intensity [g2vh(t)] via the relationship g1vh(t) = [g2vh(t) – 1]1/2, against the square of the magnitude of scattering vector (q2) provides the average value of Dr for anisotropic suspended particles in the manner of Γ1vh| = 6Dr, that is, the intercept of Γ1vh versus q2. On the other hand, the initial slope of the first cumulant under the usual DLS condition (Γ1vu) against q2 provides the average translational diffusion constant of suspended particles in the manner of Γ1vuq–2| = Dt, irrespective of the shape of suspended particles.

The dependencies of Γ1vh and Γ1vu on q2 are plotted in Figure for aqueous suspensions of the CNC sample at the concentrations of c = 1.0 and 5.0 g L–1. Although the Γ1vh data were relatively scattered because of low scattering intensity under the depolarized condition, the intercept, Γ1vh|, was able to be evaluated irrespective of the c values. Then, we might conclude that Dr = 2.3 (±0.2) × 103 s–1 for the nanosized CNC particles in the sample. On the other hand, the initial slope between Γ1vu and q2, which is shown with a thin red solid line in Figure , provides the translational diffusion constant, Dt = 1.0 (±0.1) × 10–11 m2 s–1. Moreover, it seems that the slope between Γ1vh and q2 shown with a blue thin broken line is identical to the value of Dt. This observation is confirmed theoretically because the relation Γ1vh = 6Dr + Dtq2 has been derived theoretically.[24−28]

Figure 3

Square of the magnitude of the scattering vector, q2, dependencies of first cumulants under the usual DLS condition, Γ1vu, and depolarized DLS condition, Γ1vh, of the autocorrelation functions of scattered light electric field, g1vu(t) and g1vh(t), for aqueous suspensions of the CNC sample at two concentrations of 1.0 and 5.0 g L–1.

In the case of the usual DLS condition, the q2 dependence of the Γ1vu data approached that of Γ1vh in the high q2 range examined, as shown in Figure . If one ignores coupling between rotational and translation motions, the first cumulant, Γ1vu, calculated from the autocorrelation function of the scattered light electric field under the usual DLS condition, g1vu(t), can be approximately described as followswhere S0 and S1 mean dynamic structure factors depending on the magnitude of the scattering vector, ||.[28] Because the value of S0 is much greater than that of S1 at small scattering angles leading to low || value conditions, the relationship Γ1vu = Dtq2 holds as described above. However, it is expected that the value of S1 increases gradually with increasing || and becomes much greater than that of S0 at higher scattering angles. At the condition satisfying ||L > 3, where L means the length of rodlike particles, the same relationship Γ1vu = Dtq2 + 6Dr as Γ1vh holds in accordance with Berne and Pecora.[28] The fairly good agreement between the Γ1vu and Γ1vh data observed at q2 > 8 × 1014 m–2, as shown in Figure , reconfirms that the nanosize CNC particles of the tested sample possess the average values of Dr = 2.3 (±0.2) × 103 s–1 and Dt = 1.0 (±0.1) × 10–11 m2 s–1, respectively, as characteristic transport coefficients in dilute aqueous suspensions.

Evaluation of Length and Diameter

Methods to evaluate the length, L, and width, W, of monodisperse rodlike particles dispersed in a viscous fluid have been established based on theoretical models taking account of the hydrodynamic end effect of rodlike particles using numerical calculation methods.[17,29,30] Then, the Dr and Dt values can be calculated as functions of L and W in an explicit form. Both of the Dr and Dt values are referred to determine consistent L and W precisely. According to TEM images which have been described in detail in a previous section, the values of L and W for individual primary CNC particles distributed in a range from 100 to 300 and 5 to 10 nm, respectively. Then, the values of Dr and Dt were calculated in the range, as shown in Figure , at 25 °C assuming water as a medium fluid. If one assumes a W value ranged from 7 to 8 nm, both of the experimental Dr and Dt values are perfectly reproduced by an identical L value close to 170 nm simultaneously.

Figure 4

Dependencies of rotational and translational diffusion constants, Dr and Dt, on the length, L, of monodisperse rodlike particles suspended into water at the temperature, T = 25 °C, theoretically calculated assuming several width, W, such as 6 (a), 7 (b), 8 (c), and 9 nm (d), respectively.[17] Both of the experimental Dr and Dt values shown with dotted lines are simultaneously reproduced with L ≈ 170 nm, only if the condition 7 < W < 8 nm is satisfied.

