Figure 1.2 Schematic of various electrodes used to control the electrospinning process.

      Source: Adapted from Teo et al. (2011).

      The collector influences the electric field and is also an important factor in the electrospinning process (Teo and Ramakrishna 2006; Andrady 2008; Ramakrishna 2005; Teo et al. 2011). The simplest collector is a stationary metal plate placed at a fixed distance from the tip. The fibers generally collect as a symmetric circular batch of nanofibers on the plate. Since the plate is grounded, the residual charges on the deposited fibers are dissipated and the mat has high areal density. Moving the collector surface during processing provides some control in the areal density (Andrady 2008). Collectors with grids or charged needles can be used to create patterned nanofiber membranes which consist of regions of high‐ and low‐fiber density. Low‐fiber density occurs in regions where the collector is insulated. Another common collector is a rotating metal drum/mandrel. The rotating surface leads to an even deposition of fibers and a uniform nanofiber mat. The rotating drum can further stretch the fiber leading to reduced diameters as well as introduce alignment of nanofibers. When using high boiling point solvents, e.g. DMF, a rotating collector can provide a longer time for the solvent to evaporate to prevent fiber fusing. Combining electrospinning and mechanical drawing by collecting on a rotating mandrel can affect fiber size. For example, the diameter of PEO fibers spun from chloroform could be reduced from ~1600 to 600 nm by increasing the velocity of the rotating drum (Ogata et al. 2007). Rotating mandrels are also often used to make tubular constructs for potential application as vascular grafts. For tubular constructs, the wall thickness could be controlled linearly with electrospinning time (Teo and Ramakrishna 2006; Andrady 2008; Ramakrishna 2005; Teo et al. 2011).

The material of the collector is also an important consideration that affects the packing density. When nonconductive materials are used as the collector, charge accumulates, and fewer fibers are deposited resulting in lower‐packing densities when compared to fibers collected on conductive surfaces. Even when using conductive collectors, nonconductive behavior can be observed as the fibers (insulating) collect. Sensitivity analysis indicates that the dielectric properties and surface area of the collector are dominant variables that influence fiber diameter and fiber spacing (porosity). Using an auxiliary electrode supplied with AC voltage minimizes the effect of the material on the collector because it reduces the residual change on the deposited fibers. Collecting in a liquid has also been reported and significantly affects the fiber morphology. The choice of liquid can affect the surface characteristics of the fiber (Ramakrishna 2005).

      The porosity of the collector also effects fiber deposition. Fibers collected on metal meshes had lower‐packing densities than smooth surfaces. This effect has been attributed to increased evaporation rate when using a porous collector. As the fibers dry faster, the residual charges persist and repel subsequent fibers. Notably, the topography of the deposited fiber mat will follow the texture of the collector (Ramakrishna 2005). Deposition of two‐dimensional patterned structures or three‐dimensional structures has also been observed. Honeycomb and dimpled structures have been observed using insulating collectors. Two‐dimensional and three‐dimensional patterning is attributed to charge repulsion of deposited fibers. The fibers of the three‐dimensional structures are loosely packed and easily compressed. The conditions to form such three‐dimensional structures are not well understood (Teo et al. 2011).

      Controlling fiber deposition to achieve fiber patterning can be achieved using gap electrodes or open frame collectors. The two parallel electrodes cause the electrostatic field lines in their vicinity to align perpendicular to the edges of the electrodes. The jet aligns with the field lines and deposits uniaxially aligned nanofibers. Charge repulsion of the deposited fibers limits collection of aligned fibers to ~minutes so that samples of thick aligned fibers are difficult to achieve. Arrays of multiple electrodes have been used to achieve more complex patterns, e.g. orthogonal fibers (Teo and Ramakrishna 2006; Andrady 2008; Ramakrishna 2005; Teo et al. 2011).

Alternatively, fibers can be aligned by collecting on a rotating mandrel. The fibers align along the circumference of the mandrel. Typically, high rates of rotation ~1000 rpm are used. To achieve alignment, the rotation of the mandrel must be faster than fiber deposition so that the fibers are taken up on the surface of the mandrel and wound rather than randomly deposited. By replacing a solid mandrel with a wire drum, alignment can be achieved at much lower rates of rotation ~1 rpm. In the case of a wire drum, the fibers are thought to align due to the electric field profile created by the parallel wires. Use of a thin disk with a sharp edge as a collector provides more control of the electrostatic field to align fibers. The electrostatic field lines concentrate toward the knife‐edge and the jet tends to follow the direction of the electric field. As the disk rotates (~1000 rpm), the fibers wind continuously along the knife‐edge with a pitch of 1–2 μm (Ramakrishna 2005). To improve the alignment, the fibers must be collected before the onset of the whipping instability. Auxiliary electrodes can be used to suppress the whipping instability (Carnell et al. 2008). Alternatively, using solvents with low dielectric constants and high purity can suppress the whipping instability (Ogata et al. 2007). Practically, the highly aligned fibers are achieved for a short period of time ~ minutes after which alignment decreases, which may be attributed to fiber repulsion due to charge accumulation. Therefore, for complex patterns, mechanical drawing techniques that avoid the electric field and whipping stability are preferable (Nain and Wang 2013).

      Nanofiber yarns have also been of interest. To produce yarns, electrospun nanofibers can be deposited on water. As the nanofibers are lifted off the water, the surface of tension bundles the fibers into a yarn. Collecting on water flowing in the form of a vortex is a means to achieving continuous yarn production. The disadvantage of this approach is that the yarn must then be dried. Self‐bundling nanofibers have been achieved using AC power. The jet splits and contains both negative and positive segments which bundle together midflight. Twisting the fibers can improve yarn strength. Such twisting can be achieved by collecting on two parallel ring electrodes and rotating one of the rings. Although simple, the length of the yarn is limited using this approach (Abbasipour and Khajavi 2013).

1.4 Effect of Solution Parameters

      1.4.1 Polymer Solution Properties (Molecular Weight, Concentration, Viscosity, and Elasticity)

      The effect of polymer solution properties is generally more significant than process and setup parameters on electrospinning and the resultant fibers. The solution properties, namely viscosity and viscoelasticity, surface tension, and conductivity are affected by the polymer, solvent(s), and additives (e.g. salts, surfactants).

      Although the electrospinning process is relatively easy to implement on a lab scale, many polymer solutions do not form uniform fibers. Issues electrospinning uniform fibers arise when the polymer solution is too dilute and is limited by polymer solubility or when the polymer chains are short or rigid. Electrospinning new materials is typically done ad hoc varying solution properties and process variables; there are no generalizable approaches to predict if a polymer/solvent system will form nanofibers when electrospun. Significant efforts have yielded useful semiempirical approaches for predicting electrospinnability, i.e. production uniform fibers.

It is commonly observed that viscosity influences electrospinning and resulting fiber properties. The viscosity is affected by the molecular weight of the polymer, polymer concentration,