Optimization of meta-aramid electrospun nanofibers productivity through wire-based electrospinning setup scale up

Abstract

Electrospinning process has been widely used over the past decades for manufacturing nanofibers. The control of the electrospinning parameters is crucial to obtain nanofibers (nonwoven mats) with optimized morphological properties. The aim of this study is to precisely define the electrospinnability of a meta-aramid solution through wire-based electrospinning setup processing. Experiments have been conducted following a design of experiment to study the influence of each parameter. Individual effects and/or combined interactions on obtained fiber diameter and general morphology (mean fiber distribution and nonfibrous area) have been investigated. The five studied process parameters are: applied voltage, relative humidity, temperature, distance between spinning electrode wire and substrate material, and airflow going through the spinning chamber. Each parameter was varied at three levels. Significant effects of parameters have been observed. The obtained results have allowed us to determine the influential factors and reduce the domain study.

Keywords

Nanofibers electrospinning fibrous materials technical nonwoven fabrics processing fabrication

Introduction

Electrospinning is a well-known technology for the production of electrospun fibers (ultra-fine fibers) whose diameter ranges from 10 to 2000 nm. Fundamental reviews for an understanding of the electrospinning process have been published. Nanofiber properties and their main application domains have been investigated by several authors, among whom those of Reneker et al. [1], Ramakrishna et al. [2], and Chronakis [3] may be the most exhaustive.

Due to their long lengths, nanoscale diameters, and large surface area per unit of mass, nanofibers have found the widest interest in filtration. The use of those materials can enhance the efficiency of conventional filters in the nonwoven industry, particularly for ultrafine filtration purpose. Since the 1990s, versatile single-needle laboratory-scale setups were sufficient to conduct the experimental research, with an average throughput around 0.1–1 g/h by fiber weight [4]. An annual laboratory-scaled production can be estimated between 300 g and 3 kg per year, which cannot be compared to the millions of tons produced by conventional spinning of fibers. The last report dated from May 2013 of BBC Research announced that the global market for nanofibers “is forecast to grow at a compound annual growth rate (CAGR) of 30.3% from 2012 to 2017, reaching a global revenue of $570.2 million by 2017” [5]. Following this increasing industrial demand, huge efforts have been made on scientific research in the last decades to enhance the electrospinning productivity.

To increase the electrospinning production, two methods have been developed based on multi-jet electrospinning: multineedle (MN) and needleless electrospinning (NE). Both methods are of interest, but for this study, only NE (also called free surface electrospinning) approach will be further developed. Concerning NE, a first method of electrospun mass production was presented by Yarin and Zussman [6], based on a combined effect of normal magnetic and electric field upon two distinct layers of ferromagnetic fluid and polymer solution. The second method was developed by Jirsak et al. [7] from the technical university of Liberec. In this needleless method, patented in 2005 under the name “roller electrospinning” and commercialized by Elmarco company under the name “Nanospider” since 2005, a slowly rotating roller is partially immersed in polymer solution. This equipment is designed for continuous nanofiber layer production. The roller is connected to a high voltage source and an upper collector is grounded or negatively charged. During the electrospinning process, polymer solution is coated on the surface of the rotating roller. When suitable voltage is achieved, numerous solution jets are simultaneously initiated on the top surface of the roller producing nanofibers that reached the collector. A study conducted by Petrik and Maly [8] in 2009, comparing MN and NE shows that the space and the number of jets set up spontaneously in their optimal configuration, which is a major advantage for this technique. By comparison in the MN systems, needle’s optimal configuration has to be precisely adjusted, and needles are submitted to clogging effects. To improve the productivity rate and decrease the free-contact zone between the solution polymer and ambient atmosphere the second generation (Gen2) of spinning electrode, the stationary wire electrode system, has been launched in 2010 for the 1.6 m wide working width NS 8S1600U industrial line [9]. The Gen2 has been adapted to NS LAB line since 2014, allowing direct transposition of the obtained results to the industrial line. Several research works in the open literature consider multiple parameters influences to achieve optimization of the process but only a few of these studies are dedicated to the high throughput device [10 –12].

