Optimization of electrospinning parameters for polyacrylonitrile-MgO nanofibers applied in air filtration

ABSTRACT

The present study aimed to optimize the electrospinning parameters for polyacrylonitrile (PAN) nanofibers containing MgO nanoparticle to obtain the appropriate fiber diameter and mat porosity to be applied in air filtration. Optimization of applied voltage, solution concentration, and spinning distance was performed using response surface methodology. In total, 15 trials were done according to the prepared study design. Fiber diameter and porosity were measured using scanning electron microscopic (SEM) image analysis. For air filtration testing, the nanofiber mat was produced based on the suggested optimum conditions for electrospinning. According to the results, the lower solution concentration favored the thinner fiber. The larger diameter gave a higher porosity. At a given spinning distance, there was a negative correlation between fiber diameter and applied voltage. Moreover, there were curvilinear relationships between porosity and both spinning distance and applied voltage at any concentration. It was also concluded that the developed filter medium could be comparable to the high-efficiency particulate air (HEPA) filter in terms of collection efficiency and pressure drop. The empirical models presented in this study can provide an orientation to the subsequent experiments to form uniform and continuous nanofibers for future application in air purification.

Implications: High-efficiency filtration is becoming more important, due to decreasing trends air quality. Effective filter media are increasingly needed in industries applying clean-air technologies, and the necessity for developing the high-performance air filters has been more and more felt. Nanofibrous filter media that are mostly fabricated via electrospinning technique have attracted considerable attention in the last decade. The present study aimed to develop the electrospun PAN-containing MgO nanoparticle (using the special functionalities such as absorption and adsorption characteristics, antibacterial functionality, and as a pore-forming agent) filter medium through experimental investigations for application in high-performance air filters.

Introduction

Filtration is one of the most common methods for air purification (Shi et al., 2013). Filtration is “a mechanism or operation for separating substances from a fluid by passing the fluid through a porous medium” (Sutherland, 2011). It is conducted by chemisorption or physisorption techniques for molecular filtration and also by physical mechanisms such as sieving, sedimentation, inertial impaction, interception, diffusion, and electrostatic for particle filtration (Sutherland, 2011). Among the principal types of filter media, including membrane, foam, granular, and fibrous, the latter (i.e., woven and nonwoven fibrous media) are highly popular for removing particles from gas stream (Wang, 2007). They are usually characterized by their collection efficiency, in which the smaller fibers can give better filter efficiency (Li et al., 2013). In general terms, extremely small-diameter fibers of less than 1 micron are referred to as nanofibers (Subramanian et al., 2009). To date, nanofibers, as promising filter media with a much better filtration performance than conventional fibers, have received considerable attention (Shi et al., 2013). Generally, nanofibers are produced by an electrospinning process, which is a remarkably simple method for generating highly functional nanofibers of polymers with the diameter of 50 nm to 5 mm, and introduce many prominent properties and potential applications in different fields, especially filtration (Wang et al., 2013). Electrospun nanofibrous filter media can be used in clean-air processes such as clean-room ventilation and respiratory air filters, since they have a high specific surface area, high porosity, interconnected pore structure with high permeability, potential to incorporate active chemistry, low basis weight, and small fiber diameter (Jackiewicz et al., 2013; Lackowski et al., 2013).

Figure 1. XRD pattern of PAN-MgO nanofibers.

Figure 2. SEM images in 15 trial runs.

The most important characteristic of a filter medium is collection efficiency, which is defined as the fraction of particles trapped by a filter over the total number of particles found in the air upstream of the filter. However, the flow state passing through air filter (or pressure drop caused by filter) is another important factor. So, quality factor (QF) of filter is developed. QF is the indicator of filter performance, considering both collection efficiency and pressure drop. In practice, a filter with high collection efficiency and low pressure drop is considered an effective one (filter with a high quality factor). Normally, the lower packing density, higher porosity, thinner filters, and larger fiber diameter can lead to lower pressure drop, whereas all of these properties lead to lower collection efficiency. Therefore, finding the optimum balance between these two factors is very critical (Huang et al., 2013).

