LEGO®-inspired assembly strategy for fabricating BN-CNT-BN multilayer Kevlar-based composites as high-performance temperature sensors and fire alarms

Abstract

A new generation of artificial intelligence devices is being developed that require miniaturization and higher working power which result in higher heat flux densities, thus presenting a potential fire hazard. Current fire alarm sensors normally utilize electrically conductive materials that are not compatible with the surface insulation required to work in short circuit-triggered fire disasters. Here, we propose a novel concept and fabrication methods to manufacture durable and fast-response fire alarms with an electrically insulative surface layer. We initially separate the sensing and fireproof function of the fire alarm by creating a multilayer structure, where the conductive, thermally sensitive layer is sandwiched by two fireproof layers. The sensor is fabricated via a unique LEGO®-inspired assembly strategy that creates a nanobridge to coat the fireproof layer on the thermally sensitive layer. The sensor exhibits ultrafast response and recovery times of only 113.54 ms and 111.96 ms, respectively, along with great stability and durability over several cycles. Moreover, the surface BN-ANF layer provides protection for the internal thermally sensitive layer, which insulates it from oxygen and suppress the decomposition of the carbon nanotubes, thus enabling the sensor to be applied as a fire alarm. Upon exposure to fire, the sensor exhibits a fast response speed of 3 s and a long duration of over 1200 s. The fireproofing mechanism is also applied to improve the fire safety of a Joule heating film exposed to simulated short-circuit conditions. This new design concept and fabrication strategy improve the fire safety of a composite film and offer new inspiration for developing high-performance flexible sensors.

---

Introduction

With the remarkable progress in the development of integrated circuits, a cutting-edge foundation for next-generation artificial intelligence smart devices has been established,^1–4 leading to the miniaturization of electronic components and an increase in their working power.⁵ However, these advances in integrated circuits result in higher heat flux densities^6,7 and potential heat accumulation,⁸ posing significant fire risks and potentially catastrophic consequences for human safety and the economy.^9–11 Therefore, it is crucial to develop sensors and actuators that can give timely warnings of potential fires.

Monitoring the occurrence of abnormal temperature status is the most popular and direct approach to provide early warning of fire hazard, as temperature sensors respond faster than other types of sensors that detect smoke and gas.¹² Researchers initially used graphene and carbon nanotubes as thermally sensitive materials. Graphene have been proved to be thermally sensitive and is widely adopted in temperature sensitive materials. Researchers have developed different types of temperature sensors for different applications based on graphene, such as field-effect transistors and textile-based and paper-based sensors. A. Harzheim et al.¹³ combined wide and narrow-legged strip-shaped graphene to form single-material field effect transistors, which can be used as self-powered temperature sensors with extremely high sensitivity. Wang et al.¹⁴ used dip-coating of a non-woven fabric into a graphene-based dispersion to fabricate a temperature-sensitive textile with improved thermal stability and mechanical properties. Gong et al.¹⁵ utilized direct writing or mask spraying of graphene ink onto a paper substrate. The graphene exhibits low defect density, enabling the sensor to preserve good conductivity to work directly with high performance without the need for a subsequent reduction process. Carbon nanotubes (CNT) are another kind of thermally sensitive material. CNTs can transfer into carbon layers under high temperature, resulting in rapid changes in impedance. Strategies for improving the high temperature sensing behavior of a material mainly include the functionalization of CNT and their combination with inorganic compounds.¹⁶ Wang et al.¹⁷ used functionalized CNT with amino groups and fabricated a fast response temperature-triggered fire alarm. Ma et al.¹⁸ combine CNT with inorganic nanofillers (TiO₂) to fabricate a fast response long duration fire alarm. However, these two materials are flammable and easily ignite under exposure to fire.

Therefore, to improve the intrinsic flame-retardant properties of a temperature sensor to enable its application in harsh environments, several attempts have been made in previously reported works and they have made great advances in improving the fire retardant properties of flexible sensors. In 2014, researchers discovered the thermally sensitive properties of MXene, which can be utilized as thermally sensitive materials in fire alarm applications.¹⁹ Researchers initially adopted pristine MXene to improve the flame retardancy of composites. Yu et al.²⁰ directly introduced MXene to chitosan layers, effectively suppressing smoke generation and decreasing the fire hazard. Also, modified MXene was applied to further enhance the flame retardant properties. Xue et al.²¹ intercalated PPDA into a MXene interlayer, where only 1.0 wt% MXene–PPDA enabled a huge decrease in the peak heat release rate of polylactide. Moreover, some researchers have worked on modifying the morphology of MXene. Mao et al.²² produced polymer-decorated MXene networks with a nacre-like aligned structure, where a fish scale–like C/N-doped titania network is formed to provide a barrier to fire expansion. In 2016, a new material based on graphene oxide (GO) was reported for temperature sensing and fire alarms.²³ However, GO was found to be thermally unstable and require a high triggering temperature, resulting in a short alarm time and long response time at low temperatures.^24,25 Researchers have spared no efforts on working on these problems. To improve the thermal stability, Qu et al.²⁶ adopted an intumescence-assisted strategy where black phosphorene, GO and amino groups serve as an intumescent system that produces a porous carbon layer to protect the substrate. Researchers have also tried molecular modification and an effective thermal transfer strategy to lower the triggering temperature of GO. Huang et al.²⁷ adopted the sulfhydryl groups of 3-mercaptopropyltri-methoxysilan to promote the thermal reduction behavior of the GO network at a high temperature. Meanwhile, Ma et al.²⁸ stated that BN with high thermal conductivity could promote the thermal reduction of GO, creating a faster resistance change and thus a highly sensitive response. Both of these strategies work well to optimize the triggering temperature of GO-based sensors. However, in a fire disaster triggered by an electrical current overload and short circuit, these thermally sensitive materials would cease to be conductive as the heat accumulates. A large external current would trigger a large amount of heat and destroy the sensor immediately. Therefore, optimizing the surface electrical insulation properties of the fire alarm is crucial and necessary. Boron nitride (BN), a fireproof nanofiller, has been applied as an electrical insulation material under various circumstances. Xu et al.²⁹ utilized BN as an insulator in electronic devices to optimize the breakdown voltage and reverse leakage current. Meanwhile, Abiodun et al.³⁰ fabricated BN-polytetrafluoroethylene composite and the dielectric constant optimized to be 16% at most. Therefore, BN is a great candidate for both fireproofing and as an electrical insulation nanofiller for composites materials.

