Thermally Insulating and Moisture‐Resilient Foams Based on Upcycled Aramid Nanofibers and Nanocellulose

Low‐density foams and aerogels based on upcycled and bio‐based nanofibers and additives are promising alternatives to fossil‐based thermal insulation materials. Super‐insulating foams are prepared from upcycled acid‐treated aramid nanofibers (upANFA) obtained from Kevlar yarn and tempo‐oxidized cellulose nanofibers (CNF) from wood. The ice‐templated hybrid upANFA/CNF‐based foams with an upANFA content of up to 40 wt% display high thermal stability and a very low thermal conductivity of 18–23 mW m−1 K−1 perpendicular to the aligned nanofibrils over a wide relative humidity (RH) range of 20% to 80%. The thermal conductivity of the hybrid upANFA/CNF foams is found to decrease with increasing upANFA content (5–20 wt%). The super‐insulating properties of the CNF‐upANFA hybrid foams are related to the low density of the foams and the strong interfacial phonon scattering between the very thin and partially branched upANFA and CNF in the hybrid foam walls. Defibrillated nanofibers from textiles are not limited to Kevlar, and this study can hopefully inspire efforts to upcycle textile waste into high‐performance products.


Introduction
The production of textiles consumes a large amount of water, often uses toxic chemicals for dyeing, and is responsible for large emissions of greenhouse gases.Moreover, more than 80% of the textile fibers used for clothing are either incinerated or disposed of in landfills, and less than 15% of used clothing is currently recycled. [1,2]However, textile fibers with useful properties could serve as raw materials for the production of materials for largescale applications, e.g., in the building sector. [3]OI: 10.1002/adma.202305195   Thermal insulation materials are essential to ensure and maintain a pleasant indoor climate in buildings. [4]Several of the most commonly used insulation materials are fossil-based, e.g., expanded polystyrene (EPS) and polyurethane (PU), and the energy needed to heat or cool buildings is directly related to thermal conductivity. [5]enewable, non-fossil-based materials such as wood chips and recycled paper were extensively used for thermal insulation prior to the introduction of fossil fuelbased foams, but their insulating performance is relatively poor. [6]Low-density foams/aerogels with porous 3D networks, e.g., silica and graphene aerogels, can display very low thermal conductivities, but complex preparation routes and brittleness limit their applications. [7,8]Foams and aerogels based on bio-based nanofibrils such as nanocellulose, [9][10][11][12] nanochitin, [13] and "top-down" processed porous wood-derived materials [14] have been shown to combine low thermal conductivity with mechanical resilience.It was recently demonstrated that freeze-cast nanocellulose-based composite foams with columnar or lamellar macropores or cells aligned in the freezing direction display excellent mechanical properties, fire-retardancy, and low thermal conductivities that outperform fossil-based foams such as PU or EPS foams. [15]nalysis of the different contributions to heat transfer in anisotropic nanocellulose foams showed that the dimensions and alignment of the nanofibrils have a profound influence on the thermal conductivity and that the super-insulating properties (i.e., a thermal conductivity lower than that of air, 25.7 mW m −1 K −1 ) can be mainly attributed to interfacial phonon scattering between thin nanofibrils. [16]he preparation of synthetic polymeric nanofibrils is challenging, but Kotov and coworkers showed how aramid nanofibers (ANF) could be produced by a chemical splitting route from macroscopic poly(paraphenylene terephthalamide) (PPTA) fibers, also known as Kevlar. [17]The ANF nanofibers possess similar properties to Kevlar, including very high thermal stability, and mechanical strength. [18]And ANF aerogels and foams are mechanically strong and can sustain elevated temperatures. [19,20]he chemical splitting method has recently been used for the chemical recycling of Kevlar-based products. [21]The addition of ANF can substantially improve the thermal stability and flame retardancy of hybrid or composite ANF/polyimide and ANF/CNF aerogels. [22,23]The thermal conductivity of ANF and hybrid ANF-based aerogels and foams varied between 42 and 26 mW m −1 K −1 , [20][21][22][24][25][26] which is similar to commercially available insulation materials such as EPS and PU. [5] Hybid foams of ANF produced from PPTA fibers and CNF displayed a thermal conductivity of 25 mW m −1 K −1 , which suggests that combining ANF with other nanofibrils, e.g., nanocellulose, can reduce heat transport further.[23] However, no previous study has demonstrated super-insulating hybrid foams or aerogels based on upcycled Kevlar, and the effect of relative humidity (RH) on the thermal insulation properties has not been studied.
