M98-D01 1 A Fundamental Investigation of the Formation and Properties of Electrospun Fibers S.B. Warner, A. Buer, S.C. Ugbolue Department of Textile Sciences, University of Massachusetts Dartmouth, Dartmouth, MA 02747 G.C. Rutledge, M.Y. Shin Departments of Chemical Engineering and Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139 Abstract and Objectives: The goal of this project is development of the fundamental engineering science and technology of electrostatic fiber production, or electrospinning. Electrospinning offers unique capabilities for producing novel synthetic fibers of unusually small diameter and good mechanical performance ( nanofibers ), and fabrics with controllable pore structure and high surface area. Over the course of this project, we aim to: 1. Design and construct electrospinning equipment for controllable and reproducible electrospinning. 2. Clarify the fundamental electrohydrodynamics of the electrospinning operation and connection to polymer solution or melt characteristics. 3. Evaluate the source and role of fluid instabilities postulated to be crucial for producing ultrafine diameter fibers. 4. Characterize the morphology and material properties of polymer fibers produced by electrospinning. 5. Develop techniques for generating oriented fibers and yarns by the electrospinning process. 6. Improve productivity of the electrospinning process. Introduction: Conventional fiber spinning techniques, e.g., melt spinning, dry spinning or wet spinning, rely on mechanical forces to produce fibers by extruding polymer melt or solution through a spinnerette and subsequently drawing the resulting filaments as they solidify or coagulate. Electrospinning offers a fundamentally different approach to fiber production by introducing electrostatic forces to modify the fiber formation process. Electrospinning of polymer fibers has an interesting history. It was first developed about 60 years ago [1]. A number of patents have been issued since then, directed for the most part towards the production non-woven fabrics. However, a review of the most recent literature confirms the extremely limited quantitative technical and scientific information available regarding the underpinnings of this process.
M98-D01 2 Electrospinning is a unique process in that it is able to produce polymer fibers with diameters ranging over several orders of magnitude, from the micrometer range typical of conventional fibers, down to the nanometer range. Owing to the smallness of their diameters, electrospun fibers possess unusually large surface-to-volume ratios and are expected to display morphologies and material properties very different from larger, more conventional fibers. Non-woven fabrics made of electrospun fibers offer unique capabilities to control pore size. In addition, electrospinning is a fast and simple process. Since it only requires small quantities of polymer, it can rightfully be termed a microprocessing technique. The objectives of this research are to develop a quantitative understanding of the electrospinning process, to link the processing conditions and melt/solution characteristics to the structure and properties of the final, electrospun nanofibers, to use this information to demonstrate formation of oriented yarns by electrospinning, and to increase the productivity of electrospinning so that it may be commercially viable. This research was initiated in May of 1998 through collaboration between groups at the University of Massachusetts Dartmouth (UMD) and the Massachusetts Institute of Technology (MIT). Experimental: The Electrospinning Process. The basic configuration of an electrospinning process is illustrated in Fig. 1. It consists, at the minimum, of a high voltage (HV) power supply, capable of 0-30 kv, a charged capillary fed with a polymer solution or melt, and a grounded collection device. Feed of the polymer liquid to the tip of the capillary results in the formation of a droplet at the capillary tip whose size and shape are dictated by surface tension and gravitational forces. The application of high voltage to the capillary produces a surface charge on the droplet which offsets the forces of surface tension, resulting in elongation of the drop, formation of a Taylor cone and, at sufficiently high voltage, ejection of a continuous stream ( jet ) from the tip of the cone. Both electrostatic and fluid dynamic instabilities can contribute to the basic operation of the process. With low molecular weight liquids, Rayleigh break-up of the stream exiting the Taylor cone results in small, charged droplets; this type of process goes by the name electrospraying and has enjoyed widespread commercial success in painting, printing and agricultural technologies. With polymeric fluids, viscoelastic forces stabilize the jet, permitting the formation of small diameter, charged filaments. Under certain conditions, electrostatic instabilities are believed to result in the break-up of the main filament into many smaller filaments, in a process called splaying [2-4]. Although little is currently known about this splaying process, it is thought to be responsible for the unusually small diameter fibers (as small as 40 nm [3]) which can be produced by electrospinning. The fibers are typically laid down in a random fashion on the collecting screen to form a nonwoven fabric.
