A unique device for controlled electrospinning

S.B. Mitchell, J.E. Sanders

Department of Bioengineering, University of Washington, Seattle, Washington 98195
Received 10 January 2005; revised 3 August 2005; accepted 17 November 2005
Published online 7 April 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.30673
Abstract: The purpose of this research was to develop a
system for controlled electrospinning of bro-porous scaffolds for tissue engineering applications and to use this
system to assess mesh architecture sensitivity to manufacturing parameters. The intent was to achieve scaffolds with
well-controlled ber diameters and inter-ber spacing. To
accomplish these objectives, a custom, closed-loop controlled, electrospinning system was built. The system was
unique in that it had a collection surface that was independent of the electrodes. The system allowed independent
manipulation and analysis of a number of manufacturing
parameters: distance between the electrodes, distance from
the nozzle to the collection surface, applied voltage, temperature of the melt, collection surface dielectric strength, and
collection surface area. Morphological analysis of fabricated
meshes showed that all test parameters signicantly affected
ber diameter and inter-ber spacing. Further, contrary to
what is generally accepted in the electrospinning literature,

voltage and temperature (inversely related to viscosity)

were not the most signicant parameters. Features of the
collection surface, including dielectric strength and surface
area, were more signicant. This dominance is, in part, a
reection of the unique electrospinning system used. The
collection surface, which was not connected to either of the
electrodes, substantially altered the electric eld between the
electrodes. Using the developed controlled electrospinning
system, thermoplastic polyurethane meshes with ber diameters ranging from 5 to 18 m with variability less than 1.8%
were made; inter-ber spacing ranged from 4 to 90 m with
variability less than 20.2%. The system has potential use in
biomedical applications where meshes with controlled ber
diameter and inter-ber spacing are of interest. 2006 Wiley
Periodicals, Inc. J Biomed Mater Res 78A: 110 120, 2006

INTRODUCTION

protective clothing,7 ltration,8 drug delivery devices,9

implant interfaces,10 and scaffolds for biomaterial devices or tissue substitutes.1113 Electrospinning involves applying a high voltage between a capillary
containing a polymer solution (or melt) and a collection surface. When a drop of the polymer forms at the
end of the capillary and the high voltage is applied, a
charge is induced on the surface of the droplet. When
the mutual charge repulsion exceeds the surface tension in the droplet, the droplet deforms into a conical
shape known as a Taylor cone. Once a critical voltage
is exceeded, a charged jet ejects from the apex of the
cone and deposits on the collection surface.14 16
Modern day electrospinning is usually performed
using a syringe lled with solvated or heated polymer.
The anode is connected to a hypodermic needle on the
syringe, and the cathode is connected to a collection
surface. When the high voltage is applied, the Taylor
cone forms, the polymer ejects from the cone apex,
and the bers collect on the grounded surface. Typical
variations on this theme included the following: using
a syringe pump to control the polymer ow rate17;
using a rotating mandrel or disk to collect the bers18,19; manipulating the electric eld20; and changing the collection surface material or shape.8

In biomedical implants, material architecture inuences the local tissue response. Materials with small
bers and small pore spacing may help to reduce or
eliminate the brous capsule that typically surrounds
implanted medical devices. Results from several studies suggest that ber diameters less than 15 m and
inter-ber or pore spacing less than 60 m are favorable.1 6 It is difcult, however, to fabricate bro-porous meshes with small ber diameters and spacing,
without inducing occulation (aggregation of bers).
Achieving consistent ber diameters can also be problematic.
Electrospinning is a manufacturing means for making nonwoven, small-diameter, bro-porous materials. It has been used for various applications including
Correspondence to: J.E. Sanders; e-mail: isanders@u.
washington.edu
Contract grant sponsor: National Science Foundation Engineering Research Center Program; contract grant number:
EEC-9529161
2006 Wiley Periodicals, Inc.

Key words: electrospinning; encapsulation; bro-porous;

tissue engineering; biomaterial; polymer

A UNIQUE DEVICE FOR CONTROLLED ELECTROSPINNING

Considerable research has been conducted, investigating the electrospinning phenomenon. Mathematical models have been developed to examine the Taylor
cone formation,14 15,21 polymer jet emergence,14 16 asymptotic decay,22 stabilities,17,23,24 and bending14,24 of
the jet. The dependence of surface features such as
beading and porosity on polymer, polymer concentrations, and process variables is also well investigated,12,2529 These studies are useful toward better
understanding of how electrospinning is achieved and
the material surface characteristics are generated.
It is reasonably well accepted in the literature that
the main parameters affecting mesh architectural features of interest in biomedical applications, ber diameter and inter-ber spacing, are the applied voltage
and the polymer viscosity,17,20,23,25,30 33 Polymer viscosity is typically controlled by adjusting the temperature of the melt. For most polymers, viscosity and
temperature are inversely related,34 However, the
strong dependence of ber diameter and inter-ber
spacing on voltage and polymer viscosity (inversely
related to temperature) is not quantitatively well supported. Further, in the few quantitative studies published, there are inconsistencies as to how ber diameter and inter-ber spacing, and features related to
them, are affected by applied voltage and polymer
viscosity. Increasing the voltage has been shown to
increase jet diameter35 and to increase polymer ow
rate16,17 and bending instability30,32,36; the latter two
features are correlated with ber diameter. However,
in other investigations, increasing the voltage was
shown to decrease the jet diameter,10,16,21,30,37,38 decrease the ber diameter,10,25 and enhance the alignment of the bers.8 Furthermore, Noris et al.27 noted
that ber diameter was independent of polymer solution concentration (related to viscosity), whereas others indicated that concentration had a major inuence
on ber diameter.32,33,35,39
We expect that it is the lack of control and consistency of other electrospinning manufacturing parameters in previous reports that explain the inconsistencies in the results. Part of the difculty is that the
practice of grounding the collection surface to the
lower electrode does not allow the manufacturing features to be independently controlled. An electrospinning system with the collection surface independent of
the electrodes is needed to systematically determine
the inuence of manufacturing features. Features expected of relevance would be those that induce the
greatest change in the electric eld magnitude and
direction; these include distance between the electrodes, distance from the nozzle to the collection surface, collection surface dielectric strength, and collection surface area. An important purpose of this
research was to conduct a sensitivity analysis of ber
diameter and inter-ber spacing to applied voltage,
polymer temperature, and other manufacturing pa-

