Lightweight, flexible, and multifunctional anisotropic nanocellulose-based aerogels for CO2 adsorption

CO2 adsorption is a promising strategy to reduce costs and energy use for CO2 separation. In this study, we developed CO2 adsorbents based on lightweight and flexible cellulose nanofiber aerogels with monolithic structures prepared via freeze-casting, and cellulose acetate or acetylated cellulose nanocrystals (a-CNCs) were introduced into the aerogels as functional materials using an impregnation method to provide CO2 affinity. The microstructure of the adsorbent was examined using scanning electron microscopy, and compression tests were performed to analyze the mechanical properties of the adsorbents. The CO2 adsorption behavior was studied by recording the adsorption isotherms and performing column breakthrough experiments. The samples showed excellent mechanical performance and had a CO2 adsorption capacity of up to 1.14 mmol/g at 101 kPa and 273 K. Compared to the adsorbent which contains cellulose acetate, the one impregnated with a-CNCs had better CO2 adsorption capacity and axial mechanical properties owing to the building of a nanoscale scaffold on the surface of the adsorbent. Although the CO2 adsorption capacity could be improved further, this paper reports a potential CO2 adsorbent that uses all cellulose-based materials, which is beneficial for the environment from both resource and function perspectives. Moreover, the interesting impregnation process provides a new method to attach functional materials to aerogels, which have potential for use in many other applications.


