Protonation of Surface Carboxyls on Rice Straw Cellulose Nanofibrils: Effect on the Aerogel Structure, Modulus, Strength, and Wet Resiliency

Rice straw cellulose nanofibrils from the optimal 2,2,6,6-tetramethylpiperidine-1-oxyl oxidation/blending process carrying 1.17 mmol/g surface carboxyls were protonated to varying charged (COO–Na+) and uncharged (COOH) surfaces. Reducing the electrostatic repulsion of surface charges by protonation with hydrochloric acid from 11 to 45 and 100% surface carboxylic acid most prominently reduced the aerogel densities from 8.0 to 6.6 and 5.2 mg/cm3 while increasing the mostly open cell pore volumes from 125 to 152 and 196 mL/g. Irrespective of charge levels, all aerogels were amphiphilic, super-absorptive, stable at pH 2 for up to 30 days, and resilient for up to 10 repetitive squeezing-absorption cycles. While these aerogels exhibited density-dependent dry [11.3 to 1.5 kPa/(mg/cm3)] and reduced wet [3.3 to 1.4 kPa/(mg/cm3)] moduli, the absorption of organic liquids stiffened the saturated aerogels. These data support protonation as a critical yet simple approach toward precise control of aerogels’ dry and wet properties.


■ INTRODUCTION
Cellulose nanofibrils (CNFs) have been commonly produced by shear force processing of aqueous suspensions of cellulose that is often chemically pretreated to aid fibrillation into homogeneous dispersions. The presence of charges plays an essential role in facilitating the disintegration of cellulose into nanocelluloses and the dispersion of the resulting nanocelluloses in aqueous media. The presence of charges in the fabrication and dispersion of nanocelluloses is well recognized and has been extensively studied in CNFs by 2,2,6,6tetramethyl piperidine-1-oxyl (TEMPO)-mediated oxidation. 1 To process or convert TEMPO-oxidized CNFs (TO-CNFs) into solids, a significant quantity of water must be removed where the concentration, 2−4 degrees of oxidation, 3,5 surface charge, 3,5 and protonation 5−7 have been shown to influence their structures and, in particular, their wet mechanical properties and performances. 5,8,9 TEMPO-mediated oxidation of purified rice straw (RS) cellulose, followed by blending, has produced TO-CNFs in reduced dimensions and increased surface charges with increasing oxidation levels. 10 From optimal TEMPO oxidation (5 mmol NaClO/g cellulose) and blending (37k rpm, 30 min), RS cellulose was readily fibrillated into ca. 2 nm wide and 1 μm long TO-CNFs with 1.29 mmol/g charges at 97% yield. Rapid ice nucleation (−196°C, 10 m) and lyophilization of TO-CNF suspensions at and below 0.1 w/v % concentrations produced viable self-assembled fibrils irrespective of their carboxylation levels. However, the finest and most uniform sub-micrometer fibrils were more favorably assembled from unoxidized CNFs by blending 10 and aqueous counter collision, 11 less oxidized ■ EXPERIMENTAL SECTION Materials. Cellulose was purified from Calrose variety RS by sequential organic toluene/ethanol extraction, 1.4% NaClO 2 (pH 3.5, CH 3 COOH), and 5% KOH to ca. 36% yield. 24 Hydrochloric acid (HCl, 1 N, Certified, Fisher Scientific), sodium hydroxide (NaOH, 1 N, Certified, Fisher Scientific), sodium hypochlorite solution (NaClO, 11.9%, reagent grade, Sigma-Aldrich), 2,2,6,6-tetramethylpiperidine-1oxyl (TEMPO, 99.9%, Sigma-Aldrich), sodium bromide (NaBr, BioXtra, 99.6%, Sigma-Aldrich), decane (C 10 H 22 , Certified ACS, Fisher Scientific), and chloroform (CHCl 3 , HPLC grade, EMD) were used as received. All water (pH 5.7) was from the Milli-Q water purification system (Millipore Corporate, Billerica, MA).