M98-D01 3 Taylor cone Capillary Jet HV Power Supply Observed Instability Collector Fig. 1: Prototypical electrospinner Electrospinning devices have been assembled at both UMD and MIT. Both apparatuses are equipped with an inclined pipette (Configuration 1) or a syringe pump (Configuration 2) as feed system. With the inclined pipette (tip diameter 0.8 mm), the solution is gravity-fed to the capllary tip at a rate dependent upon the angle of inclination of the pipette, and the head of fluid within the pipette. Although it permits relatively little control over the feed rate, this configuration offers the significant advantage of simplicity for preliminary trials, and has been instrumental at both UMD and MIT in identifying relevant operating parameters. The use of a syringe pump in place of the pipette provides better control over feed rate, through the use of a digitally controlled electric motor that ensures a constant volumetric feedrate. Each pump is connected to a feed capillary of 1 mm inner diameter. The capillary itself may be either conducting or insulating (in which case the HV electrode is fed directly to the solution entering the capillary). Choice of capillary material may affect both electrode efficiency and flow characteristics within the capillary. Both electrospinners have been configured with plexiglass isolation boxes, to minimize extraneous air currents. In addition, the UMD apparatus has been equipped with a Faraday cage, consisting of 0.25 inch mesh metal screen surrounding the plexiglass box, to provide isolation from extraneous electromagnetic disturbances. Collection is on a static or moving screen or plate at a lower potential (typically connected to ground). The spinnerette-to-collector distance has been varied from approximately 1-20 cm, with 10-15 cm separation distance typical. Material of construction of the collector may influence the electric field lines in the vicinity of the collector, as well as ease of release of the fiber, yarn or web; candidates considered at UMD include aluminum, steel, copper, carbon, woven and non-woven fabrics. At MIT, static aluminum and steel collectors have been used to date. In addition to the basic configurations, the feed and collection systems on the MIT device have been modified with 10 cm diameter aluminum plates to create an
M98-D01 4 apparatus with parallel plate geometry (Configuration 3). The use of opposing aluminum plates on both feed and collector provides a simpler and more stable electric field configuration around the spin line. A resistor placed between the collector and ground permits measurement of the current flow down the spin line. Different sizes of capillaries may be employed in the feed line of this apparatus. Laser Doppler Velocimetry. Laser doppler velocimetry (LDV) is a method for measuring fluid flow velocity. Crossed laser beams form an interference fringe pattern in the flowing fluid, from which light is scattered by tracer particles moving with the fluid flow. Light scattered from a tracer particle moving at velocity u exhibits a characteristic frequency shift f which is related to the observed frequencies f 1 and f 2 of the scattered beams: f = (f 1 - f 2 )/λ = 2u sin(k)/λ where k is the half angle of the intersecting beam and λ is the wavelength of laser light used. The LDV technique has been used in the filament production industry for monitoring threadline speed. At UMD, an LV system originally designed to measure water flow in a pipe has been used to measure the spinline velocity. The laser was recalibrated and positioned using a 20X microscope. Preliminary tests at UMD indicated that the technique could be used to measure the spinline velocity in streams having diameters on the order of 0.25 mm. For measurement of velocity, Configuration 2 with a 10 ml syringe and 18 gauge stainless steel needle was used. A 10 cm diameter wire ring placed around the spinline at a position 3 cm downstream from the needle tip was found to stabilize the spinline. The feed system was set on an X-Y table to permit precise alignment of the electrospinning jet in the LDV measuring volume. Typically, 5 sets of 5000 counts were used to determine spinline velocity for this configuration. Strobe Photography. Rapid variations in the spinline in the vicinity of the observed instabilities require high speed photographic techniques in order to completely freeze the action on film. Conventional video or photographic methods provide only a time averaged image. In order to study the electrospinning process with the MIT apparatus, a conventional 35 mm camera (Canon) was equipped with a strobe (GenRad 1538A) which is capable of producing flashes with durations as short as 0.5 µs. High intensity, short duration flashes are used to collect still images, while high flash frequencies aid in the dynamic visualization of the process. Solutions. Four primary sets of polymer/solvent combinations have been considered to date. In each case, the range of concentrations used for each molecular weight of polymer has been varied empirically to obtain good spinnability. Both UMD and MIT are using solutions of polyethylene oxide (PEO) for standardization trials. The UMD group has electrospun PEO (Polysciences, Inc., molecular weights 300K to 4M) dissolved in a 6:1 mixture of isopropanol:water (polymer concentrations from 3-10 wt. %). The MIT group has electrospun polyethylene glycol (PEG, Aldrich, molecular weight 10K) and PEO (Aldrich, molecular weights 100K to 2M) in chloroform (polymer
M98-D01 5 concentrations from 0.5-30 wt. %), a system for which previous efforts have been reported in the literature [5]. In addition, the UMD group has focused attention on polyacrylonitrile (PAN, Polysciences, Inc.) dissolved in dimethyl formamide (DMF, 15 wt. % polymer concentration). The MIT group has focused on polyethytlene terephthalate (PET, Aldrich) dissolved in a 1:1 mixture of dichloromethane:trifluoroacetic acid (polymer concentration 12-18 wt. %). Results Process. PET fibers spun using Configuration 1 are typical of electrospinning production. The stream exiting the Taylor cone is observed to form a stable jet, which then fans out into a cone, apparently as a result of splay. As the driving voltage is increased, the length of the single jet decreases slightly and the apex angle of the Taylor cone increases. For polymer concentrations between 14 and 18 wt.%, the voltage required to initiate dripping is about 4.5 kv and to obtain a steady jet is 6 kv; these threshold voltages are relatively insensitive to polymer concentration in this range. A lower bound estimate of the spinning velocity exiting the spinnerette in this case is 40 m/s, based on volumetric flow rate. The fibers are generally wet when they reach the collection screen and exhibit a 30% weight loss over ten minutes due to continued solvent evaporation. A non-woven fabric consisting of wet PET fibers is readily stretched about 100%; the strain at failure is significantly less for a non-woven fabric of dry PET fibers. Orientation of fibers is dramatically increased upon stretching. Fiber diameters spun from PET as described here are on the order of 2.0±0.5 µm. These fibers are clearly birefingent. An optical micrograph of oriented electrospun PET fibers is shown in Fig. 2. 100 µm Fig. 2 Individual electrospun PET fibers under crossed polars. Spun from 18 wt.% PET at 10 kv.
