Article pubs.acs.org/Langmuir
Distinct Chiral Nematic Self-Assembling Behavior Caused by Different Size-Unified Cellulose Nanocrystals via a Multistage Separation Yang Hu and Noureddine Abidi* Fiber and Biopolymer Research Institute, Department of Plant and Soil Science, Texas Tech University, Lubbock, Texas 79409, United States ABSTRACT: Cellulose nanocrystals (CNCs) are perfect rodlike nanofibers that can self-assemble and form a chiral nematic phase. We found that different self-assembling morphologies could be formed by different size-unified CNCs. This study reported a facile and new approach of fractionating raw (unseparated) CNCs in a wide particle size distribution (9−1700 nm) into a series of narrower size ranges to obtain size-unified CNCs via a well-designed multistage separation process composed of layered filter membranes with different pore size cutoffs followed by a fast pressurized filtration. The smaller size-unified CNCs readily self-assembled into polish chiral nematic phases with larger pitch value as compared to larger size-unified CNCs. Such a distinction among different chiral nematic phases and pitch values as functions of size was addressed by a mathematical evaluation, which suggested that the reduced volume fraction of the anisotropic phase as a function of both increased ionic strength and reduced crystallinity of rigid-rod-like CNCs is a critical factor. In addition, Fourier-transform infrared spectroscopy, thermogravimetric analysis, and X-ray diffraction results revealed that different size-unified CNCs exhibited particular thermal stabilities and crystallinities even though their chemical and crystalline structures remained unchanged. The discrepancies in physicochemical characteristics and self-assembling chiral nematic behavior among different size-unified CNCs may benefit the specific functionalization of cellulose materials using size-unified fibers instead of raw CNCs containing mixed small and large fibers.
■
ration.14 Under specific conditions (e.g., the presence of different electrolytes,15 orientational codispersion with gold nanorods,16 alignment under an external magnetic field,17 and metallic deposition onto nanofibers),18 CNCs exhibit improved chiral nematic LCM structures that have allowed for advanced applications such as ordered template molding chiral photonic reflectors,19 free-standing mesoporous silica films,20 and enantioselective catalysts.21 CNCs have attracted significant interests because of their renewability and other superior features. Most CNCs are rodlike aggregates in the nanoscale range (i.e., lengths between 50 and 400 nm and widths between 5 and 70 nm),22,23 and they consist of glycosidic units with a degree of polymerization (DP) distribution in the range of 150−300.24 Individual CNCs tend to aggregate into irregular particles with a particulate distribution spanning from nanoscale range to microscale range (20−2000 nm).25,26 Size is a critical parameter for both individual rodlike CNCs fibers and CNCs particles, affecting most properties of CNCs. It determines the rigidity, viscosity, and self-assembling LC phase of CNCs and thus warrants further investigation. The literature contains few studies
INTRODUCTION Cellulose is the most abundant bio-derived polymer in nature. It exists in most biomasses as a cellulosic component, and its content varies with the source from microscale algae to macroscale trees.1 Acidic hydrolysis of native cellulose fibers generates low-molecular-weight cellulose products, referred to as cellulose nanofibers/nanocrystals/nanowhiskers (CNCs), as stable milky suspensions. The use of sulfuric acid at a high concentration is one of the most common treatments to generate CNCs from cellulose, and it can lead to a relatively high yield of more than 60% when using 64% sulfuric acid and hydrolyzing cellulose at 50 °C for 5 h.2 Combining lowintensity ultrasonication with sulfuric acid at a short hydrolysis time of 1.5 h, the yield of CNCs can reach 40.4%.3 Recent studies reported on the use of mineral acids such as phosphoric acid which produced CNCs with enhanced thermal stability and biocompatibility.4,5 CNCs have been extensively used as fiber reinforcements in advanced materials.6 Further chemical modification of CNCs for surfactant adsorption,7 (2,2,6,6tetramethylpiperidine-1-oxyl)-mediated oxidation,8 cationization,9 esterification,10,11 silylation,12 and grafting with other polymers13 can lead to CNCs with interesting characteristics. In suspension, CNCs display an overwhelming ability to behave as self-assembled liquid crystalline molecules (LCMs) and form a chiral nematic phase when subjected to slow water evapo© XXXX American Chemical Society
Received: August 1, 2016 Revised: September 1, 2016
A
DOI: 10.1021/acs.langmuir.6b02861 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 1. Schematic flowchart of multistage separation of raw CNCs.