However, offering electrical insulation protection to the fire alarm without harming the fire sensing performance is quite difficult due to its working principles. A fire alarm normally detects temperature variation by generating a certain amount of heat-activated electrons, and so it requires the sensing material to be conductive to offer an electron transport pathway.³¹ However, the insert of insulation materials into the conductive network would be a barrier to the movement of electrons activated by the heat input, and thus the temperature sensing and fire alarm performance will be weakened as a result. Therefore, the use of electrical insulation is in conflict with the intrinsic working principles of a fire alarm and is hard to utilize in a working fire alarm. New approaches need to be discovered to provide electrical insulation protection to enable the fire alarm to work in an electrically conductive environment.

Herein, we propose a novel concept and fabrication methods to manufacture a durable and fast response fire alarm with an electrically insulative surface. We initially separate the sensing and protective function of the fire alarm, which is composed of a Kevlar nanofiber network hybridized with carbon nanotubes (CNT) and boron nitride (BN). Further, we develop a novel assembly strategy to connect the sensing and protective functional layer. Inspired by the assembly process of LEGO® toys, where different parts are connected by the embossed nodes of the LEGO® components, we mimic this mechanism and create connection nodes in the interlayer space by coating substrates with an aramid nanofibers-dimethyl sulfoxide (ANF-DMSO) dispersion and removing the solvent. In the process, the dispersion diffuses into the micropores between the fibers and transforms into LEGO®-inspired fiber-shaped nanobridges that connect the different layers. In this method opposite functions such as insulation and conduction can be integrated into a single composite film, and a fire alarm with a sensing layer encapsulated by a fire and electrical protective layer is fabricated in this way. The composite film demonstrates extraordinary temperature sensing performance with great stability, ultrafast response, and recovery times when compared to other reported temperature sensors. An improved fitting model enhances the reliability for further standardization and prediction. Additionally, the BN-based encapsulation layer offers fire protection to the internal sensing layer and the enhanced fire safety properties provided by the BN-ANF layer are tested and verified. Compared to bare CNT composite films, the nanobridge-assisted, encapsulated multilayer composite film shows an improved heat release rate (HRR) and total heat release (THR) performance. Additionally, even when exposed to fire, the surface insulation properties can still be maintained. Moreover, the film with a protective layer can withstand persistent burning and be reused multiple times. A fire alarm circuit designed to provide early fire warnings maintains an ultrafast response to flames, allowing repeated use of the composite film. Inspired by a fireproofing mechanism, the composite film exhibits excellent self-Joule heating performance with great stability and durability and can be operated within human-safe voltage ranges. An overcurrent test simulates short-circuit conditions, where the surface BN-ANF layer provides enhanced fire safety, allowing the film to work at higher voltages. In addition, the surface electrical insulation properties are evaluated. In summary, we propose the innovative concept of a novel LEGO®-inspired fabrication method based on the separation and re-assembly of the sensing and protection functions of the fire alarm to produce a fire alarm with superior alarm performance. This work will have find wide application in fire safety and provide guidance for further research.

Experimental section

Materials

Aramid fibers (Kevlar 1000D) were purchased from Suzhou QiCaiShi composites material Co., Ltd. Dimethyl sulfoxide (DMSO, 99.9% GC) and boron nitride (h-BN, 99.9%, metal basis, 5–10 μm) were purchased from MACKLIN. Potassium hydroxide (KOH) was purchased from Sinopharm Chemical Reagent Co., Ltd. The multi-walled carbon nanotubes (MWCNTs, >95%) were bought from the Xfnano company. Conductive silver paste was purchased from Shenzhen SINWE electronic Materials Co., Ltd. All the chemicals were used without further purification. DI water was obtained in the laboratory.

Preparation of the temperature-sensitive layer

First, the solid ANFs were dispersed. A 2 wt% aramid nanofiber (ANF) dispersion in dimethyl sulfoxide (ANF-DMSO dispersion) was prepared according to the method previously reported by Yang.³² Briefly, 2.0 g of Kevlar 1000D, 5.0 g of KOH and 47 g of DMSO were added into a conical bottle and heated at 50 °C for 1 week under vigorously stirring until no solid fibers were suspended in the solution. Then, the dispersion was blade-coated onto a glass base with a blade distance of 1.5 mm and immersed into DI water for 48 h following by freeze drying for 24 h to remove the water in the gel structure. The film was dipped into the 2% CNT-water dispersion to construct a temperature sensitive layer. The copper electrode was stuck to the surface of the CNT using conductive silver paste followed by solidification at 120 °C for 30 min.

Preparation of the fireproof layer

Using similar procedures as before, a 3 wt% ANF-DMSO dispersion was obtained. Then 0.2 g of h-BN was put into 4.8 g of DMSO via ultrasonication for 30 min. Then, 10 g of the prepared 3 wt% ANF-DMSO dispersion was added into the dispersion and stirred for 30 min and ultrasonicated for 30 min to ensure a good dispersion. Then the blade-coating, solvent exchange and freeze-drying process was repeated to obtain a fireproof layer.

