A thermally insulating vermiculite nanosheet–epoxy nanocomposite paint as a fire-resistant wood coating

Conventional fire-retardant composite coatings are typically made of organic-based materials that reduce flame spread rates. However, the associated chemical reactions and starting precursors produce toxic and hazardous gases, affecting the environment and contributing to climate change. Wood is one of the most common materials used in construction and households, and thin-film fire-retardant coatings are needed to protect it from fire. Here, we derive high-performance nanocomposite paint-based coatings from naturally occurring and highly insulating layered vermiculite. The coatings are made using different weight percentages of shear-exfoliated vermiculite nanosheets in an epoxy matrix and are brush-coated onto teak wood. A series of tests using coated wooden rods and standard fire retardancy tests confirm a reduction in flame spread and combustion velocity with minimal toxic smoke release. Samples coated with the vermiculite/epoxy nanocomposite paint resist fire propagation, and post-combustion analysis indicates their resistance to thermal degradation. Our results offer a novel and eco-efficient solution to minimize the flame propagation rate, enhancing char development, and expand the scope of applications of ultra-thin vermiculite in nanocomposite coatings as a fire retardant, exploiting its low thermal conductivity in thermal insulation systems.


Introduction
Since the discovery of graphene, a variety of 2D nanomaterials with interesting properties have been continuously evolving. In fact, their properties are controlled by the number of layers and the functional groups embedded in them. [1][2][3] The exfoliation of layered nanosheets with unique thermal and electrical properties for diverse target applications is rapidly gaining momentum. 4 Some of these layered materials are monoelemental or bi-elemental, while others have complex structures and contain many elements. 5 One such example is the phyllosilicate family. Phyllosilicates are the group of minerals that form parallel sheets of silicate tetrahedra, e.g. talc, mica, serpentine and other clay minerals. 6 Vermiculite is an exciting and naturally occurring clay-based layered material. 7 It has incredibly low thermal conductivity due to phonon scattering at ordered interfaces as well as a very low density. 8 Recently, Paolo et al. 9 reported the thermal insulation properties of vermiculite nanosheets exfoliated in acidic solvent from their 3D precursors. They found that the exfoliated vermiculite nanosheets (<0.1 W m À1 K À1 ) had ultra-low thermal conductivity and proposed possible applications in thermal insulation and protection. Therefore, vermiculite nanosheets can be used effectively as nanollers in epoxy or polymer matrices. 10,11 Wood is used in construction and interior design due to its unique characteristics such as natural insulation and lowenergy production; however, it is highly ammable. 12 Phosphorus-based coatings that form char when ignited and dramatically swell when exposed to re are widely used in passive re protection of structures, but at the cost of releasing toxic gases. 13 Layered double hydroxide nanollers have been extensively studied for re retardant applications owing to the ability of the divalent and trivalent cations to suppress ame. 14,15 Carbon nanotube (CNT) and graphene-based polymer composites are good re retardants that use the barrier mechanism to retard ame spread. 16,17 Carbon nanotubes are expensive to synthesize. Experimental difficulties in aligning high-thermal-conductivity graphene nanosheets parallel to the substrate surface and volatility in reducing heat resistance limit their use in thermal protection systems. 18 Exfoliated boron nitride (BN) nanosheets, 19 aerogels, 20 polyurethane-based vermiculite composites, 21 heterostructures 22 and clay along with polyethylenimine and chitosan 23,24 have been effectively used in thermal insulation and ame retardant applications to reduce thermal resistance, and the addition of vermiculite has been found to delay the decomposition temperature and impede oxygen supply. To the best of our knowledge, there is presently no coating material that satisfactorily protects wood against re using a sustainable two-dimensional material with ultra-low thermal conductivity. The range of sustainable materials and methods to develop next-generation re-resistant materials is continuously expanding. 25 In this work, high-aspect-ratio exfoliated vermiculite nanosheets (ex-VN) are prepared by shear exfoliation of the bulk precursor in an aqueous medium, and a vermiculite-based nanocomposite paint is formulated using epoxy as the matrix. Due to its natural abundance, low cost and ultra-low thermal conductivity, the re-retardant properties of wood are enhanced aer coating with the ex-VN/epoxy nanocomposite paint. The addition of ex-VN to the epoxy leads to ame-extinguishing properties, reduced smoke production in the combustion test, and protection of the underlying structure.