■
focused on the size fractionation of raw CNCs according to either their fiber size or particle size. A differential centrifugation technique was reported to fractionate CNCs with a narrow size distribution within 100 nm in length but overlooked large-size CNCs greater than 100 nm in length.27 High-pressure homogenization combined with sulfate hydrolysis was used to obtain CNCs with a relatively uniform size of 90 nm in length and 10 nm in width,28 while likewise no largesize CNCs were mentioned in this study. The asymmetrical flow field-flow fractionation (AF4) in conjunction with a multiangle light scattering (MALS) detector was considered a useful tool to fractionate raw CNCs into narrower size ranges. However, no evidence regarding the specific characteristics based on the size-unified CNCs after separation was shown in the research.29−31 An alternative approach to fractionate raw CNCs in terms of their extremely wide particle size range of 9− 1700 nm was developed in this study. Such a fractionation of raw CNCs was performed using a multistage separation process assembled by layered filter membranes with different pore size cutoffs. To the best of our knowledge, this multistage separation was the first application of the multilayer ultrafiltration approach to raw CNCs fractionation. As compared to other approaches, the multistage separation was facile, fast, repeatable, and easily scalable, and it may provide new uses for CNCs fractionation and subsequent functionalization. No study on the specific functionalization of size-unified (separated) CNCs has been conducted so far. All related studies are based on raw CNCs once they are isolated from the acidic hydrolysis of native cellulose. Functionalizing size-unified CNCs obtained in this study may achieve superior characters according to discrepancies in physicochemical properties and liquid crystalline structures caused by the size difference of CNCs as compared to raw CNCs. This study has paved the way to understand the effect of size difference of CNCs on their diverse characteristics to target advanced applications based on size-unified CNCs. In this study, we performed a multistage separation to fractionate raw CNCs over a wide size range into a series of narrower size intervals based on their particle sizes. The collection and material characterization of size-unified CNCs were subsequently conducted to demonstrate the effectiveness of CNCs multistage separation and specify particular characters of size-unified CNCs. Liquid crystalline phases of different sizeunified CNCs in a series of narrower size ranges were observed, and the formation of distinct chiral nematic phases and chiral nematic pitch values (P) is mathematically discussed.
EXPERIMENTAL SECTION
Materials. Raw cotton linter was harvested from cotton plants locally (Lubbock, TX). Prior to its use, scouring and bleaching procedures were performed to remove impurities according to our previous study32 to obtain highly purified cotton fiber containing 99% pure cellulose. The molecular weight (MW) of the purified cotton linter is in the range of 610−870 kDa (degree of polymerization: 3750−5350), as measured by gel permeation chromatography (GPCmax, Viscotek, Houston, TX). Sulfuric acid (Cat. # LC255503, 96%) was purchased from LabChem (Zelienople, PA) and was diluted to 63.5% for cotton linter hydrolysis. Filter membranes with pore sizes of 220 nm (Cat. SA1J788H5), 450 nm (Cat. SA1J791H5), and 800 nm (Cat. SA1J794H5) were purchased from Merck Millipore (Billerica, MA). Preparation and Multistage Separation of Raw CNCs via Sulfuric Acid Hydrolysis of Raw Cotton Linter. Purified cotton linter (11 g) was hydrolyzed in 100 mL of 63.5% H2SO4 at 45 °C for 90 min. The milky solution was centrifuged at 3000 rpm, which clearly separated the supernatant and pellet. The supernatant was discarded, and fresh deionized (DI) water was added. Centrifugation was repeated at the same rpm until the supernatant was not clear. The milky supernatant was gently separated from the pellet. The milky supernatant was subsequently neutralized with KOH (3 mol/L) and dialyzed for at least 3 days. The resultant solution after dialysis against DI water was collected in a glass vessel and allowed to stand for 3 days. The top layer of the solution was collected, and it was ultrasonicated for 5 min prior to the next study. A manual, discrete multistage separation was applied using filter membranes with different pore size cutoffs (i.e., 220, 450, and 800 nm), as illustrated in Figure 1. Individual CNCs are prone to aggregate and form irregular particles in DI water because of molecular agglomerative effects, which enables the separation of raw (unseparated) CNCs according to