From an industrial point of view, the NS LAB line that we used to produce nanofibers presents a huge interest for studying the up-scaling of nanofibers production. In this study, the results on an up-scaled 500 mm wide working width wire-based electrospinning setup are presented.

The aim of our study is to give an effective method using an optimal design of experiment to investigate a new solution of polymer to be electrospun. Our interest in this study is not about what kind of polymer solution is used, but how the control of electrospinning process can be achieved for further possible industrial-scale production. In order to determine the influence of the process parameters on physical properties of the electrospun nonwoven, our experiments have been conducted following a design of experiment (DOE) to compare the influence of each studied parameter whether separately or with their interactions. A focus has been made on the influence of the environmental parameters (humidity and temperature) and the results have been confronted to a former study conducted in 2010 by Dao and Jirsak [13]. Moreover, an estimation of the capacity of production for the NS500 has been calculated for the tested meta-aramid solution.

Experimental section (methods and materials)

Materials

Polymer solution

A couple polymer/solvent of meta-aramid polymer (polyamide-imide) in a polar aprotic solvent (1,3 dimethyl-2-imidazolidinone) at 60 Pa·s in working viscosity is provided by an industrial partner (KERMEL Company). It has been highlighted through rheological study that 60 Pa·s is the best polymer viscosity to obtain good macromolecular conformation of the polyamide-imide chains while electrospinning and sufficient viscoelastic properties [14].

NS 1WS500U “Nanospider™” electrospinning setup

NS 1WS500U (NS®) “Gen2” from Elmarco Company is used to conduct all experiments on the meta-aramid solution. In this free-surface technology, needles are replaced by a wire electrode coated with the polymer solution. The spinning electrode is a stationary wire onto which the spinning solution is continuously deposited by a special carriage. The carriage, driven by a pneumatic system, is a moving tank with a calibrated metal orifice inserted through which the spinning electrode wire is drawn. The spinning solution flows out from the orifice during movement of the carriage and is applied on the spinning electrode wire surface. Furthermore, the wire is continuously rewound so neither the evaporation of the spinning solution from the free surface nor the drying of the polymer on the wire occurs. The spinning and collecting electrodes are connected to high voltage power supply. During free-surface electrospinning operation, the NS® is secured by protective components to prevent access to the spinning chamber. When the high voltage field is applied the electrostatic force overcomes the surface tension of the solution and, therefore, the polymeric solution is attracted to the area with lower electric potential and Taylor cones and jet formation are initiated. While the filament is traveling to the collecting electrode, solvent evaporates and fibers get stretched, split, and deposited onto a moving substrate material/collector adjacent to the collecting electrode. We choose to spin the meta-aramid solution on a siliconized backing paper as a collector where the adhesion is minimized in order to remove the nanofiber layer from the substrate easily for weighting and observations.

Process parameters and choice of the variables

The obtained nanofibers are expected to have thin diameters with high homogeneity of the products, i.e. low levels of beads, residual solvent or nonfibrous area, and a narrow fiber diameter distribution, in order to validate the industrial requirements. A wide range of process parameters are available with the NS® technology. The viscosity of the meta-aramid solution fixed at 60 Pa·s has been found to be the optimum one to obtain nanofibers according to a former rheological study [14]. All other process parameters are investigated in order to obtain the best quality of production.

Reservoir of polymer solution is fully filled (40 mL) for the longest processing time, standard wire diameter of 0.2 mm is used, and 300 mm is the standard width of our silicone paper substrate/collector. The smallest metal insert with calibrated orifice (0.5 mm) and the slowest spinning electrode wire speed (0.5 mm/s) are selected regarding the solution viscosity. It has been concluded that small orifice allows a better stretching and repartition on the spinning wire and a slow speed gives the polymer solution enough time to be stretched by electrostatic field. The lowest substrate speed (15 mm/min) is selected to obtain a thicker nanofiber layer. A medium value of carriage speed (200 mm/s) is chosen to continuously renew the polymer solution on the spinning wire electrode.