Results and discussion
Vermiculite nanosheets and nanocomposite coating The as-received bulk vermiculite had a light brownish-yellow appearance, as shown in Fig. 1a. The optical microscope image shows a shiny surface due to the presence of metallic ions and large crystals (Fig. 1b). The layers in vermiculite can be separated using shear/sonic energy in solvents. Shear exfoliation is known to delaminate layered materials into thin nanosheets. 26 To avoid the aggregation or sedimentation of the exfoliated vermiculite, we used a surfactant to stabilize the vermiculite nanosheets in water; a photograph of the dispersion is shown in Fig. 1c. Here, the role of the anionic surfactant is to overcome the attractive van der Waals forces between the exfoliated nanosheets, which are stabilized due to electrostatic repulsive forces. 27 To analyze the quality of the nanosheets and lateral dimensions of the exfoliated vermiculite, TEM was performed.
As shown in the images in Fig. 2a and b, the exfoliated nanosheets are thin and transparent. The average lateral length is an essential factor in nanoller-based epoxy composite applications. We measured the lateral length of $100 vermiculite nanosheets from TEM images using ImageJ soware. The statistical distribution shown in Fig. 2c indicates that $70% of the nanosheets have lengths of over one micron, with an average lateral length of 1350 nm AE 149 nm. The structure of vermiculite is identical to that of mica and talc with a 2 : 1 composition of magnesium-based octahedral sheets sandwiched between two tetrahedral sheets of silicon, 28 as shown in Fig. 2d. It has a layered structure and weakly bonded exchangeable cations, and water molecules are present between these layers. The results demonstrate that the bulk vermiculite samples were successfully exfoliated in the water-surfactant solution using a simple shear exfoliation process.
Subsequently, a solvent-exchange procedure was carried out to exchange the solvent from water to acetone. Fig. 3 shows the ex-VN/ epoxy paint. The paint was prepared by rst dispersing the vermiculite nanosheets in the epoxy, followed by bath sonication for 30 min to achieve homogenization. Hardener was then added to the ex-VN/epoxy to make the paint, and the mixture was again sonicated for 30 min. This composite paint suspension was coated directly on the wood with a ne brush. To study the re-retardant properties of the ex-VN/epoxy nanocomposite, experiments were performed using two sets of coatings: an epoxy-only coating (control sample), and a series of coatings containing varying weight percentages of ex-VN (5%, 10%, 20%, and 30 wt%). To test the effect of the coatings on re resistance properties 29 such as thermal degradation 30 and combustion velocity, re tests were performed on coated wooden rods, including rods coated with the control sample. Brush-coating using brush strokes in a single direction yielded uniform coatings, as can be seen from the surface smoothness of the coated rod.  The thickness of the coated lm was measured using an optical microscope for samples on which 5, 7 and 15 coatings of the 5 wt% ex-VN/epoxy had been brush-painted. A representative optical microscopy image is shown in Fig. 4a. The thickness was computed from the images by measuring the thickness at 30 different points, and the average values were $85 mm for the sample with 5 coatings and $104 mm for the sample with 15 coatings. In the case of the 30 wt% samples, most of the paint is composed of the epoxy matrix, making the coating smooth and dense; the top view of the surface in Fig. 4b conrms the smoothness of the paint coating. In contrast, the surface of the control sample was rough, as shown in Fig. 4c. Thinner paints generally have a greater tendency to ow on the coated surface, aided by gravity. Therefore, when the samples are rested for curing, lumps occur at specic points on the rod surface. The brush-coated samples also exhibit a few inconsistencies since the samples were coated by hand and the wood surface was polished using sandpaper, thereby resulting in a somewhat non-uniform thickness due to the varying force applied via brush strokes on the wooden surface.