Preparation of the LEGO®-inspired multilayer composite film

The previously prepared thermally sensitive layer and the BN-ANF layer were soaked into the 1% ANF-DMSO dispersion to coat a thin layer onto the surface of the film. The different functional layers were successively stacked following the sequence of a BN-ANF layer, a temperature-sensitive layer and another BN-ANF layer. The hot-press parameter was set as 25 KN with 80 °C to ensure sufficient diffusion of the dispersion. Then, the sample was put into water for solvent exchange for 48 h and dried at 80 °C for 8 h to obtain the conductive composite film encapsulated with a protective layer.

Characterization of the morphology and analysis of the mechanical and thermal properties

The surface microstructure was observed using field-emission scanning electron microscopy (SEM, Nova NanoSEM 450, FEI). The Raman spectra were collected with a Raman spectrometer (Renishaw inVia Reflex, 532 nm excitation laser). The surface temperature distribution was measured and monitored using an IR camera (VarioCam680, InfraTech). The fire-safe effect of the hybrid films was tested with a microscale combustion colorimeter (FAA-PCFC) made by Fire Testing Technology Limited, and other burning experiments were carried out with an alcohol lamp. Thermogravimetric analysis (STA449F3, NETZSCH) was performed under an air simulated atmosphere (80% nitrogen, 20% oxygen) at heating rates of 10 °C min⁻¹ over a range from 100 to 1000 °C. The BN was dispersed in ANF using an ultrasonic cleaner (KQ-400DKB). The ANF was fabricated using an oil bath heater (DF-101S). The temperature sensing performance was tested on a 4 probe system integrated dynamic hot-cold plate system provided by Zhengzhou Ketan Instrument Equipment Co., Ltd (KT-0904T-RL). The in-time resistance data was collected with a Keithley 2600 signal source meter.

Result and discussion

Fabrication and characterization of the temperature sensor via a LEGO®-inspired strategy

Fig. 1 shows the fabrication of the ANF-based temperature sensor and fire alarm. As depicted in Fig. S1 (ESI†), aramid nanofiber yarn is initially dispersed in a DMSO solution with excess potassium hydroxide. The strong alkaline polar solvent helps break the hydrogen bonds between the poly-p-phenylene terephthamide (PPTA) molecular chains, transforming the solid materials into a disperse solution. Hexagonal boron nitride (h-BN), used as the fireproof nanofiller for the fire protective layer, is added to the ANF-DMSO dispersion for uniform distribution to improve the overall fireproofing. The dispersion is then blade-coated onto a glass substrate, with the thickness controlled by the PI tapes attached to the glass stick. The coated sol is soaked in water for solvent exchange to replace the DMSO with water which facilitates hydrogen bonding between the nanofibers and converts the sol into a hydrogel. The sample is then freeze-dried to obtain a BN-ANF composite film. The pure ANF film is obtained using the same procedure without the addition of h-BN. The surface temperature-sensitive layer, made of carbon nanotubes (CNT), is deposited on the surface of the ANF film by dipping it into a water dispersion of the CNTs. This results in different functional layers that act as building blocks for further processing and fabrication.

Fig. 1 Schematic illustration of the LEGO®-inspired fire alarm assembly process.

In the assembly process of LEGO® toys, embossed nodes are grown on the surface of the connection interface and different parts are connected via the unique LEGO® nodes. Therefore, we mimic this process by creating a nanobridge connection from nodes at the interface. The thermally sensitive layers and fireproof layers function as building blocks and are dipped into ANF-DMSO for surface coating, with a thin layer of dispersion attached to the surface. Notably, a lower concentration of ANF-DMSO dispersion was chosen for the preparation process as it exhibits lower viscosity and is more likely to flow across the interface. The functional layers are then stacked in the following sequence: BN-ANF protective layer, temperature-sensitive layer, and BN-ANF protective layer, as a sandwiched structure. Pressure is applied to maintain the stacking process, which helps remove residual dispersion and ensures good contact between the cement and each functional layer. In the previous solvent exchange and freeze-drying procedures, both a gradient protonation process or the random growth of ice crystals can lead to irregular wrinkles in the final ANF composite films. Therefore, applying pressure is necessary for the fabrication process. The stacked sample is heated under pressure, as higher temperatures reduce the viscosity of the dispersion for better permeation into the nanopores. We conducted a series experiments to clarify the most suitable temperature and pressure parameters (Fig. S2–S4, ESI†). The sample was then immersed in water again for another solvent exchange process. The surface-coated ANF-DMSO dispersion layer is then solidified, forming the LEGO®-inspired nanobridges that act as cement to connect each layer. The film is dried in an oven instead of freeze-drying, as the crystal growth of ice in the freeze-drying process can break the interlayer connection, resulting in insufficient connection between each layer.

