Sclerotization-Inspired Aminoquinone Cross-Linking of Thermally Insulating and Moisture-Resilient Biobased Foams.

Thermally insulating foams and aerogels based on cellulose nanofibrils (CNFs) are promising alternatives to fossil-based thermal insulation materials. We demonstrate a scalable route for moisture-resilient lightweight foams that relies on sclerotization-inspired Michael-type cross-linking of amine-modified CNFs by oxidized tannic acid. The solvent-exchanged, ice-templated, and quinone-tanned cross-linked anisotropic structures were mechanically stable and could withstand evaporative drying with minimal structural change. The low-density (7.7 kg m-3) cross-linked anisotropic foams were moisture-resilient and displayed a compressive modulus of 90 kPa at 98% relative humidity (RH) and thermal conductivity values close to that of air between 20 and 80% RH at room temperature. Sclerotization-inspired cross-linking of biobased foams offers an energy-efficient and scalable route to produce sustainable and moisture-resilient lightweight materials.

Introduction

The substitution of fossil-based thermal insulation materials, such as expanded polystyrene (EPS) and polyurethane (PU) foams, with insulating foams and aerogels that are based on renewable resources is needed to reduce the carbon footprint and mitigate climate change.[1,2] Biopolymer-based aerogels can be produced from, for example, alginate, chitin, pectin, and starch,[3] but wood-derived cellulose nanofibrils (CNFs) have recently emerged as the preferred material to produce low-density aerogels and foams with a high strength.[4,5] However, most biobased materials, including nanocellulose, are hygroscopic, water uptake at humid and wet conditions can result in a very large reduction of strength, and replacement of air with water in aerogels and foams can increase the thermal conductivity.[6,7]

Superinsulating CNF-based aerogels[8−10] and foams[11−14] with nanosized pores and/or highly aligned nanofibrils have been produced by, for example, freezing, followed by supercritical drying or freeze-drying. However, freeze-drying and supercritical drying are energy- and time-intensive processes that are unsuitable for large-scale production,[15] and there is a need to develop more facile and scalable processing routes that involve water removal by evaporative drying. The capillary forces during evaporative drying can result in shrinkage and even structural collapse, and there have been several attempts to increase the strength of isotropic nanocellulose-based foams and aerogels by, for example, delayed Ca-induced gelation,[16] condensation of aminosilane,[17] covalent cross-linking of dialdehyde-CNF,[18,19] noncovalent cross-linking of CNF/alginate composites,[20] and aggregation of CNF by freeze-toughening.[21]

Nature offers examples of moisture-resilient biobased materials, including insect cuticles and squid beaks that are hardened by a process called sclerotization, also called quinone tanning. During sclerotization in organisms, the chitin nanofibril-based matrix is stiffened by the formation of covalent cross-links between protein amino groups and catechol-containing dopamine via enzymatic oxidation.[22−25] Catecholamine cross-linking of chitin or chitosan has indeed been used to produce films[26−29] and hydrogels,[30,31] but preparation of moisture-resilient nanocellulose-based foams by quinone tanning has not been reported.

Here, we describe how anisotropic, ultra-lightweight, and moisture-resilient CNF-based foams can be prepared by sclerotization-inspired Michael-type cross-linking and evaporative drying. The foams were cross-linked by the reaction between oxidized catechol-containing tannin and a branched polyethylenimine (bPEI) that was adsorbed onto the carboxylated CNF. The sclerotization-inspired processing route relies on unidirectional ice-templating of aqueous bPEI-stabilized CNF suspensions, thawing in an aqueous solution of oxidized tannin, solvent exchange to ethanol, and evaporative drying in an oven. The combination of the anisotropic porous structure of aligned CNF foam walls and sclerotization of the bPEI-containing fibrillary network with oxidized tannin resulted in a strong material with thermal conductivity close to that of air in both dry and humid conditions.