TEMPO-CNFs, Protonation of Surface Carboxyls, and Assembly into Aerogels. Optimized TEMPO-mediated oxidation (5 mmol/g NaClO, pH 10) followed by shearing by high-speed blending (37.5k rpm, 30 min, Vitamix 5200, 250 mL) of RS cellulose (1.0 g) produced cellulose nanofibrils (TO-CNFs) at 0.4 w/v % in the aqueous supernatant via centrifugation (5k rpm, 15 min, Thermo-Fisher Megafuge 1.6 L) as per our previous report. 10 The aq. TO-CNFs (pH of 5.7) were divided into three aliquots and diluted to 0.13 w/v % below the 0.2 w/v % gelation concentration. One was the as-is aq. TO-CNFs, and the other two were protonated by dropwise additions of dilute HCl (0.1 M) to pH 4 and 2.8, respectively, and maintained under vigorous stirring for 1 h to prevent flocculation, then dialyzed to below 10 μS/cm. The three parallel aliquots of aq. TO-CNFs had different percentages of sodium carboxylate and carboxylic acid contents at the same degree of carboxylation. Each aq. TO-CNF suspension was degassed, concentrated (Buchi Rotavapor R-114) to 0.6 w/v %, pipetted (∼9 mL) into 1.4 cm inner diameter (ID) Pyrex borosilicate glass or polypropylene (PP) tubes to ca. a height of about 6 cm, and then frozen (−20°C, 15 h, 6 mL) and freeze-dried (−50°C, 0.05 mbar, 3 days, Free Zone 1.0 L Benchtop Freeze Dry System, Labconco, Kansas City, MO) to produce aerogels.
Characterizations. The degree of carboxylation on CNFs by TEMPO oxidation was determined by acid−base conductometric titration (OAKTON pH/Con 510 series). The original aq. CNF aliquot obtained at pH 5.7 was separated into 50 mL aliquots at a concentration of 0.05 w/v %, and 200 μL of 0.5 M NaCl was added to raise the conductivity and 50 μL of 1 N HCl to convert all surface carboxyls to carboxylic acids with excess H + in the dispersion. This acid−base titration used 0.01 M NaOH as the titrant, added in 100 μL increments. The titration curve plotting conductivity against added NaOH showed three distinct regimes, i.e., an initial negative slope of consumed free acid, a plateau region representing neutralized surface carboxylic acid to sodium carboxylate, and a positive slope indicating excess NaOH. The total carboxyl content (σ, mmol/g cellulose) was calculated as where c is the NaOH concentration (M), m is the CNF mass in the suspension (g), and v is the volume (mL) of NaOH consumed in the plateau region from v 1 to v 2 . Since all three CNF aliquots had the same degree of carboxylation, the percent of carboxylic acid on each of the three CNF aliquots was determined similarly without adding HCl. Thus, titration with NaOH converted only surface carboxylic acids on CNFs, and the titration curve showed just the second and third regimes. The nanoscale dimensions of CNFs were determined using atomic force microscopy (AFM) (Asylum-Research MFP-3D, OMCL-AC160TS standard silicon probes, 1 Hz) for the thickness (n > 100) and transmission electron microscopy (TEM) (JEOL 1230, 100 kV) for the width (n > 25). Both utilized just 10 μL of 0.0005 w/v % as-is CNF (pH 5.7) air-dried on fresh mica for AFM and glowdischarged carbon-coated grids (300-mesh copper, formvar-carbon, 100 kV, Ted Pella Inc., Redding, CA) stained with 2 w/v % uranyl acetate for TEM. The TO-CNF lengths were estimated from AFM images of individualized nanofibrils.
