**Viscosity and conductivity of electrospinning solution.** Viscosity and conductivity varied obviously with different ratios of PI/PES, while five ratios were used to test the optimal ratio as shown in Table 1, so polymer concentration is an important factor to form nano fibrous membrane during electrospinning process47. Table 1 shows that PI concentration variation is relatively connected to viscosity change, where viscosity value is 186 cP with ratio of 0:100 (PI/PES), and viscosity increases positively with enhancement of PI concentration. Viscosity tends to maximum value of 3374 cP while ratio of PI/PES is 100:0. Moreover, ratio of PI/PES adjusts to 0:100 to result in the smallest conductivity of 1.30 µS, and conductivity rises with increase of PI concentration. The maximum value of conductivity is 3.71 µS, if PI concentration adjusts to the maximum value.

Furthermore, tractive force of electronic filed influences degree of entanglement between molecular linking, so solution viscosity becomes too large while ratio of PI/PES adjusts to 75:25. The filter quality reaches to the maximum value of 2.57, accompanying with the maximum filter efficiency of 86.88%, when ratio of PI/PES adjusts to equal value.

Electric field highly influences solution conductivity during electrospinning process, and solution conductivity increases with enhancement of PI concentration. When solution conductivity is lesser, emitting flow rate become lower during electrospinning process, resulted in less amount of fibrous formation. Wide range of diameter distribution on fibrous formation occurs if solution conductivity is higher, resulted in instability of electrospinning solution

Table 1

Results of viscosity, conductivity, fiber efficiency, and filter quality with different ratios of PI/PES.

Ratio of PI: PES | Viscosity (cP) | Conductivity (µS) | Filter efficiency (%) | Filter quality (Pa− 1) |

0: 100 | 186.0 | 1.30 | 71.83 | 1.31 |

25: 75 | 500.9 | 1.93 | 82.40 | 1.77 |

50: 50 | 1077 | 2.50 | 86.88 | 2.07 |

75: 25 | 2202 | 3.03 | 81.67 | 1.74 |

100: 0 | 3374 | 3.71 | 78.46 | 1.59 |

**Sample decomposition analysis by TG.** Figure 2 shows decomposition processes of pure PI, pure PES, and PI/PES, and initial decomposition temperatures of PI and PES range above 500 ℃ while PI/PES is less than 500 ℃. The weight loss of three sample occurred at the ranges of 500 to 700 ℃, and almost 50% weight of three sample would loss during decomposition process. Therefore, these three samples acquired thermal stability because of high decomposition temperature. According to results of Table 2, the initial degradation temperatures are 549.8, 552.1, and 458.5 ℃ or PI, PES, and PI/PES, and the maximum degradation temperatures of PI, PES, and PI/PES are 568.4, 621.3, and 559.3 ℃, correspondingly. Mekuria et al., indicated that decomposition temperature of PI ranged as 500 to 700 ℃48, as matched the study of Zhou et al49. Liu et al. and Forati et al. proposed that decomposition temperature of PES ranged as 500 to 650 ℃50,51. This study depicted similar results with the above studies.

Figure 3 depicts slight weight loss at temperatures of 100 to 200 ℃, and the signal of weight loss may be defined as the vapor of dimethylacetamide (DMAC), because the boiling point of DMAC ranges as 164–166 ℃. As a result, the first weight loss occurred at temperature of 150–170 ℃ (decomposition of DMAC), and second weight loss started at temperature of 500 ℃ (decomposition of PI/PES)52. Wang et al., proposed that hot gas filters usually acquired filtration temperature of above 260 ℃53. For example, filtration temperature in waste incineration was between 200–350 ℃ in the study of Heidenreich54, exhaust gas temperature of thermal power plant operation was usually at 130–150 ℃ in the study of Liu et al.,55, operation temperature of concrete producing ranged as 150–300 ℃ in the study of Patel et al.,56, and flue gas temperature of crematorium was between 171–210 ℃ in the study of Takaoka et al.,57. Therefore, the PI/PES membrane could withstand hot gas temperature ranged 260–500 ℃, because weigh loss of PI/PES would not exceed 10% while hot gas temperature ranged 260–500 ℃.