Fine fibers of low diameter allow for rapid removal of dirt and render the medium with high density, small pore size, and high filtration efficiency, whereas the higher porosity (as the ratio of void space volume in a filter medium to the total volume of filter) can decrease flow resistance (Swanson et al., 2013). According to filtration theories, the porosity that would affect dirt holding can be inferred from fiber diameter, so that the number of pores per area of medium is inversely proportional to the quartic root of fiber diameter (Purchas and Sutherland, 2002).

The electrospun nanofiber filter medium has a high potential to be used for filtration, which is comparable to high-efficiency particulate air (HEPA) filter (Li et al., 2013). HEPA filter composed of micron-size fibers (0.5–2.0 microns in diameter) is commonly used in medical, electronic, and industrial processes for arresting very fine particles effectively, and it has a minimum particle collecting efficiency of 99.97% for 0.3 μm (300 nm) (Wadsworth and Hutten, 2007).

Electrospinning for producing nanofiber creates a high-voltage electric field between capillary tip of syringe that contains polymer solutions and grounded collector (Jørgensen et al., 2015). The morphological structure and properties of nanofibers can be changed through the many working parameters during the process. These parameters can be broadly divided into three groups, including polymer solution properties (concentration, molecular weight, viscosity, surface tension, and surface charge density), processing parameters (applied voltage, volume flow rate, collector, and tip-to-collector distance), and ambient conditions (temperature, atmospheric pressure, and humidity) (Han et al., 2013).

Studies show that concentration of solution plays an important role in determining the diameter of fibers (Noorpoor et al., 2014) and consequently porosity of media, so that there is a strong linear relationship between solution concentration and fiber diameter (Li and Wang, 2013). The applied voltage has been found to have a controversial effect on morphologies and fiber diameter, so that some studies have reported that electric field has a negligible effect on fiber size (Li and Wang, 2013) and some researchers have suggested that higher voltages can facilitate narrowing of fiber diameter (Teo et al., 2011). Others have shown that fiber diameter increases with the increase of the applied voltage (Iqbal, 2011). It has been also proven that the nozzle-to-collector distance is an important physical aspect of electrospun fiber, and there is an inverse relationship between the fiber diameter and the distance between the collector and tip of the syringe (Li and Wang, 2013). The three parameters mentioned above are considered to be the crucial factors in electrospinning process, which can induce the various different changes in fiber morphology.

There are many organic polymers that can be electrospun within the nanofibers. Among them, polyacrylonitrile (PAN) as a versatile polymer for its good physical characteristics, reasonable price, and ease of electrospinning has a potential to be applied in different fields of science (Nataraj et al., 2012). Several researchers studied the filtration performance of electrospun PAN nanofiber mats and found that PAN filter media could generally provide an efficient particle collection as high as the conventional HEPA filter (Zhang et al., 2010) (Menkhaus et al., 2010). PAN, as a carbon nanofiber (CNF) precursor, has also attracted attention of many air purification researchers (Nataraj et al., 2012). Incorporating organic/inorganic components such as nanoparticles into the electrospinning solution has also attracted much attention, because it allows for the preparation of hybrid nanofibers with special functionalities (Wang et al., 2009). In air filtration, selection of suitable catalysts embedded in nanofibers can give the filter media a specific absorption and adsorption characteristic and antibacterial functionality (Dadvar et al., 2012). Absorption properties and antibacterial effects of nanofibers containing metal oxide nanoparticles (such as MgO) have been proved in many studies (Behnam et al., 2013; Lange and Obendorf, 2012). Furthermore, the result of previous studies show that adding some inorganic nanoparticles, such as MgO, into the polymer matrix can improve the mechanical strength of membrane and create nanoscale pores (Kim et al., 2012).

