Hybrid Foams based on Multi‐Walled Carbon Nanotubes and Cellulose Nanocrystals for Anisotropic Electromagnetic Shielding and Heat Transport

Lightweight and mechanically robust hybrid foams based on cellulose nanocrystals (CNC) and multi‐walled carbon nanotubes (MWCNT) with an anisotropic structure are prepared by directional ice‐templating. The anisotropic hybrid CNC‐MWCNT foams displayed a combination of highly anisotropic thermal conductivity and an orientation‐dependent electromagnetic interference (EMI) shielding with a maximum EMI shielding efficiency (EMI‐SE) of 41–48 dB between 8 and 12 GHz for the hybrid foam with 22 wt% MWCNT. The EMI‐SE is dominated by absorption (SEA) which is important for microwave absorber applications. Modelling of the low radial thermal conductivity highlighted the importance of phonon scattering at the heterogeneous CNC‐MWCNT interfaces while the axial thermal conductivity is dominated by the solid conduction along the aligned rod‐like particles. The lightweight CNC‐MWCNT foams combination of an anisotropic thermal conductivity and EMI shielding efficiency is unusual and can be useful for directional heat transport and EMI shielding.

DOI: 10.1002/admi.202300996[3] Biomaterials commonly display several functions, combining, e.g., load-bearing capability with thermal resistance, as well as protection against microbes or biodegradability. [4]The multifunctionality of engineered materials is often controlled by phenomena at multiple length scales and may be tailored through careful control of the hierarchical structure and composition. [5,2,6]9] The EMI-S is especially important in the X-band (8-12 GHz)  as well as in the K u -band (12-18 GHz).EMI attenuation can be achieved by interaction with either the electric or the magnetic field of the electromagnetic (EM) wave, which leads to dielectric or magnetic loss, respectively.The shielding efficiency is determined by a combination of EMI reflection, absorption and multiple reflections and depends on the materials structure and interfaces as well as the conduction and polarization properties of its components. [10]Well dispersed nanomaterials such as carbon-based fillers like carbon fibers, [11][12][13] graphene, [14][15][16] reduced graphene oxide [17] and carbon nanotubes, [18][19][20][21] carbonized bio-based fibers, [22] metal nanoparticles [23][24][25] or transition metal carbides and/or nitrides (MXenes) [26][27][28][29][30][31] have been shown to result in large dielectric or magnetic loss and can thus efficiently transform microwave radiation into thermal energy.
Carbon nanotubes (CNT) are promising nanofillers for multifunctional materials with direction-dependent heat transport and EMI-S properties.CNT display a highly anisotropic thermal conductivity. [32,33]The interaction of CNT with an EM wave is dominated by dielectric loss mainly caused by two effects 1) conductivity loss due to the motion of -electrons in the CNT structure and 2) interfacial polarization loss caused by the nonuniform distribution of charge carriers between CNT and the matrix it is embedded into.It can be challenging to incorporate CNT homogeneously in a matrix [34] but cellulose nanocrystals (CNC) have been found to act as a dispersant for CNT in aqueous media. [35][41][42] Anisotropic foams based on CNC and the longer and less crystalline cellulose nanofibrils (CNF) have been shown to display a significantly higher thermal conductivity along the freezing direction than perpendicular to it [43][44][45] with anisotropy values up to 6 for CNC foams. [46]Aqueous suspensions of CNC are far less susceptible to gelling compared to CNF, which allows a wide range of foam densities. [45]Composite aerogels based on CNF and multiwalled CNT (MWCNT) together with various additives have been shown to display a high shielding efficiency (SE). [47,48]MI-S materials often require the generated heat to be managed and excessive heat gradients to be avoided.While studies on multifunctional materials investigating both heat transport and EMI-S properties are emerging, [2,22,49,50] studies on light-weight materials that combine direction-dependent thermal conductivity with a high EMI-SE are sparse. [51]Here, we have prepared multifunctional and anisotropic hybrid MWCNT-CNC foams with up to 29 wt% MWCNT via directional icetemplating.The columnar macropores were aligned in the freezing direction and the lightweight foams displayed an orientation dependent shielding efficiency with a total electromagnetic shielding efficiency (SE T ) of 41-48 dB in the X-band (8-12 GHz) for a foam with 22 wt% MWCNT.Shielding efficiency via absorption (SE A ) strongly dominated the shielding efficiency with a SE T /SE A of 0.98-0.99.The thermal conductivity of the foam was highly anisotropic, and a serial-resistor model showed that the low thermal conductivity in the radial direction can be attributed to phonon scattering.The anisotropic thermal conductivity and EMI-SE of the lightweight CNC-MWCNT foams could be useful for directional heat transport and EMI-S.