The diameter (cm), weight (mg), volume (cm 3 ), and density (ρ a , mg/cm 3 ) of 1 cm tall cylindrically shaped aerogel sections were determined, and their percent porosity (Φ, %) was calculated as where ρ c is the density of cellulose (1.6 g/cm 3 ). 25 The total pore volume (V p ), indicative of the absorption capacity of the aerogel, was calculated using the density (ρ a , mg/cm 3 ) and porosity (Φ, %) by where w s and w 0 were the weights of saturated and dry aerogel, and ρ l was the liquid density (g/mL). Welch's T-tests (a = 0.05) of unequal variance were applied for the statistical comparison of the means. The percent (%) of measured absorption over the pore volume or absorption capacity was also reported. Fourier transform infrared (FTIR) spectroscopy was conducted on milled aerogel powders pressed into pellets (1:100, w/w KBr) (Thermo Nicolet 6700, RT, 4 cm −1 resolution). Freeze-dried cellulose (−196°C, 0.1 w/v %) and milled aerogel powders were characterized by thermogravimetric analysis (TGA-50, Shimadzu, Japan, 10°C/ min, 50 mL N 2 /min) and the degree of (X-ray diffraction) crystallinity (PANalytical XRD, Ni-filtered Cu Kα, λ = 0.1548 nm, 45 kV, 40 mA). The crystallinity index (CrI) was determined using the Segal where Ι 200 was the peak intensity of the 200 lattice plane located at 22.7°2θ, and Ι am is the intensity minimum at ca. 18.7°2θ, attributable to non-crystalline or amorphous cellulose Ι. 26 The micromorphology of each CNF aerogel was characterized using a field-emission scanning electron microscope (Thermo Fisher QUATTRO S Environmental SEM-FEG, 10 mm WD, 5 kV). The aerogels were cut into radial and longitudinal cross-sections using a new razor blade. The cross-sections were then sputter-coated (Au, 20 mA, 60 s, <10 nm thickness) (BioRad SEM Coating System) and imaged by SEM. The cell wall thickness (nm, n > 30) and pore widths (μm, n > 30) were measured using the SEM software. Compressive stress−strain tests characterized the dry and wet mechanical properties of each aerogel using 1 cm tall sections placed between two glass plates for measurements in the air (21°C, 65% relative humidity) and between a glass plate and a jar containing water (pH 5.7) (Instron 5566, 2.5 kN load cell, 1 mm/min loading−unloading). The Young's modulus of each aerogel in the air and the water was calculated from the stress−strain linear elastic region, and the scaling relationship with density was calculated by where E and ρ were the respective modulus and density of each aerogel, and E s (∼150 GPa) and ρ s (1.6 g/cm 3 ) were the longitudinal modulus and bulk density of cellulose. 27 ■ RESULTS AND DISCUSSION CNFs and Protonation. Regioselective TEMPO-mediated oxidation at an optimal 0.81:1 NaClO/AGU molar ratio followed by 30 min of blending 10 converted 95% of RS cellulose into 0.4 w/v % C6 TO-CNFs with a 1.38 ± 0.38 nm thickness, a 2.49 ± 0.34 nm width, and ca. a 1 μm length (Figure 1a,b). The aqueous TO-CNF dispersion was divided into three equal volume portions; one "as-is" (pH 5.7) while the other two were protonated with 0.1 M HCl to either 4 or 2.8 pH and then dialyzed in water. Acid−base conductometric titration of the as-is TO-CNF dispersion showed a total surface carboxylation of 1.17 mmol/g, of which 0.13 mmol/g was carboxylic acid (COOH), thus was designated as 11% COOH TO-CNF. The two protonated samples at respective pHs 4 and 2.8 showed 0.53 and 1.17 mmol/g COOH to be designated as 45% COOH and 100% COOH TO-CNFs. These three aqueous aliquots of 11%COOH, 45%COOH, and 100% COOH TO-CNFs had the same degree of carboxylation and morphology but 89, 55, and 0% negative sodium carboxylate (COO − Na + ) surface charges, respectively. All three aliquots showed shear-thinning behaviors, yet the opacity increased with increasing protonation, which was consistent with the reduced optical transmittance and slight aggregation

Biomacromolecules pubs.acs.org/Biomac