Here, we upcycled Kevlar yarn into aramid nanofibers by a combination of chemical splitting and hydrothermal acid treatment (upANF A ) and prepared hybrid foams of the branched and very thin upANF A together with TEMPO-oxidized cellulose nanofibril (CNF) by directional ice templating from aqueous dispersions.The nanofibrils were highly aligned in the icetemplated foam walls, and the thermal conductivity perpendicular to the aligned nanofibrils varied between 18 and 23 mW m −1 K −1 for the hybrid foams with upANF A contents between 40 and 5 wt% in the studied RH range (20-80%).The thermal conductivity of the hybrid upANF A /CNF foams was found to decrease with increasing upANF A content (5-40 wt%), which was related to pronounced phonon scattering at the interfaces of the very thin upANF A and CNF by theoretical calculations.This work demonstrates how upcycled and renewable nanofibers with a small diameter can be integrated into super-insulating foams to achieve enhanced thermal insulation properties, and illustrates how phonon scattering can be interfacially engineered by controlling the assembly, dimensions, and alignment of the nanofibrils.

Results and Discussion
Aramid nanofibers (upANF A ) were prepared from poly(paraphenylene terephthalamide) (PPTA) (Kevlar) yarn by chemical splitting and deprotonation of amide groups in dimethyl sulfoxide (DMSO) saturated with potassium hydroxide (KOH), followed by hydrothermal treatment in acid (Figure 1a). [27]The colloidally stable upANF A aqueous dispersion could be mixed with CNF and produce well-dispersed dispersions (Figure 1b), which were freeze-cast (Figure 1c) and freeze-dried to produce foams with compositions ranging from 5 to 40 wt% upANF A .The foams are abbreviated as CNF-upANF A (X), where X represents the upANF A content in wt%.The intensity of the yellow color of the hybrid foams increased with increasing amounts of yellow upANF A (Figure 1d).
The combination of chemical splitting and hydrothermal acid treatment of Kevlar yarn resulted in branched and very thin upANF A (Figure 2a,b).TEM images (Figure S1, Supporting  S1 in the Supporting Information.c) AFM image of CNF and; d) diameter distribution for CNF.One-Way ANOVA analysis were performed on three data groups on 40 samples, see Table S2 in the Supporting Information.e)  -potentials of CNF-only, upANF A , and CNF-upANF A dispersions with different upANF A contents (5, 20, and 40 wt%).f) Viscoelastic properties of CNF-only and CNF-upANF A dispersions with a total CNF and upANF A concentration of 0.5 wt%.
Information) suggest that the main branches of upANF A have 2-6 sub-branches and the length of the sub-branches range between 0.1 and 1 μm.The average length of upANF A was estimated to be 1.9 μm (length distribution in Figure S2 in the Supporting Information), with an average thickness of ≈2.8 ± 1.7 nm (Figure 2b) estimated by atomic force microscopy (AFM).It should be noted that the thickness of upANF A is similar to CNF (Figure 2c,d; Figure S3, Supporting Information).It is important to note that the diameter of the upANF A is significantly smaller than the thicknesses previously reported for ANF prepared by chemical splitting in DMSO (4.5-30 nm). [18,19,27,28]The average diameter of CNF obtained from AFM (2.5 nm) (Figure 1d) was smaller than determined by TEM (3.4 nm) (Figure S3, Supporting Information). [29]The AFM image of the mixed dispersion (Figure S4, Supporting Information) shows that upANF A and CNF are well mixed and the components do not phase-separate.
The CNF, upANF A , and their mixtures have relatively high -potentials that vary between −40 and -47 mV at neutral pH (Figure 2e), which are of sufficient magnitude for electrostatic stabilization of the aqueous dispersions.Figure 2f shows the frequency sweep measurement, which reveals how the storage modulus (G′) and loss modulus (G′′) change with frequency for CNF-only and CNF-upANF A dispersions.The cross-over frequency, where G′ and G′′are equal, increases with the upANF A content.The sample with 40 wt% upANF A displays a fully solidlike behavior, with no cross-over frequency within the measured range.This means that the CNF-upANF A dispersions have stronger and more interconnected networks than the CNF-only dispersion, which can store more elastic energy and dissipate less viscous energy under oscillatory shear.The enhanced network formation and gelation of the CNF-upANF A dispersions with increasing upANF A content can probably be attributed to the increased degree of entanglement induced by the long and branched upANF A .The steady-shear viscosity, which measures the resistance to flow under constant shear, also increases with increasing upANF A content (Figure S5, Supporting Information), reflecting the more branched and entangled upANF A fibrillar network.The densities of the ice-templated and freeze-dried hybrid foams: CNF-upANF A (5): 6.6 ± 0.1 kg m −3 ; CNF-upANF A (20): 5.8 ± 0.1 kg m −3 ; and CNF-upANF A (40): 5.9 ± 0.2 kg m −3 , are closely related to the concentrations of the aqueous dispersions that were used for freeze-casting.The slight density difference may be due to small differences in shrinkage or moisture uptake of the different foams during conditioning (40%RH at 22 °C).The scanning electron microscopy (SEM) images of the radial cross-sections (Figure 3a-d) and axial cross-sections (Figure S6, Supporting Information) show that the ice-templating method results in honeycomb-like macropores that are oriented in the ice-growth direction.The structure and pore dimensions of the ice-templated composite CNF-upANF A foams correspond well to previous reports of ice-templated nanofibrillar foams, [10,16,30] which show that the CNF and upANF A in the mixed and welldispersed aqueous dispersions do not significantly disturb the growth of the ice crystals.The average cross-section diameter of the columnar macropores (Figure S7, Supporting Information) ranged between 25 to 40 μm for the CNF and CNF-upANF A foams with an upANF A content up to 20 wt%, which suggests that the freeze casting process is relatively unaffected at low to intermediate upANF A contents.The macropore diameter of the foams at the highest investigated upANF A content of 40 wt% (CNF-upANF A (40)), was higher and broader and ranged between 20 and 100 μm.The Brunauer-Emmett-Teller (BET) surface area of the foams that were obtained from the nitrogen adsorption-desorption isotherms (Figure S8, Supporting Information) was only slightly influenced by the upANF A con-tent and varied from 15 to 23 m 2 g −1 (Figure S9, Supporting Information).