Introduction
Adsorption is an energy-efficient and low cost method for capturing CO 2 (Pino et al. 2016;Yu et al. 2012). A suitable CO 2 adsorbent should have a high specific surface area, a high porosity with proper pore sizes, a favorable surface chemistry, and a high chemical/ mechanical stability. Many CO 2 adsorbents based on different materials, such as zeolites (Banerjee et al. 2009;Cavenati et al. 2004;Kongnoo et al. 2017;Ojuva et al. 2013), silica gels (Arellano et al. 2016), activated carbon (Fujiki and Yogo 2016;Roberts et al. 2017;Singh et al. 2017), and metal-organic frameworks (An and Rosi 2010;Millward and Yaghi 2005;Salehi and Anbia 2017), have been investigated in recent decades. However, adsorbents based on these materials have some drawbacks, such as their high cost, complex preparation methods, toxicities, and the use of corrosive materials during their preparation. Therefore, researchers continue to look for some new materials to produce CO 2 adsorbents that could be used on the industrial scale.
Biomass-derived materials have attracted attention owing to their abundant resources, renewability, and biodegradability (Jonoobi et al. 2014;Khalil et al. 2012;Oksman et al. 2003). Nanocellulose materials, including cellulose nanofibers (CNFs) and cellulose nanocrystals (CNCs), that can be extracted from various types of biomass have attracted considerable interest owing to their specific physical and chemical properties (Geng et al. 2018;Herrera et al. 2017;Nissilä et al. 2019). Recently, several studies on the preparation of highly porous nanocellulose-based aerogels using ice-templating have been reported, and some have shown interesting adsorption/absorption capabilities. In an adsorption process, gas or liquid molecules are captured on the surface of the adsorbents while in an absorption process, those molecules can enter the bulk of the absorbents and create strong interactions there (Yu et al. 2012). Yang et al. (2018) prepared aerogels containing CNFs and sodium alginate using unidirectional and bidirectional ice-templating. They found that after silane modification, the obtained aerogel was superoleophilic, which have oil absorption capacities up to 34 times its weight. Using similar techniques, Zhang et al. (2019) also reported CNF-based aerogel for selective oil/ organic solvent absorption. The aerogels shown outstanding compression flexibility and high hydrophobicity which allowed the aerogels to absorb oils that were 99 times of its weight. Gorgieva et al. reported a unidirectionally ice-templated cationic dye remover prepared from carboxymethyl cellulose and CNFs with the highest dye adsorption capacity of approximately 1.8 g/g adsorbent (Gorgieva et al. 2019). Valencia et al. (2019) prepared hybrid aerogels consisting of CNFs, gelatin and zeolite for CO 2 adsorption via unidirectional freeze-casting. The aerogel containing 85 wt% zeolite showed an CO 2 adsorption capacity of approximately 1.2 mmol/g at 308 K and 101 kPa.
Although these previous studies illustrate that nanocellulose materials can form highly porous structures with large surface areas, which is beneficial for adsorption, pure nanocellulose does not have good affinity to CO 2 molecules (Gebald et al. 2011). Thus, the sorbent-sorbate affinity needs to be improved to produce a nanocellulose-based CO 2 adsorbent that can adsorb CO 2 from a mixed gas. It has been reported that some polar groups such as nitriles (-CN), carbonyls (-C=O-), acetates (-COO-), and amides (-NHCO-) contribute to a high CO 2 solubility and CO 2 /N 2 selectivity (Park and Lee 2008). There have been several studies on the functionalization of cellulose by amine groups to improve the cellulose-CO 2 affinity. Gebald et al. (2011) synthesized amine-based CO 2 adsorbents composed of CNFs and N-(2-aminoethyl)-3-aminopropylmethyldimethoxysilane, where the highest adsorption capacity achieved was 1.39 mmol/g after 12 h. Subsequently, they found that amine-functionalized nanocellulose provided a stable CO 2 adsorption capacity (approximately 0.9 mmol/g) in ambient air during adsorption/desorption cycling tests (Gebald et al. 2013). However, in addition to an enhanced CO 2 adsorption capacity, amine-based materials require a relatively large amount of energy to regenerate adsorbed CO 2 (Gebald et al. 2011), which is an obstacle to using this kind of adsorbent on a large scale. Under this circumstance, acetyl groups could be an alternative. Kilic et al. (2007) experimentally confirmed that acetate-functionalized polymers are more CO 2 -soluble than polymers with only ether groups because the addition of an ester oxygen provides a much stronger attractive site than that of the ether group for CO 2 interactions. Moreover, the oxygen of carbonyl groups is more electron-rich and more favorable for CO 2 binding. Karimi et al. (2016) functionalized silica membranes with acetyl groups and reported that the CO 2 adsorption was increased compared to that of the unmodified membranes.
Inspired by the previous studies on nanocellulosebased aerogels and acetate-based materials for CO 2 adsorption, in this study, we demonstrated a novel method to prepare lightweight and flexible nanocellulose-based CO 2 adsorbents with functional acetate groups and anisotropic structures. The adsorbents consist of a freeze-cast CNF aerogel and acetylated CNCs in which the acetylated CNCs were attached in the anisotropic porous CNF aerogel by an impregnation process driven by capillary forces. The acetylated CNCs have a certain amount of acetyl groups on the surface; however, the integrity of CNCs is maintained. Thus, they are expected to build nanoscale scaffolds together with the CNF aerogel, which is beneficial for the CO 2 adsorption capacity. Cellulose acetate was used as the reference, which was also attached to the CNF aerogel by impregnated with different amounts. The structure and morphology of the materials and the prepared adsorbents were investigated, and the mechanical properties and CO 2 adsorption capacity of the adsorbents were examined and discussed.

Preparation of the CNF aerogel
First, 62 g of 1 wt% CNF aqueous suspension was mixed with 0.03 g of BTCA (as a crosslinker) by stirring for 1 h at room temperature. The mixture was then frozen unidirectionally by a freeze-casting setup with a freezing rate of 5 K/min (Deville et al. 2006) and freeze-dried for 72 h at a pressure of 0.064 mbar using a freeze dryer (Alpha 1-2 LD plus, Martin Christ GmbH, Germany). All the freeze-dried samples were placed in a vacuum oven at 393 K for 3 h to crosslink the aerogel via esterification between CNFs and BTCA. Finally, the crosslinked CNF aerogel was obtained for further processing.

