Melt Electrospinning of Poly(ε-caprolactone)/Poly(ethylene oxide)/Poly(ethylene glycol) Blend: Effects on Structure and Cellular Compatibility

Abstract

This work aims to investigate the effects of the melt electrospinning process on the properties of ternary blend poly­(ε-caprolactone)/poly­(ethylene oxide)/poly­(ethylene glycol) (PCL/PEO/PEG) fibers. The utterly different morphology of the specimens before and after melt electrospinning acknowledges the significant effect of melt electrospinning on the sample’s properties. However, it does not chemically alter the samples. The Scanning Electron Microscope (SEM) images showed that PEO and PEG behave differently in the PCL matrix, producing distinctly different morphologies in the respective binary blend fibers. Differences in morphology also lead to differences in the crystallization behavior of the specimen. Wide-angle X-ray diffraction (XRD) and differential scanning calorimetry (DSC) also reveal that although PEO and PEG are fundamentally the same polymer with different molecular weights, their behaviors in the PCL matrix are entirely different; PEG mostly remains in the amorphous state, while PEO can form its crystalline structure. Hence, the final morphology of the ternary melt electrospun fiber is a combination of the morphology of PCL/PEO and PCL/PEG binary blends. Preliminary biological tests showed that the porosity created by the removal of the PEO and PEG hydrophilic phases improved fibroblast metabolic activity, highlighting the potential of this system for further investigation in bone tissue engineering applications.

1. Introduction

Bone defects beyond the capacity of natural repair remain a major challenge in regenerative medicine. Although conventional methods such as autograft and allograft are widely used, limitations such as resource scarcity, donor site complications, and risk of immune response highlight the need to develop alternative solutions. In this regard, tissue-engineered scaffolds have emerged as one of the most promising approaches, especially when they can simultaneously mimic the hierarchical structure, appropriate porosity, and biological properties of bone tissue. ,

Among synthetic polymers, poly­(ε-caprolactone) (PCL) has been widely studied in biomedical applications due to its high biocompatibility, favorable processability, and good mechanical stability. However, the hydrophobic nature, limited bioactivity, and especially the very slow degradation rate of this polymer limit its widespread application in tissue engineering, particularly in bone regeneration, which requires matching the scaffold’s degradation rate to the tissue regeneration process. , Therefore, the design of multicomponent systems and structural engineering through composition and process modification has been proposed as an effective solution to improve PCL performance.

One efficient approach in this field is the use of polymer blending with hydrophilic polymers, such as poly­(ethylene oxide) (PEO) and poly­(ethylene glycol) (PEG). , Despite their chemical similarity, these polymers could exhibit kinetic behaviors due to differences in molecular weight, leading to the formation of complex, controllable morphologies in multiphase systems. , In addition, these hydrophilic components can act as sacrificial phases and, upon removal, form porous intrafibrous structures that play a key role in improving cellular penetration, nutrient transport, and biological response of the scaffold.

In addition to the composition design, the process method also plays a decisive role in the scaffold’s final structure. While solution electrospinning has been used in numerous investigations on polymer blends, − melt electrospinning of blends has received much less attention. Melt electrospinning presents several key advantages over solution electrospinning, most notably its solvent-free nature, which eliminates the risk of residual toxicity and enriches the biocompatibility of the fabricated fibers. This technique aligns with environmentally sustainable manufacturing approaches, avoids the need for postprocessing phases such as solvent evaporation or drying, and significantly improves process safety. Due to these features, melt electrospinning is a scalable and appealing method for industrial and biological applications. It is fundamental to study the properties of polymer blends during melt electrospinning, as there is limited information on this process, which limits morphology control.

In this study, for the first time, the behavior of a ternary PCL/PEO/PEG system in the melt electrospinning process has been systematically investigated; an approach that, beyond conventional studies based on single polymers or binary systems, allows for the design of multiphase structures with precise control. The main focus of this research is to elucidate the role of the molecular weight difference between PEO and PEG in the formation of morphology, crystal organization, and, ultimately, the biological response of the scaffolds. Also, the relationship between the resulting structural features and the initial biological performance, especially in improving cellular metabolic activity, has been evaluated. The results of this research pave the way for the design of engineered scaffolds with optimal performance in bone tissue engineering applications.