The CNF and upANF A nanofibrils are preferentially oriented in the freezing direction, as shown by the 2D small-angle X-ray scattering (SAXS) patterns (insets of Figure 3e). [31]The averaged degree of orientation () of CNF and upANF A nanofibers was evaluated from the azimuthal angle plot of the SAXS intensities (Figure S10, Supporting Information) using the equation  = (180 − FWHM)/180, where FWHM is the full-width half maximum of the azimuthal angle curve. [32]Figure 3e shows that the orientation degree decreases somewhat with increasing upANF A content but remains high (0.87), also at the highest investigated upANF A content of 40 wt%.The structure and orientation of CNF and upANF A in the ice-templated foam walls are schematically illustrated in Figure 3f.
The presence of a single and relatively broad diffraction peak at 20°in the X-ray diffraction pattern of upANF A (Figure 3g) suggests that the degree of upANF A is relatively low.ANF prepared by chemical splitting of PPTA fibers display three diffraction peaks at 20°, 23°, and 28°that are assigned to the (110), (200), and (004) reflections, respectively. [21,33,34]The reduced crystallinity of upANF A can probably be related to the additional hydrothermal acid treatment that results in a significant reduction in the nanofibril diameter (Figure 2a) compared to ANF prepared by only chemical splitting (Figure S11, Supporting Information).The two peaks located at 15°and 22.5°in the XRD pattern for CNF are characteristic of the cellulose type I  polymorph. [35]he XRD patterns of all the hybrid CNF-upANF A foams are dominated by the signal from the CNF due to the low crystallinity of upANF A compared to that of CNF.
The broad absorption band at 3340 cm −1 in the infrared (IR) spectra (Figure 4a) can be assigned to O-H and N-H stretching vibrations. [36,37]The absorption band at 2904 cm −1 can be ascribed to the C-H asymmetric and symmetric stretching of CNF, and the absorption band at 1036 cm −1 can be ascribed to the C-O stretching vibration of CNF.The characteristic peak at 1644 cm −1 from the amide C═O of upANF A experiences a bathochromic shift with increasing upANF A content (see the spectra of CNF-upANF A (40) and CNF-upANF A (20)  in the magnified graph in Figure 4b), which indicates the formation of interfibrillar H-bonding between upANF A and CNF. [38,39]he Raman spectra of the upANF A and the hybrid foams (Figure 4c) show similar features to those of ANF produced by chemical splitting, [23] which suggests that the additional hydrothermal acid treatment of the upANF A had a minor effect on the molecular backbone of the upcycled nanofibrils.The C-C ring stretching vibration (1184, 1279, 1512, and 1610 cm −1 ), the C-H in-plane vibration (1328 cm −1 ) and the C═O stretching (1648 cm −1 ) that were found for the upANF A could also be observed in the Raman spectra of the hybrid foams with an upANF A content of 20 wt% and above.
Thermal gravimetric analysis (TGA) (Figure 4d) shows that the thermal stability of the hybrid foams improves significantly with increasing amounts of upANF A .The derived differential TG curves in Figure 4e suggest that the thermal decomposition of CNF occurs at ≈260 °C and that upANF A decomposes at a significantly higher temperature, between 500 and 560 °C. [40]The very high thermal stability of upANF A corresponds well with previous studies that also showed a high fire resistance of hybrid CNF-ANF foams. [23]he water resilience of the foams was probed by measurements of the dynamic contact angle, water uptake, and mechanical properties as a function of relative humidity.Figure 5a shows that the hybrid CNF-upANF A (40) foam can support the added water drop and displayed a water contact angle of 54°, while the addition of a water drop resulted in a partial collapse of the CNFonly foam.The moisture uptakes of the prepared foams at different relative humidity with their standard deviations are compared in Figure 5b, the values are also listed in Table S4 (Supporting Information) for clarity.The water uptake between 20% and 50%RH was 16-10% smaller for the CNF-upANF A (40) hybrid foam compared to the CNF-only foam (Figure 5b), while the difference at higher RH is negligible.The CNF-upANF A (40) absorbs less moisture than those reported for hybrid foams with hygroscopic components, e.g., CNF-polyoxamer foams and CNFsilica foams. [11,41]Figure 5c shows that the relative humidity only had a minor effect on the compression properties of the CNF-upANF A (40) foams within the investigated RH range of 26-81%RH.The stiffness at low deformations decreases somewhat with increasing RH, while the deformation behavior at compressive strains above 20% is nearly identical at the different RHs.The moisture-resilience of the CNF-upANF A hybrid foams might be ascribed to the water-resilient properties of aramid nanofibers [42,43] and to the high degree of entanglement of the hybrid foam walls.