Temperature and humidity are controlled by air conditioner and humidifier in the laboratory working area, which are the most complex parameters to be kept at fixed values (40–50% for relative humidity and 22–28℃ for temperature). Therefore, low airflow (20–60 m³/h) and short processing time (10 min) are chosen to maintain ambient parameters at stable values. All range of distances (125–290 mm) are investigated. Voltage of 50 kV value is selected as the lower limit of voltage that initiates jet formation for a fixed distance of 290 mm, otherwise voltages from 50 kV to 100 kV are investigated.

Table 1 presents the main settings and chosen investigated values.

Table 1.

Settings.

NS settings	Range	Chosen values	Unit
Airflow	10–180	20–60	m³/h
Distance spinning electrode/substrate	125–290	125–290	mm
Voltage	0–100	50–100	kV
Relative humidity	30–70	40–50	%
Temperature	20–40	22–28	℃
Time of processing	1–480	10	min
Substrate speed	15–215	15	mm/min
Spinning electrode wire speed	0–10	0.5	mm/s
Carriage speed	150–400	≈200	mm/s
(spinning electrode and collecting electrode) Wire diameter	0.2	0.2	mm
Metal insert with calibrated orifice	0.5–0.9	0.5	mm
Width of substrate	100–600	300	mm
Reservoir volume	20–40	40	mL

Characterization of nanofibers

Surface morphology analysis of the electrospun meta-aramid nanofibers was assessed by scanning electron microscopy Hitachi S-2360N (SEM) after being gold-coated. Three SEM pictures are taken on each sample. Diameter measurements of 20 to 25 are considered for each SEM pictures at 10k magnification using Image J software, in order to calculate significant fiber diameter distribution including standard deviation. The mean fiber diameter (MFD) is calculated as an average from 60 to 75 diameter measurements of randomly selected fibers, the relative standard deviation (RSD) is determined, and coefficient of variation (CV%) is calculated according to the following equation¹⁵

CV = 100 \times \frac{MFD}{RSD}

(1)

Nonfibrous areas (NFA) were measured on each SEM pictures at 5 k magnification (Image J software). The nonfibrous area is defined (equation (2)) as the residual solvent area (beads are also counted in that measurement) and is used as an estimation of fiber morphology or nanofiber nonwoven criterion of quality (our formula is adapted from research article of Dao and Jirsak [13]). NFA is calculated for a minimum of three different SEM pictures as the sum of all beaded or solvent area on the total area of the SEM picture

NFA = 100 \times \frac{\sum (beads + solventareas)}{SEMpicturearea}

(2)

Experimental design

Among all of the process parameters five were selected and used as variables factors in the DOE: applied voltage (kV), relative humidity (%), temperature (℃), distance between spinning electrode wire and substrate material (mm), and airflow going through the spinning chamber (m³/h). The aim of the DOE is to particularly understand the influence of each factor or combination of factors on MFD and on fibers morphology. In order to investigate all the five different factors for three levels each and all possible combinations, a D-optimal design is chosen according to the DOE methods described by Goupy and Creighton [15]. This approach is the most efficient by giving the highest experiments/response ratio and taking into account all possible interactions between factors. The upper, center, and lower levels of each factor are defined in Table 2 (coded and actual values). Upper and lower values have been chosen in order to obtain the widest study domain, according to settings from Table 1. Center values are the mean values of parameters between lower and upper values, and they are of huge importance to calculate an estimation of the experimental error and model validity.

Table 2.

Levels of factors for DOE – Study domain.

		Coded levels
Factor	Unit	−1	0	1
Voltage (V)	kV	50	75	100
Relative humidity (RH)	%	40	45	50
Temperature (T)	℃	22	25	28
Distance spinning electrode/substrate (D)	mm	125	208	290
Airflow (AF)	m³/h	20	40	60

The studied responses of the DOE are: mean fiber diameter (MFD in nm), coefficient of variation (CV %), and nonfibrous area (NFA %).

Results and discussion

The main objective of this study is to obtain a qualitative index of homogeneity and quality of our electrospun production. The influence of each parameters and combination of parameters on MFD, fiber homogeneity and quality (CV and NFA) has been firstly investigated. Then, the main factors have been detected. D-Optimal DOE have been generated (thanks to Modde 6.0 Software, developed by Umetrics) and followed, for a total of 20 experiments.