Standard re testing and thermal degradation
The Underwriters Laboratory-94 (UL-94) test is a standard plastics ammability test based on the tendency of material to extinguish ame and resist dripping. 31,32 It also provides the time taken for the ame to spread. 33 Fig. 5a shows bare wood, control sample coated wood and ex-VN-epoxy coated wood. A schematic of the UL-94 experimental set up is shown in the ESI (Fig. S2 †). The sample is exposed to ame for 10 s, and the time required for the ame on the sample to be extinguished is recorded as t 1 . Aer the ame is completely extinguished, the same process is repeated, and time t 2 is noted. Oen, even aer the ame is extinguished, a reddish aerglow remains, which slowly fades away. The time taken for this glow to disappear is recorded as time t 3 . Depending upon the values of t 1 , t 2 , and t 3 obtained from the UL-94 test, a polymer composite can be given a re retardant rating of V 0 , V 1 , or V 2 . Materials that exhibit slow burning or extinguish themselves suddenly without any dripping are given the highest UL-94 ranking. 34,35 Rod-shaped wooden samples with a length of 10 cm and diameter of 10 mm were rigidly xed to the stand. In the UL-94 test, if the ame is extinguished in less than 10 s aer the rst ignition and less than 50 s aer the second ignition, then it is ranked as V 0 category, which is considered to be the best class of re-retardant materials. 36 The sample coated ve times with the 20 wt% ex-VN-lled epoxy composite was able to extinguish the ame within 10 s upon the rst exposure to ame in the UL-94 test, qualifying it as V 0 .
The control sample did not display any re extinguishing properties and had to be forcefully extinguished aer the entire length of the rod was burnt 67 s aer the rst ignition. On the other hand, the ex-VN/epoxy-coated wood sample extinguished the ame in just 10 s, while the bare wood sample extinguished the ame in 31 s. On the vermiculite nanosheets-epoxy-coated sample, the ame propagated over only 4% of the wooden rod aer the rst ignition. This more favourable time duration shows that the ex-VN/epoxy nanocomposite coating can hold the re static for a longer time by forming a physical barrier to   propagation. 37 Since combustible and non-combustible gases were also released, the total and solid mass loss of the sample were quantied by tapping the samples with minimal force to see how much black char disintegrated from the edges of the tested samples. The results are shown in Fig. 5b and c. It was noted that there was a sharp decrease ($98%) in the amount of char formed during the combustion of the ex-VN/epoxy coated samples compared to that for the epoxy-coated control samples. The control samples show a marked 68.5% increase in thermal degradation as the number of coatings increases from 5 to 15. This increase is due to the higher amount of epoxy, which is polymeric and ammable, on the substrate. In the case of the 5 wt% samples, with an increasing number of coatings, the thermal degradation is increased as well. For this lower concentration of ex-VN, a smaller number of coatings works better. The trend is different when the concentration of ex-VN is high. In the case of 20 wt% ex-VN, it was observed that the thermal degradation decreased by 24% when the number of coatings was increased from 5 to 15. This implies that the higher amount of ex-VN in the thicker coatings was able to successfully counteract the heat produced by the pyrolysis of epoxy matrix. When a higher percentage of ex-VN is used in the epoxy, larger amounts of water molecules are present, which undergo an endothermic dehydration process to remove some of the heat generated during the combustion process.
As shown in Fig. 5d, the epoxy control samples developed visible cracks along the length of the wood, and their structure deteriorated, indicating that the coating was not able to prevent the ame from penetrating into the wood structure and inducing substantial degradation. When the ex-VN/epoxy-coated wood samples were tapped aer the tests, shedding of the loosely adhered black-coloured char from their surface occurred. Although the surface of the ex-VN/epoxy coated wood turned black aer the re test, the core of the wood remained intact, and cracks did not appear, in stark contrast to the other samples. No dripping was observed for the ex-VN/epoxy coated samples.
