Controlling Topography and Crystallinity of Melt Electrowritten Poly(ɛ-Caprolactone) Fibers

Abstract

Melt electrowriting (MEW) is an aspiring 3D printing technology with an unprecedented resolution among fiber-based printing technologies. It offers the ability to direct-write predefined designs utilizing a jet of molten polymer to fabricate constructs composed of fibers with diameters of only a few micrometers. These dimensions enable unique construct properties. Poly(ɛ-caprolactone) (PCL), a semicrystalline polymer mainly used for biomedical and life science applications, is the most prominent material for MEW and exhibits excellent printing properties. Despite the wealth of melt electrowritten constructs that have been fabricated by MEW, a detailed investigation, especially regarding fiber analysis on a macro- and microlevel is still lacking. Hence, this study systematically examines the influence of process parameters such as spinneret diameter, feeding pressure, and collector velocity on the diameter and particularly the topography of PCL fibers and sheds light on how these parameters affect the mechanical properties and crystallinity. A correlation between the mechanical properties, crystallite size, and roughness of the deposited fiber, depending on the collector velocity and applied feeding pressure, is revealed. These findings are used to print constructs composed of fibers with different microtopography without affecting the fiber diameter and thus the macroscopic assembly of the printed constructs.

Introduction

Melt electrowriting (MEW) is an additive manufacturing technology that enables the fabrication of fibrous constructs in a layer-by-layer technique based on a computer-aided design model.¹ In detail, MEW processes a molten polymer in a syringe by conveying it with a defined pressure toward a conductive spinneret. An electric field applied between the spinneret and an opposed collector causes the polymer to form a Taylor cone at the tip of the spinneret leading to the generation of a molten polymer jet, which is accelerated toward the collector plate. The jet solidifies on its way to the collector, where it is deposited as a fiber with diameters in the lower micrometer range.^2–4 The computer-aided collector moving in x- and y-direction enables the fabrication of fiber-based constructs in a direct-writing manner. Using repetitive, pattern-based fiber deposition, the fabrication of porous constructs with high accuracy is possible. Different laydown patterns, such as the most commonly used box structures,^5,6 but also triangular and other shapes,⁷ enable different construct architectures. Tissue engineering (TE) is currently the most important application for melt electrowritten constructs used, for example, for cartilage,⁸ skin,⁹ nerves,¹⁰ cardiac tissue,¹¹ and blood vessel¹² regeneration. Since cells cannot be incorporated directly during MEW fabrication in case of nonphysiological melting temperatures of the utilized polymers, these constructs are seeded with cells after fabrication. Pore size and fiber diameter in cellular dimensions,¹³ which can be achieved via MEW as well as material roughness and thus surface topography, can alter cellular behavior¹⁴ and should be controlled.

Poly(ɛ-caprolactone) (PCL) is the most widely investigated polymer for MEW and, therefore, considered as the “gold standard”.¹⁵ This polymer shows excellent printing behavior and has a melting temperature of 59°C to 64°C.¹⁶ Typical pore sizes of PCL-based MEW constructs range from a few micrometers to several millimeters, while fibers have diameters ranging from 5 to 40 μm.^17–19 However, also submicrometer PCL fibers were achieved.¹⁷ Only recently, we were even able to reduce the pore size of a box-shaped MEW-construct down to 40 μm, with high consistency and precision throughout the whole construct.¹³ These PCL fiber-based constructs showed auspicious immunomodulatory, pro-healing properties solely through structural control in an in vitro study of human macrophages. Although PCL is the most studied material for MEW, more detailed investigation is still needed to better understand and control the semicrystalline polymer's printing behavior.

Thus, this study addresses the effect of adjustable printing parameters such as spinneret diameter, applied feeding pressure, and collector velocity to influence fiber characteristics on a macro- and microlevel, such as fiber diameter and topography as well as mechanical properties and crystallinity of PCL fibers.

Materials and Methods

Melt electrowriting

A custom-built MEW printer, as described in previous studies,^6,20 was used at standard room temperature conditions in the range of 20°C ± 3°C and relative humidity of 40% ± 10%. In detail, medical-grade PCL pellets (PURASORB PC 12, Lot# 1412000249, 03/2015; Corbion, Inc., Gorinchem, Netherlands) were heated to above 70°C inside a syringe to achieve a homogeneous melt. For processing, the molten PCL was heated to 85°C and the polymer was changed every 5 days. By applying pressure (0.5, 1.0, 2.0, 3.0, and 4.0 bar) to the syringe, the melt was extruded through a flat-tipped, conductive spinneret (25G d_i = 250 μm, 27G d_i = 200 μm, 30G d_i = 150 μm). Glass slides were used as grounded collector, while a positive 4 kV voltage was applied to the spinneret at a spinneret tip to collector distance of 1.2 mm.