SEM analysis

Fiber diameter characterization

Twenty experiments have been conducted, and each sample produced through NS® is observed on SEM. Diameters of fibers are ranged from 30 to 664 nm, MFD are ranged from 81 to 183 nm, and CV are ranged from 11% to 60%. SEM observations of three experiments, selected among the 20 are shown in Figure 1 to illustrate the differences in obtained results for the all study domain, with MFD measurement and distributions of fiber diameter histograms. Spinning conditions and obtained results of these three experiments are reported in Table 3.

Figure 1.

SEM pictures for MFD measurement of two experiments (N1 and N8) at the edge of the domain study and one at the center (N17) + fiber diameters distribution histograms.

Table 3.

Experimental conditions and statistics of MFD and NFA measurement.

Trial no.	Voltage (kV)	Rel. humidity (%)	Temp. (℃)	Dist. (mm)	Airflow (m³/h)	MFD (nm)	CV (%)	Min (nm)	Max (nm)	NFA (%)
1	50 (−1)	40 (−1)	22 (−1)	125 (−1)	60 (+1)	89	19	42	141	3.6
8	100 (+1)	50 (+1)	28 (+1)	290 (+1)	20 (−1)	173	25	67	292	0.02
17	75 (0)	45 (0)	25 (0)	208 (0)	40 (0)	140	11	108	173	0.5

MFD: mean fiber diameter; CV: coefficient of variation; NFA: nonfibrous area.

It can be observed that fiber diameters, diameter distribution, and CV presented in Figure 1 are strongly dependent on the spinning parameters values. Histograms of fiber diameter distribution seem to indicate a Gaussian distribution for each sample.

NFA measurements

Beads and other NFA are measured on the pictures at 5k magnification (Figure 2). Ellipses are drawn to approximate the NFA zones, and are measured thanks to Image J software.

Figure 2.

SEM pictures for NFA measurement of two experiments (N1 and N8) at the edge of the domain study and one at the center (N17).

D-Optimal design

Sixteen experiments at low (−1) and high (+1) levels for each of the factors and four at center points (i.e. middle levels (0) of the factors) have been carried out. The obtained results of D-Optimal DOE are presented in Table 4.

Table 4.

DOE factors and experiments results (highlighted lines are excluded values from the model).

	Actual factor values					Coded factor values					Responses
Trial no.	V (kV)	RH%	T (℃)	D (mm)	AF (m³/h)	V	RH	T	D	AF	MFD (nm)	CV (%)	NFA (%)
1	50	40	22	125	60	−1	−1	−1	−1	+1	89	19	3.6
2	90	50	28	125	20	0.6	+1	+1	−1	−1	129	28	6.3
3	90	40	28	125	40	0.6	−1	+1	−1	0	81	18	3.3
4	50	50	28	125	20	−1	+1	+1	−1	−1	125	36	3.5
5	100	40	25	290	20	+1	−1	0	+1	−1	101	18	10.2
6	50	50	22	290	20	−1	+1	−1	+1	−1	183	22	0.6
7	50	40	28	290	20	−1	−1	+1	+1	−1	96	21	3.9
8	100	50	28	290	20	+1	+1	+1	+1	−1	173	25	0.02
9	90	40	22	125	60	0.6	−1	−1	−1	+1	88	15	5.2
10	50	50	22	125	60	−1	+1	−1	−1	+1	171	60	6.3
11	50	40	28	125	60	−1	−1	+1	−1	+1	121	16	4.3
12	90	50	28	125	60	0.6	+1	+1	−1	+1	90	16	3.6
13	50	40	22	290	60	−1	−1	−1	+1	+1	95	16	4.9
14	100	50	22	290	60	+1	+1	−1	+1	+1	129	17	0.6
15	100	40	25	290	60	+1	−1	0	+1	+1	106	17	1.6
16	50	50	28	290	60	−1	+1	+1	+1	+1	164	25	1.1
17	75	45	25	208	40	0	0	0	0	0	140	11	0.5
18	75	45	25	208	60	0	0	0	0	+1	125	16	0.9
19	75	45	25	208	60	0	0	0	0	+1	124	20	0.8
20	75	50	28	208	20	0	+1	+1	0	−1	175	34	2.7

MFD: mean fiber diameter; CV: coefficient of variation; NFA: nonfibrous area.