Experimental Section

Materials

Carboxylated CNFs (1.1 mmol carboxyl groups per gram pulp) were obtained from (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO)-mediated oxidation and defibrillation of softwood pulp (Domsjö, Sweden). Details on the preparation procedure are found in the Supporting Information. Tannic acid (TA, Alfa Aesar), sodium periodate (NaIO4, ≥99.8%, Sigma-Aldrich), bPEI (Mw ∼ 25,000, primary/secondary/tertiary amine = 1:1.2:0.76, Sigma-Aldrich), and 2,2-dihydroxyindane-1,3-dione (ninhydrin, ACS reagent, Sigma-Aldrich) were used as received.

Preparation and Characterization of CNF and CNF–bPEI Suspensions

Suspensions of CNF and bPEI with final concentrations of 5 mg mL–1 CNF and 0.5 mg mL–1 bPEI (dry weight ratio CNF–bPEI = 10:1) in deionized water were prepared using a high-speed disperser (Ultra-Turrax, IKA, Germany). Topographical images of dry CNF and CNF–bPEI were obtained using atomic force microscopy (AFM, dimension 3100, Bruker, USA) operated in the tapping mode. The suspensions were diluted to 0.001 wt %, deposited onto a freshly cleaved mica substrate, and dried under ambient conditions. The average diameter of the CNF was estimated from height measurement of 80 particles. Light transmittance through CNF and CNF–bPEI suspensions was measured with a Genesys 150 UV–visible spectrometer (Thermo Fisher Scientific, USA) in the range of 300–800 nm using water as the background. Rheological measurements were performed on a Physica MCR 301 rheometer (Anton Paar, Austria) using a smooth cone-on-plate geometry (2° nominal angle, 25 mm diameter) and a 50 μm gap. Amplitude sweeps were performed from 0.01 to 1000% strain at a frequency of 1 Hz and a temperature of 25 °C.

Preparation of CNF–bPEI–TA, CNF–bPEI, and CNF–FD foams

CNF–bPEI suspensions (5/0.5 mg mL–1) were degassed, 3 g was transferred into cylindrical polymer molds (diameter 15 mm, height 17 mm) equipped with copper bottom plates, and freeze-cast by placing the molds onto blocks of dry ice (−78.5 °C). The frozen freeze-cast suspensions were immersed into 70 mL of a solution of 10 mg mL–1 TA in 0.1 M pH 7 sodium phosphate buffer that had been oxidized with 5 mg mL–1 of NaIO4 immediately before immersion. The sclerotization-inspired cross-linking was performed by thawing and reacting the frozen specimens in the oxidized TA solution for 20 min. The TA concentration in the thawing solution and the soaking time were optimized toward complete impregnation of the hydrogel and maximizing the average specific toughness, Young’s modulus, and strength of the final dry foam during compression after conditioning at 98% relative humidity (RH) (Figure S1). The resulting hydrogel was subsequently transferred to water/ethanol solutions with increasing ethanol content: 20/40/60/80/99.5% (v/v) ethanol. The resulting organogels were dried by evaporation by increasing the temperature from room temperature to 105 °C at a rate of 10 °C/h and then holding at 105 °C for 11 h. The final composition of the foams was estimated by comparing the weight of dry foams from sclerotization-inspired cross-linking (CNF–bPEI–TA) to foams that were solvent-exchanged and dried directly after freeze-casting (CNF–bPEI). For comparison, foams of nonmodified CNFs were prepared by freeze-casting 5 mg mL–1 of CNF suspension as described above, followed by drying using a freeze-dryer (Christ Alpha 1-2LDplus, Germany) for 72 h (CNF–FD).