Article of fully protonated TO-CNFs shown previously. 7 In the absence of the inter-TO-CNF electrostatic repulsion, the uncharged 100% COOH TO-CNF was most viscous, attributable to significant hydrogen bonding association among C6 carboxylic acids and abundant C2 and C3 hydroxyls, in addition to physical entanglements due to the high L/W aspect ratio of TO-CNFs. The rheological changes were similar to the most-viscous and least sulfonate surface charge relationship of rod-like cellulose nanocrystals, 28 corroborating more significant inter-nanocellulose associations and closer proximity with lower or no surface charges. Aerogel Properties. At a fixed 1.17 mmol/g total surface carboxylation and 0.6 w/v % TO-CNF concentration, the aerogels fabricated in the 1.4 cm ID glass tubes maintained their original height but had reduced diameters of 1.14, 1.23, and 1.31 cm or 18.5, 12.1, and 6.7% shrinkage, from 11% COOH, 45%COOH, and 100% COOH TO-CNF aliquots, showing 31.9, 23.7, and 3.8% volumetric shrinkage or 68.1, 76.3, and 96.2% volume retention (Figure 2a,b). The pronounced shrinkage from suspension-to-aerogel was attributed to the repulsion of anionic surface-charged TO-CNFs away from the negatively charged glass surface, creating a boundary layer of water near the glass with little or no TO-CNF. The fully protonated TO-CNF aerogel, i.e., 100% COOH TO-CNF, showed the best shape and volume retention, supporting minimal charge-induced TO-CNF excluded volume. Hence, the aerogel from the most-charged 11% COOH TO-CNFs with the most volumetric shrinkage was the densest (8.0 mg/cm 3 ) and least porous (99.5%) (Figure 2c). In a highly linear relationship (r = 0.991, p = 0.086), reducing surface charges by protonation to 45% COOH and 100% COOH TO-CNFs gave aerogels with Biomacromolecules pubs.acs.org/Biomac Article reduced densities of 6.6 and 5.2 mg/cm 3 and increased porosities of 99.6 and 99.7%, respectively. While volumetric shrinkage of aerogel was expected of charged TO-CNFs, the 15% reduction of the aerogel by the most charged 1.04 mmol/ g TO-CNF here was significantly less than the 25% shrinkage of aerogel from far less charged 0.5 mmol/g carboxymethyl CNFs reported. 29 Irrespective of the levels of protonation or surface charges on TO-CNFs, all three aerogels had similar morphologies, i.e., thin and smooth film-like cellular walls surrounding ca. 200 to 300 μm wide and hexagonally shaped pores (Figure 2d−f). Conversely, the external surfaces of these aerogels did vary, i.e., the mostly charged 11% COOH TO-CNF aerogel had large surface pores that appeared to be interconnected with the internal pore structure and attributable to significant ice crystal growth by water in the TO-CNF-excluded boundary. The partially charged 45% COOH TO-CNF aerogel showed the most heterogeneous surface morphology with pores, smooth sections, and ridges. The uncharged 100% COOH TO-CNF aerogel had a smooth and non-porous surface as if mirroring the glass surface, corroborating its minimal TO-CNF-excluded volume (Figure 2g−i). Closer examination revealed that the mostly charged 11% COOH TO-CNF aerogel had the thickest (740 ± 338 nm) cellular walls and the smallest (192 ± 51 μm) pores, while the uncharged 100% COOH TO-CNF aerogel had 36% thinner walls (475 ± 183 nm) and 60% larger pores (312 ± 113 μm) (Figure 2j,k). As salts are known to impact ice crystal growth by disrupting hydrogen-bonding networks, 31 larger ice crystals are expected in the uncharged 100% COOH TO-CNF aerogel, leaving slightly larger pores and porosity and more compacted cellular walls. The thicker cellular walls of the 11% COOH TO-CNF aerogel support that electrostatic inter-TO-CNF repulsion, which impedes close inter-fibrillar associations. Upon freezing and lyophilization, those intracell wall spaces or free volume spaces are consistent with our previously reported intra-lamellar 10 nm mesopores. 14 Since all three TO-CNFs had the same level of carboxylation, nanoscale dimensions, and fixed 0.6 w/v %, only eliminating the surface charge by protonation was responsible for tighter internanofibril compaction of the cellular walls in the 100% COOH TO-CNF aerogel.  (Figure 3a). Notably, the distinct C�O stretching band at 1725 cm −1 signified hydrogen-bonding carboxylic acid groups, 6,30 while isolated and non-hydrogen-bonding carboxylic acid was found at ca. 1740 cm −1 as with the open-ring and carboxylated periodatechlorite-oxidized CNFs. 3 Hence, the uncharged 100% COOH TO-CNFs could self-organize randomly via hydrophilic and hydrophobic interactions without surface charges upon freezing/freeze-drying.