Constraints

At lower distance (125 mm) between spinning and collecting electrodes maximum voltage (100 kV) was not possible to set. We choose to minimize the high voltage value by 10 kV in order to run this experiment.

Modeling

We proceed to the calculations and principal statistics like the standard deviation, p-values, and R². A first-degree model with interaction is generated for evaluation of the responses

y = a_{0} + \sum a_{i} x_{i} + \sum a_{iJ} x_{i} x_{J}

(3)

where y is one of the responses, x_i represents a level of factor i, x_J represents a level of factor j and a₀ is the constant, a_i are the effects of the factors, and

a_{iJ}

are the second-order interactions (a₀, a_i, and

a_{iJ}

are the coefficients of the polynomial). R² is defined as

R^{2} = Sumofsquares (Model) / SumofSquares (Total)

(4)

This definition of R² means that the closer the model reproduces the observed responses, closer is the sum of squares of the calculated responses to the sum of squares from the model and to the total sum of squares. Therefore, R² is used as an indicator of the quality of the model. Several fits of the model have been checked, excluding some experiments and the less significant interactions in order to obtain the best coefficient of determination R² (close to 1). Experiment nos. 1, 11, 17, and 20 have been excluded. As shown in Table 5, the calculated R² reaches the value of 0.968, which means that model explains 96.8% variability of the response. The acceptance probability (p-values) of the coefficients is fixed at 0.05; when p-value is less than 0.05, the factor has significant impact on the mean fiber diameter.

Table 5.

Coefficients and significance of the model for MFD.

Term		Coefficient		p-Value (Prob>\|t\|)
Constant		a ₀	125.35	5.73 × 10⁻⁹
V	x ₁	a ₁	−6.06	0.0139
RH	x ₂	a ₂	25.29	2.13 × 10⁻⁵
T	x ₃	a ₃	−0.27	0.8969
Dis	x ₄	a ₄	11.18	0.0012
AF	x ₅	a ₅	−1.98	0.3506
V*RH	$x_{1} x_{2}$	$a_{12}$	−3.95	0.0577
V*AF	$x_{1} x_{5}$	$a_{15}$	−9.10	0.0036
RH*T	$x_{2} x_{3}$	$a_{23}$	−11.03	0.0038
RH*AF	$x_{2} x_{5}$	$a_{25}$	−13.68	0.0015
T*Dis	$x_{3} x_{4}$	$a_{34}$	10.13	0.0017
		F	Significance	R ²
		45.79	<0.0001	0.968

Coefficients analysis

Partial least square (PLS) linear regression is a standard approach to approximate the solutions of overdetermined systems and to compare the experimental data to the obtained mathematic model. Coefficients of the factors and significance probability (p-value) were obtained by PLS linear regression for our reduced DOE. They are reported in Table 5.

According to equation (1), the obtained first-degree model with interaction is:

y_{MFD} = 125.35 - 6.06 x_{1} + 25.29 x_{2} - 0.27 x_{3} + 11.18 x_{4} - 1.98 x_{5} - 3.95 x_{1} x_{2} - 9.10 x_{5} - 11.03 x_{2} x_{3} - 13.68 x_{2} x_{5} + 10.13 x_{3} x_{5}

Eleven coefficients have been determined in our model for the 16 experiments included. The value of the constant a₀ is similar to the predicted responses at the center of the study domain (Trial nos. 18 and 19), and we can validate that first-degree model with interaction. Fisher F-test with low probability level (p model < 0.0001) also verified significance of the regression model. According to results in Table 5, MFD is significantly affected by three factors (p < 0.05): relative humidity, distance, and voltage, with relative humidity having the most pronounced effect. Several second-order interactions are significant as well: RH*AF, T*Dis, V*AF, and RH*T. As significant probability of V*RH is close to the probability level, we can assume that it also has slight influence on MFD. Graph of the effects in Figure 3, allow us to examine more precisely the importance of each coefficient.

Figure 3.

Effects of factors and interactions at center level (standard error bar indicates confidence intervals CV at 95%).