For the determination of the critical translation speed (CTS), a series of parallel lines were printed, while the collector velocity was increased stepwise (10 mm min⁻¹ per step) every four lines. The lowest speed with all four lines being visually straight under stereomicroscopic assessment (Discovery V20; Carl Zeiss Microscopy GmbH, Oberkochen, Germany) was noted as CTS (n = 4 each spinneret).

Diameter and topography of platinum sputtered (Leica EM ACE600 sputtering unit; Leica Microsystems, Wetzlar, Germany) PCL fibers were measured under a Zeiss Crossbeam 340 scanning electron microscope (SEM; Carl Zeiss Microscopy GmbH). Fibers of all combinations of the three nozzles (25G, 27G, and 30G), five different pressures (0.5, 1.0, 2.0, 3.0, and 4.0 bar), and six collector velocities relative to CTS (1.2 × , 1.5 × , 2.5 × , 5.0 × , 10.0 × , and 15.0 × CTS) were determined directly during SEM imaging. Sixteen parallel fibers with a length of at least 50 mm and with three fibers deposited on top of each other were printed before the fiber diameters of eight arbitrarily chosen fibers were measured at their central area. This experiment was repeated three times (N = 3, n = 8).

X-ray diffraction measurements

Dense fiber arrays with 100 parallel fibers, a spacing of 50 μm in between and 7 fibers on top of each other, were printed on glass slides. X-ray diffraction (XRD) patterns were recorded on a diffractometer (D5005; Siemens, Karlsruhe, Germany) with Cu-K_α radiation (λ = 0.154 18 nm) and the following parameters: voltage of 40 kV, current of 40 mA, 2ϴ range from 10° to 60°, step size of 0.02°, and scan rate of 8 s per step. The crystallite size was calculated using the Debye-Scherrer equation.

Atomic force microscopy measurements

Before MEW fiber production, glass slides were cut into pieces (11 × 11 mm) that were reassembled without gaps on the collector. Fiber arrays (40 parallel fibers, 5 fibers on top of each other) were printed onto the glass slides, and the fibers were carefully cut with a knife along the glass edges to collect small pieces of glass with printed fibers on top. The quantitative mechanical characterization of the PCL fibers were performed under ambient conditions using a MultiMode 8 Scanning Probe Microscope (Bruker AXS, Santa Barbara) equipped with a Nanoscope V controller and a 120 μm piezoelectric scanner. The type of atomic force microscopy (AFM) probe used was TAP150A, designed for measuring the Young's moduli (E) in the range of 5 to 500 MPa. The spring constants of the silicon cantilevers were calibrated by using thermal tuning, while the radius of the AFM tips were determined by measure of a tip characterizer sample (Ti roughness sample and two-component polymer sample consisting of polystyrene and polyolefin elastomers) and using the tip qualification function in NanoScope Analysis software. The Young's modulus of elasticity was extracted from the recorded force-distance curves by application of hertz model (Supplementary Fig. S1, Equation 1). At least two randomly selected fibers were analyzed by AFM to determine the elastic modulus and fiber roughness (root mean square [RMS]). RMS average of height deviation was taken from the mean image data plane using the Bruker software NanoScope Analysis.

Box-shaped PCL construct fabrication

Box-shaped constructs with two fiber surface topographies were directly written as 3D structures by alternating layer deposition via 0° and 90° layers (10 × 10 mm²) with a similar G-code-based motion path and filament deposition onto the collector plate as has been previously described.¹⁷ The spacing between fibers on each layer was set to 300 μm to finally reach a pore size of 150 μm between fibers with different topographies. After each layer in x- and y-direction, the pressure-velocity pair was changed. In between each layer, a jet stabilization was performed in an off-sample with the changed settings. While acceleration voltage, collector distance, heating temperature, and spinneret diameter were the same as described above, applied feeding pressure and collector velocity were varied as follows: 0.5 bar and 484 mm min⁻¹ (1.5 × CTS) or 2.0 bar and 1763 mm min⁻¹ (15 × CTS).