Table 2

Degradation temperatures of PI, PES, and PI/PES at 5% and maximum % weight losses.

Sample | TD5% (oC) | Maximum degradation temperature (oC) |

PI | 549.80 | 568.40 |

PES | 552.09 | 621.34 |

PI/PES | 485.50 | 559.26 |

**Sample decomposition analysis by DSC.** Figure 4 depicts decomposition excursion of pure PI, pure PES, and PI/PES, and initial decomposition temperature starts at approximate 400 oC for three samples. The heat flow of three samples ranged as PI, PI/PES, and PES from high to low values, because PI/PES was made of half portions of PI and PES, resulted in heat flow of PI/PES between PI and PES. Moreover, an endothermic peak occurred at temperature below 100 oC, and this peak assumed as moisture because of the boiling temperature of moisture20. Figure 5 shows decomposition excursion of PI/PES by TG and DSC, and the initial decomposition starts approximately above 500 oC. Therefore, results of Fig. 5 also conformed PI/PES acquired thermal resistant of high temperature (usually above 500 oC), and the PI/PES membrane could be used as a potential material of hot gas filter.

**Filter quality at different experimental conditions by Taguchi method (larger the better).** There are four parameters, such as ratio of PI and PES, flow rate, voltage, and spinning time, applied to Taguchi method for producing high efficiency PI/PES nanofibrous membrane, and each parameter contains three levels, including ratio of PI and PES (25:75, 50:50, 75:25), flow rate (0.10 ml/hr, 0.15 ml/hr, 0.20 ml/hr), voltage (17.5 kV, 20 kV, 22.5 kV), and spinning time (20 min, 40 min, 60 min). The symbol of S/N is an important factor in calculation of Taguchi method, and the meaning of S/N indicates benefits and drawbacks of analyzed index, while higher S/N ratio indicates better quality of analyzed index. If the target goal was filter quality, larger-the-better was used to analyze the results of various fibrous membrane. Nine sets of experimental conditions were used to obtain the optimal filter quality, as shown in Table 3.

Table 4 shows evaluation of S/N and factor influence order for factors by using equation of larger-the-better. According to result of Table 4, factor influence order arranges as spinning time, flow rate, ratio of PI and PES, and voltage, because each S/N values are 9.03, 2.20, 1.67, and 1.43. As a result, spinning time is the most influenced factor for filter quality of fibrous membrane, and other factor influences list as flow rate, PI and PES, and voltage.

Table 4 shows that filter quality rises with decrease of spinning time because fiber mass and fiber thickness increase with enhancement of spinning time. During particles flowed through fiber membrane, the pressure variation tended to larger between two sides of fiber membrane if fiber thickness increased, resulted in poor filter quality of fiber membrane58. Therefore, Table 5 shows filter quality of items 1, 5, and 9 are larger while the spinning time is 20 mins, and filter quality of items 2, 6, and 7 rank as middle value while the spinning time is 40 mins. Moreover, filter quality of items 3, 4, and 8 are smaller while the spinning time is 60 mins. According to results of S/N values, the optimal experimental conditions are ratio of PI and PES (50:50), flow rate (0.15 ml/hr), voltage (22.5 kV), and spinning time (20 mins), and the best filter quality is 2.90, accompany with filter efficiency of 93.8%. As a result, smaller spinning time (20 mins) could produce a membrane with the best filter quality and filter efficiency if the target goal was filter quality (the larger the better).

Table 3

Results of fiber diameter and S/N at various factors by Taguchi method.