particle size distribution of CNCs irregular particles rather than singular size of individual CNCs. Pressurized filtration through such a multistage separation was performed. The cakes that formed on each membrane filter layer were gently removed using DI water and collected in fresh DI water. As cakes may block the passage of CNCs, in order to achieve the effective passage and minimize the unwanted CNCs in each size range, the membrane was exchanged once the filtration under the syringe pressure could not proceed. The cake that was formed was redispersed in DI water and the filtration was repeated using new membrane for couple of times. Material Characterization of CNCs. Particle Size Distribution. A particle size analyzer (Zetatrac, Microtrac, Inc., USA) was used to determine the particle size distribution for each separated (sizeunified) CNCs specimen before and after multistage separation. The running parameters for the CNCs were transparency, a refractive index of 1.64, an irregular shape, and a density of 1.016 g/cm3 for a suspension containing CNCs. For the fluid (DI water), the parameters were a refractive index of 1.333, a low temperature of 20 °C with a viscosity of 1.002 mPa·s, and a high temperature of 30 °C with a viscosity of 0.797 mPa·s. The dielectric constant was set to 79. The measurements were conducted over 1 min and repeated three times. B
DOI: 10.1021/acs.langmuir.6b02861 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
Figure 2. Particle size distribution of CNCs before (a) and after multistage separation (b−e). Filter cakes collected from different filter membrane resuspended in DI water possibly contain CNCs particles with major size ranges of (b) below 220 nm, (c) 220−450 nm, (d) 450−800 nm, and (e) above 800 nm. Column bars are distribution percentages of CNCs particles, and broken lines represent the variation of size distribution of CNCs particles. Electron Microscopy. Field-emission scanning electron microscopy (FESEM, Hitachi S-4700) was used to characterize the surface morphologies of the CNCs specimens. The CNCs specimens were ultrasonicated for 5 min and then diluted to 0.001% (w:v). A 100 μL
drop of the specimen was placed onto an aluminum sample stage coated with copper tape and was allowed to dry in air. The dehydrated CNCs specimen was sputter-coated with gold for 1 min and subsequently observed by FESEM on a microscope operated at 2 C
DOI: 10.1021/acs.langmuir.6b02861 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir kV. The aspect size (length and width) of individual fibers of CNCs was measured using the GetData graph digitizer software (version 2.2). Characterization by FTIR, TGA, and XRD. Prior to these characterizations, different CNCs specimens suspended in DI water were lyophilized in a freeze-drier (Freezone 4.5, Labconco, USA). The freeze-dried specimens were then conditioned in a laboratory maintained at a temperature of 21 ± 1 °C and a relative humidity of 65 ± 2% for 48 h. Fourier-transform infrared spectroscopy (FTIR) spectra of the CNCs specimens were recorded using an FTIR Spectrum 400 (PerkinElmer, USA) over the wavenumber range from 4000 to 650 cm−1. The resulting FTIR spectra were analyzed after baseline correction and normalization. Thermogravimetric analysis (TGA) (Pyris1TGA, PerkinElmer, USA) of the freeze-dried CNCs specimens was conducted to determine the thermal behavior of the CNCs in distinct size ranges. TGA was performed between 37 and 600 °C at a heating rate of 10 °C/min under N2 flowing at 20 mL/min. The Pyris software was used to evaluate the weight loss, decomposition temperatures, and weight-loss rate (DTG). X-ray diffraction (XRD) of dehydrated CNCs specimens was conducted using a SmartLab X-ray diffractometer (Rigaku, Japan) equipped with a Cu target (λ = 1.541 867 Å) and operated at a voltage of 40 kV and a current of 44 mA. Fibril CNCs specimens were placed on glass sample slides, and the centers, which were 0.5 mm lower than the surfaces of the glass sample slides, were covered. Another clean glass slide was used to press CNCs specimens to ensure the full coverage of the center area. X-ray scans over the 2θ range from 5° to 50° at a scan rate of 1°/min were performed to analyze the crystalline morphology and crystallinity of the samples. The peak fitting of all XRD patterns was performed using the PeakFit software (PeakFit version 4, SYSTAT, USA). The fitting results were used to calculate the crystallinity and crystal size according to eq 133 and the Scherrer equation (eq 2):34
crystallinty (%) =