In this display, the effects are sorted from the largest to the smallest. For process factors, the effect plot displays the change in the response when a factor varies from his low (−1) to his high (+1) level, all other factors are kept at their averages (0). Values of the effects are approximately twice the values of the coefficients, as the coefficients are the change in the response when the factors vary from the average (0) to the high (+1) level and the variation of the factors from low (−1) to average (0) levels are not taken into account. Factors or interactions where the confidence interval includes zero are insignificant. The graph in Figure 3 indicates positive effects of factors RH (relative humidity) and D (distance) and interaction T*D, while other factors and interactions have negative effects on MFD. It also shows that AF (airflow) and T (temperature) factors have no significant effects on MFD their effects could be either negative at low level or positive at high level.

Main effect plots and interaction plots of factor on MFD

The main effect plot displays the fitted values of the response with the confidence interval at the low and high value of the selected factor and at the center point while the other factors are kept at their average values. Main effect plots for all factors are illustrated in Figure 4.

Figure 4.

Main effect plot on MFD for all factors.

The observed values of the response at the center points are plotted in Figure 4. As the observed values are within the confidence intervals, we can conclude that the linear model is valid. As shown in Figure 4, an increase in relative humidity from 40% to 50% leads to an increase of MFD of 68%. An increase of distance from 125 to 290 mm leads to an 18% increase of MFD. An increase of voltage from 50 to 100 kV leads to a decrease of 7% of MFD. The increase of airflow from 20 to 60 m³/h or of the temperature from 22℃ to 28℃ seems to have no impact on MFD when every other factor are kept at their average values. Table 6 summarizes the observations made in Figure 4, reporting values of MFD for each level of factor and the induced variations of MFD.

Table 6.

Main effect on MFD for all factors.

Factor	Low (nm)	Center (nm)	High (nm)	% Variation low–center	% Variation low–high	SD (%)
Relative humidity	90	120	151	+33	+68	5.3
Distance	111	121	131	+9	+18	5.5
Voltage	125	120	116	−4	−7	5.1
Airflow	121	121	121	0	0	6.2
Temperature	121	121	121	0	0	5.2

Interactions analysis

When we obtain an interaction between two factors, the effect of one of the factors depends on the level of the other factor. The plotting in Figure 5 displays the change in MFD when one factor varies, and the second factor is set at both its low and high level. When the two lines are parallel there is no interaction between these two factors. When the two lines cross each other there is a strong interaction.

Figure 5.

Interaction profiles for all factors.

All plots in Figure 5 indicate that there are strong interactions between some factors of our model: V*AF > D*T > RH*AF > RH*T > RH*V, which proves the importance of studying the influence of the combination of factors on MFD instead of individual influence of factors on MFD.

Response surface modeling

RSM of mean fiber diameter

To complete this analysis, RSM is conducted to optimize and predict MFD of our electrospun meta-aramid fibers. Four-dimensional surface plots are displayed in Figure 6 in order to observe the influence of all factors on MFD.

Figure 6.

Response surface modeling of MFD at fixed voltage of 75 kV.

According to Figure 6, thinner (MFD = 70 nm) and thicker (MFD = 180 nm) fibers are obtained when airflow value is low (20 m³/h). This is probably due to the fact that at low airflow no interference appeared in the electrical field. Meanwhile, variations of diameters are important (70–180 nm), because airflow is not important enough to obtain homogeneous ambient parameters. Relative humidity, in particular, has to be controlled at small airflow because it has the strongest impact on MFD. When relative humidity increases, two combined effects occur on MFD. One is due to the fact that our polymer solution is hydrophilic, therefore when relative humidity increases evaporation rate of the solvent is faster but some water may remain within the fiber. Secondly, an increase of relative humidity may interfere with the electrical field by reducing his strength which reduces the stretching of the polymer solution. These two combined effects caused an increase of MFD.