Item | Ratio of PI/PES | Flow rate (ml/hr) | Voltage (kV) | Spinning time (min) | y1 | y2 | y3 | S/N |

1 | 25:75 | 0.10 | 17.5 | 20 | 2.25 | 1.82 | 1.98 | 5.97 |

2 | 25:75 | 0.15 | 20.0 | 40 | 1.29 | 0.95 | 1.56 | 1.57 |

3 | 25:75 | 0.20 | 22.5 | 60 | 0.92 | 0.48 | 1.79 | -2.38 |

4 | 50:50 | 0.10 | 20.0 | 60 | 1.05 | 0.96 | 0.82 | -0.59 |

5 | 50:50 | 0.15 | 22.5 | 20 | 3.17 | 2.41 | 3.12 | 10.52 |

6 | 50:50 | 0.20 | 17.5 | 40 | 1.06 | 1.00 | 1.02 | 0.23 |

7 | 75:25 | 0.10 | 22.5 | 40 | 1.28 | 1.21 | 1.37 | 1.94 |

8 | 75:25 | 0.15 | 17.5 | 60 | 0.97 | 0.91 | 0.94 | -0.40 |

9 | 75:25 | 0.20 | 20.0 | 20 | 1.56 | 2.18 | 3.43 | 7.26 |

Table 4

Evaluation of S/N and factor influence order for factors.

| Ratio of PI/PES | Flow rate | Voltage | Spinning time |

1 | 1.72 | 2.44 | 1.93 | 7.91 |

2 | 3.39 | 3.90 | 2.75 | 1.25 |

3 | 2.93 | 1.70 | 3.36 | -1.12 |

**Δ** | 1.67 | 2.20 | 1.43 | 9.03 |

Factor influence order | 3 | 2 | 4 | 1 |

Table 5

Filter efficiency and quality of various fibrous membranes at different conditions.

Item | Conditions | Pressure loss (Pa) | Average penetration | Filter efficiency (%) | Filter quality (Pa− 1) |

1 | PI(25%)/PES(75%) 0.10 ml/hr;17.5 kV༛20 min | 0.98 | 0.14 | 85.95 | 2.02 |

2 | PI(25%)/PES(75%) 0.15 ml/hr;20 kV༛ 40 min | 2.62 | 0.04 | 95.67 | 1.27 |

3 | PI(25%)/PES(75%) 0.20 ml/hr;22.5 kV༛60 min | 7.52 | 0.01 | 99.38 | 1.06 |

4 | PI(50%)/PES(50%) 0.10 ml/hr;20 kV༛ 60 min | 5.88 | 0.004 | 99.57 | 0.94 |

5 | PI(50%)/PES(50%) 0.15 ml/hr;22.5 kV༛20 min | 0.98 | 0.06 | 93.82 | 2.90 |

6 | PI(50%)/PES(50%) 0.20 ml/hr;17.5 kV༛40 min | 6.87 | 0.001 | 99.88 | 1.03 |

7 | PI(75%)/PES(25%) 0.10 ml/hr;22.5 kV༛40 min | 3.92 | 0.01 | 99.33 | 1.29 |

8 | PI(75%)/PES(25%) 0.15 ml/hr;17.5 kV༛60 min | 6.21 | 0.003 | 99.69 | 0.94 |

9 | PI(75%)/PES(25%) 0.20 ml/hr;20 kV༛ 20 min | 1.96 | 0.02 | 98.05 | 2.39 |

**Factors influence on filer efficiency of nano fibrous membrane.**

**Factor of spinning time.** Table 6 shows that all filter efficiencies reach 99% at various spinning time, and all penetration efficiencies appear very low value, as shown in Fig. 6. The thickness of fiber membrane enlarges with increase of spinning time, because the mass per unit area increases accordingly with spinning time, resulted in most aerosol particles can be block. However, enhancement of fiber membrane caused increase of pressure loss, and filter quality decreased positively with pressure loss. As the result, 20 min was recognized as the optimal condition of spinning time.

Moreover, Fig. 6 depicts that penetration efficiency approaches to 0% while particle sizes are ranged below 100 nm, and less 5% of penetration efficiency occurs while particle sizes are ranged 100 to 500 nm. Therefore, Fig. 6 presents the same results of Table 6, and the best spinning time is 20 min.

