Nanofiber Engineering of Microporous Polyimides through Electrospinning: Influence of Electrospinning Parameters and Salt Addition

The electrospinning of high-performance polyimides (PI) has recently sparked great interest. In this study, we explore the effect of the electrospinning parameters — namely polymer concentration, voltage, tip-to-collector distance and flow rate — and salt addition on the diameter, morphology, and spinnability of electrospun PI nanofibers. Three different polyimides of intrinsic microporosity (PIM-PIs) with high Brunauer–Emmett–Teller (BET) ranging from 270 to 506 m 2 g -1 , and two microporous polyimides, were synthesized through the polycondensation of 4,4′-(hexafluoroisopropylidene)diphthalic anhydride (6FDA) and aromatic diamines. The addition of tetraethylammonium bromide (TEAB) salt considerably increased the conductivity of all the PI solutions, significantly improved spinability, and resulted in thinner fibers. We also used molecular dynamic simulations to investigate the macromolecular mechanism of improved spinnability and fiber morphology in the presence of an ammonium salt. The small droplets detached from the parent droplet, followed by the rapid evaporation of the ions through the hydration effect, which facilitated the electrospinning. The resulting uniform nanofibers have great potential in environmental applications due to the presence of microporosity and hydrophobic pendant trifluoromethyl groups, which enhance the sorption performance of the fibers for hydrophobic species.


Introduction
Electrospinning is a method for generating fibers through the formation of a jet from a charged polymer or non-polymeric system under an electrical field [1]. The resultant fiber morphology and diameter vary depending on the intrinsic characteristics of the polymer, the solution properties, and the electrospinning parameters, as well as the used spinneret system [2,3]. Various nanostructures, from beads to bead-free fibers, can be produced by tuning these parameters through a shift from electrohydrodynamic spraying to spinning [4]. While fibers are suitable for applications that require flexibility, such as sorption or wound dressing, beads are mostly used for drug delivery. Beaded-fibers are formed by the capillary breakup of the electrospinning jet by surface tension, which reduces the mechanical properties of the resultant fibers. Thus, the engineering of fiber morphology (i.e., controlling the fiber diameter and morphology, as well as spinnability) is a technological bottleneck for the development of electrospun materials.
Polyimides of intrinsic microporosity (PIM-PIs) are porous polyimides with Brunauer-Emmett-Teller (BET) surface area exceeds 200 m 2 g -1 and contain pores with pore size below 2 nm [5,6]are highperformance polymers that are synthesized by a conventional cycloimidization reaction between dianhydrides and diamines ( Fig. 1) [7]. In this regard, 4,4′-(Hexafluoroisopropylidene)diphthalic anhydride (6FDA) has been widely used as a building block for the synthesis of PIM-PIs since it allows producing solution-processable polymers [8]. These PIM-PIs have gained considerable interest in the fields of gas and liquid separations, heterogeneous catalysis, and adsorption-based applications, such as water treatment and hydrogen storage, due to their high surface area, chemical and thermal stability, and tunable molecular structures [9].
PIM-PIs can be electrospun from their concentrated solutions in organic solvents [10], and the resultant nanofibrous mats are employed mainly for oil adsorption [11], lithium-ion batteries [11,12], supercapacitors [13], and protective clothing applications [14]. We recently developed hierarchically porous fluorinated polyimide-based nanofibrous sorbents by electrospinning, which have been deployed in an oil spill remediation scenario to limit ecological damage, as well as being used for the removal of non-polar solvents [15]. Particularly, the presence of microporosity in nanofibers endows them great application potentials in adsorption, separation, catalysis since the microporous network can allow diffusion of molecules of interest. For further reading, a comprehensive review of electrospun polyimide nanofibers has been published [16]. The spinnability of polymers is highly dependent on the properties of their solutions, such as polymer concentration and solution conductivity. The uniformity of the electrospun fibers can be tuned by adjusting the electrospinning process parameters to produce uniform nanofibrous structures in a narrow size range [17][18][19][20][21]. Likewise, the addition of salt significantly improves the spinnability of polymer of intrinsic microporosity-1, [22] poly(3-hydroxybutyrate-co-3-hydroxyvalerate) [23] and polystyrene solutions [24]. Improved fiber morphology and spinnability with the addition of salt has been attributed to the accelerated evaporation rate of the water [25].
In this study, we investigate the influence of the electrospinning parameters and salt addition on the morphology and diameter of the resulting electrospun nanofibers of PIM-PIs (Fig. 1). All of the studied polymers exhibit porosity, and high BET surface with exceptions for 6FDA-mPDA and 6FDA-DABA.