0.1 and 1% (w:v), and individual CNCs may aggregate into irregular particles with different particle sizes at high concentration. Unseparated CNCs exhibited a wide size range from 9 to 1700 nm, whereas the separated (size-unified) CNCs were obtained in a series of narrower size ranges after the multistage separation: below 220 nm, 220−450 nm, 450−800 nm, and above 800 nm. The most intense peaks represent the most abundant CNCs in the corresponding size range after the multistage separation. For example, the size range above 800 nm, i.e., the CNCs particles collected on the filter membrane with a pore size cutoff of 800 nm, exhibited a most common size of 818 nm (13%). Similarly, the sizes of the most common CNCs particles that passed through the membrane of 450 and 220 nm are shown for the size ranges of 450−800 nm (Figure 2d), 220−450 nm (Figure 2c), and below 220 nm (Figure 2b). Each relevant CNCs abundance for the specific size range after the multistage separation was dramatically enhanced as compared to the unseparated CNCs. The CNCs in the size fraction below 220 nm for unseparated CNCs were increased from 77.83% to 95.68% after CNCs consecutively passed through membrane with size cutoffs of 800, 450, and 220 nm. Likewise, CNCs in the size fraction of 220−450 nm for unseparated CNCs increased from 16.07 to 35.52%. Large CNCs in the size fractions of 450−800 nm and above 800 nm were greatly increased from 4.55% and 1.55% for unseparated CNCs to 54.03% and 35.25% in the fractionated samples, respectively. It is noticed that some unwanted CNCs were still present in the specific size range which should not include these unwanted CNCs. This was a result from the formation of a cake on the filter membrane preventing the passage of filterable CNCs particles. However, the use of new filter membrane exchange to repeatedly filter the cakes formed on the membrane appeared to have minimized these unwanted CNCs. Although the separation accuracy did not reach 100%, the data from particle size analyzer show certain positive results. On the other hand, this multistage filtration is considered a simple, rapid, and easy strategy to scale up the separation of CNCs in the future as compared to other reported methods,27,28,31 and thereby developing such a CNCs separation method to provide parameters and fundamental concept for future commercialization is of great significance. Material Characterization of Unseparated and Separated CNCs by FESEM, FTIR, TGA, and XRD. Two CNCs specimens, the full-size (unseparated) and below-220 nm (separated) specimens, were observed using FESEM. The surface morphologies of these CNCs show clearly irregular rectangular shape with a low aspect ratio (Figure 3). Axial ratios of CNCs from different cellulose sources or by different isolation methods of acid hydrolysis showed significant discrepancy. Spherelike CNCs and CNCs with other morphologies in addition to rodlike shape or large aspect ratio were reported.25,35 The CNCs isolated from cotton cellulose in this study showed irregular rectangular shape rather than a narrow rodlike shape. The low-speed centrifugation (3000 rpm) instead of the traditional high-speed centrifugation (>10 000 rpm), used in the isolation of CNCs from acid hydrolysis of cotton cellulose, may be the major reason that leads to such a rectangular shape with a relatively low aspect ratio.35,36 For unseparated CNCs samples in a wide size range, the CNCs exhibited various sizes in the range of approximately 30−350 nm in length and 15−75 nm in width. Nevertheless, the CNCs passing through 220 nm filter membrane exhibited more uniform and smaller sizes in the range of 36−60 nm in
∑ Acryl ∑ Acryl + ∑ A amph
(1)
;
Bhkl =
Kλ cos θ (Δ2θ)2 − (Δ2θinst)2
(2)
where ∑Acryl is the sum of the integral area of all crystalline peaks and ∑Aamph is the sum of integral area of all amorphous peaks; Bhkl is the average crystal width of a given plane (hkl); K is a constant of 0.9; λ is the wavelength of the incident X-rays with regards to the XRD instrument (λ = 1.541 867 Å); θ is the center angle (in degrees) of the specific peak; Δ2θ (in radians) is the full width at half-maximum (fwhm) of the corresponding peak and can be read from the fitting results; and Δ2θinst is the instrumental broadening (0.0018 radians). Chiral Nematic Phases of CNCs as a Function of Size. Rodlike CNCs can self-assemble into chiral nematic phases. Polarized optical microscopy (POL, Eclipse LV100, Nikon) was used to observe the self-assembly of CNCs in various size ranges that would most likely exhibit distinct liquid-crystal behaviors. One 0.5 mL drop of CNCs was deposited onto a glass slide and was allowed to dry in air. The POL images were observed and captured. The characteristic parameters of the chiral nematic liquid crystal phases and the pitch (P) value were measured and compared using the GetData graph digitizer software (version 2.2).