Although electrospinning has developed quickly with the occurrence of coaxial, side-by-side and triaxial processes (Yu et al., 2015a, 2015b), the mainstream of this technology is still the single-fluid electrospinning of a working solution containing a guest functional ingredient and a host filament-forming polymer matrix. Therefore, we considered the fabrication and optimization of the mentioned electrospinning parameters for PAN nanofibers containing MgO nanoparticle in air filtration. In our study, we attempted to perform the optimization process using the statistical analysis of experimental data. The empirical models obtained in this study can be used in subsequent experiments to form uniform and continuous nanofibers for future application in air purification.

Materials and methods

Materials

Polyacrylonitrile (PAN) polymer, with an average molecular weight (Mw) of 80,000, was obtained from Polyacryle Co. (Isfahan, Iran), and 99% N,N-dimethyl formamide (DMF) solvent and magnesium oxide (Mw = 40.3 g/mol, size 100 nm and 99% metal basis) were purchased from Merck Co. (Darmstadt, Germany).

Optimization of electrospinning parameters

In order to achieve the desired fiber diameter and porosity as well as uniform and bead-free fibers, optimization of electrospinning parameters, including solution concentration, applied voltage, and tip-to-collector distance, was done and other electrospinning variables were set at constant values in all the experimental runs. Utilization of the experiments design and analysis of the experimental data were carried out by Design-Expert software (version 7; DX7; Stat-Ease, Minneapolis, MN, USA) to predict the optimum condition of the studied factors in electrospinning process. In total, 15 experimental runs were performed (Table 1). Experimental procedure for optimization of the electrospinning process consists of eight steps as follows:

1. Determining the boundary values for studied parameters. The boundary values for the studied electrospinning parameters (i.e., solution concentration, electric voltage, and nozzle-to-collector distance) were obtained through pilot studies to form continuous fibers without breaking up into droplets. They were also set based on the operating conditions of our electrospinning setup.

2. Making the experiment design. In order to assess the influence of process conditions on the fiber diameter and porosity of media and to optimize and predict the average fiber diameter and porosity of electrospun PAN-MgO, experiment design was made via Design-Expert software (version 7). Optimization of electrospinning parameters was conducted through the response surface methodology (RSM) based on central composite design (CCD) to obtain the desired values for response/dependent variables.

Table 1. Design of experiment and average fiber diameter (nm) and porosity (%) of filter media.

	Electrospinning Parameters
Experiment	Concentration	Applied Voltage	Tip-to-Collector Distance	Response Variables
standard no.	(8–16 wt%)	(10–20 kV)	(10-15 cm)	Average Fiber Diameter (nm)	Porosity of Medium (%)
1	16	20	10	263.33	57.87
2	16	10	15	313.58	40.03
3	8	20	15	69.05	29.46
4	8	10	10	46.44	36.53
5	9.6	15	12.5	27.78	38.15
6	14.4	15	12.5	113.39	44.73
7	12	12	12.5	105.51	37.49
8	12	18	12.5	45.36	43.25
9	12	15	11	43.02	51.17
10	12	15	14	98.50	48.09
11	12	15	12.5	54.80	44.23
12	12	15	12.5	68.10	45.69
13	12	15	12.5	84.36	44.23
14	12	15	12.5	72.40	45.69
15	12	15	12.5	52.22	44.23

Determination of optimum levels of working parameters in chemical process by conventional methods is highly time-consuming and requires a large number of experiments. In addition, the conventional methods do not consider the combined effects of all the factors involved in response variable. RSM, as a combination of mathematical and statistical techniques, is suitable where several independent process parameters influence a response variable, and it is usually used for evaluating the relative significance of factors affecting the response variable with a limited number of planned experiments. Typically, it comprises making the experimental design, response surface modeling through regression, and optimization (Lenth, 2009).

• (3) Preparation of electrospinning solution. The PAN-MgO solutions with the weight ratio 3:1 were dissolved in DMF to obtain solution concentrations of 8–16 wt%. Accordingly, the solutions were stirred using a magnetic stirrer (MR Hei-Standard; Heidolph) at room temperature for 20 hr and then by an ultrasonic vibrator (Sono Swiss, SW 6H) for 2 hr.