We performed a simple experiment to determine the depth of the structure disintegration. Sharp wooden pins were inserted at various points on the surface of the tested samples, as shown in Fig. 5e, and the depth to which the pin penetrated was measured. When the bare wood sample was pierced, it was unable to withstand the minimum pressure when the pin was inserted and broke into pieces. This experiment suggests that the pin was able to travel through the entire diameter of the rod, i.e., 10 mm. In the case of the epoxy-coated samples, the pin at the centre point travelled a distance of >1 mm at the edges that were exposed to the ame. The same test was also performed on the ex-VN-epoxy coated samples. It was observed that at all points of the sample, the pin travelled <1 mm. This test leads to the conclusion that the vermiculite nanosheets are able to successfully reduce the penetration of the ame into the wood sample.
Flame extinguishing time and ame-spread characteristics Fire tests were conducted on the epoxy and ex-VN/epoxy nanocomposites with different weight percentages of the nanoller to measure the ame extinguishing time, combustion velocity and ame height. The set-up used to measure the re-resistance properties is shown in the inset of Fig. 6a. The epoxy-coated sample showed no extinguishing properties and had to be extinguished forcefully 60 s into the experiment. Photographs of the samples for the ame height measurements are shown in Fig. S5. † The sample with the optimum weight percentage of exfoliated vermiculite nanosheets in epoxy showed the highest resistance towards catching re when ignited for the rst time. This response is due to the good coverage of the vermiculite nanosheets embedded in the epoxy and the reduction of thermal conductivity to the ultra-low regime upon exfoliation. The 10 wt% sample took 5 s to self-extinguish, while the 20 wt% and 30 wt% samples took 22 s and 10 s to self-extinguish, respectively, as shown in Fig. 6a. The observed variation in the self-extinguishing time at higher weight percentages of ex-VN is due to the compromise in the formation of a compact char layer on the wooden surface. The attractive van der Waals forces between the vermiculite nanosheets cause aggregation at higher weight percentages. This leads to larger tortuous paths at the optimum loading and shorter paths at higher loading (Fig. S9 †). At higher weight percentages, this aggregation of the layered material affects the barrier formed by char, thereby creating pathways through which heat can travel and sustain the pyrolysis reaction. Also, the dispersion of higher loadings of nanosheets in epoxy could increase the viscosity resulting in inhomogeneous lm formation at higher weight percentages, leading to variation in the self-extinguishing time. Interestingly, the aerglow time for the ex-VN lled epoxy nanocomposite samples was much longer than that of the epoxy control sample. This result indicates that not only does the lm coating resist re, but also holds the structure intact for a longer period without disintegrating. During the combustion of wood with phosphorus-based coatings, the smoke density is high and toxic chemicals are released. 38 As shown in Fig. 6b, the epoxy control sample shows a high combustion velocity of 0.015 cm s À1 , which is reduced by 40% in the 10 wt% sample and by 68.3% in the 30 wt% sample. There is a reduction in the spread of ame with increasing vermiculite nanosheet content in the epoxy, which gives rise to a mechanism in which the heat is dissipated gradually during the combustion process. Fig. 6b shows the ame test results, revealing that there is a $62% decrease in the ame height for the 10 wt% samples, a $20% decrease for the $20 wt% samples, and a $27% decrease for the 30 wt% samples.