111

rameters. This information is potentially useful toward design efforts to engineer meshes of specic
architectures for particular applications and toward
the design of a feedback-controlled electrospinning
manufacturing system. It was of further interest to
establish possible consistencies in ber diameter and
inter-ber spacing, using a well-controlled electrospinning device.
To conduct such sensitivity analysis, rst, a system
that allows control over all of these manufacturing
parameters simultaneously was needed. Systems previously described in the literature typically allowed
for precise control of the applied voltage, with only
gross estimates of polymer feed rate, viscosity, collection distance, or other features. Further, to eliminate
the inuence of solvent evaporation on the results, a
system that spun from a polymer melt was necessary.
The objective of this research was threefold: (1) to
develop an electrospinning system that allowed
closed-loop control of the following parameters: distance between the electrodes, distance from the nozzle
to the collection surface, applied voltage, temperature
of the melt (inversely related to the viscosity), collection surface dielectric strength, and collection surface
area; (2) to conduct a sensitivity analysis of ber diameter and inter-ber spacing to these electrospinning
manufacturing parameters; (3) to determine the consistency in inter-ber diameter and ber spacing that
can be achieved with this controlled electrospinning
system.

MATERIALS AND METHODS

Instrumentation
Mechanical system
A custom electrospinning system that allowed closedloop control of the features of interest, listed above, was
developed. The system allowed the electrodes to be positioned independent of the collection surface. In addition to
avoiding the inuences of grounding (or charging) the collection surface, this conguration enabled the exploration of
the inuence of collection surface characteristics on mesh
architecture and ber morphology. In the future, independence of the collection surface and electrodes, through position control of the collection surface during electrospinning, may allow the development of variable architecture
with complex, three-dimensional shapes.
The primary components of the electrospinning apparatus
were the polymer chamber, electrodes, enclosure, collection
surface, and x-y-z translation stage (Fig. 1). The polymer
chamber held 16 cm3 of polymer melt at a controlled temperature and pressure. The chamber was heated via band
heaters (model NHW00158, Tempco, Wood Dale, IL) placed
around the outer surface of the chamber. Temperature was

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

112

MITCHELL AND SANDERS

Figure 1. Custom electrospinning system. The electrospinning device consists of the polymer chamber (a), electrodes (b,c),
enclosure (d), collection surface (e), and three-axis translation stage (f). [Color gure can be viewed in the online issue, which
is available at www.interscience.wiley.com.]
monitored using platinum resistance temperature transducers (RTD) (model 1PT100KN815, Omega, Stamford, CT) embedded in the bottom of the chamber. A pressure control
valve (IP413 020, Omega) and a vacuum control valve
(PV104 5V, Omega) were connected to ports on the top of
the chamber, and pressure and vacuum sensors (PX142
030D5V, PX141 015V5V, Omega) were located in the lines
between the control and chamber ports. Temperature, pressure, and vacuum were controlled via closed-loop controllers as described later. The grounded electrode was positioned within the polymer melt chamber, extending out
through an upper port. Though typically the positive electrode is positioned within the melt, this practice was not
followed here because of the risk of arcing to the RTDs. A
stainless steel nozzle with a length to diameter ratio of 4.516
was positioned on the bottom of the chamber, so that the
polymer emerging from the chamber entered the enclosure.
The polymer chamber was mounted directly on top of the
polycarbonate enclosure, via high-temperature thermoplastic mounting brackets. The mounting brackets were designed in such a way that only the nozzle was exposed to the
interior of the enclosure. Iris ports on each side prevented
airow from affecting the electrospinning process, yet allowed easy entry into the enclosure. When assembled, the
polymer chamber was sealed, except for the inlets for pressure and vacuum control and the nozzle.
The positive electrode was mounted to a Delrin carriage
within the enclosure. The carriage was connected to a
0.8-mm lead screw so that the electrode distance from the
nozzle could be easily adjusted. A high-voltage power supply (RP-50 100, Del, Valhalla, NY), connected between the
positive and ground electrodes, provided a controllable potential difference from 0 to 50 kV.
The collection surface was mounted to an x-y-z stage
(404150XR, Parker Daedal, Harrison City, PA) via a polycarbonate arm that protruded into the enclosure through a port
in the rear. An accordion boot formed a seal between the
arm and the enclosure. All components within the enclosure
were constructed from dielectrics, with the exception of the
nozzle and the positive electrode.