Preparation and Structure of the Foams
Anisotropic hybrid foams of MWCNT and CNC were prepared from aqueous dispersions via directional ice-templating and subsequent freeze-drying (Figure 1).The rod-like CNC have a diameter of 3.6 nm and an aspect ratio of ≈40 while the MWCNT have a diameter of 12 nm, as determined by AFM (Figures S1 and S2, Supporting Information).Well-dispersed mixtures of CNC and MWCNT were prepared by sonication and removal of agglomerates by centrifugation prior to ice templating.Carbon nanotubes (CNT) are known to aggregate in water, but CNC have shown to disperse both single-and multi-wall CNT efficiently. [35]he MWCNT content in the resulting hybrid foams (Table 1) was controlled by the composition of the dispersion and determined by thermogravimetric analysis using a calibration curve obtained from materials with a known CNC/MWCNT ratio (Figure S3, Supporting Information), which was supported by elemental mapping (Table S4, Supporting Information).The MWCNT content of the foams was: 0 wt%, 13 wt%, 22 wt%, and 29 wt% (denoted as CNC-100, CNC-MWCNT_13, CNC-MWCNT_22 an CNC-MWCNT_29, respectively).The density of the foams varied slightly between 30.4 and 31.7 kg m −3 , which suggests that the amount of material removed by the centrifugation step was small.The cumulative volume of pores between 1.7 and 300.0 nm determined from the adsorption branch of the N 2 adsorption isotherm following the Berrett-Joyne-Halenda (BJH) method, here termed "nanoporosity", was about a factor of 10 higher for CNC-MWCNT_29 compared to the hybrid foams with lower MWCNT-content (Figure S4, Supporting Information).Indeed, the surface area was also significantly higher for CNC-MWCNT_29.
Scanning electron microscopy (SEM) images of the hybrid foams perpendicular to the ice growth direction, henceforth called the radial direction, show that the hybrid foams with a MWCNT content of up to 22 wt% displayed columnar macropores, similar to CNC foams prepared by unidirectional icetemplating [53] (Figure 2a-c).For CNC-MWCNT_29 however, the macropores are poorly defined, indicating that unidirectional ice growth was hindered (Figure 2d).The macropores were well aligned in the ice growth direction (axial direction), for CNC-100, CNC-MWCNT_13 and CNC-MWCNT_22, while the pore morphology appears to be much more random for CNC-MWCNT_29 (Figure 2e-h), which corroborate that the directional ice templating process is disturbed when the MWCNT concentration becomes too high (above 22 wt%).
The alignment of the CNC in the foams was investigated via wide angle X-ray scattering (WAXS) deriving the Hermans' orientation parameter which ranges between 0 for fully isotropic and 1 for fully aligned orientation.The Hermans' orientation parameter P̅ 2 that was extracted from the azimuthal integration of the (1 10) peak of CNC in the 2D WAXS patterns of the ice template foams decreased from 0.62 for CNC-100 to 0.38 for CNC-MWCNT_29 (Figure 2i-l; Figure S5, Supporting Information).The addition of MWCNT disturbed the alignment of the CNC during ice templating, also at relatively modest MWCNT concentrations (13 wt% and 22 wt%) where the macroscopic pore structure was relatively unaffected.The alignment of MWCNT could not be determined due to an overlay of the scattering peaks and a much lower crystallinity of the MWCNT in comparison to CNC (Figure S6, Supporting Information).