Then, the consistent Ltr and Wtr values, which reasonably satisfy both of the experimental Dt and Dr determined for the CNC sample in this study, are determined to be 170 and 7.6 nm, respectively, and fairly agree with the values of LTEM and WTEM, as shown in Figure . Differences between the values, Ltr and Wtr, and the TEM average ones, LTEM and WTEM, are calculated to be +21 and −1.6 nm, respectively. The reason for the 14% increment found in the Ltr from LTEM should be the presence of the small minor component with its length of L = 240 nm. Because the depolarized DLS (DDLS) and DLS techniques used in this study provide the weight average Dr and Dt in general, the presence of the minor component of L = 240 nm possessing higher weight than the main component of L = 150 nm slightly increases the Ltr value than the LTEM value, which is simply calculated as the number average. On the other hand, the reason for the substantial decrement of 19% found in the Wtr from WTEM is not clear at present. Because the transport quantities, Dr and Dt, determined in dilute aqueous suspensions strongly assure a high dispersibility of the examined CNC sample into water, uncontrollable processes in the drying and staining procedure during specimen preparation for TEM observation possibly induce some artifacts increasing the LTEM value, such as coagulation of primary CNC particles and excess smearing. Consequently, we might conclude that CNC particles originated from the examined sample well-disperse in dilute aqueous suspension individually, as shown in Figure .

Elazzouzi-Hafraoui et al.[21] reported the so-called cross-sectional plots obtained from small-angle X-ray scattering (SAXS) experiments for aqueous suspensions of some CNC samples. The obtained cross-sectional plot for Avicel, which is a commercial CNC sample, resulted from wood pulp via hydrolysis procedure using hydrochloric acid and is believed to possess size and structural characteristics similar to the CNC sample examined in this study, showing that its cross-sectional radius of gyration is ca. 3.8 nm. This observation strongly reveals that the average width of CNC particles of Avicel in aqueous suspensions is estimated to be ca. 7.6 nm by SAXS techniques, which is identical to the Wtr value determined above.

Viscoelastic Behavior

Modified DLS measurements[31,32] were carried to obtain dynamic viscoelastic behavior of aqueous suspensions of the CNC sample including a small amount of monodisperse spherical probe particles possessing the diameter of d = 300 nm. When the probe particles are dispersed in pure water, they show Brownian motions caused by thermal agitation, feeling only the water viscosity. However, the probe particles dispersed into a viscoelastic medium would demonstrate other types of Brownian motions feeling the viscoelastic properties of the medium even under the same thermal agitation. The aqueous suspensions of the CNC sample examined in this study were not completely transparent but slightly opaque. Then, the probe particle concentration of 1.5 × 10–2 g L–1 was chosen to get enough scattering intensity to form individual probe particles. It should be noted that the radius of the probe particles was correctly determined as a monodisperse value in a pure water suspension at the same concentration. According to previous studies, the time dependence of g1vu(t) resulted from the Brownian motions of the probe particles dispersed into a liquid medium is directly related to the t dependence of creep compliance, J(t), of the liquid medium via an equation given belowwhere kB and T means the Boltzmann constant and the absolute temperature.[31,32] In the case of pure water as a suspending medium for the probe particles, J(t) was proportional to t with a slope of the reciprocal of water viscosity (ηw = 0.885 × 10–3 Pa s at 25 °C). Because the aqueous suspensions of the CNC sample examined in this study were ranged in a very dilute regime, the determined J(t) data looked similar to that of the suspending medium, water. However, careful data analysis allowed us to recognize that the probe particles dispersed into the CNC suspensions did not provide J(t) simply proportional to t over the entire time region observed, as observed in water. This observation manifested that the aqueous suspensions of the CNC sample are viscoelastic. In accordance with the generalized Voigt–Kelvin model, J(t) for viscoelastic liquids (without instantaneous compliance) can be described with the summation of the component of zero shear viscosity (η0) and a certain necessary number of retardation processes possessing sets of retardation strength (J) and time (λ), respectively, for a mode j as given below.[32,33]

In the case of the dilute aqueous suspensions of the CNC sample, only one retardation process was enough to describe J(t) – tη0–1 data approximately for each concentration sample examined, as seen in Figure . The obtained J(t) curves were able to be converted to complex compliance (J* = J′ – iJ″, i; imaginary unit, J′; storage compliance, and J″; loss compliance) via Fourier transformation in general.[32,33] Then, J′ and J″ are given as functions of the angular frequency (ω) using the determined retardation parameters, J1 and λ1, as

Figure 5

Dependence of J(t) – tη0–1 data on time, t, for the aqueous suspensions of the CNC sample at c = 1.0, 2.0, and 3.0 g L–1 obtained by using DLS viscoelastic measurements. Solid lines imply the fit retardation curves to experimental data with one set of retardation strength, J1, and time, λ1, for each suspension as 2.3 × 10–3 Pa–1 and 1.0 × 10–4 s (c = 1.0 g L–1), 5.6 × 10–3 Pa–1 and 1.0 × 10–4 s (c = 2.0 g L–1), and 9.0 × 10–3 Pa–1 and 1.1 × 10–4 s (c = 3.0 g L–1), respectively.