Acetylation of cellulose nanocrystals
The acetylation method used in this study is based on a heterogeneous process that uses iodine as a catalyst (Abraham et al. 2016). Briefly, solvent exchange of the CNC suspension from water to acetic anhydride (Ac 2 O) through acetone was carried out prior to the modification. A CNC/Ac 2 O suspension with a concentration of 0.02 g/mL of CNC in Ac 2 O was prepared and heated to 373 K for 1 h. Iodine (0.001 g/mL of I 2 in Ac 2 O) was added as a catalyst, and the reaction ran for another 30 min. Subsequently, the suspension was cooled to room temperature, and saturated sodium thiosulphate solution was added dropwise until the color of the reactant changed from brown to transparent to stop the reaction. The mixture was poured into ethanol/water (weight ratio of 2:1) and stirred for 30 min to precipitate the acetylated CNCs (a-CNC). The precipitated a-CNCs was then further washed with ethanol (70 vol%) and distilled water. Finally, a-CNCs was collected and dispersed in acetone to make a homogenous suspension for impregnation.

Impregnation of the aerogels
First, 40 g of the a-CNC/acetone suspension at a concentration of 0.25 wt% was used for impregnation, and the same amounts of cellulose acetate/acetone solution with 0.25 wt% and 1.25 wt% of solid contents, respectively, were prepared as references. The crosslinked CNF aerogel was placed vertically in the suspension/solution until the suspension/solution was completely adsorbed by the aerogel owing to the capillary force (Movie S1, Supporting Information, SI). Because all the 40 g of the solution/suspension was impregnated into the CNF aerogel, the amount of the a-CNCs/cellulose acetate added is controlled by the concentration of the solution. The impregnated aerogel was then placed in the vacuum oven at 368 K overnight to remove acetone. The weight of the aerogel was measured before and after the impregnation process to calculate the amount of a-CNC or cellulose acetate that had been attached to the sample. The sample coding and the compositions of all samples prepared for this study are shown in Table 1.

Atomic force microscopy (AFM)
AFM was used to characterize the morphology of the CNFs used in this study. To prepare the AFM sample, the CNF suspension was diluted to 0.001 wt% of solid content, and then, a droplet was deposited on freshly cleaved mica and dried at room temperature. The sample was scanned using a Veeco MultiMode scanning probe (Santa Barbara, USA) in tapping mode. The width of the CNFs was measured according to the height in the AFM height images. More than 80 nanofibers were analyzed, and the average fiber width was then calculated.

Fourier transform infrared spectroscopy (FT-IR)
FT-IR was used to study the crosslinking of the CNF aerogel and the acetylation of the CNCs. To prepare the FT-IR samples, the crosslinked CNF aerogel was washed thoroughly with distilled water to remove the possible unreacted BTCA and then grounded and dried in a vacuum oven at 393 K. The filtered a-CNCs were also dried in the vacuum oven overnight prior to the FT-IR test. The uncrosslinked CNF aerogel and unacetylated CNCs were prepared in the same manner for comparison. Then, 30 mg of the sample was carefully mixed and pressed with 270 mg of KBr and then tested using a Vertex 80v FT-IR spectrometer (Bruker, USA). For each single experiment, 128 scans were run at a resolution of 4 cm -1 .

Titration
The degree of substitution (DS) of the prepared a-CNCs was measured by titration according to ASTM D871-96 (Zhou et al. 2016). Briefly, 0.1 g of dried a-CNCs were dispersed in 40 mL of 70 vol% ethanol at 60°C for 30 min; then, 40 mL of 0.1 N NaOH was added. The mixture was stirred for 15 min; then, it stood at room temperature for 48 h with occasional shaking. Phenolphthalein was used as a pH indicator, and 0.5 N HCl was used to titrate the mixture until the color changed from pink to faint pink. The unacetylated CNC was titrated as a blank using the same method, and the DS of the a-CNCs was calculated using the following equations (Ramírez et al. 2017): where Acyl% denotes the acyl group content. V b and V s represent the volumes of HCl added to the blank and to the a-CNC sample, respectively. N HCl corresponds to the normality of the HCl solution, which was 0.5 N.
W is the mass of the used sample.