The 100% COOH TO-CNF aerogel had the highest CrI of 0.634 compared to the slightly lower 0.605 and 0.609 CrI of the respective 11% COOH and 45% COOH TO-CNF aerogels and the 0.616 CrI of the original cellulose ( Figure  3b). The reduced CrI values for the aerogels from the as-is and partially protonated TO-CNF were expected from TEMPO oxidation and shear force blending. At the same time, the increased CrI of the 100% COOH TO-CNFs gave evidence of interfacial crystallization. The uncharged surface carboxylic acid is expected to enhance the proximity of CNFs with little inter-CNF free volume to self-assemble compactly, to form hydrogen bonds, and to recrystallize. The 11% COOH TO-CNF aerogel had the highest moisture content and the lowest T max attributable to porous, cellular walls�consistent with

Biomacromolecules pubs.acs.org/Biomac
Article enhanced water uptake by the charged groups and their counterions 9 �and a higher specific surface to heat exposure, respectively. In contrast, the lowest moisture and higher T max of the 100% COOH TO-CNF aerogel supported the notion of tenaciously assembled lamellae via compact, extensively hydrogen-bonded, and possibly recrystallized TO-CNF interfaces (Figure 3c,d).
As TO-CNF charges affected the aerogel densities and porosities due to charge repulsion against the glass, the uncharged and hydrophobic PP tubes were investigated for aerogel from the most highly charged 11% COOH TO-CNFs. The PP-molded aerogels had a statistically equal density of 8.1 mg/cm 3 (±0.7) as the glass-molded aerogel (8.0 ± 0.6 mg/ cm 3 ) (p > 0.05, Welch's T-Test). However, the PP-molded aerogel had anisotropic and heterogeneous pore sizes and shapes in both radial and longitudinal directions (Figure 4a,b); i.e., the outer pores were most irregular and largest (ca. 1 mm width) and became incrementally smaller and more compact toward the center. Furthermore, the external surface of the PPmolded aerogel was non-porous and had distinctive spin-like features (Figure 4c), in stark contrast to the isotropic, honeycomb-like, and an interconnecting open cell structure with a porous external surface of the glass-molded aerogel (Figure 4d−f). The concentrically reducing pore sizes of the aerogel were attributed to the more thermally insulating and hydrophobic PP tube while also explaining the inward collapse of the aerogel surface when wetted with a water droplet driven by increasing capillary pressure (Figure 4c,f). It is worth noting that the water absorption of fully immersed and saturated aerogels from the PP-and glass-molded 11% COOH TO-CNF aerogels was statistically equal (p > 0.05, Welch's T-Test) at approximately 104 mL/g water, filling 80% of the total ∼130 cm 3 /g pore volume. Hence, the mold materials affect the ice templating and, thus, the pore morphology and surface wetting behavior of a negatively charged CNF aerogel, but not their overall porosity and absorption capacity.
Amphiphilic Super-absorption. All three aerogels were amphiphilic and super-absorptive, rapidly taking in water, decane, and chloroform in <3 s. The absorption of water (103−166 mL/g), decane (78−163 mL/g), and chloroform (86−154 mL/g) increased linearly with increasing pore volumes of 125, 152, and 196 mL/g for the 11% COOH, 45% COOH, and 100% COOH TO-CNF aerogels, respectively (Table 1). These liquid absorptions filled 62 to 85% of the pore volumes, confirming the largely open cell and interconnected pore structure observed by SEM (Figures 2, 4). The most charged 11% COOH TO-CNF aerogel absorbed more water than decane or chloroform, indicating a more significant proportion of hydrophilic cellular wall surfaces. In contrast, similar polar and nonpolar liquid absorption by the 45% COOH and 100% COOH TO-CNF aerogels signified similar proportions of hydrophilic and hydrophobic surfaces. Such was ascribed to the random self-organization of TO-CNFs with low or no charges such that their inter-facial self-  Biomacromolecules pubs.acs.org/Biomac Article assembly was stabilized by polar and nonpolar associations or hydrogen bonding and van der Waals interactions, respectively. A fully water-saturated aerogel from the most charged or least protonated 11% COOH TO-CNF was also titrated to show no charge, indicating the charged carboxylates and their counterions to be embedded within the cellular walls and corroborating the inter-TO-CNF free volume or porosity within the cellular walls, which is consistent with the 10 nm mesopores by N 2 desorption of the highly charged (86%) and carboxylated (1.29 mmol/g) TO-CNF aerogel. 10,14 Ostensibly, protonation would reduce inter-nanofibril spacing toward higher nanofibril compaction in the cellular wall. Furthermore, the incomplete filling of the entire pore volume of the aerogels indicated some degree of closed cell pore structure regardless of the protonation extent and the liquids. Further evidence of closed cells was visualized as trapped air bubbles within a fully immersed and saturated aerogel (Figure 5a,b).