• (4) Performing the experimental runs. The trials were carried out according to the prepared study design via electrospinning process in the conditions of applied voltages: 10–20 kV, nozzle-to-collector distance: 10–15 cm, temperature: 30 °C, flow rate: 1 mL/hr, collector: covered with aluminum foil, syringe: 5 mL, and needle diameter: 1.2 mm (Table 1). The samples were then kept in a vacuum for 2 hr to be dried at room temperature before scanning electron microscopic (SEM) imaging.

• (5) Taking SEM images. Morphology studies of the fibers were carried out using a scanning electron microscope (Hitachi S 4160) after being gold-coated.

• (6) Measurement of nanofiber diameter. The diameter of electrospun fibers was measured with an image analyzer (Microstructure Measurement, Ferdowsy University). For each trial, the average fiber diameter was determined after about 50 measurements of random fibers.

• (7) Measurement of porosity. Porosity measurement was done using SEM image analysis. Many studies have introduced the image analysis as a reliable technique for characterization of porosity (Ghasemi-Mobarakeh et al., 2007). In the present study, the porosity of filter media was determined by using image analysis algorithms in which input element is a binary image of the medium. Global thresholding is one of the simple ways to create a binary image from a grayscale image where a single constant threshold is used to segment the image. All pixels up to and equal to the threshold belong to the object and the remaining belong to the background. Global thresholding is sensitive to any inhomogeneity in the gray-level distributions of object and background pixels. Therefore, local thresholding scheme could be applied to eliminate the effect of inhomogeneity (Nataraj et al., 2012). Percent porosity was determined as below:

(1)

where n is number of white pixels, N is total number of pixels in the image, and P is percent porosity.

(8) Statistical analysis. The statistical software Design-Expert version 7 (DX7) was used to analyze the direct and interaction effects between parameters. This allowed for offering a polynomial model of independent variables for predicting the response variables and determining the optimum condition for the experimental variables to maximize or minimize the value of dependent ones.

Filtration testing

For air filtration testing, the nanofiber mat was produced based on the suggested optimum conditions for variables of electrospinning to produce nanofiber of desirable fiber diameter and porosity range. Nanofibers for filtration are too thin and usually a supporting structure is required to reinforce their weak mechanical strength (Zhang et al., 2010). Therefore, a nonwoven spunbond polypropylene sheet (Baftineh Co., Iran), with grammage of 18 gsm and thickness of about 100 µm, was selected as substrate for collecting nanofibers, because of its suitable cost and low pressure drop. Five layers of substrate coated with nanofibers were compressed to form a single mat for testing. HEPA filter performance test and pressure drop testing for both fabricated mat and substrate were conducted using step aerosol photometer (DOP Solutions Ltd., UK) and pressure drop test setup penetrometer (Beasat Industrial Complex, Iran) in accordance with MIL-STD-282 (Department of Defense, 1956), respectively. The step aerosol photometer can continuously generate monodispersed 0.3-µm DOP (dioctyl phthalate) particles in 100 to 0.00001 µg/L mass concentration and flow rate nominally 28 L per minute. It is equipped with a scanning probe that is placed on the surface of understudy filter. The photometer is comparing the real-time concentration of the upstream and downstream. It can provide the digital read-out for the mass concentration or penetration percent. It indicated that the results of HEPA filter performance test were in compliance with acceptance criteria: efficiency (%) ≥99.975 or penetration (%) ≤0.025 through a visual and audible alarm (British Standards Institution, 2009).

The pressure drop across the filter at the nominal air volume flow rates was measured before the filter is loaded with test aerosol. The pressure drop testing apparatus is operated by the extraction pump and completed by devices to measure and regulate the air volume flow rate (International Organization for Standardization, 2003).