Electromagnetic Interference Shielding Performance and Electrical Properties
The electromagnetic interference shielding efficiency of the foams has been evaluated as a function of orientation and frequency (Figure 3).Thin slices of ≈9 mm were cut along the axial direction of CNC-MWCNT_22,CNC-MWCNT_13, and CNC-100 foams, and their EMI-SE was tested along the radial direction with different rotation orientation angles , which indicates the angle between the electric field of the electromagnetic (EM) wave and the oriented macropore channels (Figure 3a).The EMI-SE for CNC-100 was as expected very low and below the detection limit (Figure S7a, Supporting Information) but CNC-MWCNT_13 and CNC-MWCNT_22 showed a maximal total shielding efficiency (SE T ) averaging 28 dB (23-33 dB) and 44 dB (41-48 dB), respectively for the X-band range of 8-12 GHz (Figure 3b), making it suitable for wideband applications.
The SE T of CNC-MWCNT_22 for  = 0°was significantly higher than for  = 90°, averaging at 44 dB and 27 dB, respectively, while EMI measurements at an angle of 45°and 135°r esulted in an intermediate SE T value of 31 dB (Figure 3c).Similar trends were observed for CNC-MWCNT_13, (Figure S7b, Weight percent of MWCNT in the foam determined by thermogravimetric analysis, for further explanations see Figure S3 (Supporting Information); b) Bulk density, calculated from the dimensions and mass of minimum six foams; c) From the adsorption branch of the N 2 adsorption isotherm, see Figure S4 (Supporting Information); S BET for Brunauer-Emmett-Teller [ 52] surface area, calculated over P/P 0 = 0.24.

Figure 2.
Foam alignment: SEM images of all four types of foam with increasing MWCNT-content.The directionally frozen foams were cut perpendicular and parallel to the freezing direction and then freeze-dried, resulting in radial a-d) and axial e-h) cuts.i-l) Azimuthal integration of (1 10) peak of CNC in the 2D WAXS pattern with the corresponding Legendre series expansion fit from which the Hermans' parameter was extracted.For further explanations, see the "Methods" section.

Supporting Information). Our results corresponds well to previous work on anisotropic CNF-based aerogels and foams
containing MXenes [30] or MWCNT, [48] exhibiting a 40%-50% lower SE for the transversal ( = 90°) compared to the sagittal ( = 0°) orientation.The orientation dependency can be related to the opposing internal electric field that is generated upon exposure to the external electric field of the electromagnetic wave.Previous studies revealed that a stronger internal electric field is created when the external electric field of the electromagnetic wave is aligned parallel to columnar macropores (sagittal orientation; 0°) in comparison to a perpendicular alignment (transversal orientation; 90°). [48]A stronger internal electric field can weaken the external electromagnetic wave more efficiently and result in a higher shielding efficiency.Comparison with previous work also suggests that the magnitude of the EMI-SE is directly related to the MWCNT content and density of the foam or aerogel. [47,48]easurements at higher frequencies of 12-18 GHz (K u -band) on CNC-MWCNT_22 (Figure 3d) showed an average SE T of 41 dB, 26 dB and 30 dB for 0°, 90°and 45°/135°, respectively, while CNC-MWCNT_13 (Figure S7c, Supporting Information) showed higher SE T in the 12-18 GHz range than in the 8-12 GHz range.The total shielding efficiency SE T combines the shielding via absorption (SE A ) and reflection (SE R ).Microwave absorbers should preferably display a high SE A and low SE R , which indeed was found for both CNC-MWCNT_22 (Figure 3e; Figure S7d, Supporting Information) and CNC-MWCNT_13 (Figure S7e, Supporting Information) with a shielding via absorption to total shielding efficiency ratio of 0.98 and 0.99 respectively.We speculate that dissipation of the electromagnetic radiation in the cellular microstructure may contribute to the strong domination of absorption of the shielding efficiency.
The electrical properties along the axial direction of the foams were assessed at ambient conditions using a relatively simple set-up, [54] as illustrated in Figure S8 (Supporting Information).CNC-100 showed a resistivity too high to be measured, which given the high porosity of the foam and the low electrical conductivity of cellulose, [55] is to be expected.The CNC-MWCNT hybrid foams displayed an electrical conductivity on the order of 10 −3 , 10 −4 and 10 −5 S m −1 for CNC-MWCNT_29, CNC-MWCNT_22, and CNC-MWCNT_13, respectively.The electrical conductivities of the foams investigated in this study are in the same range as reported for CNC-MWCNT composite foams [54] but lower than when single-or few walled carbon nanotubes were incorporated in nanocellulose foams and aerogels [56][57][58] or when the MWCNT content was much higher [48] or the foam density was higher. [47]