Table 6

**Results of penetration and filter efficiency at different spinning time.** (Conditions: PI/PES (50:50), 0.15ml/hr, 22.5kV)

Spinning time (min) | Pressure loss (Pa) | Penetration efficiency (%) | Filter efficiency (%) | Filter quality (Pa− 1) |

20 | 0.98 | 0.005 | 99.54 | 5.56 |

30 | 1.63 | 0.001 | 99.92 | 4.89 |

40 | 1.63 | 0.002 | 99.79 | 4.17 |

50 | 2.94 | 0.001 | 99.85 | 2.26 |

60 | 3.27 | 0.003 | 99.74 | 1.96 |

**Factor of flow rate.** Table 7 shows that filter efficiency increases with flow rate because flow rate influences the volume of electrospinning solution in the spinneret. The volume of electrospinning solution is too less to generate stable Taylor cone when flow rate is slow. Moreover, some electrospinning solution may coagulate at the spinneret, and the distribution of fibers generate uniformed on the collector plate. The fiber diameter decreases with slow flow rate, because duration of tractive force by electronic filed is much longer. Oppositely, the mass volume of electrospinning solution gathers quickly in the spinneret when flow rate is fast, so the electronic filed is not enough to force solution generate smaller diameter and pore of fibers. Therefore, filter efficiency reaches over 90% at the conditions of 15 and 20 ml/hr. Although filter efficiency approaches 98% while flow rate sets as 20 ml/hr, filter quality is influenced vividly because value of fiber mass per area is high at high flow rate value, resulted in enhancement of pressure loss.

According to result of Table 4, spinning time and flow rate are the top two factors in this study by Taguchi method. Moreover, fiber mass per area increases positively time flow rate and spinning time resulted in high block efficiency of particle aerosol, but enhancement of fiber mass per area induce increase of pressure loss resulted in low value of filter quality. Therefore, 0.15 ml/hr is determined as the best condition of flow rate to present high value of filter efficiency and quality.

Table 7

**Results of penetration and filter efficiency at different flow rates.** (Conditions: PI/PES (50:50), 22.5kV, 20min)

Flow rate (ml/hr) | Pressure loss (Pa) | Penetration efficiency (%) | Filter efficiency (%) | Filter quality (Pa− 1) |

0.05 | 3.27 | 0.19 | 81.35 | 0.52 |

0.10 | 1.96 | 0.20 | 80.47 | 0.83 |

0.15 | 3.27 | 0.06 | 93.84 | 0.88 |

0.20 | 5.88 | 0.02 | 98.15 | 0.69 |

0.25 | 4.58 | 0.12 | 88.17 | 0.48 |

Figure 7 shows influence of penetration efficiency and flow rate, and penetration efficiency approaches to above 20% for 200–500 nm particle aerosol at flow rates of 0.05 and 0.10 ml/hr resulted in low filter efficiency, because low flow rate only can generate low fiber mass membrane. Moreover, the fiber membrane contains less value of fiber mass and can not block aerosol with small particle size of less than 50 nm. Moreover, the filter efficiency is also low at flow rate of 0.25 ml/hr, compared with flow rates of 0.15 and 0.20, because the solvent of solution can not evaporate efficiently at high flow rate. Moreover, diameter of fiber membrane increases obviously at high flow rate, and the pore of fiber also enhances accordingly. Therefore, the fiber membrane at flow rate of 0.25 can not effectively block aerosol at particle sizes of 200–500 nm and less than 50 nm. Summary of various factors test on filter efficiency and quality, the optimal filter efficiency and quality valued as 93.82% and 2.90 in this study.

Liu et al., proposed PM2.5 removal efficiency of polyacrylonitrile filter membrane valued as 98.69%17; Zhang et al., described PM2.5 removal efficiency of metal–organic framework filters valued as 88.33%59; Zhang et al., depicted PM2.5 removal efficiency of polyimide nanofiber membrane valued as 99.50%18; Wang et al., depicted PM2.5 removal efficiency of silk nanofibers valued as 98.80%60. Compared with the above studies, PI/PES membrane depicted high filter efficiency and high temperature resistance in this study.