Results and discussion

Morphology of fibers

The X-ray diffraction (XRD) pattern of the hybrid nanofibers is shown in Figure 1 to show the coexistence of MgO with PAN in nanofibers. The SEM images in Figure 2 show the morphology of the fibers obtained from different experimental runs. As can be seen in Table 1, the maximum average fiber diameter is obtained in experiment standard 2 (STD 2), with the concentration being 16 wt%. The minimum average fiber diameter belongs to STD 5 (concentration 9.6 wt%, voltage 15 kV, and distance 12.5 cm).

The highest bead (solidified droplets) number of fiber was related to experiment standards 3 and 4. The lowest concentrations (8 wt%) had the highest tendency to form beads. Beads are generally formed because of aggregation of solvent molecules. At high solution concentrations and more chain entanglement, solvent molecules are distributed among entangled chains and their tendency for congregating decreases (Pilehrood et al., 2012). This means that lower concentrations favor smaller diameter fibers, but not necessarily the higher-quality fibers. In line with our findings, Tan et al. (2013) concluded that diameter of PVP/TiO2 (polyvinylpyrrolidone/titanium dioxide) composite nanofiber increases with the increase of PVP polymer concentration (Tan et al., 2013). Gu et al. (2005) reported that spindle-like beads formed for fibers were obtained from lower electrospun PAN solution (Gu et al., 2005). They also found that the average fiber diameter was much larger at higher concentrations than at lower concentrations, which is consistent with the results obtained in the present study. This was attributed to the fact that higher solution concentration would have more polymer chain entanglements and less chain mobility, leading to harder jet extension and higher disruption during the electrospinning process and favoring the formation of a large-diameter fiber (Hasanzadeh et al., 2013). A mixture of beads and fibers will be commonly obtained, when polymer chain entanglement is insufficient (happened in low solution concentration).

Maximum porosity was achieved in STD 1. A weak positive correlation was seen between average nanofiber diameter and percent porosity (Pearson correlation, r = 0.29; IBM SPSS Statistics 22; IBM, Armonk, NY, USA), so that the larger fiber diameter gave the higher porosity. Dadvar et al. (2012) revealed that adding MgO nanoparticles (wt%) to PAN solution can result in higher total pore volume of electrospun media compared with the metal oxide–free ones (Dadvar et al., 2012). Fiber diameter has a major role in forming the pore structure (Bagherzadeh et al., 2014). However, previous works did not specifically report on the effect of PAN-MgO nanofiber diameter on porosity obtained by SEM image analysis algorithm. For example, Bagherzadeh et al. (2013) conducted a three-dimensional pore structure analysis of nano/microfibrous scaffolds using confocal laser scanning microscopy (CLSM) (Bagherzadeh et al., 2014). They showed that lower electrospun concentration of polycaprolactone can lead to decrease in average fiber diameter, increase in total fiber volume, and therefore decrease in percent porosity measured by CLSM. Kwon et al. (2005) also found that decrease of fiber diameter of electrospun nano- to microfiber fabrics of poly l-lactide-co-ɛ-caprolactone can decrease the porosity obtained by mercury intrusion porosimetry (Keun Kwon et al., 2005). It’s worth to add that image analysis algorithm measures porosity in the cross-section of surface layer, not in the total volume of mat.