The most charged 11% COOH TO-CNF aerogel absorbed less water at pH 2, i.e., 89.5 mL/g or 72% of capacity, than at pH 5.7 and 10, or 103 and 105 mL/g (ca. 85% of capacity), respectively (Figure 5c), indicating constriction of internanofibrillar spaces in the cellular walls by protonation at pH 2. Hence, fewer closed cells are accessible and interconnected with the primary open cell pore morphology through nanopore channels within cell walls at pH 5.7 and 10. The pH-dependent water absorption strongly supports the role of the embedded charged groups and their counterions in creating cellular wall porosity, which was previously associated with water uptake into charged and non-crystalline domains 8,9 and, now, nanoscopic pore interconnectivity and transport.
Repetitive water absorption by squeezing or compressing (ε > 90%) up to 10 cycles trended oppositely for the as-is and fully protonated TO-CNF aerogels, i.e., from an initial absorption of 103 and 187 mL/g, respectively, to converge at 119 mL/g at the seventh absorption, then slightly higher absorbed by the as-is aerogel (125 mL/g) than the fully protonated aerogel (114 mL/g) at the 10th absorption ( Figure  5d). The denser (8.0 mg/cm 3 ) as-is aerogel exhibited better water-activated shape recovery following squeezing and absorbed 21% more by the 10th cycle, indicating the opening of some closed cells and inter-nanofibril spaces within the cellular walls. Conversely, the nearly 40% reduction in water absorption of the fully protonated aerogel from the 1st to the 10th cycle was consistent with its lower density (5.2 mg/cm 3 ) and less wet-resilient structure.
Regardless of the protonation-dependent shape recovery property, all aerogels showed excellent wet stability under acidic condition of pH 2, remaining intact for at least 30 days (Figure 5e). Protonation of surface-charged TO-CNFs prior to self-assembly into aerogels further improved pH stability in near-neutral to even basic conditions, i.e., the fully protonated TO-CNF aerogel remained intact at pH 5.7 and 10 for 30 and 2 days, respectively, while the as-is TO-CNF aerogel disintegrated within 48 h at pH 5.7 and 12 h at pH 10. Hence, while surface-charged TO-CNF aerogels function well under acidic aqueous and organic media, protonation has been proven to be highly effective in reducing inter-CNF electrostatic repulsion to facilitate compact, tenaciously wet-stable, and pH-stable aerogels.
Compressive Behaviors. The densest (8.0 mg/cm 3 ) aerogel from the most charged 11% COOH TO-CNF showed   Table 2) and a stress−strain loading curve with three regimes, i.e., initial linear elastic behavior (ε < 10%), plastic deformation (ε, 10−30%), and densification (up to ε = 80%) (Figure 6a). The less-dense aerogels from partially and fully protonated TO-CNFs had significantly lower moduli of 8.2 and 1.5 kPa/(mg/cm 3 ), respectively, accompanying deformation over wider strain regions. Notably, it was only upon densification (ε > 30%) that the least dense 100% COOH TO-CNF aerogel showed stress. The relative moduli of these aerogels were proportional to their relative density, or The compression strength at maximum loading (ε = 80%) also lowered with charge-dependent density reduction from 3.6 kPa/(mg/cm 3 ) to 2.9 kPa/(mg/cm 3 ) for the respective as-is and fully protonated aerogels, although it was far less affected than the moduli. The effect of protonation on compressive properties was measured by the effect size, 33 which indicates the relationship between the mediating variable, density, and the outcome, moduli or strength, of the aerogels ( Table 2). The larger the effect size, the greater the magnitude of the difference between the average dry moduli and strengths, i.e., while both lowered significantly with decreasing density, the dry moduli experienced the most significant reduction (f = 7.07, r = 0.998, p < 0.001).