Characterizations.
The electrospun fibers was analyzed by scanning electron microscopy (SEM, Magellan FEI). Before analysis, the fibers were coated with Pt (5 nm) using a K575X sputter. The mean fiber diameter was calculated from the respective SEM images over approx. 100 fibers using ImageJ (NIH, US). The infrared spectra of the polyimides were recorded for powder samples using a Varian 670-IR ATR-FTIR spectrometer. The spectra were recorded as the average of 128 scans at 4 cm -1 resolution.
1H Nuclear magnetic resonance (NMR) spectra of the polyimides were recorded on a 500 MHz Bruker AVANCE-III spectrometer in deuterated dimethylsulfoxide (DMSO-d6) solvent, and each spectrum was the sum of 128 scans. Number average molecular weight (Mn), molecular weight (Mw) and polydispersity index (PDI) of the polyimides were obtained by gel permeation chromatography (GPC) in an Agilent 1200 system using polystyrene standard calibration and DMF as the eluent. The thermal stability of these polymers was evaluated using a thermal gravimetric analyzer (TGA) and a Model Q5000 system (TA Further purification of the polymer powder was achieved by re-precipitation of the polymer solution in methanol. The polymers were characterized by 1 H NMR, FTIR, BET, and TGA (Figs. S1-4). The physical characteristics and solubility of the polyimides are given in Table 2 and Table S1.
6FDA-mPDA ( Table 1). The solutions were then transferred into 1 mL disposable plastic syringes (HSW, Henke Saas Wolf GmbH) equipped with metallic needles (23G x 1'', monojet, standard hypodermic needle, Covidien). Afterward, the syringes were horizontally secured in a Legato TM syringe pump in a closed chamber made of Plexiglas. The fibers were gathered on a 100 mm × 100 mm metal plate covered with a piece of aluminum foil. The experiments were carried out at a relative humidity of (50±5) %. During the electrospinning, the temperature was (21±1) o C. a With respect to the polymer content.

Molecular Dynamics Simulations.
The electrospinning process of the 6FDA-TrMPD polymer and the effect of the TEAB salt were studied using molecular dynamics simulations. We applied the all-atom OPLS-AA force field to describe the polymer, TEAB salt, and the DMF solvent [26]. This force field has been successfully applied to describe the properties of DMF in previous studies [27,28]. The simulations also include an external electric field [29], and the interactions within the polymer [30]. Electrostatic potential fitted atomic charges (CHELPG) [31] were assigned based on quantum chemical computations employing the ωB97M-V [32] density functional and the def2-TZVP [33] basis set, as implemented in the Q-Chem 5.3 [34]software.
We performed the molecular dynamics simulations using the LAMMPS code [35]with timestep of 0.25 fs. Constant pressure simulations were performed using Nose-Hoover barostat, employing the Martyna-Tuckerman-Klein formalism [36]. The time constants were 100 fs and 1000 fs for the temperature and pressure, respectively. The atomic coordinates for the initial configuration were set up using the Packmol [37] and the Moltemplate [38] software suites, while the Ovito [39] program was used to visualize the results. Constant temperature simulations were performed using the Berendsen thermostat [31].
The initial droplet configuration consisted of a polymer of 30 monomer units, which were folded by simulated annealing and 1448 DMF solvents, while 4 TEAB molecules were also added to investigate the salt effect. The droplet was pre-equilibrated at 300K and was immersed in 3000 N2 molecules in a 30·30·100 nm periodic simulation box (Fig. S5). The whole system was equilibrated for 300 ps at 300K and 1 atm.
A unidirectional electric field of 0.8 V/nm was applied in the electrospinning simulations, and the nitrogen bath gas was thermostated at 600K to provide heat for the evaporation [25]. The applied electrical field and the temperature are certainly higher than those in the experimental conditions, but they were necessary due to the available time scales in the molecular dynamics simulations. Nevertheless, a previous study has shown that these settings give realistic results [25].
We carefully selected the diamines to show the difference between polyimides bearing functionalities and non-functionalized polyimides and to explore the effect of using planar versus contorted diamines.
The obtained polyimides had different BET surface areas, ranging from 36 to 506 m 2 g -1 , with the highest surface area for 6FDA-DMN, which was constructed from the contorted DMN ( Table 2).
The presence of methyl groups at the ortho position to the imide linkage can lock the imide bond rotation and produce a more rigid polymer backbone, which leads to an increase in the BET surface area and fractional free volume (FFV) [40].  Measured by GPC in DMF with polystyrene as the calibration standard. b 5 wt% weight loss decomposition temperature measured by TGA by a ramp rate of 5 o C min -1 to 800 o C (Fig. S3). c Measured from the N2 adsorption isotherm, which was obtained from Micrometrics ASAP 2020 at -196 o C (Fig. S4).