Controller
A Labview (National Instruments, Austin TX) virtual instrument (VI) was used to control the temperature, voltage,

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

vacuum/pressure behind the melt, and stage position. For

temperature measurement, signals from the RTDs were converted to voltages with signal conditioning modules (OM5IP-600-C, Omega, Stamford, CT), digitized with a multifunction data acquisition board (PCI-MIO-16XE-10, National
Instruments, Austin TX), and monitored on a PC (450 MHz
Optiplex GX1, Dell, Round Rock, TX). To control the duration of heating so as to achieve set point temperatures,
pulse-width-modulated digital outputs from the controller
were sent to solid-state control-output modules (ACO5-C,
Omega, Stamford, CT). The duty cycle was determined using the average temperature as a percentage of the set point.
Lower percentages increased the duty cycle.
Vacuum and pressure were controlled with a proportional
closed-loop feedback control scheme. Signals from the vacuum sensor and pressure sensor were digitized and monitored on the PC. Signals sent to the vacuum control valve
and pressure controller were proportional to the differences
between the set points and their respective instantaneous
values. A greater difference increased the time the valves
were open.
Components from a library of Labview virtual instruments (Motion Toolbox, Parker Compumotor, Rohnert
Park, CA) were embedded in the VI for motion control of the
x-y-z stage. Control command inputs from the VI were sent
to an AT6400 four-axis controller (Compumotor) that converted the data for use in a step indexer (Zeta4, Compumotor). On initialization, all axes were positioned with respect
to a known position. Monitoring the position with the
AT6400 allowed for pseudo-feedback control of axis position.

Calibration
To calibrate the temperature transducers, the polymer
chamber was lled with water or olive oil, and then, heated
using the band heaters via digital control from the computer.
The temperature inside the chamber was monitored using a
mercury thermometer. Thermometer values and corresponding transducer voltages were recorded with a calibration VI, while the temperature was increased to 100 or to
250C (for water and olive oil, respectively) and then decreased to room temperature. These data were used to estimate the duty cycle for the closed-loop controller.

A UNIQUE DEVICE FOR CONTROLLED ELECTROSPINNING

To calibrate the pressure transducer, lab-forced air was

applied to both the transducer and a vacuum gauge through
a medical grade pressure regulator. Pressure was increased
from 0 to 172.5 kPa, while voltages were recorded at 34.5 kPa
intervals, using a calibration VI on the computer.
Sensitivities for each transducer were determined by using a linear least-squares regression to the voltage verses
measured value data. The sensitivities were used in the
controller VI for feedback control. The feedback controls
were then evaluated for percent overshoot, time to peak, and
steady state error, under typical operating conditions.

Mesh production and analysis

An analysis of the system parameters was performed
using meshes electrospun from a melt. Melt spinning was
chosen over solvent spinning, so as to reduce inconsistencies
related to solvent evaporation rates and solution concentration changes that may result during electrospinning from
solution.

Controlled parameters
In this electrospinning system, there were three classes of
variables to consider: the device-specic variables, the polymer-specic variables, and the collection surface-specic
variables. The primary device-specic variables were the
applied voltage (V), distance from the electrode to the nozzle
(De), nozzle length (NI), and nozzle diameter (Nd) (note: the
nozzle is usually referred to as capillary in the electrospinning literature). A consistent NI:Nd ratio of 4.5, considered
optimal based on previous research,16 was used. The only
polymer-specic variable considered for melt spinning was
the polymer viscosity. Viscosity of the melted polymer was
dependent on the melt temperature (T), and since temperature was easily controlled in this system, temperature was
used as the parameter, instead of viscosity. The collection
surface-specic variables were dielectric strength (DS) and
surface area (SA).
A separate experiment was conducted to determine
whether the viscosity indeed decreased when melt temperature increased. Polymer of 5.75 g was placed in a nozzleshaped chamber of 2 mm diameter and 11.73 mm length.
The nozzle was heated in the thermal-controlled melt chamber described earlier and brought up to a specied temperature225, 234, or 243C over a 15-min interval. Polymer
was collected for a 5-min interval and then weighed. Viscosity was calculated as the ratio of the shear stress at the
nozzle wall to the apparent shear rate at the nozzle wall:
a

P R4
L 2Q

where, P is the pressure gradient across the nozzle (g*(H

L)), H is the average height of polymer above the nozzle,
L is the nozzle length, R is the nozzle diameter, and Q is the
volumetric ow rate (M/(T)). The polymer density () was
determined experimentally by weighing a precise volume of
molten and condensed polymer, ensuring it was free of all

113

TABLE I
Analyzed Parameter Values
Parameter
De (mm)
(mm/mm)
V (kV)
T (C)
DS (kV)
SA (mm2)

Analyzed Values
130
0.39
25
225
1
6082

150
0.53
30
234
54
15,394

170
0.66
35
243
84

190

210

126

air bubbles. Three trials were conducted at each temperature. A test statistic of 0.05 was used to make comparisons
between groups.
Unique to this device, the distance between the collected
bers and the nozzle was independent of the electrode position. Thus, the device-specic parameter of collection surface distance from the nozzle (Ds) was also considered. This
parameter was normalized to De for consistent comparison
with values of De (Ds/De ). For analysis of the electrospinning parameters of electrode distance (De), ratio of collection surface distance to electrode distance (), applied
voltage (V), and temperature (T), polystyrene Petri dishes
(100 mm diameter; 15 mm depth) were used for the collection surface. To determine the effect of the collection surface
dielectric strength (DS), additional tests were conducted
using Teon, Pyrex, and additional polystyrene Petri
dishes (0.8 mm thickness and 1.75 mm thickness). More tests
using 150 mm 15 mm polystyrene Petri dishes were
conducted to analyze the effect of collection surface area
(SA) on the end product mesh. Table I summarizes the
experimental parameters and the ranges over which they
were varied and analyzed. For most of the parameters, these
values also represent the range over which the electrospun
jet was stable enough to collect bers. The upper and lower
bounds for , V, and T were determined by systematically
changing the parameter within a 170-mm electrode spacing
range until a consistent mesh could no longer be collected.
The upper and lower bounds for De were established using
the upper and lower bounds for and systematically changing the De until a consistent mesh could no longer be collected.
The interactions between variables were assessed. All parameters other than those under study were held constant
for each of these evaluations: The ratio was varied at three
levels within ve levels of De. V was varied at three levels
within three levels of and three levels of De. T was varied
at three levels within three levels of . The collection surface
dielectric strength was varied at four levels within three
levels of . The collection surface area was varied at two
levels within three levels of .
The mass ow rate (MFR) was measured for a subset of
the test parameters. The subsets were as follows: measurements were made at ve levels of De, three levels of V, three
levels of , and two levels of SA. Thus, a total of 75 interactions were evaluated.