At higher airflow value (60 m³/h) MFD is equally dependent on relative humidity, temperature, and distance. Range of MFD is narrower (100–160 nm), which means that process conditions are more stable. Airflow is high enough to renew the ambient conditions and give homogeneous humidity and temperature. Distance strongly influences MFD, with MFD increasing when distance between spinning electrode wire and substrate material increases. Indeed, when distance increases electrical field decreases, which reduces the stretching of fibers. Other values of voltage are not been displayed, but the general aspect remains the same. The only difference is that all MFD values increased at 50 kV and decreased at 100 kV. Voltage has also a strong influence on MFD, which decreases when voltage increases. RSM is an important tool to further understand the combined influences of factors. To fully complete this study, and choose the best combination of process parameters to obtain minimum MFD, CV%, and NFA values, another RSM has been conducted for NFA.

RSM of NFA

Several zones are of interest to produce thinner MFD. We now have to determine the zones to avoid, which means to keep NFA as low as possible. RSM is conducted to predict NFA on our electrospun meta-aramid fibers. Four-dimensional surface plots are displayed in Figure 7 in order to observe the influence of all factors on NFA.

Figure 7.

Response surface modeling of NFA at fixed voltage of 75 kV.

According to Figure 7, most of the lower NFA values correspond to the thinner MFD zones. Airflow, temperature, and relative humidity are the most influential factors for NFA. Voltage has no influence on NFA.

The observation of the response surface modeling of MFD and NFA had allowed us to reduce the domain of process parameters. We selected 0 to 5% of NFA as the acceptance values to obtain good nanofiber layer morphology, with a range of MFD from 90 to 150 nm.

When we applied 50 kV voltage, the MFD globally increased, so we selected a reduced domain of 75 to 100 kV for applied voltage. An increase of 10% of the relative humidity leads to an increase of 68% of the MFD, but it also leads to a better homogeneity of the nanofiber layer with limited NFA (0–5%). So in order to limit the MFD we choose to reduce the humidity range at 40–45%. The temperature has no impact on MFD, but a limited impact on NFA at high airflow (60 m³/h). We decided to keep a room temperature of 25℃ ± 1℃. An increase of distance leads to an increase of MFD and has no effect on NFA. We selected the range of distance where MFD comprised between 90 and 150 nm, which means a new range of distance from 125 to 208 mm. When airflow is up to 60 m³/h, the NFA is comprised between 0 and 5%, thus airflow should be maintained at 60 m³/h.

The domain of process parameters has been reduced to new levels as reported in Table 7.

Table 7.

Acceptance domain of experiments.

Factors	Minimum	Maximum	Units
Applied voltage	75	100	kV
Relative humidity	40	45	%
Temperature	24	26	℃
Distance SE wire/substrate Material	125	208	mm
Airflow	60	60	m³/h

Productivity

The productivity of our meta-aramid polymer nanoweb through NS® equipment has been calculated based on the quantity of nanofibers deposited on collector after 8 h of continuous spinning. The spinning conditions are reported in Table 8.

Table 8.

Spinning conditions for productivity calculation.

Time (h:min:s)	V (kV)	RH (%)	T (℃)	D (mm)	AF (m³/h)	Width (mm)	Length (mm)
08:00:00	75	42.4	29.2	250	30	300	7200

Twelve samples of 1 cm² area have been weighted and an approximation of the nanofiber layer weight has been calculated. About 0.63 g of nanofiber have been collected on the siliconized paper used as collector. The production rate was about 0.08 g/h for the meta-aramid solution in our process conditions. MFD of produced nanofibers was 120 nm (CV = 27%).

The increase in productivity from the traditional electrospinning process (with a single needle) is interesting with a production rate 8 times more important for the NS®. The NS® LAB we use is a laboratory line device and is composed of only one “spinning module” in the spinning chamber. The two “industrial lines” of Elmarco have respectively 4 and 8 spinning. The production rate on these devices could be expected to be respectively 4 and 8 times more important than the production rate obtained on NS® LAB, thus an effective increase of productivity (32 times and 64 times more) is observed compared to the traditional electrospinning process.

Limitations

The present study has been conducted on a specific domain and predictions of nanofibers diameters can only be applied for parameters within this domain. However, the domain has been chosen as wide as possible. Complementary experiments have to be conducted to improve the productivity, and process parameters should be investigated to find most influential parameters on NS® LAB productivity.