Anisotropic Heat Transport
The thermal conductivity in the radial and axial direction of the prepared foams was determined as a function of relative humidity (Figure 4a,b).The thermal conductivity in the radial direction (Figure 4a) is 4 to 7 times lower than the thermal conductivity in the axial direction (Figure 4b), which is consistent with previous studies on anisotropic nanocellulose-based hybrid foams. [59,60]e thermal conductivity in the axial direction (Figure 4b) increased with increasing relative humidity (RH) while the thermal conductivity in the radial direction (Figure 4a) displayed a U-shaped RH-dependence.Previous studies on hygroscopic CNC and CNF-based anisotropic foams [53,61] found a similar RH-dependence of the radial thermal conductivity, which was related to a competition between a humidity-induced swelling that increased phonon scattering at low and intermediate RH, and a replacement of air with water at higher relative humidity.The incorporation of MWCNT in the foam showed insignificant effects on the water uptake with increasing relative humidity (Figure 4c), which suggests that the humidity-dependent mechanisms in the CNC-MWCNT hybrid foams are similar to pure CNC foams within the studied composition range.The radial thermal conductivities of the hybrid foams increased with increasing MWCNT content but it should be noted that the radial thermal conductivities remained similar or lower than the values for common insulation materials. [62]While no other study investigating the thermal conductivity of hybrid foams of CNC and MWCNT could be found, Li et al. [58] reported a thermal conductivity of 30-80 mW m −1 K −1 for CNF-SWCNT foams with a share of 17-58 wt% SWCNT at non-defined temperature and humidity conditions.
The parallel resistor model with volume fraction-weighted sums of the solid and gas contributions to the heat transport were recently shown to be able to estimate the thermal conductivity of CNC-based foams in the radial direction in good agreement with experimental results. [53]The model includes the water uptake, the Knudsen effect and phonon scattering as shown in Equation 1: with Φ mp , Φ np , Φ solid and Φ H2O representing the volumetric shares, and  mp ,  np ,  solid and  H2O representing the thermal conductivities of the macropores (mp), nanopores (np), CNC (solid) and water (H 2 O), respectively.The thermal conductivity of the solid contribution including phonon scattering ( solid ) was assessed following [63] : with  representing the thermal conductivity of an individual particle (CNC) along the radial direction, d as diameter of the particle and R k representing the interfacial thermal resistance (Kapitza resistance).To estimate the thermal conductivity of the hybrid foams with the two solid components CNC and MWCNT, the latter equation was transformed to   [64] and 3000 W m −1 K −1 , [65] respectively and d CNC and d MWCNT represent the diameter of CNC and MWCNT, which were 3.6 and 12 nm (Figures S1 and S2, Supporting Information), respectively.The interfacial thermal resistance R k at the CNC-CNC interface and the CNC-MWCNT interface was assessed using a geometric mean approach [64] following: for the CNC-CNC interface, and for the CNC-MWCNT interface.As illustrated in Figure 4d, d and  represent the diameter and thermal conductivity of the nanoparticles, respectively, while d t and  t give the total length and thermal conductivity of a system consisting of two nanoparticles that are aligned to each other with a gap (g) that is set to 0.37 nm based on previous estimates for CNC. [53]For  tCNC a previously estimated value of 0.27 W m −1 K −1 [64] was used while the mean value between  tCNC and  tCNT (0.55 W m −1 K −1 [66] ) was chosen for  tCNC-MWCNT (0.41 W m −1 K −1 ).The relatively good correlation between the theoretical estimates and the experimental data for the radial thermal conductivities at 50% RH (Figure 4e) suggests that the relatively simple parallel resistor model is appropriate to estimate the heat transport in the CNC-MWCNT hybrid foams.Although MWCNT have a far higher thermal conductivity (≈4000 × ) than CNC, the composite CNC-MWCNT foams only displayed a slightly higher thermal conductivity in the radial direction than the pure CNC foams due to the high interfacial thermal resistance between CNC and MWCNT.The model slightly overestimates the thermal conductivity of the pure CNC foam (with ≈2-3 mW m −1 K −1 ).The theoretical estimates suggest that CNC-MWCNT composites with even lower radial thermal conductivities could be within reach if the interfacial resistance could be optimized by, e.g., achieving a very high degree of alignment of CNC and MWCNT also at high MWCNT contents resulting in increased anisotropy.
The axial thermal conductivity was found to be much higher than the radial thermal conductivity, ranging between 110-200 mW m −1 K −1 for CNC-100 which is in very good agreement with the values found for anisotropic CNC-based foams with similar densities. [53]The ice crystal growth-induced formation of the columnar macropores and the alignment of CNC within the foam walls [42,46] is the prerequisite for this anisotropy.The continuous foam wall network along the axial direction can transfer the heat less hindered than the pore-rich radial direction.On the nanoparticle scale, the heat transfer along the covalent bonds in the direction of the long axis (axial direction) of the aligned CNC is higher than along the weaker bonds such as hydrogen bonds or van der Waals forces in the transversal (radial) direction. [67]Indeed, modelling of individual CNCs suggest a thermal conductivity of 5700 mW m −1 K −1 along the axial direction and 720 mW m −1 K −1 along the radial direction. [68]Additionally, the particle alignment entails a lower density of internal interfaces along the axial direction of the foam compared to the radial, resulting in less incidents of phonon scattering. [69]The addition of carbon nanotubes increased the axial thermal conductivity but had little effect on the anisotropy.EMI shielding via absorption entails the conversion of energy from the electromagnetic wave into thermal energy since the dipoles created upon the external electric field of the electromagnetic wave collide upon the attempt to align with the oscillating field. [70]SE A is dominating the EMI-S of the CNC-MWCNT composites, and the generated heat is expected to favorably dissipate along the axial direction due to the higher thermal conductivity along that direction.