The schematic illustration of the structure is shown in Fig. 2a. As depicted in the image, two BN-ANF layers cover the internal temperature-sensitive layer, which consists of CNTs deposited on a base ANF layer. Scanning electron microscopy (SEM) was utilized to observe the morphology of each layer. Fig. S5 (ESI†) shows the surface morphology of the BN-ANF protective layer, where the BN flakes are uniformly distributed on the surface of the ANF film, with sizes ranging from 5–10 μm, corresponding to the standard size given in the information of the purchased chemical. The boron nitride nanoflakes distribute uniformly on the surface of the ANF film, which is proved by TEM and EDS mapping of the SEM images (Fig. S6 and S7, ESI†). Fig. S8 (ESI†) shows the skeleton of the ANF film. The discrete nanofibers in the DMSO capture protons in the water solution and combine to assemble a fiber network with numerous pores. Fig. S9 (ESI†) shows the morphology of CNTs, which form a continuous electrically conductive network used to sense temperature changes and transfer current into heat. Fig. 2b–e show the cross-sectional morphology between each layer. Fig. 2b provides an overall view of the BN-ANF layer, the ANF layer, and the CNT layer. The interfaces of each layer are observed in Fig. 2c–e. Energy dispersive spectroscopy (EDS) mapping was used to confirm the composition of each layer, as shown in Fig. S10–S14 (ESI†). Fig. 2c shows the interface between the BN-ANF protective layer and the base ANF base layer, which is initially filled with ANF dispersion. As the dispersion permeates the pore structure between the neighboring ANF and BN-ANF layers, it solidifies upon encountering water. With the removal of DMSO, the discrete nanofibers converge to form LEGO®-like nanobridges that connect the two layers. Fig. 2d shows the cross-sectional surface of the top BN-ANF protective layer and the CNT-assembled temperature-sensitive layer. Compared with Fig. 2c and e, it is evident that the ANF layer and the CNT layer exhibit similar structural characteristics, with wire-like building blocks cross-linked to form the layer. Thus, the same fabrication mechanism was applied between the CNT-sensitive layer and the BN-ANF functional layer. The previously filled dispersion solidifies and transforms into nanobridges connecting the two functional layers. Fig. 2e shows the nanobridge between the CNT-sensitive layer and the ANF base layer, that fabrication of which was different from the previously mentioned mechanism. As COOH@MWCNTs are used for electrical functionalization, the carboxyl groups that branch from the CNTs form a chemical link with the amide groups on the PPTA molecular chain. Therefore, at the interface of the two layers, the carbon nanotubes link with the ANFs on the upper surface to form a hybrid nanobridge for a tight connection between the two layers. The fabricated sample exhibited an ultralow thickness of only 431 μm (Fig. S15, ESI†). The Raman spectra of each layer are shown in Fig. 2f. From top to bottom, the Raman spectra of the CNT, BN-ANF, and ANF are listed in order. The peaks at 1320 cm⁻¹ and 1560 cm⁻¹ correspond to the D band and the G band of the carbon nanotube.³³ The spectra of BN-ANF and ANF show similar characteristic peaks, where peaks for the C=C stretching (1182, 1275, and 1325 cm⁻¹), amide I (1570 cm⁻¹), and amide II radial vibrational modes (1651 cm⁻¹) are observed.³⁴ A sharp peak at 1367 cm⁻¹ corresponds to the E_2g vibration peak of hexagonal BN,³⁵ confirming the co-existence of ANF and BN in the material. In summary, with the assistance of SEM and Raman spectroscopy, we observed the morphology of the cross-sectional surface, confirming the composition of each layer, and clarifying the mechanism of the hetero-interface binding strategy.

Fig. 2 Characterization of the sandwiched composite film: (a) cross-sectional scheme of the sandwiched composite film. (b) Cross-sectional SEM image of the sandwiched composite film. (c) The interface between the protective BN-ANF layer and base ANF layer. (d) The interface between the protective BN-ANF layer and CNT layer. (e) The interface between the protective base ANF layer and CNT layer. (f) The Raman spectra of each functional layer.

The temperature sensing properties of the composite film

CNTs are carbon-based electrically conductive materials that are widely utilized for electrical functionalization due to their excellent affinity with organic bases.^36–38 In 1996, T. Thio first discovered that the electrical resistance of CNTs varied with temperature,³⁹ indicating their potential application as temperature sensors. Unlike metals, the electrical resistance of carbon-based conductive materials (such as graphene, carbon nanotubes, and graphite) decreases with increasing temperature. Therefore, through dip-coating, we successfully established a self-assembled conductive network on the surface of ANF as a thermally sensitive material. Copper conductive electrodes were affixed to the CNTs using silver paste, with a protective BN-ANF layer covering the entire structure except for the exposed electrodes. We fabricated a sample that only had the bottom protective layer and monitored the temperature variation of the conductive layer as a function of the source temperature. The temperature measurement results are shown in Fig. 3a. As the source temperature increased, the surface temperature of the thermally sensitive layer also rose, showing a strong linearity with the source temperature (R² = 0.9995). This confirms the great quality of the binding that exists between the sensitive layer and the protective layer, which is facilitated by the nanobridges providing a stable heat transfer channel. Importantly, the assembly of the BN-ANF protective layer did not introduce unpredictable temperature changes to the thermally sensitive layer, preserving sensor accuracy.

Fig. 3 Temperature sensing performance of the sandwiched composite film. (a) The temperature of thermally sensitive layer corresponding to the temperature of the surface protective layer. (b) Dynamic test of the temperature sensor by exposure to a surface temperature increased stepwise. (c) Cyclical durability test of sensor via repeated cold-plate and hot-plate exposure. (d) Experimental and linear fitting of the relationship between surface temperatures and electrical response. (e) and (f) Response and recovery time evaluation at the beginning and the end of the cyclical durability test. (g) The response time and corresponding temperature variation compared with other reported works.^40–47