Porosity measurement
The porosity of the prepared aerogels was calculated as (Sehaqui et al. 2011): where the density of the aerogel, q * , was calculated by dividing the weight of the aerogel by its volume. The density of the solid material, q, was calculated as: where x c , x A and x B denote the weight percentages of CNFs, the impregnated material (cellulose acetate or a-CNCs), and BTCA, respectively. q c , q A and q B correspond to the solid densities of cellulose, the impregnated material, and BTCA respectively, which were 1.46 g/cm 3 , 1.28 g/cm 3 , and 1.65 g/cm 3 (Sehaqui et al. 2011).

Scanning electron microscopy (SEM)
The morphology of the aerogels was investigated using a scanning electron microscope (JSM-IT300 InTouchScope, JEOL, Japan). Samples were cut by a sharp blade that was perpendicular and parallel to the freezing direction, respectively. The cut surfaces were coated with gold using a Leica EM ACE200 coater (Wetzlar, Germany) prior to the observation to avoid charging, and the secondary electron images were captured.

Brunauer-Emmett-Teller (BET) surface area
The BET surface area was measured using a BET analyzer (Gemini VII 2390a, Micromeritics Instrument Corp., Norcross, USA). The samples were degassed at 368 K for 24 h under vacuum before the BET surface area measurements were performed.

Compression testing
The mechanical properties of the aerogels were examined by compression testing using a Q800 dynamic mechanical analyzer (TA Instruments, USA) with the compression configuration. Samples were cut into a 1-cm cubic geometry and tested in both axial and radical directions. The experiment was carried out when equilibrated at 303 K with a 0.01 N preload and a strain rate of 10%/min. The elastic modulus (E) of the sample was calculated according to the slope of the initial linear part of the stress-strain curve, and the specific elastic modulus (E s ) of the samples was calculated as shown below: CO 2 adsorption measurement The capacity for CO 2 adsorption of the impregnated aerogels was studied using a BET analyzer (ASAP 2020 Plus, Micromeritics Instrument Corp., Norcross, USA) as a function of pressure. Approximately 20 mg of the sample was loaded and degassed at 393 K for 24 h, and the degassed sample was precisely weighed and transferred back to the analyzer. The CO 2 isotherm was then measured at 273 K in a pressure range of 0-101 kPa. The CO 2 forward-step change breakthrough curve of the aerogels was determined by loading the sample in a steel column (15 mm in inner diameter, 300 mm in length). The sample was first treated under a 0.5 L/min nitrogen flow at 368 K for 24 h. After cooling, a mixture of N 2 /CO 2 with 10% CO 2 (AGA Gas AB, Sweden) was fed at a flow rate of 0.3 L/min at room temperature. The CO 2 concentration was monitored by an IR 1507 fast-response CO 2 infrared transducer (CA-10 Carbon Dioxide Analyzer-Sable Systems International, USA).