Mesh collection
To conduct the analysis, meshes were manufactured from
thermoplastic polyurethane (Estane 58315, Noveon, Cleve-

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

114

land, OH) while each parameter was varied. For the analysis
of the device-specic parameters, to assure that the temperature of the melt was consistent throughout testing, the
following protocol was used for each test. Approximately
5 g of polymer was placed in the electrospinning melt chamber and heated until a slow steady ow emerged from an
1-mm nozzle at atmospheric pressure (15 min). Once the
ow was steady, the nozzle tip was cleaned. When a fresh
droplet formed on the tip, voltage was applied between the
polymer drop and the positive electrode, electrospinning a
continuous ber from the nozzle. The continuous ber was
collected on a surface (inverted Petri dish) that was positioned between the nozzle and the positive electrode. To
analyze the polymer-specic parameter (related to polymer
viscosity), electrospinning was conducted at different polymer temperatures.
To ensure a consistent thickness of the sample, during
electrospinning, the collection surface was translated in the
horizontal plane in concentric circles with diameter decreasing at a rate of 5 mm/s. Fiber collection began at the outer
edge of the surface, and then, the circle diameter was decreased in 2-mm increments until the center of the surface
was reached. Fiber collection was continued, when the surface returned to the original position, in concentric circles of
increasing diameter. Average sample thickness was 0.3 mm.
To determine the MFR, bers were collected for 300 s, and
then, the sample mass was measured. Care was taken such
that all bers were collected on the surface, throughout the
tests. The MFR was the sample mass divided by 300 s.

Mesh analysis
Since the focus of this research was to produce predictable
and repeatable scaffolds for biomaterial and tissue engineering applications, an appropriate method needed to be developed to analyze the end product mesh. The properties of
the mesh of interest were the ber diameter and inter-ber
spacing. Samples of each manufactured test mesh were
mounted on 14-mm OD, 12-mm ID Teon frames, using
Titebond polyurethane glue (Franklin International, Columbus, OH). Care was taken to prevent straining the sample during the mounting process. The samples were embedded in dyed OCT (Tissue-Tek, Sakura, CA) and ash frozen.
The OCT was dyed to enhance the contrast between the
translucent OCT and the translucent polymer bers so as to
simplify image processing. For each sample, 20 consecutive,
5-m thick sections were taken perpendicular to the horizontal plane of the sample and placed on glass slides for
light microscopy analysis. At least 16 images from the rst,
last, and two intermediate sections from each sample series
were taken. The images were digitized using Image Pro
Plus (Media Cybernetics, Silver Spring, MD) and then
analyzed using custom algorithms written with the Image
Processing Toolbox in Matlab (Mathworks, Natick, MA).
To assess ber diameter, a semi-automated method was
used. The digitized images were imported, converted to
grayscale, and thresholded such that only the ber crosssections remained in the image. Specs resulting from inconsistencies in the embedding medium were removed. To
determine the appropriate threshold level for each image,
algorithms in the Matlab image processing toolbox were

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

MITCHELL AND SANDERS

used. The algorithms, based on an intensity level specied

by the user, identied peaks and valleys in the image and
lled in the enclosed regions. Once the user was satised
with the image processing, the properties of each region
(ber) were determined.
The properties of each ber were determined using the
regional properties algorithms in the Matlab image processing toolbox. An ellipse was t to each ber cross-section, and
the minor-axis length was determined. The minor-axis
length was considered to be the ber diameter. An average
inter-ber spacing was determined using the stereological
relationship for the mean free distance between particles in
a two-dimensional plane, L 0.5(NA)1/2.40,41 The interber spacing was dened as the edge-to-edge distance between bers.

RESULTS
Calibration
The x-y-z stage has a resolution of 98 m, a repeatability of 1.3 m, a positional accuracy of 21 m, and
a straight-line accuracy of 6.75 m/100 mm on all
three axes. The system had the capability of 60 mm/s
velocity with a maximum acceleration of 20 m/s2.
Total range of motion was 400 mm on all axes.
The performance of the temperature controller was
dependent on the processing temperature for the polymer of interest. Four melt temperatures (180, 225, 234,
and 243C) were evaluated in calibration testing. The
time to peak temperature ranged from 180 to 420 s,
steady-state error ranged from 0.10 to 0.79C, and
percent overshoot was between 0.10% and 1.03%. The
average full-scale output (FSO) error was 0.22% for all
temperatures evaluated.
The pressure controller for a step input from 0.0 to
103.0 kPa had a steady-state error of 6.0 kPa, a time to
peak of 5 s, and a percent overshoot of 0.01%. The
average FSO error for pressure control was 0.41% for
all tests conducted.