Mechanical Properties
The compression mechanical properties along the axial direction for both CNC-100 and the hybrid CNC-MWCNT foams with 13 wt% and 22 wt% MWCNT are characterized by a linear region at low strains, followed by a stress maxima and an extended plastic yield region at higher strains (Figure 5a).The CNC-MWCNT_29 was very weak and did not display a measurable stress maximum.The Young's modulus and the stress maxima of the foams decreased with increased MWCNT content (Table 2).
Carbon nanotubes are very strong and stiff and have been used to reinforce polymer composites [71][72][73] and foams. [74,75]Studies on the compression behavior of nanocellulose-carbon nanotube composite foams and aerogels are sparse and the specific Young's Modulus presented in them differ strongly.The specific Young's modulus of CNC-CNT Pickering foams with either MWCNT or SWCNT [76] was very low and ranged between 0.4 and 2 kN m kg −1 .Zeng et al [48] presented a CNF-MWCNT foam with a very high MWCNT content of 67% that revealed a specific Young's modulus of 5 kN m kg −1 .Zhu et al [47] investigated how the MWCNT content in the range of 2 wt%-30 wt% influenced the mechanical properties for a composite based on CNF, cationic CNC, sodium alginate and MWCNT.The specific Young's modulus ranged between 218-250 kN m kg −1 with the highest value obtained for a MWCNT concentration of 5 wt%.Li et al [58] and Xu et al [77] presented mechanically robust CNF-CNT composite aerogels and foams with a maximum CNT concentration of 58 wt% and 70 wt%, but the Young's modulus was not reported.The wide range of mechanical properties reported in previous studies suggests that the preparation method, characteristics of the nanocellulose and CNT and other additives all have an influence and makes a direct comparison with our work difficult.The CNC-MWCNT_13 and CNC-MWCNT_22 foams investigated in this study displayed a higher Young's modulus that is similar or larger than most previously reported nanocellulose-carbon nanotube composites.Transmission electron microscopy showed that the MWCNT in the foam walls are well distributed in the CNCmatrix, but no directionality could be identified (Figure 5b).The misalignment of the MWCNT supports the conjecture that is could be possible to increase the anisotropy of the thermal conductivity and possibly also increase the Young's modulus if would be possible to enhance the alignment of MWCNT.It should be noted that the contrast of the MWCNT is larger than for CNC and it was difficult to identify individual CNC in the foam wall section.