A substantial reduction in smoke release and other harmful chemical fumes is observed. This reduction in smoke can be attributed to the inorganic nature of the vermiculite nanoller. Its water content and intercalated ions enhance the formation of the char layer post-combustion. In the absence of the coating, the char layer formed is quite weak and oen cracks into powdery ash, allowing gases to be released from the wood surface and form smoke. However, the presence of the coating leads to the formation of an enhanced thick char layer, which acts as an insulation barrier, preventing the gases from escaping and mixing with the air to form smoke. 8,39 As shown in Fig. 6c, as the thickness of the coating was increased, the re extinguishing time shows a decreasing trend with increasing wt% of vermiculite nanosheets, reecting the reretarding behaviour. Comparing the $104 mm thick ex-VN/ epoxy nanocomposite lms to the control sample, we see a $79% decrease for the 5% sample and an $83% decrease for the $20% sample.
Wooden rods of varying diameter were coated ve times using the 10 wt%, 20 wt%, or 30 wt% ex-VN-epoxy nanocomposite. The optimum re-retardant effect in terms of combustion velocity and ame spread was observed for the 5 mm diameter rods. The thin 3 mm rod samples have the advantage of having less fuel for combustion. The 30 wt% nanocomposite coating shows superior re-retardant properties, with the ame self-extinguishing within 2 s when ignited in the case of the 3 mm wooden rods, while the 5 mm samples took 8 s longer to extinguish. The ame spread is faster on the 3 mm wood samples, indicating a high combustion velocity. Hence, the 5 mm wooden rods can hold the ame static for a longer time and prevent it from spreading, as the cross-section of the sample is larger. As a result, the ame spreads gradually on the samples with larger cross-sections for the same coating thickness.
The reason that the exfoliated vermiculite nanosheets work as a re-retardant lm coating on wooden surfaces is a combination of three important factors. The mechanism is schematically depicted in Fig. 7. Firstly, vermiculite exhibits exceptionally low thermal conductivity compared to most inorganic layered materials. The high aspect ratio and low thermal conductivity (k) values of exfoliated vermiculite allow it to act as a physical barrier forming interconnections over the wood surface. 9 The dispersion of the compact ex-VN network is shown in Fig. S10. † FTIR spectra suggest that weak interactions and graing exist between the ller and the epoxy (Fig. S11 †). This would force ammable gases/products to take a tortuous path to react with the surface beneath. 37 In addition to thermal conductivity, thermal diffusivity and thermal effusivity are important properties for heat transmission through the material and its surroundings. Thermal diffusivity through a material is expressed as: where a is the thermal diffusivity (m 2 s À1 ), k is the thermal conductivity of the nanoller ($0.1 W m À1 K À1 ), 9 r is the density of the nanoller (225 kg m À3 ) and C P is the specic heat capacity of the nanoller ($108 J kg À1 K À1 ). By substituting these values into the equation, the thermal diffusivity of vermiculite is calculated to be $4.11 Â 10 À6 m 2 s À1 . This diffusivity value is much lower than the reported values. 40 During combustion, an ultra-small amount of thermal energy is transmitted through the material. Thermal effusivity is dened as: and its value is 1558 W (s) 1/2 m À2 K À1 . Although this value is slightly higher than the reported values, 40 in terms of thermal Paper Nanoscale Advances conduction and mass transfer, the low-regime thermal conductivity leads to enhanced re-retardant properties. 41 Heat ux is an important factor in the combustion analysis of materials, particularly wood-like structures, and is best described by the re triangle (fuel, oxygen and heat). It is understood as the ow of energy per unit area per unit time (kW m À2 ). The temperature of a lighter ame is different in different areas of the ame. The high temperature zone of the lighter ame, i.e., the non-luminous zone, was used for the combustion analysis of wood. The temperature of the non-luminous zone is above $1500 K. We found that most of the lighter heat ux values in the literature were greater than $50 kW m À2 . 42 Thus, we used 50 kW m À2 as the approximate heat ux value to study the re-retardant property of the wooden rods. Additionally, most materials are ignited with a heat ux between 10 and 20 kW m À2 . 43 In cone calorimetry studies, heat ux values in the range of 10-100 kW m À2 are typically used to study the peak heat release rate (PHRR), mass loss rate and ignition time of FR materials. Our observations of reductions in the mass loss, ame height, combustion velocity and smoke density corroborate the cone calorimetry measurements. These parameters were effectively used to evaluate the re-retardant properties of vermiculite nanosheets-epoxy-coated wood.