Parameterization
ANOVAs performed on all data revealed that there
was a signicant (p 0.05) effect for all the test parameters with respect to ber diameter and inter-ber
spacing. In addition, there was no apparent combination of parameters that would reduce the number
needed for predicting the ber diameter and interber spacing features of the end product. Here, parameters are expressed normalized to their maximum
tested values. The range of testing within a parameter
was limited by the stability of the electrically spun jet.
For the distance between electrodes (De), broporous

A UNIQUE DEVICE FOR CONTROLLED ELECTROSPINNING

meshes outside of the indicated parameter range (Table I) could not be collected. Fibroporous meshes
could not be collected at voltages (V) less than 25 kV.
Voltages above 35 kV resulted in highly unstable jets,
which could only be collected at De 170 and
0.39. For all combinations of De and V, meshes could
be collected only within the range for indicated in
Table I. Temperatures below 225C resulted in polymer melts that were too viscous to spin, whereas
temperatures above 243C overheated the polymer.
The collection surface values were limited by the
availability of acceptable Petri dishes for ber collection. Data presented here are for the parameter combination, within each test variable, that had the greatest inuence on the ber diameter and inter-ber
spacing.
Viscosity testing at controlled temperatures showed
that viscosity decreased with increased temperature.
Viscosities at 225, 234, and 243C averaged 678.3
(149.9), 136.2 (35.2), and 53.8 (4.5) cP, respectively. The viscosity at 225C was signicantly higher
than that at 234 or 243C; the viscosity at 234C was
signicantly higher than that at 243C (p 0.05).
Contrary to the data reported in the literature, increasing the temperature (T) (which decreased the
viscosity) caused an increase in ber diameter [Fig.
2(a,b)]. A concave curve related ber diameter to applied voltage. Both temperature and applied voltage
had less effect on ber diameter compared with the
features of the collection surface, that is, dielectric
strength and surface area. The dielectric strength (DS)
and surface area (SA) had opposing effects. The diameter increased with increasing dielectric strength,
whereas the diameter decreased with increasing surface area. Fiber diameter was nonlinearly related to
distance from the nozzle (De), a parameter previously
investigated for its inuence on ber diameter.10,15
Inter-ber spacing increased with increasing temperature, but was not strongly dependent on applied
voltage [Fig. 3(a,b)]. The surface area of the collection
surface had the most signicant effect of all parameters tested, decreasing with inter-ber spacing. The
collection surface dielectric strengths effect was to
increase inter-ber spacing. Both distance between the
electrodes and the ratio between electrode distance
and collection surface distance exhibited nonlinear
trends with respect to inter-ber spacing.
The complexity of the interdependence of parameters is exemplied in Figure 4. Holding the electrode
distance constant and varying the collection surface to
electrode distance ratio resulted in three different
trends for the dependence of ber diameter on applied
voltage. Only the high ratio ( 0.66) showed a
strong relationship. The other two ratios ( 0.53 and
0.39) had a concave curve and a convex curve, respectively.
The average percent change of ber diameter and

115

Figure 2. (a,b) Fiber diameter dependence on manufacturing parameters voltage (V), temperature (T), ratio of collection surface distance from nozzle to distance between the
electrodes (), surface area (SA), dielectric strength (DS), and
distance between the electrodes (De).

inter-ber spacing for all parameter tests is shown in

Figure 5. Data from all the tests were used to generate
this graph. Contrary to the published literature, applied voltage and viscosity (controlled by temperature) did not have the greatest inuence (indicated by
average percent change) on ber diameter and interber spacing. Within the six parameters of interest, the
collection surface had the greatest impact on ber
diameter and inter-ber spacing. The dielectric
strength of the collection surface induced the greatest
change, and the collection surface area induced
slightly less of a change in both parameters.
The effects of the parameters were compared quantitatively. The average percent change in ber diameter was at least 14 and 15% greater for dielectric
strength and surface area, respectively, than that for
any other parameter. For inter-ber spacing, the induced change was at least 34 and 13% greater for
dielectric strength and surface area, respectively. To
determine if the difference in percent change was
Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

116

MITCHELL AND SANDERS

Figure 4. Fiber diameter dependence on voltage for three

different values of . Result shows the complexity of the
interdependence of the parameters, with three different
shaped curves depending on the value of .

Figure 3. (a,b) Inter-ber spacing dependence on manufacturing parameters voltage (V), temperature (T), ratio of collection surface distance from nozzle to distance between the
electrodes (), surface area (SA), dielectric strength (DS), and
distance between the electrodes (De).

signicant, Student t tests were performed. The null

hypotheses were that the percent change in ber diameter and inter-ber spacing induced by one of the
four parameters voltage (V), distance between the
electrodes (De), temperature (T), or the ratio () was
not different from the change induced by the collection surface dielectric strength (DS) or surface area
(SA). The percent change in ber diameter induced by
DS and SA was signicantly (p 0.05) different from
the change induced by V, T, or . However, it was not
signicantly different from that induced by De. DS
induced a greater change in inter-ber spacing than V,
T, or . The change induced by SA on inter-ber
spacing was not signicant when compared with the
change induced by the other parameters.
Though the ANOVA analysis results demonstrated
that there was no combination of parameters that
would reduce the number needed for predicting the
ber diameter and inter-ber spacing of the end product, it was postulated that MFR was related to at least
Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

some of the test parameters. An increase in MFR

would be expected to increase the ber diameter.
The MFR decreased slightly with increasing and
increased with increasing voltage. Increasing SA increased the MFR, whereas increasing DS resulted in a
parabolic relationship with the MFR (Fig. 6).
Correlations between the MFR and ber diameter,
and MFR and inter-ber spacing were analyzed to
further investigate the relationships among the parameters. Table II shows the correlation coefcient for
MFR verses ber diameter and inter-ber spacing for
all test parameters, except one held constant (, DS, or
V). MFR was well-correlated to the ber diameter for

Figure 5. Relative inuence of the six test parameters. Features of the collection surface, surface area and dielectric
strength, had the greatest inuence on ber diameter and
inter-ber spacing.

A UNIQUE DEVICE FOR CONTROLLED ELECTROSPINNING

Figure 6. Mass ow rate dependence on voltage (V), ratio

of collection surface distance from nozzle to distance between the electrodes (), surface area (SA), dielectric
strength (DS), and distance between the electrodes (De).

and V but not for DS. The inter-ber spacing correlated poorly with MFR for all three test parameters.