To quantitatively evaluate the sensing performance of the composite film, tests were conducted under various working conditions. A dynamic temperature test was performed to monitor real-time resistance changes with varying source temperatures. Starting from room temperature, the temperature was raised in increments of 20 °C every 120 s and maintained for 120 s before the next increment, up to 200 °C. The real-time response curve is shown in Fig. 3b, where the response is defined as (R − R₀)/R₀, with R₀ representing the initial resistance under room temperature conditions. When compared with the temperature curve, the response curve exhibited a consistent trend. The resistance showed distinguishable and stable plateaus throughout the temperature variations, demonstrating high thermal sensitivity and stability. Fig. 3d analyzes the relationship between the response and temperature. According to previous research, for a self-assembled film containing CNTs, the mathematical model should consider the contact electrical resistance between the CNTs prior to the intrinsic resistance of the CNTs themselves, which should be ignored in the calculation.⁴⁸ Therefore, an exponential model was applied to fit the temperature-response model for higher accuracy, although both fitting strategies showed excellent reliability. The exponential model achieved an excellent R² value of 0.9995, ideal for standardization and prediction in practical applications. Furthermore, the durability and stability, crucial for evaluating sensor properties, were assessed by repeatedly subjecting the sensor to hot-plate and cold-plate cycles, as depicted in Fig. 3c and e and f. The sensor exhibited repeatable responses over 10 cycles without significant degradation, indicating excellent stability and durability under harsh thermal cycling conditions. Throughout the cycling process, ultrafast response and recovery times of 113.54 ms and 111.96 ms, respectively, were observed initially. Even after 10 cycles, the response and recovery times remained at 130.91 ms and 90.5 ms, respectively, confirming sustained sensor stability. Compared to previously reported temperature sensors, this sensor demonstrated superior performance (Fig. 3g).

The fire retardant properties of the sensor and its application as a fire alarm

The surface protective layer endows the composites with extraordinary fire retardant properties where boron nitride acts as a fireproof nanofiller. Therefore, a comparison of the flame retardant properties of the multilayer composite with an unprotected sample needs to be conducted. Fig. 4a and b depict burning tests conducted on temperature sensors with and without a protective layer, respectively. Upon direct exposure to fire, the sample without the protective layer immediately ignites and undergoes violent combustion with the conductive layer detaching from the base film. In contrast, the sample with the protective layer initially turns white upon contact with fire and maintains this appearance throughout the burning process. The residual remnants of both samples are shown in Fig. S16 (ESI†). The thermally sensitive CNT layer detaches from the ANF base layer after the violent combustion. However, the protected sample exhibits a different behavior; its fire-exposed surface turns white, consistent with the presence of boron nitride, confirming that the BN-ANF layer significantly enhances flame retardancy. Thermogravimetric analysis (TGA) (Fig. 4c) reveals that the unprotected sample undergoes a sharp mass loss at 350 °C, corresponding to CNT degradation, while the protected sample maintains its mass until temperatures exceed 600 °C, indicating ANF degradation. To further investigate the surface carbonization and protection mechanisms, we conducted micro-combustion calorimetry (MCC) experiments (Fig. 4d and e). Fig. 4d demonstrate the heat release rate (HRR) analysis. The unprotected sample exhibits intense heat release at 400 °C due to the CNT decomposition observed in the TGA. In contrast, the protected sample shows significantly lower heat release rates at this stage, indicating suppression of CNT decomposition by the BN-ANF layer. However, at 600 °C, the protected sample exhibits a higher HRR due to BN-ANF layer destruction, exposing the internal CNT layer to air and subsequent rapid decomposition. This confirms that the BN-ANF layer provides effective surface protection. The total heat release (THR) analysis (Fig. 4e) shows that samples with and without protective layers exhibit THR of 6.85 and 8.93 kJ g⁻¹, respectively. The BN-ANF layer improves flame retardancy, enabling prolonged resistance to fire exposure.

Fig. 4 Flame retardant ability of the sandwiched composite film and its application as a fire alarm. (a) and (b) The burning tests of the sample with and without a surface BN-ANF protective layer. (c) Thermal gravimetric analysis of the sample with and without a surface BN-ANF protective layer. (d) The heat release rate during the combustion of the sample with and without a surface BN-ANF protective layer. (e) The total heat release during the combustion of the sample with and without a surface BN-ANF protective layer. (f) The variation of electrical resistance during the combustion of the sample with and without a surface BN-ANF protective layer. (g) and (h) Fire alarm application of the sample with a BN-ANF protective layer. (i) Duration test of the fire alarm.

The exceptional properties of the temperature sensor and its improved fire retardant properties indicate its potential for use as a fire alarm. We monitored the resistance change of the middle conductive layer during the combustion process compared with a sample without a protective layer. As shown in Fig. 4f, the sample without protection immediately exhibits increased resistance leading to an open circuit, due to the violent combustion-induced separation of the conductive and base layers. Conversely, the protected sample responds to the fire and, upon removal of the fire, shows resistance recovery to initial levels, indicating the potential for repeated fire exposure. Leveraging the exceptional temperature sensing and flame retardant properties of our material, we designed a simple fire alarm device (Fig. S17, ESI†), connecting sensors to a circuit comprising alarm lights, fire alarm sensors, switchable resistances, and an optical coupler. The sensor's triggering threshold was set to a 50% response to abnormal temperatures. At this threshold setting, the triggering temperature is calculated as 400 °C according to Fig. 4a and d, where the presence of this abnormal temperature can be confirmed to be an occurrence of a fire emergency. Fig. 4g–i and Movie S1 (ESI†) illustrate tests where exposure to fire triggers slow burning, activating the circuit to illuminate the alarm lights within 3 s. The sensors’ robust flame retardancy ensures persistent alarm performance, capable of maintaining alerts for up to 1200 s after fire exposure. We have compared the performance of our fire alarm sensor with other previously reported works and it demonstrates a better performance than other previously reported works^49–54 (Table S6, ESI†).

