The combustion behavior of epoxy-based multifunctional electrolytes

Multifunctional or structural electrolytes are characterized by ionic conductivity high enough to be used in the electrochemical devices and mechanical performance suit-able for the structural applications. Preliminary insights are provided into the combustion behavior of structural bi-continuous electrolytes based on bisphenol A diglycidyl ether (DGEBA), synthesized using the techniques of reaction induced phase separation and emulsion templating. The effect of the composition of the structural electrolytes and external heat flux on the behavior of the formulations were studied using a cone calorimeter with gases formed during testing analyzed using FTIR. The composition of the formulations investigated was changed by varying the type and amount of the ion conductive part of the bi-continuous electrolyte. Two ionic liquids, 1-ethyl-3-methylimidazolium bis(fluorosulfonyl)imide (EMIM-TFSI) and 1-butyl-3-methylimidazolium tetrafluoroborate (BMIM-BF 4 ), as well as a deep eutectic solvent (DES) based on ethylene glycol and choline chloride, were used. The results obtained confirm that time to ignition, heat release rate (HRR), total mass loss, as well as the composition of the gases released during tests depend on the composition of the formulations. Addition of liquid electrolyte is found to reduce the time to ignition by up to 10% and the burning time by between 28% and 60% with the added benefit of reducing the HRR by at least 34%. Gaseous products such as CO 2 , CO, H 2 O, CH 4 , C 2 H 2 , N 2 O, NO, and HCN were detected for all formulations with the gases SO 2 , NH 3 , HCl, C 2 H 4 , and NH 3 found to be for certain formulations only.

supercapacitors and a matrix in composites. To date, the majority of the effort has been directed at improving the performance of individual components and the overall mechanical and electrochemical performance of structural energy storage devices. 1,[5][6][7][8] However, as the technological advances in multifunctional energy storage devices in general and more specifically structural supercapacitors matures, a focus on mechanical properties and electrochemical performance alone is insufficient, as additional questions arise which require answering. One such question concerns the fire safety of these devices and their constituents.
It is difficult to overestimate the safety issues surrounding energy storage devices, especially after the number of related fire accidents that have been reported, 9 mainly caused by thermal runaway of batteries. Not surprisingly, the latter has led to a significant amount of published research concerning batteries, directed at understanding the mechanism of thermal runaway, batteries' performance when exposed to fire, the influence of their individual components, and their fire dynamics. [9][10][11][12][13][14][15][16][17] Indeed, the majority of work on the fire safety of energy storage devices has been directed at batteries, most likely as they represent the main large scale commercially used type of energy storage device. It has been reported that lithium-ion batteries generate significant amounts of heat, with total heat release (THR) values strongly dependent on the state of the charge of the battery. 10,14,17 However, high temperature and the amount of heat released are not the only reported danger associated with lithium-ion batteries under fire conditions; the toxic gases emitted during the fire is another key factor which cannot be overlooked. 14 Such gases include not only CO 2 and H 2 O but also flammable and toxic ones, that is, CH 4