Results and discussion
The morphology of the CNFs used to prepare the CNF aerogel was characterized by AFM, as illustrated in Fig. 1a. Their width distribution is shown in Fig. 1b, and the average width was determined to be 8.5 ± 6.1 nm. The CNCs used for acetylation were fully characterized in our previous study with a width of 5.0 ± 1.5 nm and a length of 122.6 ± 53.3 nm (Butylina et al. 2016). As reported by Kilic et al. (2007), polymers containing acetate groups have a considerable higher CO 2 solubility than those with only ether functionalities because the acetate groups have three binding modes that can interact with CO 2 molecules, i.e. two with a carbonyl oxygen and one with an ester oxygen, as shown in Fig. 1c. Thus, the acetylated CNCs were prepared in this study as the functional materials for CO 2 adsorption. The successful acetylation was confirmed using FT-IR, as illustrated in Fig. 1d. Three peaks are commonly used to identify the acetate group. The carbonyl stretching vibration is at 1751 cm -1 (C=O), the C-H in-plane bending of -CO-CH 3 is at 1377 cm -1 and the stretching of C-O in the acetyl group (-COO-) is at 1232 cm -1 (Hu et al. 2011;Uschanov et al. 2011). The prominent peaks for the a-CNC sample indicate that heterogeneous acetylation was successful. No other adsorption was observed between 1840 and 1760 cm -1 ; and therefore, there was no unreacted acetic anhydride left. The absence of the carboxylic group peak at 1700 cm -1 indicated no acetic acid byproducts (Hu et al. 2011). The degree of substitution of the a-CNCs was determined to be 1.6 by titration. After the preparation of the CNF aerogel via freezecasting and freeze-drying, the CNF aerogel was crosslinked with BTCA and then impregnated with an a-CNCs/acetone suspension or cellulose acetate/ acetone solution to prepare the CO 2 adsorbents. The entire procedure is shown schematically in Fig. 2. The crosslinking process was performed to strengthen the CNF aerogel to help it maintain intact during the impregnation process. Figure 3a shows the FT-IR spectra of the CNF aerogel before and after crosslinking. A new adsorption peak appeared in the crosslinked sample at 1730 cm -1 , which corresponded to carbonyl stretching (C=O) in the ester group. It suggests that ester groups were formed between the The impregnation process is a simple yet robust method to attach functional materials to aerogels that we developed in this study. Movie S1 shows a representative impregnation experiment. Driven by capillary forces, the to be impregnated material can be adsorbed by the CNF anisotropic aerogel in a couple of seconds depends on the length of the aerogel and the viscosity of the impregnated material. Moreover, we demonstrate that not only the polymer solutions (cellulose acetate) but also the nanoparticle suspensions (a-CNCs) can be impregnated, which provides a unique method to produce porous materials with special functionalities. Finally, a CO 2 adsorbent with a length as long as 10 cm can be obtained after drying, and there is no obvious shrinkage in the aerogels after the impregnation with 0.1 g CA, as illustrated in Fig. 3b. The aerogel impregnated with 0.5 g CA shrank a bit more due to the increase in the viscosity of the impregnating solution. However, the cylindrical shape of the aerogel was remained. The density, porosity, and specific surface area of the prepared adsorbents are shown in Table 2. The porosities of all the adsorbents are higher than 98%. The specific surface area of CNF-X-a-CNC reached 21.04 m 2 /g, which was higher than CNF-X-0.1CA, and higher than that of CNF-X-0.5CA, which is the lowest likely owing to the relatively large amount of impregnated cellulose acetate (0.5 g) filled part of the pores in the CNF aerogel.
The microstructure of the adsorbents was investigated by SEM and is shown in Fig. 4. It is obvious from Fig. 4a i -d i and Fig. 4a ii -d ii that all adsorbents have a distinct monolithic structure with anisotropic pores, which is formed during the freeze-casting Fig. 2 Process of preparing the acetate-functionalized nanocellulose-based CO 2 adsorbents used in this study Fig. 3 a FT-IR spectra of the native CNF aerogel and crosslinked CNF aerogel and b illustration of the aerogels after freeze-drying, crosslinking, and the impregnation process process owing to the preferred growth of ice crystals when frozen unidirectionally (Lee and Deng 2011). Typically, in the CO 2 adsorption process, gas containing CO 2 will pass through columns filled with adsorbents. Therefore, the microstructure of the adsorbent is important to ensure a rapid and smooth adsorption process. It is reported that the monolithic structure has potential advantages in CO 2 adsorption applications because of the improved mass-transfer, low pressure drop at high flow rates and uniform flow distribution (Rezaei and Webley 2010;Svec and Huber 2006). In Fig. 4, there was no collapse of the  pores observed in the impregnated adsorbents, indicating that the impregnation process did not damage the anisotropic structure. However, the width of the pores in CNF-X-0.5CA (Fig. 4c ii ) showed a decreasing trend compared to that of the pores in CNF-X (Fig. 4a ii ) and CNF-X-0.1CA (Fig. 4b ii ). This phenomenon may be a result of the increase in viscosity of the impregnation solution. With an increase in the amount of cellulose acetate dissolved in a certain amount of acetone, the impregnation solution containing 0.5 g of cellulose acetate became more viscous than the one containing 0.1 g of cellulose acetate. Therefore, CNF-X-0.5CA suffered more of a pressure gradient when dried in the oven, which led to a decrease in the pore width. Comparing the magnified images of CNF-X (Fig. 4a iii ) with the impregnated adsorbents, the cell walls of aerogels impregnated with cellulose acetate (Fig. 4b iii and Fig. 4c iii ) were more compact than that of CNF-X, suggesting that small pores on the cell wall of CNF-X may be filled with cellulose acetate during the impregnation process. Such a filling effect is much less prominent in CNF-Xa-CNC (Fig. 4d iii ), which corresponds to the specific surface area results ( Table 2). Instead of filling up small pores on the cell wall, the a-CNCs could build a nanoscale scaffold together with the CNF aerogel owing to their integrity, which is beneficial for the CO 2 adsorption properties.