Characterization of CNF–bPEI–TA, CNF–bPEI, and CNF–FD Foams

The apparent densities of the foams were determined by measuring their weight with an analytical balance (Sartorius, Germany) and their volume with a digital caliper after conditioning for at least 48 h at 23 °C, at 50 or 98% RH. Infrared spectra of dry foams were recorded on a Varian 670-IR FTIR spectrometer (Varian, USA) equipped with an ATR detection device with a single reflection diamond element. 1H → 13C cross-polarization magic-angle spinning (CP-MAS) NMR spectra were collected on an AVANCE-III NMR spectrometer (Bruker, USA) at a magnetic field strength of 14.1 T (Larmor frequencies of 600.1 and 150.9 MHz for 1H and 13C, respectively) using 4.0 mm zirconia rotors at an MAS rate of 14.00 kHz. Acquisitions involved proton with a 90° excitation pulse of 4 μs and matched spin-lock fields that obeyed a modified Hartmann–Hahn condition νH = νC + νr (where νH and νC are the cross-polarization radiofrequencies of 1H and 13C, respectively, and νR is the sample spinning rate). A contact time of 500 μs was used, and SPINAL-64 proton decoupling was performed at 60 kHz. A total of 32,768 signal transients with 2 s relaxation delays were collected for each sample. Chemical shifts were referenced with respect to neat tetramethylsilane. Scanning electron microscopy (SEM) was performed using a tabletop TM 3000 scanning electron microscope (Hitachi, Japan) at an accelerating voltage of 5 kV on foams that had been rapidly frozen by immersion in liquid nitrogen, followed by cryo-fracturing and freeze-drying. Energy-dispersive X-ray spectroscopy (EDXS) analysis of the foams was performed using the same instrument at an accelerating voltage of 15 kV. High-resolution SEM images were recorded on a JEOL JSM-7401F (JEOL Ltd., Japan) using an accelerating voltage of 1 kV and a working distance of 8 mm. The average thickness of the CNF–bPEI–TA foam walls was estimated by analyzing the wall thickness of 44 foam wall samples by image analysis of SEM images using ImageJ. The modified ninhydrin assay[32] was performed by suspending and mixing 11 mg of the foam in 2 mL of ethanol and 1 mL of ninhydrin solution (1.5% in ethanol, w/v), heating the mixture to 80 °C for 25 min, and then allowing the mixture to separate and cool to room temperature. The supernatant was diluted 10-fold with deionized water and its absorbance at 570 nm was measured on a Genesys 150 UV–visible spectrometer (Thermo Fisher Scientific, USA) against a blank containing the same amount of ninhydrin. The amount of primary amino groups in bPEI before and after sclerotization is proportional to the optical absorbance of the solution.[33] The degree of alignment (Hermans’ orientation parameter) of the CNF in the foams was determined using X-ray diffraction and the details are found in the Supporting Information. The mechanical properties of the foams in compression were measured using an Instron 5966 universal testing machine (Instron, USA) equipped with a 100 N load cell. The cylindrical foams were compressed at a strain rate of 10% min–1. The foams were conditioned either at 50% RH and 23 °C in a humidity- and temperature-controlled room or at 98% RH over a saturated solution of potassium sulfate at 23 °C for at least 48 h prior to measurement. The dimensions and weight of each specimen were determined directly before the measurement and their apparent density was calculated. The Young’s modulus was determined from the slope of the elastic part of the stress–strain curve. The strength was determined as the stress at 10% strain. The toughness was calculated from the area under the stress–strain curve between 0 and 70% strain. The average value and standard deviation for each material are reported based on the measurements of three to seven specimens. The stability of the foams against repeated changes in RH was determined by measuring their volume before and after five cycles of conditioning for 2 h at 50 and 90% RH, respectively, in a Climacell Evo humidity chamber (MMM group, Germany). Nitrogen sorption measurements were performed using an ASAP 2020 (Micromeritics Instrument Corporation, USA). The surface area of the foams was estimated using the Brunauer–Emmett–Teller (BET) model.[34] The cumulative pore volume and the average pore diameter in the foam walls were estimated using the Barrett–Joyner–Halenda model.[35] The foams were degassed at 80 °C for at least 840 min. The water uptake of foams was estimated by determining the mass change under controlled RH and temperature using a BP 210S high-precision balance (Sartorius, Germany) inside a Climacell Evo humidity chamber (MMM group, Germany). The foams were initially conditioned at 40 °C and 20% RH before the weight was determined at 22 °C and 20/35/50/65/80% RH every 30 s for up to 6 h until steady state was reached. The thermal conductivity of the foams perpendicular to the fibrils (radial, λr) and along the fibrils (axial, λa) at 22 °C and between 5 and 80% RH was measured using the TPS 2500 S Thermal Constants Analyzer (Hot Disk AB, Sweden) in the anisotropic mode.[36] Details on the measurements are found in the Supporting Information.