, CO, and
HCl, [17][18] where the flammable hydrocarbons being products of the decomposition of the electrolyte and separator present in the battery. 18 Research into the fire safety of energy storage devices includes studies of the individual components, 16,[18][19] full lithium-ion batteries, and battery packs. [14][15]17 There is also a significant number of reports on the fire properties of fiber-reinforced composites, from both an experimental and modeling perspective. [20][21][22][23][24][25][26][27] Interest in studying the behavior of fiberreinforced composite materials exposed to the fire is due to their wide range of applications, from sports equipment and medical prosthetics, to ships and aircraft, as well as less usual applications such as unmanned aerial vehicles (UAVs), space launchers, and satellites. 28 Often, composite materials are used to replace metal, since composites provide a combination of high mechanical performance, lightweight, and a resistance to corrosion. However, unlike metals and alloys, when subjected to fire, composite materials can produce volatile gases and vapors, both flammable (CO, CH 4 , etc) and nonflammable (CO 2 , H 2 O, etc), together with fumes and smoke. 29 The production of flammables gases can lead to increased heat, assisting in the growth and spread of fire, while toxic gases and smoke can result in reduced visibility and pose a serious hazard to both the environment and human health. The literature shows that tests have been performed under various conditions with the help of number of instruments, from tubular furnaces to cone calorimeter 24,30 apparatus, looking at mass loss, the kinetic mechanisms of thermal decomposition, heat release rate (HRR), and the release gases from the oxidation. 22,24,31 In most cases, the matrix is the main source of volatile products which could reduce the HRR due to the endothermic nature of the decomposition reactions of the organic materials. However, the composition of the volatile products emitted during thermal decomposition depends on the nature of the matrix and the heating process, as well as the atmosphere. A wide range of matrices are used in fiber-reinforced composite materials with epoxy being one of the most widely studied. To improve their fire performance, the effect of number of factors has been investigated, from the type of curing agent used 32 to the presence and type of fire retardants. 33 However, structural electrolyte contains not only epoxy resin but also a liquid electrolyte, 7,34-38 with bicontinuous epoxy-based electrolytes containing ionic liquid (IL) and deep eutectic solvent (DES) showing great potential for application in structural supercapacitors, as they exhibit a good balance of mechanical performance and ionic conductivity. 7,[34][35][36] ILs are salts with a melting temperature below room temperature and a unique combination of propertiessuch as high ionic conductivity, low vapor pressure, a wide voltage window, as well as their safety aspects when compared to organic electrolytes. Due to the potential of the ILs for various applications including energy storage, the question of their safety has received a lot of attention. [39][40][41][42] It has been shown 39,42 that despite ILs being difficult to ignite their associated HRR can be as high as 8000 kW/m 2 , with the HRR and toxicity of the emission produced depending significantly on the composition of the IL. 42 Despite the interest shown in the fire safety of structural energy storage devices, to the best of our knowledge, there is no information

| DES preparation
The DES was synthesized by mixing ChCl and EG in 1:3M ratio 43 at ionic conductivity of synthesized DES was determined to be 7.58 mS/cm which is in agreement with data reported in the literature. 43 2.3 | Preparation of the structural electrolytes 2.3.1 | Synthesis of structural electrolytes via reaction induced phase separation DGEBA was dissolved in the required amount of IL followed by the addition of the measured amount of iPDA. Next, the mixture was stirred until a homogeneous solution was obtained following degassing. Prepared formulations were cured using horizontal silicon moulds to produce plaques with dimensions 100 mm × 100 mm and thickness of 3 to 4 mm. For curing the following cycle was used: 1. dwell at room temperature for 22 hours; 2. ramp to 60 C at 2 /min; dwell at 60 C for 1 hour; 3. ramp to 80 C at 2 /min; dwell at 80 C for 2 hours.
Samples were cooled down to the room temperature, removed from the moulds and post cured using the following cycle: 1. ramp to 120 C at 6 C min −1 ; 2. hold at 120 C for 2 hours.
Samples were cooled in the oven to a room temperature prior to their removal from the oven.

| Structural electrolyte using medium internal phase emulsion approach
Medium internal phase emulsions (MIPEs) were prepared using a glass reaction vessel equipped with a glass paddle rod connected to an overhead stirrer. The continuous phase was prepared by dissolving surfactant in the hardener and after a solution was formed the epoxy was added, ensuring the weight ratio of DGEBA to iPDA was 1:4. The mixture was stirred using the overhead stirrer until a solution was obtained. Internal phase (DES; 50 vol%) was added dropwise under continuous stirring at 500 rpm. After all the internal phase was added, the stirring rate was increased to 2000 rpm for 2 minutes to further homogenize the emulsion. The prepared emulsions were transferred into the horizontal silicon mould as above to obtain plaques with the same dimensions. The MIPEs were polymerized using the curing cycle described in the previous section.
The structures of the all main compounds and compositions used for synthesis of the formulations studied are presented in Figure 1 and Table 1, respectively.

| Cone calorimeter test
The thermal decomposition and the combustion behavior of the different formulations were studied in a cone calorimeter, 44   F I G U R E 2 Schematic layout of the ISO 5660 cone calorimeter used elsewhere. [35][36]54 The properties of the formulations studied are listed in Table 2.