The mechanical properties of the adsorbents were characterized by compression tests. The resulting stress-strain curves in both the axial and radial directions are illustrated in Fig. 5, and the E, E s , and yield strength data of the adsorbents in the axial direction are shown in Table 3. The stress-strain curves in the axial direction shown in Fig. 5a display the typical compression behavior of foams, including linear elastic, plateau, and densification regions. Owing to equipment limitations (maximum load is 18 N), the CNF-X-0.5CA and CNF-X-a-CNC experiments could not reach the same strain as those of CNF-X and CNF-X-0.1CA. Compared to those of CNF-X, both E and the yield strength of the aerogels were enhanced after impregnated with cellulose acetate. The E was 9% and 60% higher by impregnated with 0.1 g and 0.5 g cellulose acetate, respectively. This enhancement is attributed to the higher integrity of the cell walls in the impregnated aerogels. Interestingly, CNF-X-a-CNC demonstrated the highest E (328.9 kPa) and E s (19.8 kNm/kg) among all the samples with a relatively low amount of impregnated a-CNCs (0.07 g), indicating that the a-CNCs provided a remarkable reinforcing effect caused by the very high rigidity of the CNCs. Moreover, the E s of CNF-X-a-CNC outperformed previously reported cellulosebased aerogels (4-18 kNm/kg) (Fan et al. 2018;López Durán et al. 2018;Zhang et al. 2019), and it was in the range of expanded polystyrene and polyurethane foams (10-100 kNm/kg) (Gibson and Ashby 1999). Figure 5b shows the compression behavior of the samples in the radial direction, and there was no significant difference observed. All the samples showed very high flexibility, and their elastic strain could reach 60%. The CO 2 adsorption properties of the adsorbents were first evaluated by adsorption isotherms measured at 273 K. Figure 6a shows the experimental data (points) and polynomials fitted to the data (curves). As shown from Fig. 6a, modified CNF-X-0.1CA and CNF-X-a-CNC adsorbed much more CO 2 than CNF-X did at a given pressure. The CO 2 adsorption capacities at 101 kPa were 1.14 mmol /g and 1.05 mmol/g for CNF-X-a-CNC and CNF-X-0.1CA, respectively. The slightly higher CO 2 adsorption capacity of CNF-X-a-CNC may be due to the nanoscale scaffold formed by the a-CNCs, which helps the aerogel to maintain more surface area after impregnation and provides more effective physisorption sites for CO 2 on the surface; this corresponded to the BET surface area data in Table 2. However, CNF-X-0.5CA only showed a slightly better specific capacity than CNF-X, which can be attributed to the decrease in surface area (less adsorption sites) and increase in density simultaneously. The decrease in surface area can be due to the filling effect of CA on the cell walls of the aerogels as discussed above, and it can also be a result from the impregnation process where some delicate porous structure may lose.
CO 2 forward step change breakthrough experiments were also conducted to evaluate the performance of the adsorbents in an adsorption column. As shown in Fig. 6b, the measurements were conducted using a 90 kPa/10 kPa N 2 /CO 2 gas mixture fed to the column under atmospheric pressure, and the CO 2 concentration at the outlet was detected. The amount of adsorbed CO 2 estimated from the breakthrough curves as illustrated in Fig. 6c should correspond to the adsorption capacity obtained from the adsorption isotherms at 10 kPa. CNF-X is considered a carrier material for functional materials because it adsorbs a very low amount of CO 2 (0.007 mmol/g) at 10 kPa, which was determined from the fitted polynomial in Fig. 6a. Therefore, the specific capacity from adsorption isotherms (q iso ) at 10 kPa was calculated using: where Q ad and Q CNF-X are the measured adsorption capacity of the corresponding adsorbent and CNF-X,  . 6 a Experimental adsorption isotherms (points) and fitted polynomials (curves) of CNF-X-a-CNC, CNF-X-0.1CA, CNF-X-0.5CA, and CNF-X measured at 273 K in the pressure range of 0 to 101 kPa. b Schematic illustration of the CO 2 forward step change breakthrough equipment. c Forward step change breakthrough curves of CNF-X-a-CNC, CNF-X-0.1CA, CNF-X-0.5CA, and CNF-X respectively, and m is the mass of the functional material in the adsorbent. The specific capacity from the breakthrough experiments (q bt ) was calculated as: where v gas is the molar flow rate of the mixed N 2 /CO 2 gas (mmol/s), c 0 and c 1 are the concentrations of CO 2 (vol%) in the inlet and outlet, and t ads is the time when c 1 /c 0 reaches 1. Figure 7 summarizes the specific CO 2 adsorption capacity of the functional materials, i.e. cellulose acetate and a-CNCs, in the related adsorbents calculated from both adsorption isotherms and breakthrough measurements. It shows that the CO 2 adsorption capacity of the functional materials from both experiments has similar trends in which the a-CNCs in CNF-X-a-CNC had an approximately 45% higher CO 2 adsorption capacity than cellulose acetate did in CNF-X-0.1CA. This indicates that the use of nanomaterials to form a nano-scaffold on the surface of the aerogel provided more physisorption sites for CO 2 , which contributed to the higher adsorption efficiency of a-CNCs. The cellulose acetate in CNF-X-0.5CA had the lowest specific capacity, which implied that the thicker layers formed by more cellulose acetate may not be fully accessible to CO 2 . Notably, although Fig. 6a shows that the entire CNF-X-a-CNC adsorbent had a similar adsorption capacity as that of CNF-X-0.1CA at 10 kPa, and the mass difference between the impregnated a-CNCs (0.07 g) and the impregnated cellulose acetate (0.1 g) caused the difference in their specific capacities.