Conclusions

Free-surface electrospinning of meta-aramid solution was processed. This study presents a complete methodology from conception of design of experiments to responses analysis (interpretation of coefficients, interactions, and response surface modeling). Fibers with diameters in the range of 30–664 nm, mean fiber diameter in the range of 81–183 nm, and coefficient of variation in the range of 11–60% were obtained. MFD and NFA were investigated for five varying parameters: applied voltage, relative humidity, temperature, distance between spinning electrode wire and substrate material, and airflow going through the spinning chamber. A first-degree model with interactions has been validated, and the mean fiber diameter obtained at 96.8% has been explained. Three parameters (relative humidity, distance, and voltage) have significant effects on the mean fiber diameter. Main effect of an increase of relative humidity from 40% to 50% or of the distance from 125 to 290 mm results in an increase of the MFD of respectively 68% or 18%. Main effect of an increase of voltage from 50 to 100 kV was a decrease of MFD of 7%. Strong interactions between factors of our model were observed: V*AF > D*T > RH*AF > RH*T > RH*V, which shows the importance of studying the influence of a combination of factors on MFD over individual factor influence. It has been found that relative humidity played a major role on MFD and NFA. Study of the response surface modeling allowed us to reduce our experimental domain to obtain the best compromise between low MFD and minimum NFA. Complementary study for higher values of airflow could be added to our design of experiments to complete the study. Another study has to be dedicated to the improvement of meta-aramid fiber productivity through NS®.

Footnotes

Acknowledgement

The authors would like to thank the KERMEL Company for supplying the materials (polyamide-imide solutions).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the KERMEL Company.

References

Reneker

Yarin

Zussman

et al.

Electrospinning of nanofibers from polymer solutions and melts. Adv Appl Mech 2007; 41: 343–346.

Ramakrishna

Fujihara

Teo

W-E

et al.

An introduction to electrospinning and nanofibers, Singapore: World Scientific, 2005.

Chronakis IS. Nano-microfibers by electrospinning technology: Processing, properties and applications. In: Micromanufacturing engineering and technology. New York: Elsevier, 2010, pp. 264–286.

Thompson

Chase

Yarin

et al.

Effects of parameters on nanofiber diameter determined from electrospinning model. Polymer 2007; 48: 6913–6922.

BCC Research Nanotechnology Report. Global markets and technologies for nanofibers. NAN043C, May 2013.

Yarin

Zussman

. Upward needleless electrospinning of multiple nanofibers. Polymer 2004; 45: 2977–2980.

Jirsak O, Sanetrnik F, Lukas D, et al. Method of nanofibres production from a polymer solution using electrostatic spinning and a device for carrying out the method. Patent WO2005024101 CZ, 2005.

Petrik S and Maly M. Production nozzle-less electrospinning nanofiber technology. In: MRS proceedings, 2009, p. 1240.

Cengiz

Dao

Jirsak

. Influence of solution properties on the roller electrospinning of poly(vinyl alcohol). Polym Eng Sci 2010; 50: 936–943.

10.

Sukigara

Gandhi

Ayutsede

et al.

Regeneration of Bombyx mori silk by electrospinning. Part 2: Process optimization and empirical modeling using response surface methodology. Polymer 2004; 45: 3701–3708.

11.

Yordem

Papila

Menceloglu

. Effects of electrospinning parameters on polyacrylonitrile nanofiber diameter: An investigation by response surface methodology. Mater Des 2008; 29: 34–44.

12.

Han

Nurwaha

et al.

Free surface electrospun fibers: The combined effect of processing parameters. Polym Eng Sci 2013; 54: 189–197.

13.

Dao AT and Jirsak O. Roller electrospinning in various ambient parameters. In: NANOCON, Olomouc, Czech Republic, 12–14 October 2010.

14.

Oertel A, Khenoussi N, Schacher L, et al. Impact of rheological properties of polymer solution on electrospinning process. In: 13th AUTEX world textile conference 2013: Book of abstracts, Dresden, Germany, 22–24 May 2013.

15.

Goupy

Creighton

. Introduction to design of experiments with JMP examples, 3rd ed. Cary, NC: SAS Institute, 2007.