Figure 6a shows that the thermal conductivity along the radial direction ( r ) of all the ice-templated hybrid CNF-upANF A foams is lower than the value for air (25.7 mW m −1 K −1 at 22 °C) [44] over the entire investigated relative humidity range (20-80%RH).
The heat transport properties of the ice-templated hybrid foams are anisotropic, and the thermal conductivity in the axial direction ( a ) is significantly higher (Figure S12, Supporting Information), which corresponds well to the heat-transport properties of previously investigated ice-templated nanocellulose foams. [9,16]The lowest  r (17 ± 1 mW m −1 K −1 ) that was achieved for CNF-upANF A (40) at 50%RH is similar in magnitude to previously reported bio-based nanofibrillar foams [16] and in fact lower than the CNF-only foam investigated here (22 ± 2 mW m −1 K −1 ).
The influence of relative humidity on the radial thermal conductivity of anisotropic CNF foams was recently analyzed in detail and attributed to a competition between phonon scattering that increases when moisture-induced swelling increases the separation distance between the fibrils, and an increase of the solid conduction as air is replaced with water. [16]The relatively small reduction in water uptake of the hybrid foams compared to the  (20), CNF-upANF A (40), and CNFonly foams at 22 °C.The thermal conductivity of air and isotropic ANF foam ( ANF ). [21]The error bars represent the relative uncertainty of the thermal conductivity from each set of measurements at the relevant relative humidity (see the Experimental Section).b) Estimated solid thermal conductivity of hybrid CNF-ANF systems at 50%RH as a function of ANF content for the upANF A with an average diameter of 2.8 nm (this study) compared to ANF with diameters of 5.5 and 10 nm, respectively.c) Comparison of thermal conductivity and density of the foams presented in this work with previously reported ANF-based aerogels/foams.pure CNF-foams (Figure 5b) and the similar dependence on RH of the radial thermal conductivity suggests that mechanism for the RH-dependence is similar for the hybrid foams as for the pure CNF-foams and highlights the importance of phonon scattering.
The relative importance of phonon scattering [16] for heat transfer of the CNF-upANF A hybrid foams in the radial direction at 50%RH was therefore further estimated.Equation (1) [15] approximates the solid thermal conductivity including phonon scattering of a particulate material,  p p = with  representing the bulk thermal conductivity of the particles, d representing the diameter of the particle, and R k representing the Kapitza resistance at the interface between two particles.The solid contribution of the thermal conductivity including phonon scattering of a composite material with the two solid components, CNF and upANF A , ( p (hybrid)), can be expressed as Equation 2 p ( hybrid ) where the bulk thermal conductivity of the CNF-upANF A composite,  fibers , equals P CNF •  CNF +P upANFA •  upANFA .P CNF and P upANFA represent the relative volumes of CNF and upANF A (P CNF + P upANFA = 1), and P CNF-CNF-interfaces and P CNF-upANF-Ainterfaces are the relative fractions of CNF-CNF and CNF-upANF A interfaces, where it is assumed that each upANF A only interfaces with CNFs: P CNF-upANFA-interfaces = 2 × P upANFA .A CNF can also interface with another CNF: P CNF-CNF-interfaces = 1−P CNF-upANFA-interfaces . CNF was assumed to be similar to the thermal conductivity of a single cellulose nanocrystal (CNC) in the radial direction (720 mW m −1 K −1 ), [45] and  upANFA is assumed to be equal to the thermal conductivity of a Kevlar fiber (3500 mW m −1 K −1 ). [46]The diameters of CNF, d CNF (2.5 nm), and upANF A d upANFA (2.8 nm), were determined by AFM (Figure 2b,d).The Kapitza resistance for two CNF particles (R K_CNF ) and for CNF and upANF A particles (R K_CNF-upANFA ) was estimated by Equation ( 3) and Equation ( 4), respectively The d t_CNF and d t_CNF-upANFA are the distances that two CNF particles, and one CNF particle with one upANF A particle, respectively, span when they are placed parallel to each other with a defined gap (0.56 nm), which corresponds to approximately the thickness of two layers of water molecules. [16] t is the thermal conductivity of a system consisting of two particles with a defined gap (0.56 nm).Here,  t_CNF is assumed to be similar to the radial thermal conductivity of a CNC thin film along the radial direction (270 mW m −1 K −1 ). [45]The value of  t_upANFA is assumed to be the thermal conductivity of a ANF thin film along the radial direction (29 mW m −1 K −1 ). [47] t_CNF-ANF was calculated by the equation  t_CNF-ANF = ( t_CNF +  t_upANFA )/2.The parameters are specified in Table S5 in the Supporting Information The estimated Kapitza resistance for CNF and CNF; R K_CNF = 1.35 × 10 −8 m 2 K W −1 , was found to be about 2.6-fold lower than that for CNF and upANF A ; R K_CNF-upANFA = 3.50 × 10 −8 m 2 K W −1 .The significantly higher Kapitza resistance, and thus larger phonon scattering, for the hybrid system is primarily related to the reduction in the solid thermal conductivity when upANF A replaces CNF with similar diameters.