Screening of main effect and interaction of factors on average fiber diameter

Table 2 shows the individual effect of solution concentration, applied voltage, and spinning distance on average fiber diameter, while keeping two other parameters constant. The relationship between input variables and response variable was shown using correlation coefficient. Solution concentration has the highest positive correlation with fiber diameter and the strongest effect on fiber size compared with other independent variables (r = 0.77, P < 0.05). As can be seen, the lower concentration gives lower fiber diameter. This might be due to the fact that the higher solution concentration would give the liquid a higher viscosity, so that it resists solution jet elongation and gets thinner (Wong, 2010). These findings are consistent with the results of many studies and indicate that polymer solution plays the most important role in determining the fiber diameter (Hasanzadeh et al., 2013; Amiraliyan et al., 2009). Fiber diameter was inversely related to applied voltage and positively related to spinning distance (r = −0.09 and r = 0.16, respectively). In this study, a very weak negative relationship was detected between fiber diameter and applied voltages. The increased applied voltage can enhance the electric field strength, therefore creating a higher electrostatic repulsive force on the polymer jet and decreasing the fiber diameter. Besides, a higher surface charge removes the solution more quickly from the tip of needle and increases the fiber diameter (Hasanzadeh et al., 2013; Ziabari et al., 2009). A nonsignificant, weak, and positive correlation between fiber diameter and electrospinning distance was also observed in this study (r = 0.16, P > 0.05). Previous studies have demonstrated that increasing the tip-to-collector distance can give more time for the solvent to evaporate, which can result in the thinner fiber diameter (Hasanzadeh et al., 2013; Zhang et al., 2005), which is inconsistent with our finding. In general, it can be concluded that the influence of applied voltage and spinning distance was not as significant as that of solution concentration. At the specified applied voltage and distance, fiber diameter is strongly associated with concentration, which is in line with the results obtained by Boland et al. (2001).

Table 2. Correlation coefficients between electrospinning parameters and response variables.

	Response Variables
Electrospinning Parameter	Porosity	Diameter
Solution concentration (wt%)	0.66*	0.77**
Applied voltage (kV)	0.26	−0.09
Electrospinning distance (cm)	−0.49	0.16

Note. *Correlation is significant at the 0.01 level. **Correlation is significant at the 0.05 level.

There was a positive relationship between solution concentration and percent porosity when keeping applied voltage and spinning distance constant in a way that increase in concentration leads to increase in porosity (r = 0.66, P < 0.01). Applied voltage and percent porosity also had a positive correlation (r = 0.26, P > .05) (weaker than the relationship existing between concentration and porosity). The correlation between spinning distance and percent porosity was negative (r = −0.49, P > .05). Air permeability of electrospun polyacrylonitrile nanoweb was studied by Abuzade et al. (2012). They concluded that higher polymer concentration at constant value of applied voltage and spinning distance can result in higher percent porosity and thereby higher air permeability per unit web weight (Abuzade et al., 2012). Kwon et al. (2005) found that the percent porosity measured by mercury intrusion porosimeter increases with the increase in average diameter of electrospun PLCL (poly-lactide-co-ε-caprolactone) fibers (Keun Kwon et al., 2005). Ziabari et al. (2008) studied the parameters affecting pore structure of electrospun PVA (polyvinyl alcohol) nanofiber and found that the increase of applied voltage and spinning distance when solution concentration was fixed at given value resulted in increase of the percent porosity measured by image analysis (Ziabari et al., 2008).

In order to consider the interaction effects of input variables on fiber diameter and porosity, the model suggested by analysis of variance (ANOVA) is presented in Table 3. Insignificant terms not included in the models are aliased according to the suggestion of DX software.

Table 3. Response to surface quadratic mode of diameter and porosity.

Response Variable	Model	Equation
Diameter (nm)	Quadratic	Diameter = −485.51 + 71.84A* − 90.59B** + 113.65C*** − 4.32AB + 1.09AC – 3.52BC + 4.15B^2
Porosity (%)	Quadratic	Porosity = +234.32 + 10.31A + 13.81B − 57.86C + 0.18AB – 0.19BC – 0.48A^2 − 0.43B^2 + 2.4C^2

*A = concentration; **B = voltage; ***C = distance.

Model fitting was done with the help of DX software. It was revealed that a quadratic model for fiber diameter and porosity gave the best fit, and the models were found to have insignificant lack of fit (Kohli and Singh, 2011). The nonsignificant values of lack of fit for the models showed that the developed models were valid (Ramakrishna and Susmita, 2012).