Here, we used the one-dimensional (1D) steady-state equation to probe the heat ux values through the vermiculiteepoxy-coated wooden rod. We considered the 1D steady-state because the rod dimensions seems to be 1D-like in nature.
The heat ux formula for 1D conduction is: where DT is the temperature difference (T 2 À T 1 ), T 2 is the temperature of the wood on the ignited side, and T 1 is the temperature of the other side of the wooden rod. Here, T 2 is $1000 K (close to the lighter temperature) and T 1 is $300 K (room temperature). L is the length of the wooden rod (here, the length up to the point of char formation was measured, i.e., $1 cm), and k is the thermal conductivity of ex-VN-epoxy composites ($0.1 W m À1 K À1 ). Aer substituting these values, the steady-state heat ux through the wooden rod was found to be $7 kW m À2 , which is lower than that of most of the nanomaterials used for the study. Secondly, the presence of water molecules, which make up $10% of the composition between the layers of vermiculite, helps to reduce the release of toxic gases by reacting with them to produce less harmful byproducts; it also reduces the heat release by absorbing heat for evaporation. 8 Thirdly, under the conditions in which the ex-VN/epoxy nanocomposite coating catches re, char formation occurs. The pyrolysis chain reaction is not sustained, thereby reducing thermal degradation to a great extent. During the combustion, vermiculite layer creates more char due to the metal ions (Mg 2+ , Al 2+ ) in its structure. 44,45 The produced char forms a protective layer on the substrate, which cuts off the oxygen supply, thus inhibiting the combustion reaction and extinguishing the ame. The black char between the ignition source and the material that has not undergone combustion acts as an oxygen-trapping barrier. 46,47 As a result, the trapped oxygen is unable to reach the wooden surface beneath, and the combustion process does not take place in the absence of oxygen. It has also been noted that the ability of vermiculite to retain its structure is maintained even aer repeated ignitions.
To probe the re-retardant mechanism, the surfaces of a wooden rod coated with ex-VN and a control sample were analyzed using electron microscopy, and the images are shown in Fig. 8a and b. The magnied regions clearly show that the char on the coated wood region has a distinct morphology compared to that of the control sample. Aer the combustion test, the uncharred region of the epoxy-controlled sample has a smooth morphology, whereas the ex-VN/epoxy-coated sample has a rough surface due to the random distribution of vermiculite nanosheets.
Vermiculite has a variety of metal ions and few interchangeable cations in its structure. To conrm the presence of nanosheets in the epoxy, we conducted energy dispersive X-ray analysis (EDAX) and elemental mapping; the resulting images are shown in Fig. 9a, b and S3 † shows elemental mapping of the un-charred and charred regions of the control and ex-VN/epoxycoated samples. These results conrm the presence of Si, Al, Mg, and Ti metals, which are part of the vermiculite structure, and other interchangeable cations like Ca and K. We carried out elemental mapping on the charred portion. We knew that the charred portion was formed by lightweight carbon residue. Interestingly, we found the elements of vermiculite in the charred region in addition to carbon, and an increased concentration of element oxygen, which indicates the presence of a compact layer of char in which metal is oxidized.