The Joule heating performance and improved fire safety of the composite film

Inspired by the fireproofing mechanism of the BN-ANF encapsulation layer, the temperature sensor can also be employed as a flexible Joule heater for personal thermal management. Initially, the surface electrical insulation properties were evaluated (Fig. S18, ESI†). The sensors exhibit a white color and have integral electrical insulation properties between each vertical layer, while the middle conductive layer still exhibits unaffected low resistance. Thus, the Joule-heating performance of the composite film was investigated, and its heating stability and durability were evaluated experimentally. Fig. 5a shows the heating performance under different working voltages. The test is conducted at room temperature, and the dynamic response of the temperature to input voltage is measured. The input voltage starts at 1 V, increasing by 1 V every 60 s stepwise until 5 V is reached, which is the safe working voltage of the composite film. A higher voltage results in ignition and fire. As the working voltage gradually increases, the surface temperature of the film rises continuously and reaches equilibrium instantly. When the voltage is increased incrementally to 1 V, 2 V, 3 V, 4 V, and 5 V, the corresponding saturated surface temperatures are 33.2 °C, 52.3 °C, 87 °C, 136.2 °C, and 194.2 °C, respectively. According to the Joule heating laws, the heat produced by electrical power, expressed as the surface temperature of the heater, should be proportional to the square of the input voltage. Therefore, the linearity between the surface temperature and the square of the input voltage was considered and fitted, with the results shown in Fig. 5b. The surface temperature exhibits excellent proportionality to the square of the input voltage, with an R² value of 0.99958, which is consistent with the theoretical results. This great linearity indicates that the temperature can be accurately adjusted by the input voltage, showing great potential for temperature control. Fig. 5c shows the durability test of the composite film. The driving voltage was set to 3 V and maintained for 10 minutes. As shown in the figure, the temperature remains steady without fluctuating or declining, demonstrating good stability, especially under high-temperature working conditions. The cycling stability was measured by repeatedly turning the power source on and off while monitoring the working status of the film heater (Fig. 5d). The driving voltage was set to 3 V, and 10 on–off cycles were applied to the heater. During the cycles, no degradation is observed, and the response temperature remains stable throughout the test. Overall, the composite film heater shows good stability and durability throughout the testing process. Additionally, the film heater can reach high temperatures at relatively low working voltages, ensuring human safety and allowing it to be driven by portable power sources for more convenient application.

Fig. 5 Joule heating performance of the sandwiched composite film: (a) temperature of the sandwiched composite film under stepwise-increased power input. (b) Experimental and linear fitting of the relationship between surface temperature and input voltage. (c) Temperature stability of the sandwiched composite film at a constant voltage of 8 V analyzed using surface IR for 600 s. (d) Cyclical durability test with an input voltage of 3 V. (e) The resistance stability of the conductive layer before and after the assembly of the protective layer. (f) and (g) Fire safety test of the heater with a bare surface and a BN surface protective layer under the same input voltage.

In daily use, the exposure of the CNT heating surface to the ambient environment can lead to dust accumulation and other contamination, resulting in degradation of the heating performance. Therefore, the surface BN-ANF layer, assembled by nanobridges, serves as a protective layer to extend the lifespan and maintain the properties of the heaters. As shown in the IR images in Fig. 5e and Fig. S19 (ESI†), two heaters with identical resistance were compared, one with a BN surface protective layer and one without. The results demonstrate that the generation of the nanobridge does not cause surface damage to the conductive network. When the same driving voltage is applied to both heaters simultaneously via a parallel circuit, no significant temperature difference is observed, indicating that the surface protective layer does not affect the heaters’ performance. In addition, power source faults are common and must be considered for safety to protect the heaters from dangerous accidents. Among these faults, short circuits are the most hazardous, causing devices to experience localized voltages several times higher than the regulated safe value.⁵⁵ We simulated a short circuit situation by applying voltages double the regulated safe value and compared the performance of heaters with and without the BN-ANF protective layer. As shown in Fig. 5f, the two samples were connected in parallel to ensure the same voltages were applied. When powered on, the sample with the protective layer continued to function at the extremely high voltage of 10 V, which is double the safe working voltage. Meanwhile, the sample without a protective layer immediately caught fire and burned out. Therefore, introducing the protective layer significantly improves the fire safety of the composite material in Joule heating applications.

Conclusion

We present a novel fabrication method for assembling a composite film with distinct functions inspired by the assembly process of LEGO® toys. An electrically conductive layer is deposited on the composite film and encapsulated on both sides by BN-ANF layers using the LEGO®-inspired nanobridge assembly strategy. The composite film shows exceptional temperature sensing capabilities, characterized by remarkable stability and ultrafast response times compared to previously reported sensors. Moreover, the BN-ANF protective layer enhances flame retardancy by mitigating the degradation of the CNTs conductive layer, positioning it as a promising fire alarm sensor. A dedicated circuit designed for fire alarm applications achieves rapid response times of less than 3 seconds and sustained durability exceeding 1200 seconds. Using the discovered fireproofing mechanism, the composite film demonstrates excellent Joule heating performance at relatively low working voltages, with outstanding stability and durability. The BN-ANF surface layer exhibits effective fire protection under short-circuit conditions. This study introduces new avenues for designing and fabricating fire alarms, offering significant potential in enhanced fire safety applications.

Author contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

This work is supported by the National Science Foundation of China (No. 52375567), Key Research and Development Program of Hubei (Grant no. 2023BAB096) and Wuhan Intellectual Property Innovation Special Project. The authors acknowledge the Nano Fabrication and Measurement Laboratory of the Collaborative Innovation Center for Digital Intelligent Manufacturing Technology and Application, and engineers in the Center of Micro-Fabrication and Characterization (CMFC) of WNLO for the support with the materials properties test. The authors thank the Analytical and Testing Center of HUST for providing SEM, EDS measurements and the Experiment Center of the School of Mechanical Science and Engineering of HUST for structural and material characterizations. The authors extend their gratitude to Mr Lang Jiang from Shiyanjia Lab (https://www.shiyanjia.com) for providing invaluable assistance with the MCC analysis.