| Effect of the composition on time to ignition, flame time, and mass loss rate
Time to ignition is one of the crucial parameters in the study of materials and devices as it shows how long a sample can be exposed to heat before it ignites and initiates a flame. It was observed that among formulations studied, neat epoxy (DGEBA) was the last to ignite for both of the heat fluxes considered (35 and 50 kW/m 2 ). However, neat DGEBA also burned the longest. It can be seen ( Table 3) that even though IL containing structural electrolytes ignite faster, they also burn faster.
All the structural electrolytes studied have highly porous structure ( Figure 3) which affects heat transfer through the structure. As the thermal conductivity of the IL is lower compared to that of epoxy, [55][56][57][58] heat transfer from the top epoxy face through the thickness of the bicontinuous structure of the structural electrolyte is lower compared to a sample of neat epoxy. Hence, the top surface of the structural electrolyte samples exposed to the cone heater is heated rapidly and ignites before the neat epoxy sample. The reduction in burning time with the addition of IL is probably due to the lower epoxy content in structural electrolytes, the high thermal stability of ILs, 42,59 and the bicontinuous nature of the structural electrolytes microstructure. The reduction in epoxy content also led to the increase in the remaining sample mass after the test (Table 3, Figure 4). This further confirms that the primary source of the ignition and thermal decomposition is the epoxy resin. Note that, it was impossible to weigh residuals after the test as char from the epoxy was dispersed in the thermally oxidized IL and spread over the sample holder ( Figure S2).
It can also be seen from Table 3   Note also that the heat capacity of EMIM-TFSI is higher than that of BMIM-BF 4 , 60 which suggest that EMIM-TFSI could be responsible for reducing the local temperature, performing the role of a thermal barrier, and as a result increasing the amount of residual char (Table 3). Another reason for the observed differences in behavior of the formulations is their physical properties, more specifically their densities. The density of the neat epoxy was calculated to be roughly 1 g/cm 3 (Table 3), whether both ILs had densities >1 g/cm 3 , and more specifically, 1.52 and 1.21 g/cm 3 for EMIM-TFSI and BMIM-BF 4 , respectively. As a result the formulations containing EMIM-TFSI had a higher density compared to neat epoxy and structural electrolyte containing BMIM-BF 4 (Table 3). It has been reported 61 that the physical properties of polymers have a significant effect on polymer flammability and specifically that polymers with lower density will reach the critical pyrolysis flux before polymers with a higher density. Even it is reasonable to assume that it is also valid for porous polymers. 61 Figure 4 shows the effect of composition on the specific mass loss rate (SMLR) and the remaining weight of formulations studied.
The SMLR curves for all of individual formulations studied can be found in Figure S2. It can be seen ( Figure 4) that some of the formulations have a peak at an early stage of the experiment, which is most likely caused by the evaporation of the water absorbed by the ILs. 62 If present, the early peak is followed by the main peak, which for structural electrolyte containing BMIM-BF 4 is less sharp in comparison to other formulations ( Figure 4A). However, the amplitude of the SMLR curves for all formulations is similar with the main stage of the thermal decomposition occurring between 200 and 600 seconds. Out of all of formulations studied, polyMIPE showed the lowest remaining weight of just 2 wt%, followed by the neat epoxy (4 wt%) and formulations containing ILs ( Figure 4B, Table 3). The behavior observed for poly-  Table 3). As mentioned earlier, four processes can be observed during the testing, and subsequently, the HRR curve can be divided into four zones. The first is ignition, which is the part of the curve before the beginning of the peak (at around 200 seconds, The results presented shown that zones 1, 2, and 3 correspond mainly to the decomposition of the epoxy, which was also confirmed by the exhaust gas emission and is discussed in the next section; while zone 4 can be attributed to decomposition of the liquid electrolyte. It should also be noted that the HRR for neat epoxy is in a good agreement with data reported in the literature. 65 The pHRR in the case of 65DGEBA_IL 2 is slightly delayed, while for samples containing IL1, pHRR is observed with no obvious delay despite the variation in the epoxy resin content. The observed behavior is due to the difference in the thermal conductivity of the ILs ( Table 4)  It can be seen that the amount of the gaseous emissions depended strongly on the composition of the formulations studied. Introduction of liquid electrolyte leads to a significant reduction in the amount CO F I G U R E 9 Evolution of the main exhaust gases in case of different structures at the heat flux of 35 kW/m 2 released during tests (Figure 9), suggesting that the main contributor to the formation of CO is epoxy resin. This is in a good agreement with data available in the literature where it has been suggested that CO is a product of the first step of the pyrolysis reaction of the polymer ðPolymer + O 2 ! T CO + H 2 O + Other productsÞ: [65][66] If enough oxygen is available, CO would slowly oxidase forming CO 2 (CO + 0.5O 2 ! CO 2 ).
However, this process is limited to the amount of oxygen present, and the initial polymer pyrolysis stage could be faster than the following oxidation. For all formulations studied, the curves showing the formation of H 2 O follow a similar pattern to CO which is expected.