Results and Discussion

Processing of Nanocellulose-Based Foams by Sclerotization-Inspired Cross-Linking

Nanocellulose-based cross-linked foams were produced by a process that mimics insect cuticle sclerotization, where enzymatically formed quinones react with amino groups on the proteins.[37] The amine-rich protein-coated chitin nanofibril matrix of insect exoskeletons was mimicked by adsorbing a positively charged polyamine (bPEI) onto negatively charged carboxylated CNFs (Figure a). A ratio of 10:1 (CNFdry–bPEIdry) was chosen as the lowest possible concentration of bPEI at which no collapse and only minimal shrinkage of the porous structure during evaporative drying were observed. The 5/0.5 mg mL–1 CNF–bPEI suspension was transferred into a freeze-casting mold, where unidirectional ice growth results in an anisotropic structure and aligns the CNF in the cell walls (Figure b). The frozen specimens were thawed in an aqueous solution of oxidized TA (Figure c), where the catechol groups on the tannin had been transformed into highly reactive quinones by chemical oxidation (Figure d).[38] The quinones react with the bPEI amines and cross-link with the hydrogels (Figure e). The cross-linked hydrogels were subsequently solvent-exchanged to ethanol and dried at an elevated temperature (Figure f), where additional amide bonds between carboxylate groups on CNF and bPEI amines are formed.[39] The shape of the lightweight (7.7 ± 0.2 kg m–3) CNF–bPEI–TA foam (Figure g) was preserved after evaporative drying, which shows that the sclerotization-inspired cross-linked hydrogel could withstand the induced capillary forces during drying.[40] SEM images of the CNF–bPEI–TA foam cross sections revealed that the macroporous honeycomb structure that formed during ice templating[41] was preserved in the oven-dried foam (Figure h). Hermans’ orientation parameters obtained from X-ray diffraction measurements showed that the degree of CNF alignment within the foam walls was not affected or disrupted by the addition of bPEI, the aminoquinone cross-linking, or the solvent exchange and evaporative drying (Table S1). High-resolution SEM images show that the foam walls in CNF–bPEI–TA foams are thin with an average thickness of 550 ± 90 nm (Figure i).

Figure 1

Processing of CNF–bPEI–TA foams: (a) a suspension of 5/0.5 mg mL–1 CNF–bPEI is transferred to a freeze-casting mold and is (b) ice-templated. (c) The frozen specimen is thawed in an aqueous solution of (d) oxidized TA where highly reactive quinones (e) form cross-links with the bPEI amine groups. (f) After solvent exchange to ethanol, the foams are dried at an elevated temperature. Representative SEM image of the porous structure of (g) CNF–bPEI–TA foam in (h) axial direction and a (i) high-magnification SEM image of a single CNF–bPEI–TA foam wall.

The addition of bPEI to an aqueous suspension of CNF results in the formation of aggregates between CNF particles and bPEI as well as clusters of excess bPEI, as shown by comparative tapping mode AFM height images (Figure a). The size of the CNF–bPEI aggregates ranges between 60 and 300 nm, while the average diameter of the CNF was 3.0 ± 0.6 nm (Figure S2). Figure b shows that the bright birefringent areas of CNF and CNF–bPEI suspensions observed between crossed polarizers are fewer and smaller in the CNF–bPEI suspension compared to the CNF suspension, which suggests that the interaction of bPEI with CNF disrupts some of the ordered CNF domains.[42] The aggregation of CNF and bPEI is also evident from the decrease of the transmission of visible light through the suspension of 5/0.5 mg mL–1 CNF–bPEI compared to the 5 mg mL–1 CNF suspension (Figure c).[43] The large decrease in the transmittance of light at smaller wavelengths suggests that the size of the aggregates is below 400 nm. The electrostatic interaction between protonated bPEI amines and deprotonated CNF carboxylates and the entropy gain caused by counterion release[44,45] could contribute to the adsorption of bPEI onto carboxylated CNFs.

Figure 2

Characterization of 5 mg mL–1 CNF and 5/0.5 mg mL–1 CNF–bPEI suspensions: (a) AFM images of dried suspensions and (b) images between crossed polarizers. (c) UV–vis transmission spectra and (d) storage (solid lines) and loss (dotted lines) moduli of CNF and CNF–bPEI suspensions at pH 11 as a function of strain.