2. Materials and Methods

2.1. Materials

The poly­(ε-caprolactone) (PCL) Capa 6500 was obtained from Perstrop (Canada) with a mean molecular weight of 50,000 g/mol in granular form. Poly­(ethylene glycol) (PEG) and Poly­(ethylene oxide), with M _n 20,000 and M _v 600,000 g/mol in flakes and powder, respectively, were purchased from Sigma-Aldrich, USA. Table summarizes the glass transition (T _g), melting (T _m), and crystallization (T _c) temperatures of the neat polymers. Values are obtained from differential scanning calorimetry (DSC) under a nitrogen purge; T _g and T _m are extracted from the first-heating scan and T _c from the subsequent cooling scan (see Section for conditions).

1. Thermal Properties of Neat Polymers: Glass Transition (T _g), Melting (T _m), and Crystallization (T _c) Temperatures.

polymer	T g [°C]	T m [°C]	T c [°C]
PCL	–51.4 ± 0.7	63.9 ± 1.0	27.5 ± 0.3
PEO	–51.9 ± 1.7	69.3 ± 0.0	42.5 ± 1.6
PEG	–55.1 ± 2.0	69.1 ± 0.1	43.9 ± 1.9

2.2. Sample Preparation

2.2.1. Melt Blending

In this study, ternary blends consisting of PCL, PEO, and PEG were prepared with three different compositions to explore the effect of blend ratio on the final structure and properties. Specifically, blends with PCL/PEO/PEG with weight ratios of 90/5/5, 67/16.5/16.5, and 50/25/25 were initially prepared and evaluated. To better understand the individual contribution of each secondary polymer (PEO or PEG), corresponding binary blends of PCL/PEO and PCL/PEG were also studied, with ratios of 90/10, 75/25, and 50/50 (w/w). The tested compositions were determined using JMP Statistical Discovery software (version 18) and based on the Design of Experiments (DOE) approach to cover a wide range of component ratios and to investigate the effect of each component on the behavior of the system (details are provided in Figure S1 of Supporting Information). In this paper, the ternary blend with a composition of 67/16.5/16.5 (PCL/PEO/PEG) and the binary blends with compositions of 75/25 (PCL/PEO and PCL/PEG) were selected for detailed analysis. The selected compositions were chosen due to their balanced characteristics, representing a middle ground among the compositions studied. They avoid the issues observed in both lower and higher amounts of dispersed components, such as unclear morphology or excessively large cavities after dissolution. Furthermore, the X-ray diffraction (XRD), Fourier transform infrared (FTIR), and DSC results for these compositions exhibit appropriate peak intensities, making them the most suitable candidate for discussing the physicochemical properties of the ternary and respective binary blends and the subsequent melt electrospun samples.

To prepare the samples PCL, PEO, and PEG were dried at 40 °C in a vacuum oven overnight to avoid hydrolytic scission during melt processing. Subsequently, 15.41 g of PCL was poured into the Brabender-30 mL batch mixer (Plasticorder DDRV501, Brabender). The compounding was conducted at 70 °C using a screw speed of 50 rpm and beneath a nitrogen blanket. Next, the molten PCL was gradually mixed with 3.79 g PEO and 3.79 g PEG for 15 min under the same conditions to prepare the ternary blend. In preparing binary blends, 17.25 g of PCL was mixed with 5.75 g of PEO or PEG. Additionally, the neat polymers went into melt mixing before electrospinning for comparison purposes. The estimated average shear rate under the employed processing conditions is 25 s^–1. After cooling the samples, they were cut into small pieces to prepare for the melt electrospinning process.