Repeatability, predictability, and accuracy

The smallest average ber diameter achieved for an
electrospun mesh in these initial experiments was
5.32 2.71 m, and the smallest average inter-ber
spacing achieved was 4.39 1.41 m. Repeatability
for scaffold production had a mean error of 1.8% and
20.2% for ber diameter and inter-ber spacing, respectively. In all repeated sample tests combined,
there was no signicant difference for diameter (p
0.39) over the range of 8.6314.13 m or for spacing
(p 0.20) over the range of 44.75 82.85 m.
DISCUSSION
A system that allowed control over a range of electrospinning manufacturing parameters was devel-

117

oped. A sensitivity analysis was conducted to determine the role of each feature toward ber diameter
and inter-ber spacing, two features of primary interest for biomedical application of electrospun meshes.
Such a system has the potential to provide an insight
into why there are conicting results in the literature
concerning parameter inuence on mesh morphology.
The system also allowed the repeatability in ber diameter and inter-ber spacing, under closed-loop control, to be evaluated.
The repeatability for ber diameter and inter-ber
spacing achieved with this system is far better than
that reported for other electrospinning systems to
date.8,10,17,30,32,33,42 44 This result is a reection of the
well-controlled manufacturing system used. Further,
the result here, all test parameters having a signicant
effect on ber diameter and inter-ber spacing, points
to the importance of system control in this application.
A lack of thorough system control may help to explain
discrepancies in the literature concerning how manufacturing parameters affect scaffold architecture.
A total of 75 interactions among parameters were
investigated here, and thus, the parameter space was
well-studied. However, to more thoroughly characterize the dependence of specic parameter sets on ber
diameter and inter-ber spacing and to thus generate
more data points for Figures 2 4, and 6, additional
studies would need to be carried out.
Fiber diameter range and inter-ber spacing range
depend on the polymer selected, the viscosity control
method (melting) used, and the collection surface material. Different selections, for example Poly(l-lactide,
-caprolactone) with a solvent, allow for smaller ber
diameter ranges,45 more consistent with those reported to reduce brous encapsulation in vivo.3
Contrary to the literature,10,25 in this investigation,
the ber diameter increased when the applied voltage
was increased. This discrepancy is an expected reection of the closed system used here, as opposed to the
open systems described in previous research. The type
of system affects the relationship between applied
voltage and MFR. It is a general consensus among
researchers that increasing the applied voltage increases the force on the polymer, which in turn deforms the droplet of polymer and initiates ber spinning. It has also been reported that increasing the
TABLE II
Correlations of Fiber Diameter and Inter-Fiber Spacing
with Mass Flow Rate with All Parameters
Except One Held Constant
Parameter not Held
Constant

Fiber Diameter

Inter-Fiber
Spacing

DS
V

0.97
0.21
0.84

0.39
0.06
0.35

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

118

voltage/force increases the MFR of the polymer16,17

and also increases the strain rate on the polymer.16
Thus, if the ow is limited, as in a closed system, then
the polymer strain rate increases with applied voltage,
and smaller bers will result. However, in an open
system, if the voltage is increased, the resulting ber
diameter will depend on the viscosity of the solution.
With high-viscosity solutions, the frictional forces will
limit the ow, resulting in an increased strain rate and
smaller ber diameters with increased voltage. But,
with low-viscosity solutions, as used here, the ow
rate increases while the strain rate remains constant,
resulting in larger ber diameters with increased voltage.
Results here for the correlation between polymer
temperature (inversely related to viscosity as conrmed by the viscosity tests reported here) and ber
diameter are also inconsistent with the literature.35,39,46,47 It is reasonable to expect the ber diameter to decrease with increased temperature, because
the polymer viscosity is decreased, and thus, the strain
rate is increased. However, if the viscosity becomes
sufciently low, the frictional forces in the uid are no
longer a factor, and the MFR will increase with increased temperature, resulting in an increase in ber
diameter. This explanation is conrmed by the high
correlation shown in Table II between MFR, ber diameter, voltage, and relative distance from the nozzle.
Consistent with reports in the literature,10,15 ber
diameter increased with increasing distance from the
nozzle. However, unlike in the literature, the decay
was not asymptotic. The difference here is expected, a
result of where the analysis was conducted. Previous
reports in the literature have focused on analysis of
the stable region of the jet, not the end product mesh.
Fiber diameter is reported to reduce in excess of 100
times9 from the jet secondary bending stability to the
collection surface, in effect, removing the asymptotic
decay characteristic. Thus, the lack of an asymptotic
relationship here is not necessary in conict with the
literature.
It is reasoned that the collection surface had the
most signicant effect on ber diameter of the parameters investigated because of its effect on the electric
eld. The system described here was unique in that
the collection surface was independent of the electrodes, and the collection surface was positioned between the electrodes. The collection surface likely distorted the electric eld (Fig. 7), causing it to deect
around the surface, and thus, causing a change in
voltage. The interdependence of the parameters collection surface dielectric strength, distance between
the electrodes, and applied voltage on ber diameter
supports this interpretation. The electric eld distortion is most likely responsible for the collection surface
dominance in the sensitivity analysis. If the electrospinning device is viewed as a closed-loop system,
Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

MITCHELL AND SANDERS

Figure 7. Expected eld lines between electrodes. The eld

lines distort in the presence of a collection surface (right
image) compared with that when no collection surface is
present (left image).

