Sustainable Carbon Xerogel in the Production of Electromagnetic Radar Absorbing Composites in the Ku-Band

Carbonaceous porous materials are strong candidates for producing low-cost, lightweight, and sustainable electromagnetic (EM) absorbing materials. This work investigates the production of radar-absorbing materials (RAM) by a simple method. Sustainable carbon xerogel (CX) was synthesized from tannin, a biosourced molecule. CX and commercial carbon nanotube (CNT) were embedded in a silicone matrix at proportions of 10 and 15 wt.% of CX and 0.1 wt.% of CNT to produce flexible composites. The morphology and structure of the carbonaceous materials were evaluated by field-emission scanning electron microscopy (FEG-SEM), Raman spectroscopy, X-ray diffraction (XRD), textural properties by N 2 adsorption-desorption isotherms, and mercury porosimetry. The electromagnetic characterization of the composites was analyzed by a vector network analyzer (VNA) in the Ku-band. The results demonstrated that an increased concentration of CX in the composite improved reflection loss reaching -43.19 dB at 13.79 GHz.


Introduction
Recently the rapid development of technologies for electronic devices has increased the electromagnetic pollution 1-3 promoting a growing research interest in developing new radar-absorbing materials (RAM) 2 . However, producing materials with suitable properties such as low density, great absorbing properties, and low cost is still a challenge 1,2 . At the same time, there is an increasing stimulation worldwide to produce sustainable materials due to the natural resource scarcity, climate change, and global warming as addressed by COP 26 4 .
Tannin is a natural phenolic extracted from plants, which is cheaper than synthetic phenolic molecules and is a biosource widely employed to produce resins 5 , foams 6 , and gels 7 , providing a sustainable approach to preparing green materials 8 . Organic biobased gels are synthesized from polymeric reactions, based on a sol-gel process from the tannin-formaldehyde system 9,10 . The polymerization reaction is performed between the natural phenolic precursor (tannin) and the crosslinker (formaldehyde) in an aqueous solution. The gelation process happens when the formed polymeric chains grow to form a macromolecule, becoming insoluble. The final product, called aquagel is composed of solid, spherical interconnected nodules enclosed by a large porosity filled with water and by-products 7 . Drying is a critical step because the liquid should be removed without generating surface tension, thus avoiding shrinkage and loss of porosity 7,11 . The evaporation of the liquid phase at room conditions (subcritical drying) generates a highly porous material named xerogel 11,12 . In this work, the surfactant F-127 Pluronic was applied to reduce the effects of surface tension during subcritical drying. This technique is the simplest and cheaper compared to supercritical drying and freeze-drying methods, which produce aerogel and cryogel, respectively 12 . For these drying methods, solvent exchange is mandatory, and reactors for supercritical conditions or freeze-dryers are needed, increasing the production costs. Furthermore, tannin is a natural phenolic, which is price-competitive compared to synthetic sources. Tannin is about 39 and 86 times cheaper than resorcinol and phenol, respectively 13 . Therefore, the synthesis of sustainable xerogels is a straightforward process with a high potential for large-scale production, and the pyrolysis of such materials provides a highly porous carbon with the desired properties for RAM, for example, low specific mass and cost along with a sustainable feature 11,14 .
This work uses a sustainable carbon xerogel (CX) as a RAM filler and measures it in the frequency range of the Ku-band . Carbon nanotube (CNT) was added later to increase conductivity. The synthesis as well as the morphological and structural characterization of sustainable CX was performed.

Methods
The gel was prepared by a simple polycondensation route in a sol-gel process 11 . The Pluronic F-127 surfactant (1.5 g) was dissolved in 18 g of deionized water: ethyl alcohol (1:1). Then 3 g of tannin powder and 6 g of formaldehyde were added under magnetic stirring until a homogeneous solution was obtained. The solution was transferred to a sealed reagent bottle and kept in an oven at 85 °C for five days for polymerization and for five more days at room condition to dry completely. The material was carbonized in a tubular oven at 900 °C for 120 minutes under nitrogen flow to produce the CX. The full production process of CX is shown in Figure 1.