Interestingly, and surprisingly, the thermal conductivity for the hybrid foams in this study decreased with increasing ANF content, which is opposite to the behavior of foams based on CNF and ANF fibers produced by chemical splitting. [23]It should be noted that the relative importance of phonon scattering scales with the number density of interfaces (Equation ( 2)) and thus the diameter of the ANF (Figure 6b).Increasing the diameter from 2.8 nm, which corresponds to the size of upANF A investigated in this study, to 5.5 and 10 nm, which correspond to diameters previously reported for ANF produced by chemical splitting, [25,28] results in a significant increase of the thermal conductivity.The thermal conductivity of the hybrid system of CNF and ANF with a diameter of 5.5 nm is predicted to be independent of the ANF content, while the hybrid system of CNF and ANF with a diameter of 10 nm displays an increasing thermal conductivity with increasing ANF content, which indeed has been observed previously. [23]These results show that it is essential to obtain numerous interfaces between very thin fibrils or particles to achieve a very low thermal conductivity of low-density hybrid foams.The importance of the heterogeneous interfaces between CNF and upANF A , and the thickness of the upcycled aramid nanofibrils for phonon scattering and heat transfer within the foam walls is illustrated in Figure S13 in the Supporting Information.
This study shows that upcycling of Kevlar yarn by a combination of chemical splitting and hydrothermal acid treatment can result in very thin aramid nanofibrils that can be mixed with CNF and produce super-insulating ice-templated hybrid foams.The radial thermal conductivity of the ice-templated CNF-upANF A (40) hybrid foam is significantly lower than other polysaccharide-based hybrid/nanocomposite low-density foams and aerogels. [10,11,41][22][23][24][25][26] Thermographic recordings (Figure S14, Supporting Information) illustrate the efficient thermal insulation in the radial direction of the CNF-upANF A (40) foams compared to EPS.
The upcycling and repurposing of Kevlar into very thin nanofibrils and lightweight hybrid foams with super-insulating properties presented here could open up other upcycling pathways for fibril-rich waste, e.g., cotton, silk, [48] and performance garments.

Conclusion
We have upcycled Kevlar textile yarn into very thin aramid nanofibers by a combination of chemical splitting and acid hydrothermal treatment and used the upANF A together with cellulose nanofibrils to prepare ice-templated, lightweight, hybrid foams.The anisotropic hybrid foams that contain up to 40 wt% of upANF A display super-insulating properties perpendicular to the aligned nanofibrils at room temperature over a wide relative humidity range, from 20% to 80%RH.The upANF A -containing hybrid foams displayed higher thermal stability and were significantly more moisture resilient compared to pure CNF foams.The superinsulating properties over a wide relative humidity range and the decrease of the thermal conductivity with increasing upANF A content was related to the increased phonon scattering when the very thin upANF A replaces CNF with similar diameters.Estimates of the solid thermal conductivity of nanostructured CNF-upANF A materials show that the phonon scattering contribution is expected to be significantly reduced for ANF with larger diameters of 5.5 and 10 nm, which corresponds to the dimensions of ANF prepared by chemical splitting.This work has highlighted how significant amounts of upcycled and very thin nanofibers from Kevlar textile can be combined with bio-based nanomaterials to produce lightweight super-insulating foams of potential interest for thermal management and packaging applications.