Statistical methods

Statistical analysis was performed using the software “SigmaPlot” (Version 12.5). First, normal distribution of the data was proved using the Kolmogorov test. This was followed by applying a one-way ANOVA with post hoc Tukey test. Significant differences are indicated with p < 0.001.

Results and Discussion

Systematic analysis of process parameters on fiber diameter and morphology

Fiber diameter analysis

Not only size²¹ and shape²² of the spinneret but also collector velocity, melt temperature, applied pressure, and many other process parameters influence the resultant fiber diameter in MEW.^18,23 Printing above the CTS, where the jet speed matches the collector speed, can also be used to further reduce the diameter due to mechanical stretching.¹⁸ As the interplay between these parameters is not fully understood yet, this study aims to gain a better understanding of printing PCL fibers utilizing MEW with diameters below 40 μm, mainly down to the single-digit micrometer range, using spinneret diameters below the most commonly used 22G nozzle.¹⁵ To limit the number of parameters, only adjustments of applied feed pressure and collector velocity were performed for different spinneret diameters while the voltage, spinning distance, and melt temperature were kept constant at 4 kV, 1.2 mm, and 85°C, respectively.

The CTS decreased with increased feeding pressure as well as an increased inner diameter of the spinneret (25G > 27G > 30G) due to the correlation between pressure and mass flow (dm/dt) through the spinneret (Supplementary Fig. S2A).²⁰ At collector velocities above the CTS, the fiber was subjected to mechanical stretching that led to reduced fiber diameters with increased collector velocities for each spinneret (Supplementary Figs. S2B–S5). Spinnerets with larger inner diameters (25G > 27G > 30G) resulted in increased mass flow (dm/dt) through the spinneret and, thus, in thicker fibers. Furthermore, with increased applied feeding pressure, a higher amount of molten polymer (dm/dt) was pressed through the spinneret and thus deposited on the collector as fiber with a larger diameter. Consequently, for each analyzed spinneret, the largest fiber diameter was achieved for the lowest collector velocity and highest applied feeding pressure (4 bar, 1.2 × CTS), while the thinnest fiber was printed with the highest collector velocity and lowest pressure (0.5 bar, 15 × CTS) resembling the highest and lowest mass flow (dm/dt), respectively. As a result, the printed fibers exhibited diameters ranging from about 30 to 1 μm (Table 1). In accordance to literature, unstable printing conditions with the so-called “long-beading” phenomenon and huge fiber diameter fluctuations²⁰ were observed, especially for the thickest spinneret used (25G) and high feeding pressure values between 2 and 4 bar, which is represented in Supplementary Figure S2 by relatively high standard deviations in the fiber diameter measurements. Overall, the presented results are in accordance with a former study on thicker PCL fibers (>50 μm in diameter) reporting similar effects of collector velocity and applied feeding pressure on melt electrowritten fibers.²⁴

Table 1.

Maximum (fØ_max) and Minimum (fØ_min) Fiber Diameters for 25G, 27G, and 30G Spinnerets

	fØmax (4 bar, 1.2 × CTS), μm	fØmin (0.5 bar, 15 × CTS), μm
25G	33.64 ± 2.60	1.85 ± 0.17
27G	15.91 ± 1.11	0.10 ± 0.08
30G	5.88 ± 0.67	1.02 ± 0.17a

Experiments were performed with constant parameters: 4.0 kV acceleration voltage, 1.2 mm spinning distance, and 85°C PCL temperature (N = 3, n = 8).

^a 2 bar and 15 × CTS, since printing with 4 bar and 15 × CTS was not possible.

CTS, critical translation speed; PCL, poly(ɛ-caprolactone).

As our experiments show, it was not possible to print continuously straight fibers without pulsing and jet breakdown with the 30G spinneret using a feeding pressure below 2 bar and increasing collector velocities, while maintaining the spinning distance (1.2 mm), as well as the applied voltage (4.0 kV). Hochleitner et al. demonstrated the need for an equilibrium between feeding pressure and the electrical field to prevent fiber pulsing.²⁰ Besides, especially for the 30G spinneret with its small mass flow (dm/dt) compared to thicker spinnerets, the CTS values are high and, therefore, also require higher collector velocities for the fiber diameter reduction. Hence, an adjustment of the electrical field would possibly result in feasible printing conditions for the 30G spinneret even for pressures below 2 bar. However, in our study, spinning distance and electrical field were kept constant for all three spinnerets for ease of comparison to maintain comparability.

