Epoxy emulsion, waterborne amine curing agent and perlite were purchased from Chengdu Jindingchuan Technology Co., Ltd. The solid content and epoxide equivalent weight of epoxy emulsion were (52 ± 1)% and 1000 (g/equiv.), respectively. The solid content and amine value of the curing agent were (60 ± 2)% and (180 ± 4) mg KOH/g, respectively. APP was purchased from Clariant (Exolit AP 422), PER was supplied by Perstorp, and ME was provided by Sichuan Golden-Elephant Sincerity Chemical Co., Ltd. Titanium dioxide (TiO₂, R902P) was purchased from Chemours.

The components of the waterborne coatings were listed in Table1. Epoxy emulsion was firstly mixed with APP, PER, ME, titanium dioxide and different content (1 wt%, 3 wt%, 5 wt% and 7 wt%) of perlite for around 30 min by a high-speed impeller to obtain homogeneous mixture, after adding the fixed ratio

curing agent and some additives, the mixture was mixed again for another 20 min. The resultant coatings were then brushed on the Q235 steel substrate with the dimension of 170 mm * 170 mm * 1.2 mm around 9 - 11 times until the average thickness of the coatings was approximately 2.8 mm.

The electrically heated, temperature-programmed furnace was selected to compare the thermal insulation properties of the coatings with and without perlite. Temperature of furnace and the backside of steel plate were measured with aids of K-type thermocouples.

Thermogravimetric analysis (TGA) was employed to measure the thermal properties of the waterborne fire resistive coatings using a TA Q5000 thermogravimetric analyzer under nitrogen atmosphere. The heating rate and the flow rate of nitrogen were set at 10˚C/min and 50 ml/min, respectively.

The heat release of the coatings was recorded with an FAA micro-scale combustion calorimeter (MCC). The sample weight was approximately 5 mg.

The element composition of the char was measured with the help of Thermo Scientific Escalab 250Xi X-ray photoelectron spectroscopy (XPS) using Al Ka radiation (1486.6 eV).

Fire protection performance test was carried out by a self-made test furnace whose average temperature was set according to ISO 834 (i.e. T = 345lg(8t + 1) + 20, where T represented the average furnace temperature, in degrees Celsius and t stood for the testing time, in minutes), so as to compare the variation of time when the temperature of the sample boards’ backside reached 500˚C, under the condition of being exposed to heat. The backside temperatures of steel boards coated with intumescent coatings using different formations were plotted as a function of time, which was shown in Figure 1.

As can be seen in Figure 1, the shape for all temperature profiles was similar for all the coating formulations. The fire resistant time was only 70 min without perlite when the temperature of the sample boards’ backside reached 500˚C. In the beginning, there was little difference in the temperature of each coating

sample since the temperature of all coating samples rose quickly with time. Once perlite was added, the temperature rates gradually slowed down, when 5 wt% perlite was loaded, the fire resistant time could reach 114.4 min, exhibiting satisfied fire resistant property.

Thermogravimetric analysis (TGA) was applied for rapid evaluation the thermal stability of intumescent coatings. The TGA curves of the coatings were shown in Figure 2, and the relative date were tabulated in Table 2. When the temperature was lower than 200˚C, all of the coatings showed the similar thermal behaviour, since the coatings began to degrade to a certain degree. Then the TGA curves differed with each other as the temperature rising, and E-0 coating without perlite degraded faster than others, leaving 34.61% of residue at 800˚C. In addition, the initial decomposition temperature of the coating was increased by up to 14.2˚C once 5 wt% perlite was loaded. In the meanwhile, it can be noticed that the residual char was gradually increased with the increase of the addition of perlite from 39.70% to 45.15% at 800˚C, indicating that the addition of perlite can enhance the stability of the residual char at high temperature.

Micro-scale combustion calorimeter (MCC), i.e. pyrolysis-combustion flow calorimeter, which was developed by Richard E. Lyon [11], was applied to measure the combustion performance of materials with a few milligram sizes, such as textile fabric [12] and extruded polystyrene foams [13].

The heat release rate (HRR) curves of the coatings obtained from MCC were presented in Figure 3, and the corresponding data were shown in Table 3. The peak heat release rate (PHRR) of pure EP resin was as high as 169.4 W/g, and it was dramatically declined after being made into coatings. It can be noticed that

the first peak of heat release rate of the coatings was delayed after perlite was added, exhibiting that perlite developed the initial decomposition temperature of the coatings, which was in accordance with TGA results. Although the value of PHHR was slightly increased after 3 wt% perlite was loaded, the total heat release of E-1 to E-7 was decreased, illuminating that perlite facilitated enhancing the flame redardancy of the coatings.

The element composition of the char was measured by virtue of XPS analysis, and that of E-0 and E-5 was listed in Table 4. It was shown that the value of C/O was increased by adding perlite, indicating that char of E-5 exerted better oxidation resistance at high temperature [14].

T_onset: Initial decomposition temperature at 5% weight loss.