Electrospun polyimide nanofiber-coated polyimide nonwoven fabric for hot gas filtration

Materials

Needle-punched PI nonwoven fabric filter materials with a thickness of 2.1–2.5 mm and a weight of 450–550 g/m² were kindly supplied by Lantian Huanbao Co. Ltd (Jiangsu, China). Pyromellitic dianhydride (PMDA), 4,4-oxidianiline (ODA), and dimethylformamide (DMF) were brought from Sinopharm Group (Tianjin). All the raw materials were used as received.

Samples preparation

The PI nanofibrous membrane-coated PI nonwoven fabrics filter materials were prepared following the scheme as shown in Figure 1. First of all, PAA precursor solutions were prepared through polycondensation of equimolar amounts of PMDA and ODA in proportionally increasing volume of DMF, as shown in Table 1. All of the composites contained a solid concentration of 16.7 wt%. The reaction was carried out in three-necked round-bottom flask immersed in an ice bath with a temperature of 2–4°C for 8 h. Then obtained transparent solution was then transferred to a 10 ml of syringe and electrospun under 16 kV of voltage and 20 cm of gap between a spinneret having an inner diameter of 0.45 mm and a flat metal collector covered with PI nonwoven fabric. The PAA solution was fed at a speed of 1.5 ml/h. The prepared samples were overnight dried at 60°C to remove the residual DMF. Finally, PAA nanofiber-coated PI nonwoven fabrics were heated to 120 and 260°C at a rising rate of 1°C/min, followed by an annealing at each stage for 60 and 90 min, respectively. Uncoated commercial PI nonwoven fabric was set as blank sample and noted as C0.

OPEN IN VIEWER

Figure 1.

Scheme of preparation of PI nanofiber-coated PI nonwoven fabric filters. ODA: 4,4-oxidianiline; PAA: polyamic acid; PI: polyimide; PMDA: pyromellitic dianhydride.

OPEN IN VIEWER

Table 1.

Monomer amounts for synthesizing precursor PAA solution of each sample.

Sample	PMDA (molar/gram)	ODA (molar/gram)	DMF (ml)
C1	0.01/2.18	0.01/2	25
C2	0.02/4.36	0.02/4	50
C3	0.03/6.54	0.03/6	75
C4	0.04/8.72	0.04/8	100

DMF: dimethylformamide; ODA: 4,4-oxidianiline; PAA: polyamic acid; PMDA: pyromellitic dianhydride.

Characterization

Infrared spectra were recorded on a Nicolet iS 50 spectrophotometer in the transmission model. Electrospun PAA nanofibers from C3 and the corresponding heated counterpart had a spectral result based on 30 scans with a 1 cm⁻¹ resolution across a wavenumber interval between 4000 and 400 cm⁻¹.

Thermogravimetric analysis (TGA) was carried out to study the thermal stability of electrospun nanofibers using a thermogravimetric analyzer equipment (TG209F3, Netzsch). Samples were loaded in alumina pans and heated at a rate of 10°C/min from 30 to 800°C under dry nitrogen atmosphere.

To determine the viscosity of PAA solutions having an ambient temperature (20°C), a Bolin Rheometer with rotational cylinder was being used. The distance between the two metal plates was 0.3 mm and tests were set to 8 min.

The morphologies of electrospun nanofibers were observed by scanning electron microscopy (SEM, Hitachi TM3030) at an acceleration voltage of 10 kV. Prior to its observation, the samples were dried and sputter coated with a thin Pd layer.

Pore size and distribution were analyzed using a TOPAS pore size meter (PSM165) at an air flow of 2000 l/h. Circular samples with diameters of 2.3 cm were held on a metal grid and wetted by the contrast liquid Topor having a surface tension of 16 mN/m.

Filtration properties of samples were measured using a TOPAS filter testing stand (ATC 131). Prior to tests, dioctylmethane aerosol containing PM of 300 mg/m³ was prepared. Circular samples with a diameter of 20 cm were mounted in a testing chamber and the coated face suffered with PM at an air flow of 3.4 m³/h. Filtration efficiency was recorded as the removal capacity of filters to various sizes of particles. Different air flow velocities (0.8–8 m³/h) were used to determine the pressure drop over filters.

Imidization confirmation

Imidization of PAA was confirmed by Fourier-transform infrared spectroscopy (FTIR) analysis and TGA tests. In the spectrum of PAA nanofiber membrane (Figure 2(a)), characteristic bands were observed at 3250, 3050, 1650, and 1602 cm⁻¹, which were attributed to OH stretching (in carboxyl group), NH stretching (in amide group), OH bending, and NH bending, respectively (Socrates, 2001). These bands were hardly identified in the spectrum of heated counterpart. Moreover, new bands appeared at 1372, 1778, and 1717 cm⁻¹, corresponding to C–N stretching in imide ring, asymmetric and symmetric stretching of C = O in imide ring, respectively (Lee et al., 2014; Miao et al., 2013). All these changes in FTIR spectra indicated the imidization of PAA to PI after a heating process.

OPEN IN VIEWER

Figure 2.

(a) FTIR spectra of PAA and PI nanofibrous membranes and (b) TGA and DTG curves of sample C3. DTG: derivative thermo-gravimetry; PAA: polyamic acid; PI: polyimide.

The TGA and corresponding derivative thermo-gravimetry (DTG) curves of the heated nanofibers are shown in Figure 2(b). All the samples displayed similar TGA and DTG patterns. For the clarity, only the results of sample C3 were showed. The membranes decomposed initially at 532°C (T_d5%) and had a highest weight decomposition temperature (T_dmax) as high as 591°C, similar to the results of pure PI fibers reported in previous work (Çakmakçı and Güngör, 2013; Wang et al., 2016). In addition, merely 1% of weight loss occurred in the heated samples at 260°C, implying a relatively high thermal stability of PI nanofibers required for hot gas filtration.