Fiber topography analysis

So far, the influence of spinneret, feeding pressure, and collector velocity on the topography of melt electrowritten PCL fibers have not been studied systematically. To achieve this, SEM images of single, printed fibers produced with the 25G, 27G, and 30G spinneret, different feeding pressures, as well as collector velocities are taken not only to depict a decreasing fiber diameter with decreasing applied pressure and increasing collector velocity but also to exhibit changes in the topography of the deposited fibers (Supplementary Figs. S3–S5).

The influence of process parameters on the fiber topography is particularly evident in Figure 1, where the topography of fibers printed with the three different spinnerets (25G, 27G, and 30G), but the same applied pressure (4 bar) and an adjusted collector velocity, are shown in higher magnifications. Especially for the low collector velocities, the fibers showed typical spherulites, which are characteristic for a super-oriented topography of semicrystalline polymers like PCL.²⁵ These structures with round or polygonal shapes occur when the melt of a polymer with the ability to crystallize is cooled below the melting point (T_m).²⁶ We further observed a more rough fiber surface with increasing collector velocities. As suggested by reports on other melt spinning techniques, the polymer jet is considered to be stretched during its solidification, which leads to polymer chains aligning perpendicular along the length of the filament.^27,28 Solidification processes take place faster for fibers with lower material content, that is, with smaller fiber diameter, which in turn explains a more pronounced occurrence of perpendicularly aligned lamellas for smaller fibers. Supplementary Figure S6 shows a reproducible morphological fiber pattern at day 0 and 5 using the same PCL polymer that maintained at a constant temperature of 85°C throughout the 5-day timeframe. This indicates that morphological differences are solely based on changes in adjusted parameters (spinneret, applied pressure, and collector velocity) and not due to crystallization processes that could occur during the permanent heating of PCL.

FIG. 1.

SEM images of single fibers (stack of 3 fibers) printed with the 25G, 27G, and 30G spinnerets and applied pressure of 4 bar, while the collector velocity was increased stepwise. Fibers show changes in their topography depending on the applied pressure and collector velocity. Scale bars correspond to = 5 μm. SEM, scanning electron microscope.

The fiber diameter studies revealed parameters for the deposition of straight and continuous PCL fibers with a uniform diameter. A diameter of 5 μm was chosen and achieved by using different spinnerets and adjusted feeding pressure and collector velocity. The 5 μm fibers generated with different parameters were analyzed regarding their topography (Fig. 3). Since fibers with a diameter of 5 μm can be printed with the 30G spinneret only under one parameter set (4 bar, 1.5 × CTS), these fibers were excluded for the following comparison. In addition to SEM analysis, the fibers were examined with XRD for crystallization and with AFM for the determination of fiber roughness and Young's modulus (Figs. 2 and 3). Contrary to the conventional determination of Young's modulus on whole MEW constructs by means of tensile testing, this analyzis method was not possible with single fibers, so that the mechanical fiber properties were deteremined by AFM measurements.

FIG. 3.

Overview of 5 μm fibers printed under different conditions. The combinations of the 25G, 27G, and 30G spinnerets with the respective applied pressures and collector velocities to obtain fibers with 5 μm in diameter are shown. The following parameters were further specified for these 5 μm fibers: Crystallite size (s_cry; in nm), Young's modulus (E; in MPa), roughness (R; in nm), average fiber diameter (fØ, in μm), and collector velocity (v_col, in mm/min). Scale bars correspond to = 2 μm.

FIG. 2.

Crystallite size, Young's modulus, and fiber roughness of 5 μm fibers printed under different conditions. Printing parameters were adjusted for the 25G and 27G spinneret. While the crystallite size was determined with XRD measurements, Young's modulus and fiber roughness were determined using AFM measurements. *p < 0.001. AFM, atomic force microscopy; XRD, X-ray diffraction.

The detected rise in crystallite size with increasing collector velocity is assumed to occur due to mechanical stretching leading to lamellas perpendicular aligned to the printing direction. The 5 μm fibers have almost identical Young's moduli when considering experimental errors. However, there is a distinct trend for the roughness of the fibers depending on the printing conditions, mainly mediated by the collector velocity. While the fiber roughness is around 15 nm for collector velocities around 500 mm min⁻¹, this significantly decreases down to 5 nm with increasing collector speed. As two parameters, the applied pressure and collector velocity, need to be adjusted simultaneously to print fibers with 5 μm in diameter, it was impossible to clearly identify which of the parameters is responsible for changes in crystallite size and fiber roughness.