M98-D01 6 Fig. 3 Unperturbed droplet and electrohydrodynamic cone-jet observed in electrospinning. The droplet deforms into the cone-jet upon application of an electric field. The polymer liquid is 10 wt.% PEG (MW=10,000) in chloroform. The applied electric field is 11kV. Better control over process variables, for purposes of developing a quantitative model of the electrospinning process, is possible using Configuration 3. With PEO, it is possible to demonstrate (see below) that, under the conditions described earlier, splaying is not observed. This permits better control and reproducability of the cone-jet formation during electrospinning. Fig. 3 illustrates the formation of a droplet at the tip of the spinnerette in the parallel plate geometry, and deformation of this droplet into a Taylor cone with a liquid jet emerging from its apex. Using high speed photography, the dynamic instability of the PEO jet is captured. Fig. 4a illustrates the Rayleigh instability which is observed for sufficiently dilute polymer solutions, and which results in droplet formation. Fig. 4b illustrates a different instability observed to occur in PEO solutions of higher concentration or molecular weight. In this case, the jet remains intact but winds in a spiraling motion further downstream. This phenomenon was first reported by Taylor in his seminal work on electrified jets [6] and is conjectured to occur as a result of charge relaxation. Significantly, this rapidly-rotating spiral jet phenomenon is indistinguishable from splay to the naked eye.
M98-D01 7 (a) (b) Fig. 4 (a) Rayleigh instability, 0.5 wt.% PEO (MW=100,000). (b) Charge relaxation instability, 0.5 wt.% PEO (MW=2,000,000). Fig. 5 shows the relationship between velocity and distance from the cone apex, measured for a 15 wt% solution of PAN in DMF, using LDV. The dotted lines show the range of measured velocities observed at each location, while the solid line shows the mean velocity. The mean velocity of the jet as it approaches the region where splay is thought to set in is about 15 m/s. 30.00 25.00 20.00 15.00 10.00 mean V min V max V 5.00 0.00 0.0 20.0 40.0 60.0 80.0 Distance from the Cone Apex (mm) Fig. 5 Velocity of electrospun jet of PAN/DMF (15:85) at 12 kv. Scanning electron microscopy is used to obtain quantitative measurement of fiber diameters. A typical electron micrograph is illustrated in Fig 6 for PAN fibers; similar resolution has been obtained for PET fibers.
M98-D01 8 Fig. 6 SEM of electrospun PAN fibers. The fibers are observed to vary in diameter both along a fiber and from one fiber to the next. Diameter measurements taken from micrographs such as this indicate that the fiber diameters are log-normal distributed, as illustrated in Fig. 7. Fig. 7 Diameter distribution of PAN fibers electrospun at 15 kv from a 15 wt. % DMF solution. Conclusion and Outlook: This report summarizes results obtained between May and September of 1998. The emphasis at both UMD and MIT during this period has been on the construction and instrumentation of complementary electrospinning apparatuses for the purposes of characterizing the electrospinning process and both the fibers and fabrics so-produced.
M98-D01 9 The results obtained to date indicate that the necessary quantification of important process and product variables is feasible. Furthermore, revealing observations concerning the development of important fluid dynamic and electrostatic instabilities have been captured. Additional efforts are underway to improve characterization of the polymer solutions themselves, in terms of surface tension, viscosity, conductivity and permittivity. With these variables under our control, quantitative modeling of the electrospinning process should be feasible, starting from existing equations of electrohydrodynamic theory (MIT). The use of multiple spinnerettes and continuous collection of nanofiber on a conveyor belt capable of speeds between 6 and 60 in/min are being explored to boost productivity and orientability (UMD). A computer-controlled Instron model 5569 has been recently acquired and will be used to characterize the mechanical properties of electrospun fiber, yarns and non-woven fabrics (UMD). Additional techniques are being explored for the purpose of characterizing the structure and morphology of the fibers themselves. These include X-ray diffraction, differential scanning calorimetry and solid state NMR. References: [1] Formhals, A., US Patent 1,975,504 (1934). [2] Doshi, J. and Reneker, D.H., J. Electrostatics, 35, 151 (1995). [3] Srinivasan, G. and Reneker, D.H., Polym. Int., 36, 195 (1996). [4] Reneker, D.H. and Chun, I., Nanotechnology, 7, 216 (1996). [5] Jaeger, R., Bergshoef, M. M., Martin-i-Batlle, C., Schoenherr, D. and Vansco, G. J., Macromol. Symp., 127, 141 (1998). [6] Taylor, G. Proc. Roy. Soc. London A, 313, 453 (1969)