Morphology and pore structure

Four electrospun PI nanofibrous samples displayed various morphologies as shown in Figure 3. This was related with the viscosities of PAA precursor solutions. When a low quantity of PMDA and ODA (0.01 mol) was used, the resultant PAA solution showed a low viscosity (0.92 Pas at a shearing rate of 0.72 1/s), resulting in a breaking of spinning jet and the formation of droplets (Arai and Kawakami, 2012). Therefore, sample C1 showed a structure of beads connected with nanofibers (Figure 3(a)). Increasing the amount of monomers, the synthesized PAA solutions showed higher viscosities, indicating a greater polymerization degree of PAA. The precursor PAA solution for preparing sample C2, C3, and C4 had a viscosity of 1.49, 8.22, and 21.22 Pa s, respectively. Consequently, the density of beads was greatly reduced in sample C2 (Figure 3(b)). When PAA solution was elastic enough (sample C3), nanofibers of homogeneous diameters were generated (Figure 3(c)). However, further increase in viscosity (sample C4) made the solution difficult to be stretched. As a result, nanofibers with larger dimension and uneven distribution were observed in Figure 3(d) (Patanaik et al., 2010).

OPEN IN VIEWER

Figure 3.

SEM micrographs of electrospun PI nanofibrous membranes: (a) C1, (b) C2, (c) C3, and (d) C4.

Figure 4 shows the pore dimension and distribution in each sample. Generally, nanofibers coating led to a decrease in pore size and increased the portion of small pore. The proportion of pores with dimensions less than 10 µm was over 60%. This is because nanofibrous mats covering the PI nonwoven fabrics have much smaller pores between the nanofibers, as seen in their SEM micrographs (Figure 3). However, this phenomenon was not found in sample C1, which contained a majority of pores (56.5%) with dimensions ranging from 14 to 22 µm (Figure 4(b)). In comparison to the blank sample (Figure 4(a)), sample C1 did not show significant changes in pore size and corresponding distribution. It may be related with the morphology of electrospun membrane as discussed previously. Sample C1 was composed by beaded nanofibers (Figure 3(a)) that deposited on the surface of nonwoven fabrics. A low nanofibers network density did not contribute to the change of pore structure of filter.

OPEN IN VIEWER

Figure 4.

Pore size distributions of commercial PI filter (a) C0 and PI nanofiber-coated nonwoven fabrics: (b) C1, (c) C2, (d) C3, and (e) C4.

Air filtration properties

The prepared samples were tested with PM removal efficiency for particles with different sizes (0.25, 0.4, 0.6, 0.85, 1.5, 2.5, and 4 µm). The PM2.5 removal efficiency of commercial PI nonwoven filter was only 81.4% as shown in Figure 5. As compared, sample C1 did not greatly enhanced filtration efficiency (85.2% of PM2.5 filtration efficiency). It is due to the lack of nanofiber web formation on the surface. In contrast, the nanofiber-coated samples (C2, C3, and C4) exhibited significant improvement in PM removal efficiency. The beads appearing in sample C1 had diameters of ∼2–3 µm, which were much larger than the nanofibers (∼300–500 nm, Figure 3) observed in other samples. Herein greater available specific surface areas were thought to be obtained in them. In addition, the nanofiber-coated filters had smaller pores and higher portions of small pores as discussed above. These structures played a crucial role in removing particles from gas (Liu et al., 2015). Lower size of pores and higher amounts of small pores led to an increasing enhancement in PM removal efficiency, as shown in Figure 5. Sample C4 showed the highest PM2.5 removal efficiency (97.2%). Moreover, its filtration efficiencies for 0.25 and 0.4 µm of airborne particles were increased to 18.1 and 41%, respectively. This suggests electropsun PI nanofibers coating is an effective method to improve the filtration efficiency of commercial PI nonwoven filters.

OPEN IN VIEWER

Figure 5.

Filtration efficiency of PI nanofiber-coated commercial PI nonwoven fabrics.

Another important desirable parameter of air filtration is low pressure drop of air flow, which is favorable for maintaining a high filtration flux and saving energy. It was reported that energy consumption was in proportion to the pressure drop through the filters and usually accounted for 70% of the total life cycle cost of the air filters (Li et al., 2014). However, there is usually a trade-off between high removal efficiency and low pressure drop at high air flow velocity. Filtration pressure drops of PI nonwoven filters coated with different morphologies of PI nanofibers under a variety of air flow rates were compared, as shown in Figure 6. To be mentioned, the configurations of coated samples were not changed after filtration test in different wind velocity, indicating the fastness between the commercial PI fabrics and PI nanofibers.

OPEN IN VIEWER

Figure 6.

Filtration pressure drops of PI nanofiber-coated commercial PI nonwoven fabrics at different air flow velocities.

There were no significant differences in pressure drop between the nanofiber-coated samples (sample C2, C3, and C4) and uncoated one. The pressure drop over sample C4 was 82 Pa at a gas velocity of 7.94 m³/h, only 6 Pa higher than the blank samples. Two factors contribute to this result. On the one hand, thickness of nanofibers coating (∼1 µm) is negligible compared to the nonwoven fabric underneath. There is a lot of empty space between the nanofibers. On the other hand, the drag force from nanofibers on the air flow is reduced due to the “slip” effect, thus reducing the pressure drop (Zhang et al., 2016). Sample C2 even had a lower pressure drop than the uncoated one. It may be due to the irregularities in the weights per square meter of nonwoven mats, which are hardly controlled during the practical production. It was noted that sample C1 showed highest pressure drop (92.5 Pa under 7.94 m³/h). The possible reason is the beads from top layer clogged the pore in nonwoven fabrics.