The composite samples were made with silicone rubber and 10 wt.% CX, 15 wt.% CX, 10 wt.% CX + 0.1 wt.% CNT, and 15 wt.% CX + 0.1 wt.% CNT. All fillers were mechanically stirred for 5 minutes, and the mixtures were poured into molds with dimensions of 15.80 x 7.95 mm and thicknesses ranging from 2 to 6 mm, producing flexible sustainable composites, as shown in Figure 2. The maximum filler concentration (15 wt.% CX) was defined as the saturation point. Concentrations greater than 15 wt.% of the CX did not allow the silicone rubber to cure. Pure CX composites were named accordingly to their concentrations of 10CX and 15CX. Samples containing carbon nanotube were named 10CX/0.1CNT and 15CX/0.1CNT.
The morphology of CX and CNT particles was evaluated through field-emission gun scanning electron microscopy (FEG-SEM; Mira3-Tescan). The graphitization was analyzed through Raman spectroscopy (Horiba-LabRam HR Evolution) using a green laser with a wavelength of 514.5 nm. The X-Ray diffraction (XRD) patterns were performed in a Rigaku diffractometer Ultima IV, using Cu Kα radiation (λ=0.15418 nm). The surface areas and the micropore volume of the CX and CNT samples were determined using the Brunauer-Emmet-Teller (BET) 39 and Dubinin-Radushkevic 40 methods from N 2 adsorption/ desorption isotherms at -196 °C using a Micromeritics ASAP 2020 Plus for samples previously degassed for 24 hours at 200 °C. Additionally, the porous texture of CX was also performed by mercury porosimetry intrusion-extrusion curves (0.0035 MPa to 414 MPa) using the Micromeritics Instrument Corp, AutoPore III. The electromagnetic properties of the composite were evaluated by a vector network analyzer (VNA; N5235A Keysight Technologies), using a Ku-Band (12.4-18 GHz) rectangular cross-section waveguide kit (P11644A). The dielectric properties (ε=ε'-jε'') 14,33,41 were obtained using the Nicolson-Ross-Weir (NRW) method, while the reflectivity was measured by the VNA using a metallic plate, as already explained elsewhere 42 .

Morphological, chemical, and textural properties
The morphology of sustainable CX and the commercial CNT was evaluated by SEM images. Figure 3a, 3b shows a tridimensional chain with interconnected and spherical nodules in a microscale range average diameter of 25 μm, typical of tannin-formaldehyde xerogels 7,11,14 . Furthermore, a large porosity between the chains, ranging from 5-10 μm is clearly seen in Figure 3b. Figure 3c presents a cylindrical structure of entangled CNT 43 without preferential orientation with the average diameter of around 22 nm.
The graphitization degree of carbon raw materials, CX and CNT, was evaluated by Raman spectra (Figure 4). The CX shows the D and G bands at 1344 cm -1 and 1590 cm -1 , while the CNT shows these bands at 1350 cm -1 and 1575 cm -1 , respectively. Another peak in the carbon nanotube spectra refers to the G' band at 2700 cm -1 .
The D originates from structural defects. The presence of this band is observed even in very low-disordered carbon. The D band in nanotubes originates from tube wall defects. The presence of vacancies and dislocations in the graphene layer is a source of defects in carbon materials 44 . The G bands correspond to the sp 2 carbon networks plane in the two-dimensional hexagonal lattice 1,43,45 . The G' band is related to the vibrational mode originating from the second-order Raman scattering process of double resonance involving the two-phonon scattering mechanism [46][47][48] . The second order of D band, the G' band, is very sensitive to the stacking order of graphene sheets along the c axis 49 .
The graphitization degree of carbonaceous materials is estimated by the intensity ratio of D and G bands (I D /I G ) 1 . Figure 4 shows the I D /I G ratios of 0.65 and 1.03 for CNT and CX, respectively. These results show a higher crystalline structure of CNT compared to the amorphous structure of CX.
The XRD patterns of CX reveal peaks at 23.3° and 43.5°, while the CNT showed the same peaks at 26.1° and 43.1°, besides an additional peak at 53.65° ( Figure 5). The first two peaks can be attributed to reflection planes (002) and (100) and the latter to (004), typical of the graphitic structure of carbonaceous materials 50-55 .    The nitrogen adsorption-desorption isotherms were performed for CX and CNT samples ( Figure 6). The CNT demonstrated a combination of isotherms types I and IV, typical of micro-mesopores materials 56 while the CX showed no significant N 2 adsorption. The H2-hysteresis loop of the CNT sample is signed to percolation in a narrow range of pores neck 57 .