According to the ANOVA results for the response surface quadratic model for fiber diameter, all of the independent variables and some of their interactions (AB, AC, BC, and B²) have a statistically significant impact on average diameter (P < 0.05). Gu and Ren (2005) studied the process optimization and empirical modeling for electrospun poly-(d,l-lactide) fibers using response surface methodology. They reported that concentration, applied voltage, and their interaction had a significant impact on average fiber diameter (at a fixed distance of 15 cm) (Gu and Ren, 2005). The production optimization of polyacrylonitrile electrospun nanofibers conducted by Hasanzadeh et al. (2005) revealed that only the P values of concentration (X₁), voltage (X₂), distance (X₃), X₁X₂, X₁², and X₂² in the quadratic model were significant (P < 0.05) (Hasanzadeh et al., 2013). The observed differences can be ascribed to the different polymer solution properties (e.g., polymer rheology and solvent thermodynamics properties), process parameters (e.g., flow rate and collection method), and ambient conditions (e.g., temperature and humidity) (Gu and Ren, 2005). The ANOVA results of porosity for the response surface quadratic model indicated that all of the studied variables and their interaction, except for AC, have a statistically significant effect on porosity.

To the best of our knowledge, there are not enough observations on optimization of electrospinning parameters for porosity of mat to make definite conclusions. Moreover, the differences between the results can be attributed to the types of experimental design and different methods and material used.

Determination of optimal conditions

DX software can help in prediction of the optimum combination of input variables to maximize, minimize, or attain a specific target for response variables. Table 4 presents some of the optimum solutions for different requirement values of fiber diameters and porosity, as the goal of optimization is set in range. Results of the present study indicate that the studied response variables always have no direct relationship and sometimes the morphological quality of fibers is lower in smaller diameters. Therefore, in the optimization process by DX, the response variable is assessed “in range,” instead of minimization or maximization. Among the different suggestions presented by DX, those combinations that can give thinner fiber, higher morphology quality, and higher porosity can be chosen (Table 4). Desirability parameter in Table 4 is an objective function, ranging from 0 to 1. For the numerical optimization, a point with the maximum desirability function is found.

Table 4. Some of the optimum solutions for different requirement levels.

Variable	Goal	Lower Limit	Upper Limit	Lower Weight		Upper Weight
Solution concentration (wt%)	In range	8	16	1		1
Applied voltage (kV)	In range	10	20	1		1
Tip-to-collector distance (cm)	In range	10	15	1		1
Diameter (nm)	In range	27.78	313.58	1		1
Porosity (%)	In range	29.46	57.87	1		1
Solutions
Solution Number	Concentration(wt %)	Voltage(kV)	Distance(cm)	Diameter(nm)	Porosity (%)	Desirability
1	14.40	15.00	12.50	116.94	45.01	1
2	12.00	15.00	11.00	46.58	51.45	1
3	8.00	10.00	10.00	45.81	36.48	1
4	12.00	15.00	12.50	62.63	44.52	1
5	9.60	15.00	12.50	31.33	38.43	1

High quality factor of filter is achieved by high collection efficiency and low pressure drop. Higher porosity and larger fiber diameter can lead to lower pressure drop, and also negatively impact on the collection efficiency. Therefore, determination of the optimum points among the effecting parameters is so important (Huang et al., 2013). Wang et al. (2008) showed that filter efficiency had a negative exponential relationship with the fiber diameter and a positive exponential relationship with solid volume fraction (equal to one minus the porosity), and filter pressure drop had a negative exponential relationship with porosity and fiber diameter (Wang et al., 2008).

Confirmation experiments

The regression equations was validated by performing the new experimental runs (three runs for each solution) according to the optimal condition obtained by regression equations and then by comparing the experimental data with the data suggested by DX software. The results indicate that experimental values and model values are closely correlated with each other, and the experimental values are in agreement with the predicted responses (percentage variation <1%). These validate the regression equations developed.

Filtration testing

Results of HEPA filter performance and pressure drop test are tabulated in Table 5 for a typical electrospun filter medium prepared by the recommendation of optimization process. As can be inferred from Table 5, the collection efficiency of the filter medium is comparable to that of HEPA filter; it means that the fabricated filter medium could obtain efficiency of 99.97% for collecting the 0.3-µm particles. Moreover, it provides lower pressure drops at any nominal flow rate in comparison with conventional HEPA filter (e.g., pressure drop 132 vs. 300 Pa for flow rate of 20 L/min). It has been noted that the high collection efficiency and low air resistance are the two main features of an effective filter.