■
RESULTS Particle Size Distribution of CNCs before and after the Multistage Separation. Unseparated CNCs across a wide size range were subjected to the multistage separation via a well-designed strategy as shown in Figure 1. They exhibited various distributions of size-unified CNCs in different size ranges as shown in Figure 2. Here, the measurement of particle size distribution of CNCs is based on CNCs particles consisting of one or several individual CNCs because the particle size measurement requires the best sample concentration between D
DOI: 10.1021/acs.langmuir.6b02861 Langmuir XXXX, XXX, XXX−XXX
Article
Langmuir
bonding pattern, all of the CNCs in the different size ranges are in the Iβ crystalline form. The vibrations at 2919 and 2853 cm−1 are assigned to aliphatic C−H stretching.38 The increase in the absorbance of these vibrations for the CNCs in the size ranges above 800 and 450−800 nm suggests that most CNCs in these size ranges contain longer aliphatic chains. The vibration at 1650 cm−1 is assigned to the O−H bending of adsorbed water, which is consistent with the peaks in the broad band from 3400 to 3200 cm−1.39 The formation of more hydrogen bonds between CNCs and water may suggest that more amorphous areas are present in larger CNCs. The peaks at 1001 and 985 cm−1 are attributed to C−O and ring stretching modes,39 which exhibit significant differences among all CNCs specimens in Figure 4b. The decrease in absorbance of these two peaks for larger CNCs (220−450 nm, 450−800 nm, and above 800 nm) suggests that the crystallinity of the CNCs may have been slightly reduced after the multistage separation. The FTIR analysis demonstrates that all CNCs specimens present a cellulose Iβ crystalline form, whereas the CNCs in the larger size ranges may contain more long-chain fragments and more amorphous areas after the separation as compared to the CNCs in the smaller-size ranges. The thermal behaviors of unseparated and separated CNCs were analyzed by TGA, and the thermograms are shown in Figure 5. In general, the weight loss of the specimens in the range of 37−100 °C is the loss of adsorbed water, and the weight loss of the specimens in the range of 100−450 °C is attributed to the loss of cellulosic component. The maximum decomposition temperature for the cellulosic component in the CNCs specimens was determined from the DTG curves in Figure 5b. The weight loss and the maximum decomposition temperature data for the cellulosic component of the CNCs were analyzed. More cellulose was decomposed in the CNCs within the larger size ranges of 100−450 °C. For example, approximately 82% of the cellulosic component was lost from the CNCs in the size range above 800 nm, whereas approximately 69% of the cellulosic component was lost for CNCs in the size range below 220 nm. Cellulosic weight losses for CNCs below 220 nm and unseparated CNCs appear highly similar, which may be the reason that the percentages of CNCs below 220 nm are close to unseparated CNCs: 96% for CNCs separated through 220 nm filter membrane and 78% for unseparated CNCs as shown in Figure 2. The maximum decomposition temperature of the cellulosic component in the CNCs specimens decreased as the CNCs size increased, which suggests that larger size CNCs are easier to decompose than
Figure 3. FESEM images of CNCs: (a, b) in full size and (c, d) below 220 nm.
length and 16−26 nm in width. The statistical analysis of aspect size of individual fibers of two CNCs specimens is summarized in Table 1. The variation amplitude in length before and after the multistage separation was remarkably decreased from 14.2% to 3.2% upon the average value of all measured fibers, while the variation amplitude in width exhibited a smaller decrease from 29.1% to 14.5%. In particular, the ratios of maximum and minimum values were both greatly decreased from 11.6 to 1.7 for length and from 5.0 to 1.6 for width before and after the multistage separation. This suggests that the fiber size of CNCs could be unified with fairly small variation amplitude via such a multistage separation. The FESEM images provided visible evidence that individual CNCs tend to self-aggregate into irregular shapes, which allows CNCs to pass through filter membrane with spherelike pores. In addition, the FESEM images further revealed that CNCs with more uniform sizes could be obtained from raw CNCs. The FTIR spectra of unseparated and separated CNCs in various size ranges are shown in Figure 4. Overall, the CNCs spectra were similar, suggesting that the chemical structure of the CNCs remains unchanged after they were fractioned. Some small discrepancies among the CNCs spectra could be identified in the ranges of 3500−2800, 1600−1700, and 1200−650 cm−1. The broad band from 3400 to 3200 cm−1 represents hydrogen bonding, where the peak at 3292 cm−1 is considered a proof of cellulose Iβ.37 Based on this hydrogen
Table 1. Statistical Analysis of Aspect Size (Length and Width) of Individual Size of Unseparated CNCs and Separated CNCs with the Irregular Particle Size below 220 nma fiber size of CNCs (nm) length and width full size particle size