Conclusion
Anisotropic, multifunctional and strong CNC-MWCNT hybrid foams with a density of 30-32 kg m −3 and a MWCNT content of up to 29 wt% were prepared by ice-templating.The preparation of the lightweight foams was facilitated by the ability of CNC to disperse the thin and fibrous MWCNT in aqueous media.SEM showed that the columnar macropores were well aligned in the ice growth direction (axial direction) for the hybrid foams up to a MWCNT content of 22 wt% and WAXS showed that the degree of alignment of CNC in the foam walls remained relatively high but decreased slightly with increasing MWCNT content.TEM showed that the MWCNT in the foam walls are well dispersed but randomly oriented.The anisotropic structure of the ice-templated MWCNT-CNC foams with the columnar macropores and the nanosized CNC aligned in the freezing direction is reflected in the anisotropic electromagnetic shielding and heat transport.The heat transport is the largest in the axial direction (along the aligned CNC particles and columnar macropores) and the shielding efficiency reached its maximum along the sagittal orientation (electric field along the aligned CNC particles and columnar macropores).
The CNC-MWCNT anisotropic hybrid foams displayed a maximum total electromagnetic interference shielding efficiency of 41-48 dB between 8 and 12 GHz for the hybrid foam with 22 wt% MWCNT, which is significantly higher than commercial conductive carbon foams of similar thickness based on polyurethane ("EA-LF500"; Leader Tech Inc, USA; Figure S7f (Supporting Information), 9.1 mm max 28 dB for 12-18 GHz).The EMI-SE of the hybrid foams increased with increasing MWCNT content and was shown to be strongly dominated by absorption (SE A ), which suggests that the CNC-MWCNT anisotropic hybrid foams could be used as directional microwave absorbers.The EMI-SE of the CNC-MWCNT anisotropic hybrid foams was orientation dependent and was ≈15 dB higher when subjected to the sagittal orientation of the external field compared to transversal orientation, to the columnar macropores.
The thermal conductivity of the anisotropic CNC-MWCNT foams was highly anisotropic with the axial thermal conductivity being 4-7 times higher than the radial thermal conductivity.The radial thermal conductivity at room temperature and 50% relative humidity increased slightly from 29 to 31 mW m −1 K −1 with increasing MWCNT content.The radial thermal conductivity could be modelled with reasonable accuracy using a parallel resistor model, which highlighted the importance of phonon scattering at the heterogeneous CNC-MWCNT interfaces.
The CNC-MWCNT foams investigated in this study showed a higher specific shielding efficiency (SSE) than other lightweight ( ≤ 350 kg m −3 ) foams and aerogels presented in literature.Also, the normalized specific shielding efficiency (nSSE) was higher than that of foams and aerogels made of polyurethane/resin/cenospheres, [78] carbon, [79] carbonized bio-based fibers, [22] carbon-reduced graphene oxide, [80,81] CNTacrylic copolymer [82,83] but lower than a CNT-polylactic acid network [84] and a carbon-graphene aerogel. [51]The axial thermal conductivity of the CNC-MWCNT foams was higher or in a similar range than the above-mentioned materials except for a carbonreduced graphene oxide foam with a thermal conductivity of 1200 mW m −1 K −1 [80] (Figure S9, Supporting Information).The lightweight CNC-MWCNT foams combination of an anisotropic thermal conductivity and orientation-dependent electromagnetic shielding efficiency with a high shielding via absorption is unusual and could enable directional heat transport and EM shielding devices.