2.2.2. Melt Electrospinning

The prepared specimens were used immediately after melt mixing for melt electrospinning to minimize any risk of morphology change. The blend was loaded into an 9 mL stainless steel syringe surrounded by a heating jacket. A cylindrical collector was linked to the positive electrode of the high-voltage power source, while the nozzle was secured to the ground wire. A schematic representation of the melt electrospinning apparatus utilized in the present study, highlighting its principal parts, is shown in Figure . The operation was started following 1 h of heating and equilibrating at 150 °C. After approximately 10 min of heating, a small amount of molten polymer was extruded until no visible air bubbles were observed in the extrudate, to reduce the amount of air inside the syringe and provide a more stable melt flow for the subsequent fiber fabrication. Following the 1 h equilibration period at 150 °C, an additional small amount of molten polymer was extruded immediately before applying the high voltage to discharge the initial melt from the needle before fiber collection. The results of time-sweep oscillatory tests showed that this heating step did not cause any signs of thermal degradation or reduced structural stability in the samples (Figure S2 of Supporting Information). The applied voltage was adjusted to 7.5 kV for the flow rate of 500 μL/h. The fibers were collected on a rotating cylinder collector at 20 rpm, 3 cm away from the 20G needle. The suffix BME in this paper stands for “before melt electrospinning”. AME refers to the samples “after melt electrospinning” or to the melt electrospun fibers. PEO seems to be too viscous in the molten state to be melt electrospun alone. Although PEG is melt electrospinnable, the final fibers are too fragile to be detached from the collector without breaking. On the other hand, PCL is readily electrospinnable, and the final fibers are flexible.

1.

A schematic view of the melt electrospinning apparatus: (1) melt-blended sample loaded stainless steel syringe, (2) stainless steel heater, (3) the temperature-controlled syringe, equipped with an in-line heater, was secured to the syringe pump to ensure a consistent flow of the molten mixture, (4) needle, (5) pump, (6) temperature controller, (7) collector and (8) high voltage supplier.

2.3. Morphological Analysis

To examine the morphology of the blends in the cross-section and surface of the samples before and after melt electrospinning, a Scanning Electron Microscope (SEM) (TM3030Plus, HITACHI) was used at a voltage of 15 kV. In order to prepare the samples, the melt electrospun fibers were immobilized into epoxy. Later, samples were microtomed using a Leica RM2165 fitted with a glass knife in a liquid nitrogen cryo-chamber. They were then immersed in water as an effective solvent for extracting PEO and/or PEG from the blends. PEO and PEG exhibit very similar solubility parameters (δ ≈ 19–21 MPa^1/2) due to their identical chemical backbone, which prevents selective dissolution of one component over the other in aqueous media. − Beyond a week of immersion, specimens were put under vacuum for 2 days to be thoroughly free of water. Finally, to better scan the PEO or/and PEG empty regions, the cross-section of the samples was coated with chromium of about 20 nm in thickness. To compare the morphology of the blends before and after melt electrospinning, blends were microtomed and soaked into water following the exact procedure employed for the fibers before the SEM test. The ImageJ software was then used to evaluate the pore size and mean fiber diameter.

The equivalent diameter of the footprints of the dispersion components after immersion in water, D _eq,i , is estimated through eq

$$D_{\mathrm{eq},i}=\sqrt{\frac{{4\times A}_i}{\pi }}$$ 1

where A _i is the measured domain area.

2.4. Fourier Transform Infrared (FTIR)

A PerkinElmer spectrometer (Waltham, MA, USA) was used to run a Fourier transform infrared (FTIR) test in Attenuated Total Reflection (ATR) mode at room temperature, with a wavenumber ranging from 600 to 4000 cm^–1, using 16 scans, with a nominal resolution of 4 cm^–1. In order to mitigate experimental errors, the test was conducted thrice, and the mean values were recorded.