The inclusion of the exfoliated vermiculite nanosheets in the epoxy matrix successfully introduces ame-retardant characteristics to the nanocomposite, such as decreased combustion velocity, ame extinguishing, decreased ame height and heat release rate, and suppression of toxic smoke release, as well as Fig. 8 Characterization of char after the combustion tests. Scanning electron microscopy images of (a) a 20 wt% ex-VN-epoxy-coated wooden rod and magnified images depicting its morphology in charred and un-charred regions and (b) control epoxy-coated wood.
reduced thermal degradation of the wooden sample. Demonstration of the re-retardant behaviour of wood with and without ex-VN in epoxy is shown in the ESI Video. † One of the most important characteristics of a re-retardant material is its ability to self-extinguish ame. As depicted in Fig. 9 and from our observations during re testing of the ex-VN/epoxy composite, the re-retardant material releases non-ammable gas via a broken char layer, which is known as the blowingout effect. 48 Vermiculite contains re-resistant compounds such as SiO 2 , Al 2 O 3 and MgO, which are responsible for the formation of inert gas from the char layer. During combustion, vermiculite releases H 2 O, CO 2 , NO and SO 2 . 49 These gases are all well-known re extinguishers.
The jetting of these gases outward from the char layer, i.e., the condensed portion, extinguishes the re. 50 During combustion, a thermally stable thin layer is formed between the wood surface and the re. This char layer insulates the wood from re and instantaneously forms non-ammable gases inside the char. The accumulation of the non-ammable gases in char leads to increased pressure, and these gases then completely extinguish the re. 49 In our experiments, there was a remarkable decrease in the thermal degradation of the wooden substrate beneath the coating due to the introduction of vermiculite nanosheets in the nanocomposite. The heat ow always chooses the path which offers less resistance. Vermiculite provides more resistance to heat ow through the coating due to its low value of through-plane thermal conductivity. For this reason, heat does not travel across the surface of the coating. Therefore, the coating can prevent the heat from penetrating into the wooden substrate, thus preserving its integrity aer the re testing. Thus, vermiculite can stop heat from reaching deeper into the samples by forming a physical barrier to heat ow. This is indicative of the fact that vermiculite can withstand heat without undergoing any considerable chemical changes or damages to its internal structure.

Materials
Bulk vermiculite samples were received from Tamil Nadu Minerals Limited, Chennai, India. Apcolite 2 pack epoxy nish was purchased from Asian Paints. Sodium dodecyl benzene sulphonate (SDBS) was purchased from Sigma Aldrich. Acetone with 99.5% purity was used in this work and was purchased from S. R. L. Chemicals. Distilled water was used throughout the experiments. Wood samples with a length of 10 cm and diameters of 3 mm, 5 mm and 10 mm were procured from local woodworks, and commercial emery paper was used to polish their surfaces before painting. A commercial lighter was used for re testing.

Preparation of exfoliated vermiculite nanosheets (ex-VN)
In this work, we used a liquid-phase exfoliation technique, namely, shear exfoliation, to delaminate the bulk vermiculite into thin layers of the parent component in a water/SDBS surfactant dispersion. 26 For this, initially, 10 g of grade 5 (ne particle, $1 mm) bulk vermiculite was added to 500 ml of water with 0.8 mg ml À1 dissolved SDBS. This dispersion was subjected to exfoliation using a kitchen blender for 1 h with an on and off timer to avoid heating effects. Aer this step, the resultant dispersion was allowed to stand overnight, and the supernatant was recovered using a Pasteur pipette for further dispersion analysis and characterization; the sediment was discarded. Centrifugation of the resultant dispersion was performed at 6000 rpm for 30 min using a Thermo Scientic Sorvall ST8R small benchtop centrifuge to recover the exfoliated vermiculite nanosheets in paste form; the nanosheets were then dried in a vacuum oven at 60 C for $8 h. The exfoliated vermiculite nanosheets were redispersed in acetone using bath sonication. This process was repeated twice to transfer the material from the aqueous solution to acetone solvent, and bath sonication was applied for 30 min to homogenize the sample.