Production of Cellulose Nanofibers (TEMPO-CNF): TEMPO-oxidized cellulose nanofibers were produced according to a reported method. [16]ever-dried softwood sulphite pulp (Domsjö dissolving pulp, 3:2 spruce/pine, Sweden) was washed with HCl (0.01 m) to remove any impurities.Followed by washing with deionized water until the conductivity of the filtered water reached a value lower than 5 μS cm −1 .The purified cellulose pulp was dispersed in water by mechanical stirring to achieve a solution with a weight percentage between 1.5 and 2.0 wt%.TEMPO (0.016 g per 1 g of cellulose pulp) and NaBr (0.1 g per 1 g of cellulose pulp) solutions were prepared in a small volume of water and then added dropwise to the cellulose dispersion.After TEMPO and NaBr were fully dissolved, 10 × 10 −3 m NaClO per gram of cellulose was slowly added with a simultaneous pH modulation to 10 using 1 m NaOH and 0.5 m HCl.The reaction was quenched by adding deionized water after 4 h.After the completion of the reaction, the chemicals were washed away using deionized water until the conductivity was less than 5 μS cm −1 and the pH was around 8. The remaining aldehyde was reduced by adding 0.1 g of NaBH 4 per gram of cellulose to a prepared cellulose dispersion (1.5-2.0 wt%) at a pH of 10.The dispersion was stirred for 3 h.The remaining chemicals were washed away by deionized water until the conductivity was less than 5 μS cm −1 and the pH was around 8. The washed cellulose was then dispersed in water with a weight percentage of 1-2 wt% for grinding.
Upcycling of Kevlar Yarn into Aramid Nanofibers (ANF): upANF A was produced by the combination of chemical splitting [17] and hydrothermal acid treatment.In a typical experiment, chopped Kevlar yarn (2.0 g) was mechanically stirred at a stirring rate of 500 rpm in 1 L of DMSO for 7 d in the presence of 3.0 g of KOH at 30 °C.The exfoliated aramid fibers were transferred to the water by slowly pouring the DMSO solution into a sufficient amount of deionized water, followed by a dialysis process using deionized water until a pH of 6 was reached.The produced upcycled ANF gel (≈4 wt%, 15 g) was hydrothermally acid treated with 20 g of mixed acid, containing 37.5 wt% of H 2 SO 4 and 8.75 wt% HNO 3 , at 120 °C for 30 min. [27]The material was then washed with deionized water until pH reached 6 and conductivity reached a value below 5 μS cm −1 .The weight percentage of the acid-treated upcycled ANF (upANF A ) was determined to be ≈0.4 wt% by thermal gravimetric analysis (TGA).The as-gained upANF A not only preserve their nanofibrous morphologies, but also become well-dispersed in water. [27]We also prepared upANF aqueous dispersion for comparative studies by the solvent change from DMSO to water.
Preparation of upANF A /CNF Foams: Water dispersions with a total weight percent at 0.5 wt% were prepared by mechanically mixing upANF A with CNF using a rotor-stator disperser (Ultra-Turrex T25, IKA) at 8000 rpm for 1 min.The prepared dispersions have a pH value between 6.7 and 7.2 and a  -potential between −45 and −38 mV.The dispersions were degassed in a vacuum with stirring for 3 min and 30 g of the dispersion was then carefully poured into the freeze-casting mold.The frozen ice-templated material was removed from the mold and freeze-dried.The thermal stability of the resulting foams were determined by TGA at weight ratios of 100:0, 95:5, 80:20, and 60:40, and denoted as: CNF-only, CNF-upANF A (5), CNF-upANF A (20), and CNF-upANF A (40), accordingly.
Moisture Uptake and Thermal Conductivity: The moisture uptake of the hybrid foams was determined by measuring the equilibrated weight of the humidified foam conditioned in a Climacell EVO temperature-and humidity-controlled chamber with different relative humidity (20%, 35%, 50%, 65%, and 80%RH) using a Sartorius BSA 124S analytical balance.For each measurement, the humidity was slowly changed over a period of 2 h, and then held constant for 4 h at 22 °C.The moisture uptake was estimated from the average mass recorded during the final hour.The dry mass was estimated from the mass after subjecting the foams to 105 °C for 3 h using a Sartorius QUINTIX224-1S balance.The estimated dry mass (m 0 ) was measured three times for each sample.