Electrospinning: Influence of Process Parameters and Salt Addition
First, we found that the optimum polymer concentration produced fibers by screening the  [4]. Similar findings have been reported for the electrospinning of poly(vinyl alcohol) [41] and polycaprolactone [42].
We also studied the influence of the flow rate on the fiber morphology and diameter. Usually, a higher flow rate results in thicker fibers due to the higher mass flow, and, above a critical value, the formation of beaded fibers may occur due to unstable jet formation. Fig. S8 shows  (w/v) resulted in beaded fibers, while the incorporation of 10 wt% TEAB directly influenced the fiber morphology, and bead-free fibers were produced. Even at lower concentrations, a transition from beads to beaded fibers was observed, demonstrating the improved spinnability in the presence of salt.   We observed a similar trend for the enhanced spinnability of 6FDA-DMN, which formed beads at a concentration of 5% (w/v) (Fig. S12). In contrast, we observed beaded-fibers after the addition of 10 wt% TEAB. Increasing the polymer content further to 10% (w/v) led to thick fibers, whereas the addition of 10 wt% TEAB resulted in thinner fibers. The addition of TEAB clearly improved the spinnability and drastically reduced the fiber diameter. For instance, the mean diameter of the fibers spun at 10% (w/v) decreased from 365±165 to 190±80 nm with the addition of 10 wt% TEAB. Likewise, a three-fold decrease in the diameter was observed for the fibers spun at 15% (i.e., from 940±205 to 340±140 nm) with the salt incorporation.
The impact of salt addition on the electrospinning of the 6FDA-TrMPD was more pronounced. We performed the electrospinning of the 6FDA-TrMPD of various concentrations in the presence of 10 wt% TEAB (with respect to the polymer) (Fig. 5). At 5 and 7.5% (w/v) of the 6FDA-TrMPD, beaded-fibers were obtained in the absence of TEAB, whereas increasing the polymer content to 10% (w/v) led to beaded-fibers. A further increase in the polymer concentration to 15% (w/v) led to thick fibers of 360±75 nm. On the other hand, the addition of TEAB significantly improved the spinnability and led to the production of fibers, even at 5% (w/v), which is the lowest polymer concentration that yielded fibers among all the polymers screened. Furthermore, because of the low polymer concentrations, ultrathin fibers were formed with a mean diameter of 40±11 nm in the presence of 10 wt% TEAB. In contrast, a two-fold increase in the polymer concentration increased the fiber diameter to 90±60 nm, while the mean diameter of the salt-free fibers was 190±65 nm. At very high polymer concentrations (i.e., 15% (w/v), the influence of TEAB addition was relatively reduced, and the mean diameter of the fibers decreased slightly from 360±75 to 320±65 nm with 10 wt% TEAB.