References

1. W. Song, Nature, 2022, 603, 585–586.
2. H. T. Zhang, T. J. Park, A. Islam, D. Tran, S. Manna, Q. Wang, S. Mondal, H. Yu, S. Banik, S. Cheng, H. Zhou, S. Gamage, S. Mahapatra, Y. Zhu, Y. Abate, N. Jiang, S. Sankaranarayanan, A. Sengupta, C. Teuscher and S. Ramanathan, Science, 2022, 375, 533–539.
3. C. Kaspar, B. J. Ravoo, W. G. van der Wiel, S. V. Wegner and W. Pernice, Nature, 2021, 594, 345–355.
4. X. Qian, X. Chen, L. Zhu and Q. M. Zhang, Science, 2023, 380, g902.
5. S. Zhang, Y. Tang, J. Zeng, W. Yuan, J. Chen and C. Chen, Appl. Therm. Eng., 2016, 93, 1135–1144.
6. G. Liang and I. Mudawar, Int. J. Heat Mass Transfer, 2019, 128, 892–933.
7. Y. Sun, Y. Tang, S. Zhang, W. Yuan and H. Tang, Renewable Sustainable Energy Rev., 2022, 162, 112437.
8. Y. Sun, F. Liang, J. Chen, H. Tang, W. Yuan, S. Zhang, Y. Tang and K. J. Chua, Renewable Sustainable Energy Rev., 2025, 208, 115064.
9. S. Zhang, J. Chen, Y. Sun, J. Li, J. Zeng, W. Yuan and Y. Tang, Renewable Energy, 2019, 135, 1133–1143.
10. S. Jia, G. Chen, S. Dai, Y. Gao, Y. Wu, Y. Qing, S. Zhang, J. Xie, Q. Chen, Y. Wang, X. Huang and Z. Su, Nano Lett., 2024, 24, 5260–5269.
11. S. T. Lazar, T. J. Kolibaba and J. C. Grunlan, Nat. Rev. Mater., 2020, 5, 259–275.
12. K. Yan, J. Wang, Y. Zong and Q. Xu, Chem. Eng. J., 2024, 489, 151315.
13. A. Harzheim, F. Könemann, B. Gotsmann, H. van der Zant and P. Gehring, Adv. Funct. Mater., 2020, 30, 2000574.
14. F. Wang, J. Jiang, F. Sun, L. Sun, T. Wang, Y. Liu and M. Li, Cellulose, 2020, 27, 2369–2380.
15. X. Gong, L. Zhang, Y. Huang, S. Wang, G. Pan and L. Li, RSC Adv., 2020, 10, 22222–22229.
16. Y. Lei, Q. N. Chan, L. Xu, E. W. M. Lee, Y. X. Lee, V. Agarwal, G. H. Yeoh and W. Wang, Adv. Compos. Hybrid Mater., 2025, 8, 112.
17. Y. Wang, J. Liu, Y. Zhao, Y. Qin, Z. Zhu, Z. Yu and H. He, Composites, Part B, 2022, 247, 110348.
18. T. Ma, Q. Zhou, C. Liu, L. Li, C. Guo and C. Mei, Polymers, 2023, 15, 3666.
19. J. Halim, M. R. Lukatskaya, K. M. Cook, J. Lu, C. R. Smith, L. A. Naslund, S. J. May, L. Hultman, Y. Gogotsi, P. Eklund and M. W. Barsoum, Chem. Mater., 2014, 26, 2374–2381.
20. B. Lin, A. C. Y. Yuen, A. Li, Y. Zhang, T. B. Y. Chen, B. Yu, E. W. M. Lee, S. Peng, W. Yang, H. Lu, Q. N. Chan, G. H. Yeoh and C. H. Wang, J. Hazard. Mater., 2020, 381, 120952.
21. Y. Xue, J. Feng, S. Huo, P. Song, B. Yu, L. Liu and H. Wang, Chem. Eng. J., 2020, 397, 125336.
22. M. Mao, K. Yu, C. Cao, L. Gong, G. Zhang, L. Zhao, P. Song, J. Gao and L. Tang, Chem. Eng. J., 2022, 427, 131615.
23. L. Dong, C. Hu, L. Song, X. Huang, N. Chen and L. Qu, Adv. Funct. Mater., 2016, 26, 1470–1476.
24. X. He, Y. Feng, F. Xu, F. Chen and Y. Yu, Chem. Eng. J., 2022, 450, 137927.
25. Z. Yu, Y. Wan, Y. Qin, Q. Jiang, J. Guan, X. Cheng, X. Wang, S. Ouyang, X. Qu, Z. Zhu, J. Wang and H. He, Chem. Eng. J., 2023, 477, 147187.
26. Z. Qu, K. Wu, C. Xu, Y. Li, E. Jiao, B. Chen, H. Meng, X. Cui, K. Wang and J. Shi, Chem. Mater., 2021, 33, 3228–3240.
27. N. Huang, Q. Xia, Z. Zhang, L. Zhao, G. Zhang, J. Gao and L. Tang, Composites, Part A, 2020, 131, 105797.
28. Z. Ma, J. Zhang, L. Liu, H. Zheng, J. Dai, L. Tang and P. Song, Compos. Commun., 2022, 29, 101046.
29. M. Xu, A. Biswas, T. Li, Z. He, S. Luo, Z. Mei, J. Zhou, C. Chang, A. B. Puthirath, R. Vajtai, P. M. Ajayan and Y. Zhao, Appl. Phys. Lett., 2023, 123, 232107.
30. S. Abiodun, R. Krishnamoorti and A. K. Bhowmick, ACS Appl. Nano Mater., 2023, 6, 3781–3796.
31. T. Ma, L. Li, M. Pan, C. Guo and C. Mei, Chem. Eng. J., 2023, 451, 138517.
32. M. Yang, K. Cao, L. Sui, Y. Qi, J. Zhu, A. Waas, E. M. Arruda, J. Kieffer, M. D. Thouless and N. A. Kotov, ACS Nano, 2011, 5, 6945–6954.
33. Y. Pei, W. Zhu, R. Yue, L. Wang, R. Li, J. Zhang and Y. Yin, J. Mater. Chem. A, 2024, 12, 739–747.
34. S. E, B. Geng, X. Ji, D. Ning, Q. Ma, R. Zhao, L. Sha and Z. Lu, Compos. Sci. Technol., 2023, 235, 109975.
35. M. H. Adegun, K. Chan, J. Yang, H. Venkatesan, E. Kim, H. Zhang, X. Shen, J. Yang and J. Kim, Composites, Part A, 2023, 169, 107522.
36. P. Hu, J. Lyu, C. Fu, W. Gong, J. Liao, W. Lu, Y. Chen and X. Zhang, ACS Nano, 2020, 14, 688–697.
37. Z. Wei, Y. Cai, Y. Zhan, Y. Meng, N. Pan, X. Jiang and H. Xia, Small, 2024, 20.
38. W. Feng, L. Zou, C. Lan, S. E and X. Pu, Adv. Fiber Mater., 2024, 6, 1657–1668.
39. T. W. Ebbesen, H. J. Lezec, H. Hiura, J. W. Bennett, H. F. Ghaemi and T. Thio, Nature, 1996, 382, 54–56.
40. H. Xiao, S. Li, Z. He, Y. Wu, Z. Gao, C. Hu, S. Hu, S. Wang, C. Liu, J. Shang, M. Liao, D. Makarov, Y. Liu and R. W. Li, Adv. Funct. Mater., 2023, 33, 2214907.
41. W. Fan, T. Liu, F. Wu, S. Wang, S. Ge, Y. Li, J. Liu, H. Ye, R. Lei, C. Wang, Q. Che and Y. Li, ACS Nano, 2023, 17, 21073–21082.
42. P. Wang, W. Yu, G. Li, C. Meng and S. Guo, Chem. Eng. J., 2023, 452, 139174.
43. T. Q. Trung, T. Dang, S. Ramasundaram, P. T. Toi, S. Y. Park and N. E. Lee, ACS Appl. Mater. Interfaces, 2019, 11, 2317–2327.
44. M. Soni, M. Bhattacharjee, M. Ntagios and R. Dahiya, IEEE Sens. J., 2020, 20, 7525–7531.
45. J. Wu, Z. Wu, Y. Wei, H. Ding, W. Huang, X. Gui, W. Shi, Y. Shen, K. Tao and X. Xie, ACS Appl. Mater. Interfaces, 2020, 12, 19069–19079.
46. M. Seifi, S. Hamedi and Z. Kordrostami, J. Mater. Sci.: Mater. Electron., 2022, 33, 6954–6968.
47. C. Kong, X. Li, E. Zhang, J. Shi, J. Ren, C. Li, H. Wang and K. Wu, Prog. Org. Coat., 2024, 186, 107989.
48. A. Lekawa-Raus, K. Walczak, G. Kozlowski, M. Wozniak, S. C. Hopkins and K. K. Koziol, Carbon, 2015, 84, 118–123.
49. C. H. Liu, C. C. Chen, Z. W. Guo, Y. K. Fuh and T. T. Li, Adv. Mater. Technol., 2023, 8, 2300914.
50. S. Wang and Y. Zhang, Colloids Surf., A, 2024, 701, 134951.
51. H. Duan, C. Wang, L. Liu, N. Sun, X. Mu, H. Ding, H. He, W. Wang, W. Wu and B. Yu, Compos. Commun., 2025, 55, 102309.
52. Q. Zeng, B. Wang, X. Lai, H. Li, Z. Chen, H. Xie and X. Zeng, Composites, Part A, 2023, 164, 107305.
53. X. Han, T. Lu, H. Wang and G. Liu, Int. J. Biol. Macromol., 2025, 303, 140651.
54. F. Yang, Y. Ran, Q. Bao, P. Tao, J. Ding, S. Zhang, J. Zhu and C. Du, J. Cleaner Prod., 2025, 492, 144884.
55. B. Yang, N. Cui and M. Wang, Internal Short Circuit Fault Diagnosis for Lithiumion Battery Based on Voltage and Temperature, in 2019 3rd Conference on Vehicle Control and Intelligence (CVCI), 2019, pp. 1–6 DOI:10.1109/CVCI47823.2019.8951717.