Experimental Section
Materials: CNC powder was purchased from Celluforce (Canada) and MWCNT with an outside diameter of 8 nm were purchased from Abcr (Germany) and used as received.The polyurethane foam with carbon Preparation of CNC-MWCNT Foams: A dispersion of CNC and MWCNT was prepared by first mixing a stock solution of CNC, MWCNT powder and DI water via stirring for 20 min.In a next step the dispersion was homogenized by sonication (Q500, Qsonica, USA with 30 min efficient time, 2 s/2 s pulse, 75% amplitude using a 12 mm tip) in an ice bath with subsequent centrifugation (Sorvall LYNX 6000, Thermo Scientific, USA) at 20 000 g for 1.5 h.The supernatant was then poured in cylindrical polymer moulds with a copper bottom which were directionally ice-templated on a block of solid CO 2 (dry ice).Some frozen foams were cut along the axial or radial direction to receive samples for SEM imaging.The frozen foams and foam pieces were freeze-dried (Christ Alpha 1-2 LD plus, Germany) at low pressure (≈0.035 mbar) for at least 48 h.
Atomic Force Microscopy: Atomic force microscopy (AFM) images were recorded using a Multimode-8 AFM (Bruker, USA).The imaging was conducted in peak-force tapping mode, using the manufacturer's ScanAsyst™ automatic optimization algorithm.The samples were prepared by the deposition of a 0.002 wt% CNC-MWCNT dispersion onto the surface of a freshly cleaved mica with subsequent drying at ambient conditions.The software Gwyddion was used to carry out the particle analysis (min 3 × 40 particles each) and image processing.The software Image J was used for further image processing.
Thermogravimetric Analysis: Thermogravimetric analysis was performed on a Discovery TGA (TA Instrument, USA) under N 2 flow with a heating rate of 10 K min −1 up to 900 °C with 3 samples each N 2 sorption: Measurements investigating the N 2 sorption of the foams were performed using ASAP 2020 (Micromeritics Instrument Corporation, USA).The foams were degassed at 70 °C for 14 h.The adsorption and desorption cumulative volume of nanopores (for diameters between 1.7 and 300 nm) and average nanopore diameters were estimated using the Barrett-Joyner-Halenda (BJH) model.
Porosity: The porosity of the foams was determined following Equation 6 with a minimum of six samples each.The skeletal density was estimated from a weighted average of the densities of the components using the following density values for: CNC = 1580 kg m −3 , CNT = 1740 kg m −3 .
Scanning Electron Microscopy: SEM images were acquired at an acceleration voltage of 15 kV in analysis mode on a TM3000 (Hitachi, Japan) electron microscope equipped with Quantax 70 EDS (Bruker, USA) (resolution: 135 eV).Image processing was carried out using the software Im-ageJ.
Powder X-Ray Diffraction: Powder X-ray diffraction (PXRD) data of both samples (CNC and MWCNT) were collected on a Bruker D8 DIS-COVER diffractometer (CuK radiation, 1 = 1.54060Å, 2 = 1.54439Å) set up in a Bragg-Brentano geometry and equipped with a LYNXEYE XE-T detector.Samples were prepared by evaporation-drying aqueous dispersions on zero-background SI plates.Wide angle X-ray scattering: WAXS was performed on the CNC-only and CNC-CNT hybrid foams with a diameter of 14.3 ± 0.2 mm using a Mat:Nordic instrument (Xenocs) with a high brilliance microfocus Cu-radiation source (Rigaku 003+).The scattering signal was recorded by a PilatuS300K detector with a total exposure time of 600 s.All measurements were made at room temperature in vacuum condition.The signal of the 1-10 peak was azimuthally integrated, corrected for variations in detector efficiency and for spatial distortions.To obtain the Hermans' orientation parameter P̅ 2 the azimuthal integration was fitted with a Legendre series expansion (Equation 8).Subsequently fitting coefficient a1 was normalized to fitting coefficient a0 and ultimately P̅ 2 was received from Equation 9.
Electromagnetic Interference Shielding Efficiency: The EMI-SE was measured using the FieldFox Microwave Analyzer (Keysight Technologies, USA), recording 10 001 points at an IF bandwidth of 10 kHz and 25% smoothing.The specimens with a size of 25.0 mm × 25.0 mm × 8.7 mm (length × width × thickness) were placed in a customized sample holder which was sandwiched between the two halves of the measurement chambers (Figure S9, Supporting Information).The dimensions of the waveguide chamber opening were 22.9 mm × 10.2 mm (length × width) for the 8-12 GHz measurements and 15.7 mm × 7.9 mm (length × width) for the 12-18 GHz measurements.Three specimens were tested for each type of foam.The recorded S-parameters were used to calculate SE T , SE R , and SE A .
Electrical Conductivity: An Interface1010 potentiostat (Gamry instruments, USA) was used in Linear Sweep Voltammetry (LSV) mode to determine the electrical conductivity of the foams along the axial direction.Cylindrical foam samples were sandwiched between two copper plates that acted as electrodes and pressure (1649 N m −2 ± 25 for CNC-100, CNC-MWCNT_13 and CNC-MWCNT_22; 593 N m −2 ± 4 for CNC-MWCNT_29) was applied by a free weight (Figure S8, Supporting Information).The electrical conductance was obtained from the slope of the ohmic region of the LSV-plot in the region of 0-1 μA.The electrical conductivity was calculated taking the height and surface area of the foam into consideration.
Thermal Conductivity: The thermal conductivity of the foams was determined using the TPS2500 S Hot Disk Thermal Analyzer (Hot Disk AB, Sweden) in anisotropic mode.A Kapton 5456 sensor (3.2 mm radius) was sandwiched between a pair of identical foams (diameter: 3.89 ± 0.03 cm; height: 2.47 ± 0.06 cm) and thermal contact was guaranteed by placing a free weight (97 g) on top of the samples (contact pressure 805 N m −2 ± 14).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 (5%-80%) and temperature control (22 °C). [85]For each relative humidity (5%, 20%, 35%, 50%, 80%) 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 input.The axial thermal diffusivity and conductivity were obtained using the software provided by Hot Disk as previously described. [61]As reported before, [86] 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 of the uncertainties of the thermal diffusivity (), the density () and the specific heat capacity (Cp wet ) following the equation below: [53] Δ =  √ ( Δ The random uncertainties of ,  and Cp wet are based on estimates of the average standard deviations (SD) of the population 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 ) which were multiplied with 1.65, relating to a confidence interval of 95%.The systematic uncertainty of  was estimated to be 5%, [87] while no systematic uncertainty was considered for  and Cp wet .ΔCp wet incorporates the water uptake at different relative humidity.
H 2 O Uptake: Foam cylinders (2.4 ± 0.1 cm in diameter) were placed in a tared weighing dish on Sartorius BCE224-1S analytical balance, which was placed within a Climacell EVO temperature-and humidity-controlled chamber.The balance was connected to a computer, which recorded the sample mass every 5 min.The temperature was constantly at 22 °C while the humidity was altered over a period of 2 h and then held constant for 4 h for each humidity condition (20%, 35%, 50%, 65%, and 80% RH).The average and standard deviation of the masses recorded in the final hour of each condition were reported.The dry condition was estimated by transferring one foam cylinder of each sample type from the chamber