The thermal conductivity was determined using the TPS 2500 S Hot Disk Thermal Analyzer (Hot Disk AB, Sweden).The transient plane source (TPS) is a transient technique, which determines the heat dissipation in a sample upon exposure to a heat pulse.Transient techniques are known to require shorter times and allow the measurement of smaller samples in comparison to steady-state techniques.The TPS was used in anisotropic mode which allows the determination of the axial and radial thermal conductivity in one measurement.The reported accuracy of TPS 2500S Hot Disk Thermal Constants Analyzer for isotropic materials is 2-5%, but the accuracy is less for anisotropic, low thermal conductivity materials. [49]herefore a Kapton 5501 sensor (6.4 mm radius) was sandwiched between a pair of identical foams (diameter: 3.7 ± 0.1 cm; height: 2.4 ± 0.1 cm) and thermal contact was ensured by placing a small weight (57 g) on top of the samples (contact pressure: 527 N m −2 ± 33).All measurements were performed with a measurement time of 10 s and a heating power of 10 mW.The samples were placed in a customized vessel that allowed relative humidity (20-80% RH) and temperature control (22 °C). [11]For each relative humidity (20%, 35%, 50%, and 80% RH), five independent measurements were performed with intervals of 15 min and three different pairs of identical foams were used.The radial thermal diffusivity and conductivity were determined using the Cp wet and the wet density of the foams as inputs.The axial thermal diffusivity and conductivity were obtained using the software provided by Hot Disk, as previously described. [16]s reported before, [12] the error bars represent the relative uncertainty of the thermal conductivity of each set of measurements at one specific relative humidity.This is estimated by a propagation analysis (Equation ( 5)) of the uncertainties of the thermal diffusivity (), the density () and the specific heat capacity (Cp wet ) [9] Δ = The uncertainties with a confidence interval of 95% of , , and Cp wet are based on estimates of the average standard deviations (SD) obtained from repeated measurements of several samples (at least 15 per sample for , at least 6 per sample for , and 5 in total for Cp wet ).The systematic uncertainty of  was estimated to be 5%, [49] while no systematic uncertainty was considered for  and Cp wet .ΔCp wet incorporates the water uptake at different relative humidity.
Specific Heat Capacity (Cp): The specific heat capacity at constant pressure (Cp) at a specific relative humidity (Cp wet ) was determined by Equation ( 6) The specific heat capacity for a dry material, Cp dry , was determined using a DSC 214 Polyma (Netzsch GmbH, Germany) on materials that had been dried in an oven at 105 °C.The DSC measurements were performed by applying the ratio method using an empty DSC pan as the reference sample and sapphire as the standard sample.All samples were first heated to 105 °C for 10 min before the Cp dry was measured between −20 and 50 °C at a heating rate of 10 K min −1 under an N 2 atmosphere.The average Cp dry at 22 °C of five independent measurements was considered for the calculations.Cp H 2 O was obtained from the literature, and the weight percent of H 2 O uptake (H 2 O w ) was determined as described above.The Cp values are plotted in Figure S15 in the Supporting Information.
Mechanical Properties: The mechanical properties in compression were measured using an Instron 5966 universal testing machine (Instron, USA) equipped with a 100 N load cell at a strain rate of 10% min −1 .The foams were conditioned at 22 °C either at 26%RH over a saturated solution of potassium acetate, or at 56%RH over a saturated solution of magnesium nitrate, or at 81%RH over a saturated solution of ammonium sulphate for at least 48 h prior to measurement.The dimensions of each cylindrical specimen were determined using a digital caliper directly before compression.Young's modulus was determined from the slope of the elastic part of the stress−strain curve using the Bluehill Universal software.The average value and standard deviation for each humidity level are reported based on the measurements of two or three specimens.
SEM, Rheology, Raman Spectroscopy, AFM, TGA, Surface Area, IR Spectroscopy, Contact Angle,  -Potential, Density, TEM, and Infrared (IR) Thermal Imaging: SEM images were acquired using a Hitachi TM3000 electron microscope equipped with a Bruker Quantax 70 EDS (resolution: 135 eV).The rheological properties of the dispersions were characterized with an Anton Paar Physica MCR 301 rheometer at 25 °C.The steady shear measurements were performed between 0.01 and 1000 s −1 .The storage modulus (G′) and the loss modulus (G′′) were measured over an angular frequency ranging from 0.1 to 100 rad s −1 at a constant strain of 2%.Raman spectra were collected using a LabRAM HR 800 instrument (Horiba).The AFM images were collected with a multimode-8 AFM instrument (Bruker, USA) using a PeakForce tapping mode.The scan size and rate were 2 × 2 μm 2 and 2.0 Hz, respectively.Nanofibrils (upANF A and CNF) were prepared by deposition on mica slides and then drying in the air.Thermogravimetric analysis was performed using a Discovery TGA (TA Instrument).Nitrogen adsorption-desorption measurements were performed using a Micromeritics ASAP 2020 instrument, and the surface area was calculated using the Brunauer−Emmett−Teller (BET) analysis method. [50]he samples were degassed under vacuum at 80 °C for 12 h before measurements.Foams were characterized using infrared spectroscopy (Varian 670-IR spectrometer).The diameter of the tubular pores was estimated by analyzing the SEM images of the cross-sections using ImageJ software (version 1.53e).Water contact angle measurements were performed on a 1 cm thick foam using a DSA25E drop-shape analyzer (Krüss GmbH).The pH measurements were conducted using a SevenExcellent pH meter (Mettler-Toledo International Inc., Ohio).The  -potentials of dilute dispersions of 0.01 wt% in 5 × 10 −3 m NaCl were determined using the Zetasizer Nano ZS (Malvern Panalytical, United Kingdom). [29]The reported values are the average of five measurements on three identical samples.The foam density was calculated from the sample weight divided by the sample volume, both values are measured after conditioning at 40%RH at 22 °C.Transmission electron microscopy (TEM) images were obtained using two different instruments: a JEM-2100F (JOEL Ltd., Japan) at a 200 kV acceleration voltage and a Themis Z (Thermo Fisher Scientific, USA) at a 300 kV acceleration voltage.To prepare the sample, a drop of dispersion with a concentration of 0.005 wt% was placed over a carbon-coated copper grid and air-dried.The JEM-2100F image shows the overall morphology and distribution of the fibers, while the Themis Z image reveals the finer details and features of the fiber at higher magnification.IR camera (Testo 872) was used to collect thermal images on foams 120 seconds after they were placed onto the heating source.