The storage and loss moduli within the linear viscoelastic region of 5/0.5 mg mL–1 CNF–bPEI suspensions were significantly higher compared to that of neat 5 mg mL–1 CNF suspensions (Figure d). Both the CNF–bPEI and CNF suspensions displayed a viscoelastic solid-like behavior at low strains (tan δ = G″/G′ < 1) that is followed by a viscoelastic liquid-like behavior at high strains (tan δ > 1) where the gel network breaks up. The CNF–bPEI suspension displays a much lower crossover strain (strain where tan δ = 1) compared to the neat CNF suspension (20% strain for CNF–bPEI; cf. 140% strain for CNF), which suggests that the entangled and volume-spanning structurally arrested colloidal network that dominates in neat CNF suspensions[46] is replaced with a more aggregated and brittle network in CNF–bPEI suspensions. The natural pH of a 5/0.5 mg ml–1 CNF–bPEI suspension was 11, corresponding to a degree of protonation of approximately 1% of the bPEI nitrogen atoms, according to the potentiometric titration of bPEI (Figure S3). Reducing the pH of the suspension to 9 (i.e. approximately 20% degree of protonation) resulted in a more liquid-like behavior (Figure S4a), which may be related to a partial disruption of the network. Reducing the bPEI concentration to 0.25 mg mL–1 resulted in a more liquid-like behavior at low strains (Figure S4b), while increasing the bPEI concentration to 1 mg mL–1 had no significant effect on the viscoelastic properties at low strains.

Chemistry of Sclerotization-Inspired Cross-Linking

The cross-linking of amine-modified CNF by oxidized TA was investigated with a combination of UV–vis, infrared (IR), and 13C cross-polarized magic angle spinning nuclear magnetic resonance (13C CP-MAS NMR) spectroscopy. The addition of sodium periodate (NaIO4) to a TA solution results in a rapid color change of the aqueous TA from slightly brown to dark green and the appearance of UV–vis absorbance peaks at 254 and 356 nm, which are related to the formation of conjugated quinones by oxidation of the catechol groups on TA (Figure a).[47] Electrophilic o-quinones are known to undergo self-crosslinking, as well as react with nucleophiles, such as amines, in a Michael-type addition reaction.[48] The incorporation of TA into the cross-linked foam was confirmed using 13C CP-MAS NMR where two additional peaks at 138 and 146 ppm indicate TA aromatic carbons (Figure b). The NMR spectra for each foam exhibit distinct peaks for the anhydroglucose carbon atoms of cellulose as well as an additional peak at 175 ppm that can be assigned to the carboxylate carbon of TEMPO-oxidized CNF. The bPEI-containing foams show an extra broad peak between 35 and 55 ppm that corresponds to aliphatic carbons. The peak around 166 ppm in the spectra of CNF–bPEI and CNF–bPEI–TA can be assigned to amidic carbons from amid bonds that probably were formed by a reaction between bPEI amines and CNF carboxylates when the foam is heated and dehydrated.[39] The formation of amides was confirmed using FTIR spectroscopy where the absorbance of the CNF carboxylate bands at 1604 and 1409 cm–1 decreases and that of the amide I, II, and III bands increases (Figure c). The cross-linked structure can withstand structural disintegration after soaking in water at pH 7 and pH 12 (Figure S5). Ninhydrin assay analysis shows that the amount of primary bPEI amine groups in the dry foams was reduced by 45% after addition of oxidized TA (Figure d), which confirms that the nucleophilic bPEI primary amine groups react with the electrophilic TA quinones.

Figure 3

Characterization of CNF-based foams: (a) UV–vis spectra of a TA solution before and after oxidation with NaIO4 and (b) 13C CP-MAS NMR spectra of CNF–FD, CNF–bPEI, and CNF–bPEI–TA foams. (c) Normalized FTIR spectra of CNF–FD and CNF–bPEI foams and their difference curves. (d) Change in the absorbance at 570 nm of ninhydrin solutions after reaction with the bPEI-containing foams.