The BET surface areas of CNT and CX were calculated from N 2 adsorption isotherms (Table 1), and the respective values are 90 m 2 .g -1 and 3 m 2 .g -1 . The porosity of carbon materials is classified according to the pore width (d pore ): d pore < 2 nm (micropores), 2 < d pore < 50 (mesopores) and d pore ˃ 50 nm (macropores) 58 . Therefore, the CNT presents a large mesoporosity and a moderate proportion of micropores (14%), while the CX showed no pores in the micro-mesoporosity range. Therefore, the porosity of CX is mainly composed of macropores (pores greater than 50 nm). Figure 7 shows the pore-size distribution of CNT from the Barrett-Joyner Halenda (BJH) model 59 . The distribution presents a narrow peak centered at 2.6 nm related to the small pores on the cylinder surface 60 and a broad pore size distribution from 5-45, with a centered peak at 22.3 nm, probably related to the hollow cylinders, corroborated by the average diameter (22 nm) found by the SEM images.
The textural properties (Table 2) and macropore size distribution of the CX were evaluated by mercury porosimetry analysis (Figure 8). The bio-based CX synthesized at a mass fraction of 11 wt.% showed a low-density value comparable to carbon aerogels from tannin-formaldehyde at a mass fraction from 26-34 wt.% and pHs between 4-6 9 . The significant porosity (52%) and the respective total pore area might serve as a pathway for the electromagnetic wave acting as an attenuating material. Figure 8 shows the mercury intrusion-extrusion curves obtained from the macroporous sustainable CX material. The hysteresis is attributed to the mercury trapped within the pores, as already observed for tannin-based gels 11,12 (Figure 8a). The pore size distribution (Figure 8b) shows peaks centered at 4 µm and 10 µm, agreeing with the SEM images, where voids in this scale range are clearly observed (Figure 3b).

Electromagnetic characterizations
The dielectric properties and reflectivity (RL) results of flexible composites from CX and CNT were evaluated by VNA. The complex permittivity of all samples 10CX, 15CX, 10CX/0.1CNT, and 15CX/0.1CNT are illustrated in Figure 9. The real (ε') and imaginary parts (ε'') of permittivity are related to the capacitive and dissipative effects in the material 29,41 .
The ε' of CX composites (Figure 9a and 9b) are observed between 5.4 and 8.1, while the ε'' had a maximum at approximately 0.5. A rise in ε' is seen when the percentage of CX increased to 15 wt.% in the polymer matrix (Figure 9b). The composites containing CNT 10CX/0.1CNT and 15CX/0.1CNT (Figures 9c and 9d) showed ε' curves between 5.7 and 9.9. The ε'' of the 15CX/0.1CNT sample increased with the frequency variation, reaching a maximum of approximately 1.5. The increase of CX concentration (15 wt.%) associated with the addition of CNT promoted the rise of ε' and ε'', improving the dielectric property. This effect can be attributed to a polarization interface due to the creation of dipoles between the CX interface and the polymer matrix 61,62 . The increase in the ε'', shown in Figure 9d, might be related to an increase in electrical conductivity, probably related to the higher concentration of porous carbon 29,63 .
Usually, materials with RL values lower than -10 dB are considered absorbers of EM radiation 45,64 . The 10CX shows a maximum RL of -11.08 dB at 16.72 GHz for a thickness of 5.80 mm (Figure 10a) and a bandwidth of 0.6 GHz. Increasing the percentage of sustainable carbon material on the composite (15CX) improves the attenuation to -13.98 dB Table 1. BET surface area (S BET ), total pore volume (V total 0.99 ), micropore volume (V DR ), mesopore volume (V meso ).
A total (m 2 .g -1 ) CX 0.40 0.87 52 1. 29 67 thick sample has been increased to 1.2 GHz. However, with the increasing concentration of sustainable material in the composite, reaching saturation (Figure 10d), a notable rise in RL was observed at -43.19 dB (13.79 GHz) and -40.17 dB (16.36 GHz), referring to the thicknesses of 5.00 mm and 5.90 mm, with a bandwidth of 1.00 and 1.42 GHz, respectively. As expected, the composites at a higher concentration of CX in the presence of CNT (15CX/0.1CNT) improve the dielectric losses (ε''≅1) 29 , which may be associated with the increased attenuation of the EM wave presented in the RL results.