Statistical Analysis: The AFM images of CNF and upANF A were processed in Gwyddion 2.62 to obtain height profiles for thickness determination of CNF and upANF A .Statistical analysis of the thicknesses of CNF and upANF A nanofibers was performed by analyzing the height profiles from at least 40 fibers (samples) in the AFM images, following D.S. Devadasan et al. [51] The pore size distribution in the radial cross-section of the foams was measured from SEM images using ImageJ software, sample size was between 30 and 45.The nanofiber thicknesses and pore sizes in the radial cross-section were performed three times on each sample.One-way ANOVA testing followed by a Tukey post-hoc test was carried out on three data groups for each sample in Origin2020.In all cases, significance was defined as p ≤ 0.05 and the obtained probability (P value) was used to estimate the variance between each set of data.

Figure 1 .
Figure 1.Schematic overview of the process to prepare hybrid CNF-upANF A foams.a) Procedure to produce a stable aqueous dispersion of upANF A from Kevlar yarn.b) Preparation of CNF-upANF A foams by freeze-casting and freeze-drying aqueous dispersions of mixtures of CNF and upANF A .c) The freeze-casting mold.d) Images of CNF-only and hybrid CNF-upANF A foams.

Figure 2 .
Figure 2. Characterizations of the nanofibrils and the aqueous dispersions.a) AFM image of upANF A and; b) diameter distribution for upANF A .One-Way ANOVA analysis were performed on three data groups on 40 samples, see TableS1in the Supporting Information.c) AFM image of CNF and; d) diameter distribution for CNF.One-Way ANOVA analysis were performed on three data groups on 40 samples, see TableS2in the Supporting Information.e)  -potentials of CNF-only, upANF A , and CNF-upANF A dispersions with different upANF A contents (5, 20, and 40 wt%).f) Viscoelastic properties of CNF-only and CNF-upANF A dispersions with a total CNF and upANF A concentration of 0.5 wt%.

Figure 3 .
Figure 3. Structure and alignment of hybrid CNF-upANF A foams.SEM images of the radial cross-section of a) CNF-only, b) CNF-upANF A (5), c) CNF-upANF A (20), and d) CNF-upANF A (40) ice-templated foams.e) The orientation degrees () of the foams are determined by azimuthal integration of 2D SAXS patterns (see inset images).f) Schematic illustration of the structure and orientation of CNF and upANF A in the foam cell walls.g) X-ray diffraction patterns for CNF-only, upANF A and the CNF-upANF A (40), CNF-upANF A (20), and CNF-upANF A (5) foams.

Figure 4 .
Figure 4. Spectroscopic and thermal characterizations of upANF A , and hybrid CNF-upANF A and CNF-only ice-templated foams.a) IR transmission spectra with; b) detailed IR spectra around 1600 cm −1 ; c) Raman absorption spectra.d) Thermogravimetric weight loss and; e) differential weight loss in nitrogen as a function of temperature.

Figure 5 .
Figure 5. Water resilience and thermal conductivity of the foams.a) The water contact angle of CNF-only and CNF-upANF A (40) foams.The inset digital images show the CNF-only and CNF-upANF A (40) foams after contact angle measurement.b) Moisture uptake versus relative humidity of only, CNF-upANF A (5), CNF-upANF A (20), and CNF-upANF A (40) foams.c) Compressive mechanical properties for CNF-upANF A (40) foams at different relative humidity.

Figure 6 .
Figure 6.Thermal conductivity and heat transport.a) Radial thermal conductivity of CNF-upANF A (5), CNF-upANF A(20), CNF-upANF A(40), and CNFonly foams at 22 °C.The thermal conductivity of air and isotropic ANF foam ( ANF ).[21]The error bars represent the relative uncertainty of the thermal conductivity from each set of measurements at the relevant relative humidity (see the Experimental Section).b) Estimated solid thermal conductivity of hybrid CNF-ANF systems at 50%RH as a function of ANF content for the upANF A with an average diameter of 2.8 nm (this study) compared to ANF with diameters of 5.5 and 10 nm, respectively.c) Comparison of thermal conductivity and density of the foams presented in this work with previously reported ANF-based aerogels/foams.