Mechanical Properties and Moisture Stability of Foams

The effect on the mechanical properties of sclerotization-inspired cross-linking was evaluated by determining the compressive mechanical properties of the foams in the direction of the cellular pores and aligned CNFs at 50 and 98% RH (Figure a–d and Table S2). Figure a shows the appearance and composition of CNF-based foams and that the apparent densities of the foams were all below 8 kg m–3 when measured at 50% RH. EDXS analysis shows that nitrogen is homogeneously distributed in the CNF–bPEI–TA foam cell walls (Figure S6). The anisotropic CNF–bPEI–TA foam displayed a low-strain linear-elastic deformation region, followed by a plastic deformation plateau region at larger strains and a densification region at very large strains (Figure b).[49] The lightweight (7.7 kg m–3) CNF–bPEI–TA foams displayed a Young’s modulus of 157 ± 40 kPa, a strength of 8.5 ± 0.8 kPa, and a toughness of 8.2 ± 0.5 kJ m–3 at 50% RH.

Figure 4

Moisture resilience and mechanical properties of foams: (a) comparison of the composition, density, and volumetric shrinkage of CNF-based foams at 50 and 98% RH. (b) Typical stress–strain curve of a CNF–bPEI–TA foam during compression in the axial direction, indicating the elastic and plastic regions and how Young’s modulus and strength were determined. (c) Specific Young’s moduli of CNF–bPEI and CNF–bPEI–TA foams at 50 and 98% RH, respectively. (d) Young’s modulus, strength, and toughness of CNF–bPEI–TA foams at 50 and 98% RH and after reducing the RH back to 50%.

Figure c shows that the specific Young’s modulus (i.e., Young’s modulus divided by density) of 20.4 kNm kg–1 at 50% RH in the axial direction for CNF–bPEI–TA foams is significantly higher compared to that for CNF–bPEI foams (12.8 kNm kg–1), which can be related to a significant stiffening of the foam cell walls by the sclerotization-inspired cross-linking. The specific Young’s modulus of the evaporatively dried CNF–bPEI–TA foams is comparable to that of silica aerogels (5–20 kNm kg–1)[50] and higher than that of nanocellulose-based freeze-dried anisotropic CNF/clay foams (12.6 kNm kg–1)[51] or isotropic evaporative-dried dialdehyde-CNF (∼8–14 kNm kg–1),[19] CNF/alginate (∼4 kNm kg–1),[20] and freeze-toughened CNF foams (∼1–2 kNm kg–1).[21]

Sclerotization-inspired cross-linking also improved the strength (8.5 kPa for CNF–bPEI–TA, cf. 5.8 kPa for CNF–bPEI) and the toughness (8.2 kJ m–3 for CNF–bPEI–TA, cf. 7.2 kJ m–3 for CNF–bPEI) of the CNF-based foams.

We have also investigated the mechanical properties of the foams at 98% RH (i.e., in almost water-saturated air) after conditioning each foam for at least 48 h over a saturated aqueous solution of potassium sulfate. The cross-linked CNF–bPEI–TA foams showed only minor isotropic compaction (11 ± 3 vol % shrinkage) after conditioning at 98% RH, resulting in an apparent density of 10.1 ± 0.4 kg m–3 and a porosity of over 99%, whereas conditioning freeze-dried CNF–FD foams at 98% RH led to partial structural collapse (Figure a). The CNF–bPEI–TA foams displayed a specific Young’s modulus of 8.9 kNm kg–1 at 98% RH, which is almost 3 times higher than that of CNF–bPEI foams (3.1 kNm kg–1). The decrease in Young’s modulus of CNF–bPEI–TA foams when changing the RH from 50 to 98% is comparable to previous reports on boric acid cross-linked CNF-based foams that exhibited a 43% decrease in Young’s modulus when changing the RH from 50 to 85%.[12] Compression tests on foams where the RH was first increased to 98% RH and then reposed back to 50% RH showed a 12% decrease in Young’s modulus and a 10 and 12% increase in strength and toughness, respectively (Figure d). We speculate that these changes are related to a release of stress in the interfibril joints upon swelling, similar to pulp fibers,[52] that leads to a slightly less aligned, but stronger bonding inside the foam cell walls upon drying. Subjecting the foams to five RH cycles between 50 and 90% RH in a humidity chamber resulted in a minor volumetric shrinkage of the CNF–bPEI–TA foams (6.4 ± 0.05%) compared to a significant contraction of the CNF–FD foams (43.3 ± 0.55%). The excellent recovery of the mechanical performance after exposure to high RH and the stability against repeated RH changes suggests that the sclerotization-inspired cross-linking minimizes moisture-induced structural degradation.