Finally, the complex modulus (G* = G′ + iG″, G′; storage modulus, G″; loss modulus) can be calculated employing the relationship G* = (J*)−1.[33]

Figure a shows the dependencies of the converted G′ and G″ – η∞ω moduli on ω for the aqueous system of the CNC sample at c = 1.0, 2.0, and 3.0 g L–1. All of the obtained viscoelastic spectra, G′ and G″ versus ω, for the system were well-described with a Maxwell model with a set of relaxation strength (G) and time (τv) and the high frequency limiting viscosity (η∞) as given by

Figure 6

(a) Frequency, ω, dependencies of G′ and G″ – η∞ω for aqueous suspensions of the CNC sample at c = 1.0, 2.0, and 3.0 g L–1, (b) concentration, c, dependencies of G, η0, and η∞ for the aqueous suspension of the CNC sample. Solid lines mean the theoretically calculated values proposing the Dr determined by DDLS techniques under the depolarized condition.

The concentration, c, dependencies of G, η0 (=Gτv + η∞), and η∞ evaluated from viscoelastic spectra seen in Figure a are shown in Figure b. The G value clearly increases with increasing c in proportion to c with the slope of ca 0.2 Pa L g–1, whereas the τv seemed to be approaching the constant value of 1.0 × 10–4 s with decreasing c. Moreover, the values of η0 and η∞ slightly increase with increasing c.

Theoretical consideration has been established for a viscoelastic behavior of suspensions of rodlike particles dispersed into viscous media.[33,34] The theory for monodisperse particle suspensions predicts the relationship between the viscoelastic relaxation time and the rotational relaxation time of dispersed rods; τv = (6Dr)−1 in an extremely dilute condition without touching between particles, and that between the relaxation strength and concentration; G = (3/5)vkBT, where v means the number density of CNC particles in suspensions.[34] Supposing the Dr is determined using DDLS techniques, the value of τv = 7.4 × 10–5 s can be calculated, which is 26% smaller than the experimental τv value via the DLS viscoelastic measurements. The experimental τv is close to the (6Dr)−1 value calculated from the dimensions of a rod with L = 190 nm and W = 8 nm, which is not so far from the value of Ltr and is placed between the L values of the major and minor components seen in Figure . On the other hand, the experimental value of Gc–1 reasonably agrees with the calculated value of 0.2 Pa L g–1, as seen in Figure b, assuming the density of the CNC particles to be 1.6 g cm–3 and the values of Ltr = 170 nm and Wtr = 7.6 nm obtained from transport coefficients above. Moreover, the proportional constants of (η0 – ηw)c–1 and (η∞ – ηw)c–1 are also theoretically evaluated to be 1.97 × 10–5 Pa s L g–1 ([η] = 22.3 mL g–1) and 4.93 × 10–6 Pa s L g–1, respectively. The agreement between experimental values and theoretical ones shown with solid lines in Figure b seems reasonable.

We consequently might conclude that CNC particles of the examined CNC sample disperse well into water and behave as individually isolated rodlike particles possessing the dimension with L ≈ 170 nm and W ≈ 7.6 nm also from the viewpoint of viscoelastic behavior of aqueous suspensions of the CNC sample, as clearly demonstrated in the TEM observation described above. The reason why the CNC sample has such a high dispersibility into water should be effective electrostatic repulsive interaction between the CNC particles because of the presence of anionically charged sulfate groups in water, which are introduced on the surface of CNC particles during the hydrolysis reaction using sulfuric acid. It looks similar to the fact that the electrostatic repulsive interaction between sulfate groups effectively prevents from hydrogen bond (re)formation between hydroxy groups on the surface of different CNC particles in the sample. At last, we conclude that the commercial prototype CNC sample examined in this study has morphological characteristics, that is, highly extended rodlike structure with approximately monodisperse length and width of 170 and 7.6 nm, respectively, which would be a promising feature as industrial materials to provide anisotropic mechanical and/or optical properties with low density of carbohydrates.