Different fiber topographies within one construct

Based on the aforementioned analysis, parameters could be identified that enable changing fiber topography on a microlevel by adjusting feeding pressure and collector velocity while maintaining a constant fiber diameter. Figure 4 shows a box-shaped PCL construct with fibers of around 5 μm in diameter, but two different fiber topographies. Here, the fibers were printed with the same spinneret but two different feeding pressure and collector velocity pairs as controlling parameters, combined in an alternating layered construct enabling a change of topography without the need for a printhead change. Some fibers show a “rough” topography, similar to elephant skin, and some a “smooth” fiber topography. Both fiber variants were combined in one construct leading to three combinations of fiber morphologies [Fig. 4(1)–(3)] at the intersections of printed fibers in x- and y-direction: smooth-smooth (1), smooth-rough (2), and rough-rough (3). As we kept the fiber diameter constant while changing feeding pressure and collector velocity, the macroscopic shape of the construct was not altered, and only changes on a microscopic level could be performed.

FIG. 4.

MEW construct with control over fiber topography using two pressure-velocity pairs (A). Pressure (0.5, 2.0 bar) and collector velocity (484, 1763 mm min⁻¹) were adjusted to print 5 μm fibers in a 150 μm pore size construct using the 25G spinneret (B). Two single constructs, each with 300 μm pore size, were printed in one construct, while after each layer (in x- and y-direction), the printing conditions (pressure and collector velocity) were changed. Whenever changing the printing conditions, the jet was stabilized with the adjusted parameters in an off-sample next to the construct. Scale bars correspond to = 50 μm (A) and 2 μm [(1), (2), (3)]. MEW, melt electrowriting.

In general, not only surface chemistry, coatings, and mechanical properties but also material roughness is well-known to affect cellular adhesion to biomaterials,²⁹ which is an important aspect for a wide range of medical implant and TE applications. Indeed, surface roughness regulates material wettability, which in turn impacts protein adsorption and cell adhesion.^30,31 Material roughness not only effects the adhesion of cells but also their morphology that was recently shown for solution electrospun PCL fibers by Metwally et al.³² Here, osteoblasts adhered to the fiber via cell filopodia showing a higher number of attached cells to porous-structured PCL fibers. Generally, cellular adhesion is increased on more rough surfaces mainly due to a higher available surface area.³³ Furthermore, also PCL degradation processes influence cell responses due to the nonenzymatic hydrolytic cleavage of ester bonds and thus the formation and release of acidic by-products, which can lead to inflammatory responses in vivo.³⁴ This degradation process correlates with the degree of PCL crystallinity and the amount of crystalline and amorphous regions within the material.

Therefore, a detailed characterization and understanding of the interface of melt electrowritten constructs, as shown in this presented study, could be a first helpful step for several TE applications to fine-tune cell–material interactions without affecting the macroscopic structure.

Conclusions

In summary, this study systematically analyzed the influence of spinneret diameter, feeding pressure, and collector velocity on the diameter and topography of deposited melt electrowritten PCL fibers and how these parameters affect the mechanical properties and crystallinity of the directly written PCL. The measurements revealed a correlation between the mechanical properties, crystallite size, and deposited fibers' roughness, depending on the collector velocity and applied feeding pressure. The topography of the fibers becomes smoother for increased collector velocities. The controlled direct-writing of PCL resulted in fibers with over an order of magnitude difference in diameter by adjusting the diameter of the spinneret, feeding pressure, and collector velocity. We succeeded in generating a MEW construct with the same macro- but a different microstructure and thus shed light on a MEW phenomenon that was not investigated in detail yet.

Authors' Contributions

C.B., G.H., J.G., and T.J. designed research; C.B. performed the major part of research; J.W. analyzed XRD measurements; G.H. provided support for MEW; V.S. and F.W. performed AFM measurements; C.B. prepared the article in consultation with J.G. and T.J.; J.W., G.H., V.S, and F.W. provided feedback on the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the European Research Council (ERC) (consolidator grant Design2Heal, Contract No. 617989) and by the German Research Foundation (DFG) (“State Major Instrumentation Programme,” funding for the SEM Zeiss Crossbeam 340, INST 105022/58-1 FUGG).

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6