---

Footnote

† Electronic supplementary information (ESI) available: Fig. S1. Preparation and functionalization process of LEGO® building blocks. Fig. S2. The sample fabricated with small pressure. Fig. S3. The sample fabricated with low temperature. Fig. S4. The sample fabricated with high temperature. Fig. S5. SEM image of the BN-ANF protective layer. Fig. S6. TEM image of BN-ANF. Fig. S7. The SEM of BN-ANF and the Boron element mapping. Fig. S8. SEM image of the ANF base layer. Fig. S9. SEM image of the CNT conductive layer. Fig. S10–S14. EDS analysis of different layer. Fig. S15. Photograph and graphic illustration of the ultrathin sample. Fig. S16. Burning residuals of the samples with or without protection layer. Fig. S17. The schematic diagram of the fire alarm circuit. Fig. S18. Electrical insulation properties of the samples with protection layer. Fig. S19. Surface Temperature Distribution with or without protection layer. Table S1. EDS analysis of the upper layer in Fig. 2e. Table S2. EDS analysis of the middle layer in Fig. 2e. Table S3. EDS analysis of the bottom layer in Fig. 2e. Table S4. EDS analysis of the upper layer in Fig. 2g. Table S5. EDS analysis of the bottom layer in Fig. 2g. Table S6. Fire alarm performance comparison with other reported works. See DOI: https://doi.org/10.1039/d5tc00815h

---