Figure 1 .
Figure 1.Foam preparation: Schematic illustration of the foam preparation with AFM images of CNC and MWCNT and digital images of the resulting CNC-100 and hybrid foams with different ratios of CNC and MWCNT.

Figure 3 .
Figure 3. Electromagnetic interference shielding efficiency (EMI-SE): a) Schematic illustration of the foam orientation in respect to the electromagnetic wave and its electric field b) total shielding efficiency (SE T ) of CNC-100, CNC-MWCNT_13 and CNC-MWCNT_22 for  = 0°at 8-12 GHz.SE T of CNC-MWCNT_22 for different orientation angles at c) 8-12 GHz d) 12-18 GHz.d) EMI-SE of CNC-MWCNT_22 for  = 0°at 8-12 GHz with the shielding efficiency by reflection (SE R ), absorption (SE A ) as well as the total shielding efficiency (SE T ).The dashed lines represent the average SE T over the measured frequency range.

Figure 4 .
Figure 4. Thermal conductivity and phonon scattering: a) Radial thermal conductivity  r , b) axial thermal conductivity  a and c) moisture uptake of the foams as function of the relative humidity.d) Illustration of the geometrical approach to calculate the interfacial thermal resistance R k .e) Radial thermal conductivity ( r ) of the foams as a function of MWCNT content for the parallel resistor model and the experimental data at 50% relative humidity.The error bars in a),b) and e) represent the relative uncertainty of the thermal conductivity measurements.The error bars in c) give the standard deviation between two measurements on two foams each.

( 3 )
with P CNC and P MWCNT representing the volumetric proportion between CNC and MWCNT (P CNC + P MWCNT = 1) and P CNC-CNC and P CNC-MWCNT representing the proportion of CNC-CNC and CNC-MWCNT interfaces (P CNC-MWCNT = 2 • P MWCNT ; P CNC-CNC = 1-P CNC-MWCNT ). CNC and  MWCNT give the thermal conductivity of CNC and MWCNT with 0.72 W m −1 K −1

Figure 5 .
Figure 5. Mechanical properties and foam wall structure: a) Representative stress-strain curves for compression along the axial direction at 295 K and 50% RH.b) Transmission electron microscopy image of a foam wall of CNC-MWCNT_13 with highlighted MWCNT.

Table 1 .
Properties of the foams.

Table 2 .
Mechanical properties of the foams along the axial direction.EA-LF500" produced from Leader Tech Inc. (USA) was used as comparative material for electromagnetic interference shielding efficiency.Deionized water was used for all experiments.