Preparation of ex-VN/epoxy nanocomposites
The coating paint was formulated using a commercial epoxy (Asian Paints) lled with various weight percentages of ex-VN. The coating was prepared via the in situ polymerization of the epoxy monomer with a hardener in the presence of the nano-ller. Acetone is used as a paint thinner and solvent medium for ex-VN. To prepare the epoxy/ex-VN nanocomposite, 50 mg of ex-VN (for 5 wt%, the total weight of the composite was 1 g) was initially added to 634 mg of epoxy and bath-sonicated (Branson, 40 kHz, 70 W) for 30 min to disperse the nanoller in the epoxy. Aer this step, 316 mg of hardener (3 : 1 ratio of epoxy to hardener) was added to the ex-VN/epoxy monomer, and bath sonication was applied again to disperse the hardener throughout the system. This process was repeated to prepare the 10, 20 and 30 wt% nanocomposites. Acetone was used in the nanocomposite preparation process for homogenization and for thinning the formulated paint before coating it on the wood. Upon the addition of ex-VN into the epoxy matrix, there was a notable change in the colour of the epoxy. The epoxy-hardener is pure white in appearance, and the ex-VN/epoxy ink is essentially light brown. The mixing of these two results in a change in the colour of the matrix; the coating mix is light brown in colour and becomes russet or dark brown when the weight percentage is high. Fig. 9 The morphology and elemental composition of 20 wt% ex-VN/ epoxy-coated wood before (a and b) and after (c and d) combustion tests.

Specimen preparation and brush-coating
All the teak wood samples were treated with electro-coated SiC grain emery paper of 220 grit size to improve the adhesion. The as-received epoxy polymerized with a cross-linking agent was also prepared as per the mixing requirements and used as the control sample. The preparation of the control coating did not require bath sonication as it could be homogenized with mechanical stirring. All the samples were coated using at brushes. The interval between the application of each coat was approximately 2 min, and the specimens were cured at room temperature for 24 h.

Characterization techniques
A eld emission scanning electron microscope (FE-SEM, Thermo scientic Apreo S) was operated at 15 kV to obtain the morphology, elemental composition, and mapping of the charred and uncharred regions of the coated samples. A turbopumped sputter coater (Quorum, Q150T Plus) was used to cover the sample surface with Cr before placing the samples under an electron microscope to avoid charging effects, with a sputtering time of 25 s and a lm thickness of $7 to 10 nm. Transmission electron microscopy (TEM) analysis of the exfoliated vermiculite nanosheets was carried out using a Hi-Resolution Transmission Electron Microscope (HRTEM, JEOL Japan, Model: JEM-2100 Plus). Images were also obtained using an Olympus BX51 optical microscope. Thermogravimetric analysis (TGA) was carried out using a simultaneous thermogravimetric analyzer (STA 7000, HITACHI) in an Ar atmosphere with a rate of 10 C min À1 . Fourier transform infrared spectra were obtained using an IRTracer 100 (Shimadzu) in ATR mode. A custommade Underwriters Laboratory-94 re retardant standard test was performed on wood samples 10 cm in height and 10 mm in diameter to characterize the fabricated ex-VN/epoxy nanocomposites.

Conclusions
In this work, we have demonstrated an eco-efficient reretardant coating using epoxy lled with shear exfoliated vermiculite nanosheets on a wooden surface. A simple and scalable brush-coating technique was used, and a lm thickness of $100 mm was achieved. Fire tests were performed on samples with varying weight percentages of vermiculite nanosheet ller (5-30 wt%) in the epoxy matrix to test the effect of the vermiculite loading. The introduction of the low-k vermiculite ller into the matrix was found to reduce the ame height and, consequently, the heat released during combustion by up to $62% in the case of the 10 wt% sample. The thermal degradation tests showed a decrease of $42% due to the addition of exfoliated vermiculite nanosheets. A 68.3% decrease in combustion velocity was observed for the samples with 30% vermiculite by weight in their matrix. It was also observed that the samples without vermiculite showed no self-extinguishing properties and burned to completion, thereby leading to the complete disintegration of the wooden structure beneath the coating. This work opens up new opportunities for using insulating layered 2D materials to resolve common issues related to re retardants and to develop alternative environmentally friendly, high-performance materials.

Conflicts of interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to inuence the work reported in this paper.