Table 5. Results of HEPA filter performance and pressure drop test.

		Pressure Drop (Pa)
Filter media	Electrospinning Conditions	Flow Rate:20 L/min	Flow Rate:40 L/min	Flow Rate:60 L/min	Flow Rate:80 L/min	Collection Efficiency Compared With HEPA filter
		132	282	440	602	Pass
Electrospun nanofibrous membrane (10 layers)	Solution concentration: 12wt.%
	Applied voltage: 15 kV
	Spinning distance: 13 cm
	Time: 10 min
Substrate		7	17	31	48	Fail

Conclusion

An experimental study was carried out to assess the electrospinning parameters in order to produce MgO nanoparticle–embedded PAN nanofibers for future application in air purification. The aim of this study was to provide a prediction scheme for domain of the intended parameters in which the targeted PAN-MgO fiber diameter and porosity with the high quality morphology of fibers could be achieved. The effects of processing variables, including polymer concentration (wt %), applied voltage (kV), and nozzle collector distance (cm), on the average diameter and porosity of electrospun nanofibers were investigated using the response surface methodology featuring and regression analysis. Morphological assessment of electrospun fibers indicated that lower solution concentration favors the thinner fiber, but not necessarily the higher-quality morphology. Regression analysis emphasized that solution concentration is a significant factor that affects the average nanofiber diameter and porosity. Porosity of media was directly related to fiber diameter, and the larger diameter of PAN-MgO gave the higher porosity. High regression coefficients between the independent variables and the responses indicated excellent evaluation of experimental data by polynomial regression model. Generally, the response surface methodology could satisfyingly determine the collaborations between the input variables and outputs. Confirmation experiment indicated that the experimental values were in a good agreement with the predicted ones. The optimal conditions were achieved for predetermined diameters of electrospun fibers, and they were in accordance with the previous studies. According to the obtained optimization results, we were able to develop the filter medium comparable to HEPA filter in terms of collection efficiency and pressure drop. In general, for reaching the desired and optimum characteristics of a nanofiber filter medium, the right selection of electrospinning conditions is so crucial. The empirical models developed in this study can provide an orientation to the subsequent experiments to form uniform and continuous nanofibers for future applications in air purification.

Acknowledgment

This study was part of a Ph.D. dissertation and a research project approved by Tehran University of Medical Sciences (ID: 92-01-27-2180). The authors also thank Fanavaran Nano-Meghyas R&D Co. for its helpful assistance in the electrospinning process.

Additional information

Notes on contributors

Somayeh Farhang Dehghan

Somayeh Farhang Dehghan is a Ph.D. candidate in occupational health at the School of Public Health, Tehran University of Medical Sciences, in Tehran, Iran.

Farideh Golbabaei

Dr. Farideh Golbabaei is a professor in the Department of Occupational Health at the School of Public Health, Tehran University of Medical Sciences, in Tehran, Iran.

Bozorgmehr Maddah

Dr. Bozorgmehr Maddah is an associate professor in the Department of Chemistry at Imam Hossein University in Tehran, Iran.

Masoud Latifi

Dr. Masoud Latifi is a professor in the Department of Textile Engineering at Amirkabir University of Technology in Tehran, Iran.

Hamid Pezeshk

Dr. Hamid Pezeshk is a professor in the Department of Mathematics at University of Tehran in Tehran, Iran.

Mahdi Hasanzadeh

Mahdi Hasanzadeh is a Ph.D. candidate in the Department of Textile Chemistry Engineering at University of Guilan in Rasht, Iran.

Farhang Akbar-Khanzadeh

Farhang Akbar-Khanzadeh is a professor in at the Department of Public Health and Preventive Medicine, University of Toledo, Toledo, Ohio, USA.