then the current must ow from the ground electrode

through the jet and collection surface to the positive
electrode. Increasing the dielectric strength in this system conguration increased the resistance between
the electrodes, decreasing the current drawn between
the electrodes. Thus, the work available to strain the
polymer was reduced. The result was an increase in
ber diameter.
The physical size of the collection surface also affected the electric eld between the electrodes. Placing
a dielectric between the electrodes deected the electric eld. Increasing the surface area of the collection
surface increased the degree of eld deection. This
would tend to increase the secondary instability of the
spinning jet, which would result in smaller ber diameters. In addition, increasing the secondary instability (hence wave propagation frequency) decreased
the mesh inter-ber spacing, as conrmed in Figure 2.
The repeatability achieved with this system, 1.8%
for ber diameter and 20.2% for inter-ber spacing, is
expected within the needs for tissue engineering applications. Consistent ber diameters may be important to tissue response,3 and a consistent architecture
will ensure meshes with relatively uniform mechanical properties without locally weak regions. A next
step will be to conduct implant testing to evaluate in
vivo performance using the meshes as coatings for
biomaterial devices, scaffolds for tissue engineering
applications, or transporters for large populations of
cells.
CONCLUSIONS
A controlled electrospinning system was fabricated
to evaluate the sensitivity of bro-porous mesh archi-

A UNIQUE DEVICE FOR CONTROLLED ELECTROSPINNING

119

8.

Doshi J, Reneker DH. Electrospinning process and applications

of electrospun bers. J. Electrostatics 1995;35:151160.
Kenawy el-R, Bowlin GL, Manseld K, Layman J, Simpson DG,
Sanders EH, Wnek GE. Release of tetracycline hydrochloride
from electrospun poly(ethylene-co-vinylacetate), poly(lactic
acid), and a blend. J Control Release 2002;81:57 64.
Buchko CJ, Chen LC, Shen Y, Martin DC. Processing and
microstructural characterization of porous biocompatible protein polymer thin lms. Polymer 1999;40:73977407.
Li WJ, Laurencin CT, Caterson EJ, Tuan RS, Ko RK. Electrospun nanobrous structure: A novel scaffold for tissue engineering. J Biomed Mater Res 2002;60:613 621.
Huang L, Nagapudi K, Apkarian RP, Chaikof EL. Engineered
collagen-PEO nanobers and fabrics. J Biomater Sci Polym Ed
2001;12:979 993.
Matthews JA, Wnek GE, Simpson DG, Bowlin GL. Electrospinning of collagen nanobers. Biomacromolecules 2002;3:232
238.
Yarin AL, Koombhongse S, Reneker DH. Taylor cone and
jetting from liquid droplets in electrospinning of nanobers.
J Appl Phys 2001;90:4836 4846.
Taylor G. Electically driven jets. Proc R Soc Lond A 1969;313:
453 475.
Larrondo L, St John Manley R. Electrostatic ber spinning from
polymer melts. I. Experimental observations on ber formation
and properties. J Polym Sci Polym Phys Ed 1981;19:909 920.
Shin YM, Hohman MM, Brenner MP, Rutledge GC. Experimental characterization of electrospinning: The electrically
forced jet and instabilities. Polymer 2001;42:99559967.
Theron A, Zussman E, Yarin AL. Electrostatic eld-assisted
alignment of electrospun nanobres. Nanotechnology 2001;12:
384 390.
Martin GE, Cockshott ID, Fildes FJT. Fibrillar lining for prosthetic device. US patent 4044404: Imperial Chemical Industries
Limited; Aug 30 1977.
Deitzel JM, Kleinmeyer JD, Hirvonen JK, Beck Tan NC. Controlled deposition of electrospun poly(ethylene oxide) bers.
Polymer 2001;42:8163 8170.
Larrondo L, St John Manley R. Electrostatic ber spinning from
polymer melts. II. Examination of the ow eld in an electrically driven jet. J Polym Sci Polym Phys Ed 1981;19:921932.
Spivak AF, Dzenis YA. Asymptotic decay of radius of a weakly
conductive viscous jet in an external electric eld. Appl Phys
Lett 1998;73:30673069.
Hohman MM, Shin M, Rutledge G, Brenner MP. Electrospinning and electrically forced jets. I. Stability theory. Phys Fluids
2001;13:22012220.
Spivak AF, Dzenis YA, Reneker DH. A model of steady state jet
in the electrospinning process. Mech Res Commun 2000;27:37
42.
Fong H, Chun I, Reneker DH. Beaded nanobers formed during electrospinning. Polymer 1999;40:4585 4592.
Bognitzki M, Czado W, Frese T, Schaper A, Hellwig M, Steinhart M, Greiner A, Wendorff JH. Nanostructured bers via
electrospinning. Adv Mater 2001;13:70 72.
Norris ID, Shaker MM, Ko FK, MacDiarmid AG. Electrostatic
fabrication of ultrane conducting bers: Polyaniline/polyethylene oxide blends. Synth Met 2000;114:109 114.
Srinivasan G, Reneker DH. Structure and morphology of small
diameter electrospun aramid bers. Polym Int 1995;36:195201.
Jin HJ, Fridrikh SV, Rutledge GC, Kaplan DL. Electrospinning
Bombyx mori silk with poly(ethylene oxide). Biomacromolecules 2002;3:12331239.
Shin YM, Hohman MM, Brenner MP, Rutledge GC. Electrospinning: A whipping uid jet generates submicron polymer
bers. Appl Phys Lett 2001;78:1149 1151.
Yarin AL, Koombhongse S, Reneker DH. Bending instability in
electrospinning of nanobers. J Appl Phys 2001;89:3018 3026.

tectural features of interest in biomedical applications,

ber diameter and inter-ber spacing, to manufacturing parameters: distance between the electrodes, distance from the nozzle to the collection surface, applied
voltage, temperature of the melt, collection surface
dielectric strength, and collection surface area. Results
showed that, contrary to what is generally accepted in
the literature, applied voltage and polymer viscosity
(inversely related to temperature) were not the most
dominant parameters inuencing mesh architecture.
Features of the collection surface, including the dielectric strength and the surface area, were more dominant. This nding is an expected result of the unique
system design used, in which the collection surface
was independent of the electrodes and positioned between the electrodes. The collection surface altered the
magnitude and direction of the electric eld. No combination of parameters reduced the number needed
for predicting ber diameter and inter-ber spacing.
However, polymer MFR was well-correlated to ber
diameter for all parameters held constant, except for
the ratio (collection surface distance to nozzle:distance from electrode to nozzle) and voltage. Repeatability using the closed-loop controlled electrospinning system was 1.8% for ber diameter and 20.2% for
inter-ber spacing. In vivo testing is a next step for the
evaluation of these bro-porous meshes for biomaterial or tissue engineering applications.
Assistance conducting viscosity testing from Ari Karchin
is appreciated.