Thermal Transport Properties of Foams

The thermal conductivity of the anisotropic CNF–FD and CNF–bPEI–TA hybrid foams has been determined as a function of RH (5–80%) both perpendicularly to the fibrils (radial, λr) (Figure a and Table S3), and along the fibril direction (axial, λa) (Figure S7). λr is always lower than λa of the foams because of the intrinsic anisotropy of the foam structure and the alignment of the CNF in the foam cell walls.[53,54]Figure a shows that the λr of cross-linked and evaporatively dried CNF–bPEI–TA foams (25.6–33.8 mW m–1 K–1) is close to or even slightly lower than that of air (26 mW m–1 K–1 at 22 °C)[55] between 20 and 80% RH. In general, the thermal conductivity of porous materials with small pore sizes (<1 mm) depends predominantly on the solid and gas conduction contributions.[56] The mesopores in the CNF–bPEI–TA foams (Table S3) are expected to result in a low gas contribution because of the Knudsen effect[57] and phonon scattering could reduce the solid contribution to the thermal conductivity.[58]

Figure 5

Thermal transport properties of CNF–FD and CNF–bPEI–TA foams: (a) radial thermal conductivity as a function of RH. The dashed red line indicates the thermal conductivity of air (26 mW m–1 K–1 at 22 °C). (b) Radial thermal conductivity of the CNF–FD and CNF–bPEI–TA foams compared to the isotropic thermal conductivity of traditional insulation materials.[59,60]

The radial thermal conductivity of the cross-linked CNF–bPEI–TA foams is relatively insensitive to changes in the RH within the RH range of 20–80%. Replacement of air with water normally results in an increase in the thermal conductivity, which suggests that the smaller increase of λr of the CNF–bPEI–TA foams (Figure a) compared to the CNF–FD foams (Table S3) with an increase of RH from 50 to 80% may be related to a lower water uptake of the cross-linked material (Figure S8).

When comparing the λr of the CNF-based foams to traditional thermal building insulation materials, the thermal insulation performance of the cross-linked and evaporatively dried CNF–bPEI–TA foams (26 mW m–1 K–1) is significantly lower than traditional cellulose-based insulating materials (40–50 mW m–1 K–1) and fossil-based EPS, extruded polystyrene (XPS), and mineral wool (all 30–40 mW m–1 K–1) and similar to high-performance PU foams (20–30 mW m–1 K–1) (Figure b).[59] It is clear that the low thermal conductivity, especially at high RH, and the ability to process the materials using evaporative drying make the sclerotization-inspired cross-linked CNF–bPEI–TA foams of interest as high-performance biobased insulation materials.

Conclusions

We have developed a scalable and sustainable route, inspired by insect cuticle sclerotization chemistry, to prepare moisture-resilient, strong, and thermally insulating biobased foams. Cross-linking a carboxylated CNF/polyamine bPEI ice-templated foam by a Michael-type addition reaction between oxidized tannin and bPEI amine groups and additional amide bonds that form upon dehydration strengthened the structure and made it possible to produce lightweight foams by evaporative drying with minimal structural shrinkage after solvent exchange to ethanol. The resulting anisotropic and lightweight foam had a high specific Young’s modulus in the axial direction at 50% RH (20.4 kNm kg–1), retained much of its mechanical strength at 98% RH, and restored the mechanical performance once reposed back to 50 from 98% RH. The radial thermal conductivity of the anisotropic foams was close to that of air over a wide range of RHs (25.6–27.7 mW m–1 K–1 between 20 and 80% RH) and significantly lower than that of fossil-based insulation materials, such as conventional EPS/XPS insulation. The high strength and low thermal conductivity of the biobased foams at both low and high humidity combined with the ability to produce low-density materials by a bio-inspired, cross-linking, and evaporation-based processing route is of potential interest for thermal insulating and packaging applications.