Molecular Modeling Studies
The effect of salt addition on the electrospinning of polyimides was explored through molecular dynamics (MD) simulations. The size and shape evolution of the droplet with and without TEAB salt is depicted in Fig. 9. We investigated the droplet size by clustering and counting the corresponding atoms with a 5 Å cutoff radius, while we used the radius of gyration (RG, Eq. 1) to analyze the spatial extent of the droplet.
where mi and ri are the mass and distance from the center of mass of each atomic nuclei, respectively.
We investigated the particle shape using the covariance of the atomic coordinates measured from the center of the atomic coordinates, as it is implemented in the Molmod library. We computed the shape factor (S) as defined in Eq. 2.
where λ 1 , λ 2 and λ 3 are the eigenvalues of the covariance matrix in descending order. Averaging of the smaller eigenvalues was specifically defined due to the ellipsoid shape of the particles. S=1 indicates a spherical particle, which increases with its distortion into an ellipsoid shape. Fig. 7a shows a remarkable increase in the droplet evaporation rate with the addition of TEAB salt.
The electrospinning process begins similarly, irrespective of the salt addition. The radius of gyration ( Fig. 7b) indicates the spatial increase of the droplet, while the shape factor (Fig. 7c) clearly shows that this was caused by the distortion of the elongated ellipsoid shape due to the external electric field. The extent of the distortion was smaller when TEAB was added to the system, which we postulate was the result of the higher evaporation rate.

Fıg. 7.
Evolution of the droplet size and shape through molecular dynamics simulations: a) the number of atoms in the droplet; b) radius of gyration; c) shape factor.
However, the evaporation mechanism depends considerably on the presence of the TEAB salt. With TEAB salt addition, the electrical field accelerates the ions, and DMF was removed rapidly in the form of secondary droplets from the tip of the ellipsoid shape particle (Fig. 8a). On the contrary, without TEAB salt, DMF was evaporated at the molecular level (Fig. 8b), which explains the remarkable difference in their evaporation rates. Similar effects have been observed in the case of a polyethyleneglycol/H2O/NaCl system. [25] Interestingly, the addition of TEAB salt resulted in a less compact polymer structure after the electrospinning process (Fig. 8c) compared to that without TEAB salt (Fig. 8d). The interactions between the polymer and the salt lead to a less compact polymer structure because the electrical field accelerates the ions, and thereby extends the polymer chain. In this regard, a similar finding was reported for the electrospinning of polyethylene oxide in the presence of sodium chloride [25]. The authors performed molecular dynamic simulations and found that some sodium ions coordinate with the ether oxygen group in the PEO chain. Under the electrical field, these ions are accelerated, resulting in the stretching of PEO chains while improving the fiber morphology. Furthermore, this supports the fast evaporation of solvent molecules from the fiber matrix as a result of the formation of wide pores.

Conclusion.
The influence of process parameters and salt addition on the electrospinning of five different porous polyimides was studied. The fiber morphology and diameter were affected by the electrospinning parameters, and their adjustment enabled the formation of thinner and uniform nanofibers. The impact of salt addition on the spinnability and fiber morphology was substantial, boosting the solution conductivity and enhancing the spinnability. With the addition of TEAB salt, the formation of fibers even at low polymer concentrations could be succeedded. In contrast, at the same concentrations in the absence of salt, electrosprayed beads were formed. The higher salt content decreased the fiber diameter and led to nanofibers as thin as 40±11 nm. We observed the most considerable changes for the electrospinning of 6FDA-TrMPD. Molecular dynamics simulations revealed the detachment of secondary droplets from the primary droplet due to the evaporation of ions and the hydration effect.
This phenomenon rapidly accelerated water evaporation and therefore improved the spinnability of the solution. Our results reveal that when salt ions are accelerated by an electric field, the polymer chain follows the motion of the ions, and ultimately improving the fiber morphology. These uniform, porous nanofibers featuring pendant trifluoromethyl groups have a high potential to be employed in various applications, including water treatment. Furthermore, the presented approach can be employed for the electrospinning of other polymers to produce uniform thin fibers at lower polymer concentrations.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.