9.

10.

11.

12.

13.

14.

15.
16.

17.

18.

19.

20.

21.

References
1.

2.

3.

4.

5.

6.

7.

Annis D, Bornat A, Edwards RO, Higham A, Loveday B,

Wilson J. An elastomeric vascular prosthesis. Trans Am Soc
Artif Intern Organs 1978;24:209 214.
Wilson GJ, MacGregor DC, Klement P, Weber BA, Binnington
AG, Pinchuk L. A compliant Corethane/Dacron composite
vascular prosthesis: Comparison with 4-mm ePTFE grafts in a
canine model. ASAIO 1993;39:M526 M531.
Sanders JE, Stiles CE, Hayes CL. Tissue response to singlepolymer bers of varying diameters: Evaluation of brous
encapsulation and macrophage density. J Biomed Mater Res
2000;52:231237.
Brauker JH, Carr-Brendel VE, Martinson LA, Crudele J, Johnston
WD, Johnson RC. Neovascularization of synthetic membranes
directed by membrane microarchitecture. J Biomed Mater Res
1995;29:15171524.
Nakayama Y, Nishi S, Ishibashi-Ueda H, Matsuda T. Surface
microarchitectural design in biomedical applications: In vivo
analysis of tissue ingrowth in excimer laser-directed micropored scaffold for cardiovascular tissue engineering. J Biomed
Mater Res 2000;51:520 528.
Matsuda T, Nakayama Y. Surface microarchitectural design in
biomedical applications: In vitro transmural endothelialization
on microporous segmented polyurethane lms fabricated using an excimer laser. J Biomed Mater Res 1996;31:235242.
Gibson PW, Schreuder-Gibson HL, Rivin D. Electospun ber
mats: Transport properties. AIChE J 1999;45:190 195.

22.

23.

24.

25.
26.

27.

28.
29.

30.

31.

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

120

32.

33.

34.

35.
36.

37.

38.

39.

MITCHELL AND SANDERS

Deitzel JM, Kleinmeyer J, Harris D, Tan NCB. The effect of

processing variables on the morphology of electrospun nanobers and textiles. Polymer 2001;42:261272.
Zong X, Kim K, Fang D, Ran S, Hsiao BS, Chu B. Structure and
process relationship of electrospun bioabsorbable nanober
membranes. Polymer 2002;43:4403 4412.
Lu Q-W, Hernandez-Hernandez ME, Macosko CW. Explaining
the abnormally high ow activation energy of thermoplastic
polyurethanes. Polymer 2003;44:3309 3318.
Demir MM, Yilgor I, Yilgor E, Erman B. Electrospinning of
polyurethane bers. Polymer 2002;43:33033309.
Fridrikh SV, Yu JH, Brenner MP, Rutledge GC. Controlling the
ber diameter during electrospinning. Phys Rev Lett 2003;90:
1 4 (144502).
Hohman MM, Shin M, Rutledge G, Brenner MP. Electropinning and electrically forced jets. II. Applications. Phys Fluids
2001;13:22212236.
Larrondo L, St John Manley R. Electrostatic ber spinning from
polymer melts. III. Electrostatic deformation of a pendant drop
of polymer melt. J Polym Sci Polym Phys Ed 1981;19:933940.
Liu H, Hsieh Y-L. Ultrane brous cellulose membranes from
electrospinning of cellulose acetate. J Polym Sci Part B: Polym
Phys 2002;40:2119 2129.

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

40.
41.
42.

43.
44.

45.

46.

47.

Russ JC, Dehoff RT. Practical Stereology, 2nd ed. New York:
Kluwer Academic; 2000. p 127.
Underwood EE. Quantitative Stereology. Reading: AddisonWesley; 1970. p 148 188.
Lee KH, Kim HY, Khil MS, Ra YM, Lee DR. Characterization of
nano-structured poly(epsilon-caprolactone) nonwoven mats
via electrospinning. Polymer 2003;44:12871294.
Buer A, Ugbolue SC, Warner SB. Electrospinning and properties of some nanobers. Text Res J 2001;71:323328.
Theron SA, Zussman E, Yarin AL. Experimental investigation
of the governing parameters in the electrospinning of polymer
solutions. Polymer 2004;45:20172030.
Mitchell SB. Electrospinning controlled architecture scaffolds
for tissue engineering and the effect of scaffold mechanical
properties on collagen synthesis in tissue engineered mitral
valves. Doctoral Dissertation, University of Washington, 2004.
Mo XM, Xu CY, Kotaki M, Ramakrishna S. Electrospun P(LLACL) nanober: A biomimetic extracellular matrix for smooth
muscle cell and endothelial cell proliferation. Biomaterials
2004;25:18831890.
Boland ED, Wnek GE, Simpson DG, Pawlowski KJ, Bowlin GL.
Tailoring tissue engineering scaffolds using electrostatic processing techniques: A study of poly(glycolic acid) electrospinning. J Macromol Sci Pure Appl Chem 2001;38A:12311243.