Conclusions
Lightweight and flexible CNF-based aerogels with monolithic structures were prepared via freeze-casting. The obtained aerogels were further functionalized by being impregnated in a-CNC/acetone suspension or cellulose acetate/acetone solution. Owing to the anisotropy of the monolithic structure, the aerogels were soft and flexible in the radical direction, while the specific elastic modulus was as high as 19.75 kNm/kg in the axial direction. Introduction of acetate groups increased the affinity of the aerogels to CO 2 , which allowed the functionalized aerogels to be used as CO 2 adsorbents. Unlike the cellulose acetate-forming compact layer on the surface of the aerogel, a-CNCs built a nanoscale scaffold on it. This delicate nanoscale scaffold not only significantly improved the mechanical performance of the adsorbents but also provided more physisorption sites on the aerogel; therefore, CNF-X-a-CNC had a better CO 2 adsorption capacity than CNF-X-0.1CA. Even though CNF-X-a-CNC had the highest CO 2 adsorption capacity among the adsorbents prepared in this study (1.14 mmol/g at 273 K and 101 kPa), more work is needed to make the most of this potential.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds Fig. 7 CO 2 adsorption capacity of a-CNCs and cellulose acetate in the corresponding aerogels estimated from the forward step change breakthrough curves and from the adsorption isotherms the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.