Mao et al.[18] very recently reported that the rotational and translational diffusion coefficients were determined to be Dr = 4.0 × 103 s–1 and Dt = 4.4 × 10–12 m2 s–1, respectively, for CNC particles contained in the same sample as examined in this study. Because the reported Dr value was miscalculated by the authors and the correct value can be calculated to be Dr ≈ 670 s–1 from the data shown in the literature,[18] their (correct) Dr and Dt are substantially smaller than the values determined in this study and give much longer L values than the LTEM and Ltr, as seen in Figure a. However, the reason for the observed disagreement between their Dr and Dt values and that determined in this study is not clear at present. Because the TEM image of CNC particles shown in the literature[18] looks quite similar to that in Figure in this study, the Dr and Dt determined in this study seem to be more reliable than their values.

Conclusions

In this study, transport properties for a CNC sample in the white dry powder shape resulted from chemical pulp via hydrolysis procedure by sulfuric acid, supplied by InnoTech Alberta Inc. as a commercial prototype sample, were determined using dynamic light scattering techniques in aqueous suspensions at room temperature of 25 °C. The obtained average rotational, translational diffusion coefficients and viscoelastic relaxation times for the CNC sample at dilute conditions reveal without contradiction that CNC particles contained in the sample possess almost a monodisperse length and width of 170 and 7.6 nm, respectively. These structural quantities well-correspond to that evaluated TEM images, which demonstrated a vast number of thin needle- or spindle-like shape objects recognized as primary CNC particles, obtained from a highly dilute aqueous suspension of the CNC sample. Then, the CNC particles in the sample keep high dispersibility into water even after the (complete) drying out process necessary as a producing one.

Experimental Section

Materials

A CNC sample made from the hardwood chemical pulp sourced from Populus tremuloides Michx. was kindly supplied by InnoTech Alberta Inc. (Edmonton) in the shape of white dry powder. Highly deionized water with the specific resistance higher than 1.8 × 105 Ω m, which was generated by a Direct-Q UV3 (Merck Millipore, Darmstadt), was used as a medium fluid to prepare aqueous suspensions of the CNC sample. Poly(styrene) spherical latex particles (PL3) possessing the mean diameter of 300 nm with a sharp narrow distribution was purchased from Sigma-Aldrich Co. LLC (St. Louis) in the shape of an aqueous suspension at the concentration of 10 vol % and was added into aqueous CNC suspension as probe particles to measure viscoelastic behavior using DLS techniques.

The concentration of the CNC sample in aqueous suspension used for TEM observation was ranged from c = 0.1 to 0.01 g L–1 and chose a suspension sample at c = 0.05 g L–1 as an adequate one for the purpose and also the statistical analysis of the obtained TEM images. Dilute aqueous suspensions of the CNC sample were prepared at concentrations of c = 1.0 and 5.0 g L–1 for depolarized and usual DLS measurements to determine Dr and Dt values. These concentrations were much lower than the previously reported critical concentrations for usual CNC samples prepared from chemical pulp to show liquid crystalline phases.[10,11] The prepared aqueous suspensions were sonicated for longer than 1 h using a usual sonicator bath to disperse CNC particles effectively and were filtrated through a hydrophilic membrane filter with the pore size of 5 μm to remove coarse dust and CNC particle aggregates. The probe PL3 particles were added to aqueous CNC suspensions at a low concentration of 1.5 × 10–2 g L–1.

Methods

A JEM-1400 Plus transmission electron microscope (JEOL Co. Ltd., Tokyo, Japan) was operated to observe CNC particle specimens at the acceleration voltage of 80 kV. The prepared aqueous suspension of the CNC sample was dropped on lacey film-coated copper grids (SPI Supplies/Structure Probe Inc., West Chester) supported by carbon films. For negative staining, 2% aqueous uranyl acetate solution was dropped on the grid containing CNC particles, and the excess solution was removed using a filter paper followed by air drying.

DLS experiments for Dr and Dt measurements were carried out using an ALV/CGS-3 compact goniometer system equipped with a 7004/UBS correlator (ALV-Laser, Langen) and an Nd:YAG laser (λ = 532 nm) as a light source. Because the incident laser source was vertically polarized, the depolarized and usual measuring conditions were obtained by inserting a horizontally polarizing optical device or not. Then, the usual DLS condition detects all of the scattered light from the vertically polarized incident laser source. Auto correlation functions of scattered light intensities were recorded at 25 °C as functions of time at several scattering angles ranged from 30 to 150°.

DLS experiments for viscoelastic measurements were carried out using a DLS-7000HL DLS photometer (Otsuka Electronics Ltd., Hirakata) equipped with an R9880U-01 photomultiplier tube (Hamamatsu Photonics K.K., Hamamatsu), an LSI correlator (LS Instruments AG, Fribourg), and a He–Ne laser (λ = 632.8 nm) as a light source. All of the measurements were carried out at 25 °C at the fixed scattering angle of 60°.