All three aerogels showed significantly reduced wet compression modulus and strength (Table 2, Figure 6b), which was expected due to the plasticizing effect of absorbed  Biomacromolecules pubs.acs.org/Biomac Article water in the inter-nanofibril regions, 8 and is consistent with Eshelby's theory of a dry foam softened upon wetting by the inclusion of a high surface tension liquid 34,35 and mediated by protonation-dependent density. The least indiscernible dry and wet moduli of the 100% COOH TO-CNF aerogel confirmed the negligible plasticization effect of water. Furthermore, they confirmed that protonation eliminated electrostatic inter-TO-CNF repulsion, enabling the most extensive self-assembly of TO-CNFs in the cellular walls. Hence, protonation-dependent density changes were ascribed to reduced electrostatic repulsion between glass and surface-charged CNFs. In the absence of charges, ice crystals grew larger and the excluded volume diminished, giving aerogels with larger pores and CNFs that associated more tenaciously into thinner and more compact cellular walls. Conversely, the decane-saturated as-is 11% COOH TO-CNF aerogel was a hardened gel as in a compliant matrix stiffened by liquid inclusion. 34 Since liquid absorption was primarily driven by surface wetting and capillary flow in the pore structure, the rigid organo-gel was attributable to the more hydrophilic cellular surfaces of the 11% COOH TO-CNF aerogel, which absorbs more water than decane or chloroform (Table 1). Hence, the stiffening of the as-is aerogel in decane was ascribed to a repulsive hydrophobic effect, while the softening in water was a plasticizing effect. As a result, evaporative air drying of both decane-and chloroformsaturated aerogels returned both with full shape retention, while strong meniscal forces collapsed the water-saturated aerogel into a film-like structure.
Incremental cyclic compressive loading in water (pH 5.7) of the as-is aerogel released air bubbles underneath the upper glass plate (Figure 6c), evident in the opening of some closed cells (Figure 5a,b). Stepwise loading−unloading in the water from 0% to 40, 60, and 80% compressive strain showed distinct linear elastic regions and full shape recovery hysteresis upon unloading (Figure 6d). Cyclic wet compressions of the as-is aerogel from 0 to 60% strain, i.e., the first seven cycles hourly and the eighth cycle at 24 h and the ninth cycle at 32 h showed decreasing strength from an initial 5.1 kPa (1 st cycle) to 2.9 kPa by the eighth cycle and then lastly to 2.3 kPa (Figure 6e). The aberration at ε ∼ 25% loading during the ninth cycle after 32 h indicated structural buckling of the cellular structure, which was consistent with the loss of structural integrity after 48 h of static immersion at pH 5.7 (Figure 5e).

■ CONCLUSIONS
This work uses the electrostatic repulsion of varying surfacecharged TO-CNFs (1.38 nm thick, 2.49 nm wide, ca. 1 μm long; 1.17 mmol/g carboxylation) to elucidate the effect of charge on inter-TO-CNF associations and self-assembly via ice-templating into aerogels. Reducing CNF surface charges by protonation most prominently produced the least dense (5.2 mg/cm 3 ), most porous (196 mL/g), and best shape-retained aerogels with the most homogeneous and open pore structure and the smoothest external surfaces. Irrespective of protonation levels, all three aerogels were amphiphilic, superabsorptive, and stable at pH 2 for up to 30 days. The aerogel from uncharged CNF absorbed water, decane, and chloroform equally and in highest quantity, whereas that from the most charged CNF absorbed the least but 20−32% more water than the organics. Despite being the most rigid and strongest aerogels in the air and in water by the scaling relationship, embedding those accessible surface charges from the most charged TO-CNFs into the less packed, and thus thicker cellular walls, resulted in the densest aerogel with the lowest water and pH stability.
Conversely, the absence of electrostatic repulsion from the most protonated and unchanged TO-CNF maximized their proximity to assemble compactly into much thinner and more crystalline cellular walls, confirming the central hypothesis that reducing charged surfaces and interfaces during ice-templating and lyophilization minimizes interfacial spaces and enhances the close and compact associations of TO-CNFs. These findings demonstrate that the porous structure and absorbent and compressive properties of hierarchically self-assembled nanocellulose materials can be tuned and enhanced by aqueous protonation of charged TO-CNFs. The aqueously stable hydrogel and organically stiffened organo-gel also demonstrate the potential for water-based applications and further chemical functionalization or processing to produce new materials for advanced performance end uses.