Based on the above discussion, Figure 11 illustrates the possible absorption mechanisms of electromagnetic wave (EW) for the composites. First, when the EW reaches the interior of the samples, multiple reflections can occur. The pores on CX may also contribute to scattering the EW. At this point, the EW energy is transformed and dissipated 16 .
Secondly, the CNT has a considerably low specific mass of 0.28 g/cm 3 , representing a large volume even at small mass fractions. Thus, inserting only 0.1 wt.% of CNT can significantly increase the number of interfaces between CX/ CNT, and the synergic effect between CX and CNT favors the formation of interfacial polarization 65 , which would explain the improvement in the RL parameters in the composite. It is also important to notice that heterogeneous interfaces associated with different materials can cause interfacial polarization 66 or Maxwell-Wagner-Sillars effect 67,68 . Besides, the interfacial polarization will be enhanced due to the elevated porosity of the CNT and CX. Concurrently, the conduction loss cannot be discarded because the alternating current conductivity increases with ε'' enhancing conduction loss.
Finally, vacancies and lattice defects in the carbon structure contribute to dipole polarization. The defects in CX and CNT will be an additional source of dipole polarization center. The Raman spectra display a D band near 1350 cm -1 , corresponding to the disordered structure of amorphous carbon, Figure 4. The D-band presence corroborates the  existence of defects that will become dipole polarization centers 2,16,69 . Thus, the dominant dielectric loss mechanism stems from interfacial polarization, dipole polarization, multiple reflections, and conduction loss 70 .
Additionally, the thickness of the CX/CNT composites, when it exceeds 5.00 mm, plays a relevant role in affecting the intensity of the RL 71 . The incident EW cannot be totally absorbed. A significant part of the EW is reflected on the metal plate interface. When the absorber layer thickness is equal to one-quarter electrical wavelength, the incident and reflected waves are 180° out of phase, resulting in a total mutual cancelation at the air-absorber interface. This phenomenon can be explained by the quarter-wavelength matching model. The absorber thickness (t m ) for the cancellation effect can be expressed according to the quarter-wavelength matching model 1,45,64,71-73 : ( ) where t m is matching thickness, f m the peak frequency related to the minimum RL value, λ the wavelength of the EW, c the velocity of light, ε r the complex permittivity and μ r the complex permeability of the microwave absorber. Figure 12 shows the relationship between the absorber thickness and the more intense RL peaks of the 15CX/0.1CNT sample, where the black line refers to a 3λ/4 plot. Therefore, the t m values are evidently consistent with the effect of quarter wavelength (λ/4) cancellation, corresponding to a minimum RL value. The RL curves of the most intense peaks of the 15CX0.1CNT composite show that the peak position changes to a lower frequency as the thickness increases from 5.00 to 5.90 mm, following the rule of variation of the absorption peak with the thickness and frequency, indicating that the attenuation of the EW by microwaves can also include the contribution of absorption and destructive interference 74,75 . Table 3 shows that the characteristics of the sustainable composite are comparable to carbon aerogels from synthetic phenolic sources. The CX composite is among the materials with the highest absorption capacity (99.99%) and presents a slight difference from the composite with the highest absorbing potential, Graphene-CNT Aerogel (99.999%) 43 . However, the CX produced in this work is synthesized from a natural phenolic using a simple synthesis route, that is, dried under ambient temperature and pressure, resulting in a sustainable and low-cost composite, while the other materials were synthesized using more expensive synthetic products, in addition to synthesis routes with extensive stages, including the supercritical drying, increasing the cost of the final product.
Therefore, the sustainable composite 15CX/0.1CNT has relevant characteristics such as RAM. The mixture of sustainable CX with CNT probably promoted the pathway network on the material's surface, improving the absorption performance and quarter-wave matching. Additionally, the porosity of the material can also contribute to the attenuation due to multiple reflections and wave scattering that may occur within pores, therefore increasing the absorption capacity 1,2,14,43 .

Conclusions
This paper demonstrates an alternative route to producing sustainable, flexible, low density, and low-cost composites in the production of RAM. CX was synthesized from a natural polyphenolic, tannin, replacing synthetic phenolics such as resorcinol and phenol. The increase in the concentration of porous CX without adding CNT in the polymer matrix favored an increase in RL intensity, reaching values over 90% of absorption of the EW. The sustainable 15CX/0.1CNT composite showed EW absorption properties, resulting in an