2.5. X-ray Diffraction (XRD)

The crystalline structure of the specimens was assessed employing an Empyrean diffractometer (Malvern Panalytical) with Cu Ka radiation generated at 45 kV and 40 mA, a 2θ spectrum of 3°–60° with a 0.0131° step size. This test was performed to identify crystalline and amorphous phases and evaluate the structural order of the samples, thereby providing an accurate examination of the crystallinity characteristics and phase structure of polymer systems in the binary and ternary samples.

2.6. Thermal Properties

The thermal and crystallization behavior of neat polymers, binary, and ternary blends was probed by DSC (Q1000, TA Instruments, New Castle, DE, USA). An aluminum pan and lid were used to encapsulate five to ten milligrams of the samples. The samples were subjected to the following procedure under a nitrogen atmosphere (50 mL/min)

• Starting point= −90 °C

• Heating rate = 10 °C/min

• End point= 150 °C

• Equilibrium

• Cooling to −90 °C with the same ramp

• Heating to 150 °C with the same ramp

The crystallinity percentage, X _C, of the specimens was determined using eq

$$X_{\mathrm{c}}=\frac{\mathrm{\Delta }{H}_{\mathrm{m}}}{\omega \times \mathrm{\Delta }{H}_{\mathrm{m}}^0}\times 100$$ 2

Where ω represents the mass fraction of the polymer, the experimental melting enthalpy of a fraction, ΔH _m, is determined as the area under the melting peak. ΔH _m ⁰ is the specific melting enthalpy of a fully crystalline polymer, which is 139.5 J·g^–1, 205 J·g^–1, and 197.7 J·g^–1, for PCL, PEO, and PEG, respectively.

2.7. In Vitro Biological Evaluation

2.7.1. Cell Culture and Scaffold Seeding

Passage 3 of IMR-90 mCherry fibroblasts expressing red fluorescent protein (RFP) were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (PS) (all sourced from Gibco, Thermo Fisher) as previously described. The cells were cultured in T75 culture flasks (Sarstedt, TC Flask T75, Stand, Vent. Cap, Germany) and maintained at 37 °C in a humidified atmosphere containing 5% CO₂, following previously established protocols.

When the cells reached 90% confluence, they were first washed with 3 mL sterile phosphate-buffered saline solution (PBS) (Gibco, Thermo Fisher, USA), then detached with 0.25% trypsin (Gibco, Thermo Fisher, USA). Fresh DMEM medium containing 10% FBS and 1% PS was immediately added, and the suspended cells were centrifuged for 5 min at 1500 rpm. The obtained fibroblast cell pellet was resuspended in 3 mL of fresh complete medium (DMEM supplemented with 10% FBS and 1% PS), counted using a hemocytometer, and adjusted to a cell concentration of 2.5 × 10⁵ cells/mL to seed on the scaffolds.

The scaffolds prepared by melt electrospinning were precisely punched into 9 mm diameter disks. Next, the scaffolds, which had previously been placed in deionized water for a week, were sterilized under ultraviolet radiation for 15 min. Afterward, scaffolds were sterilized in DMEM containing 1% antibiotics (PS) to remove any contamination and then placed in each well of a 48-well nonadherent cell culture plate (SARSTEDT AG & Co.). Subsequently, 200 μL of fibroblast cell suspension (containing 50,000 cells per scaffold) was added to each well and incubated for 24 h under humidified conditions and 5% CO₂ atmosphere.

To ensure that only cells attached to the melt electrospun scaffolds were examined, the scaffolds were transferred to new plates with fresh media 1 day after cell seeding. Then, the scaffolds were maintained under standard culture conditions for 1, 3, 7, 14, 21, and 28 days, and the medium was changed every 3 days.

2.7.2. Metabolic Activity Measurement (Alamar Blue Assay)

Alamar Blue assay (Invitrogen, USA) was performed to determine the cellular metabolic activity. At designated intervals, the growth medium was substituted with the fresh DMEM high-glucose medium containing 10% (v/v) Alamar Blue reagent. 200 μL of assay solution was added to each scaffold in 48-well plates, and the scaffolds were incubated for 2 h at 37 °C in a humidified atmosphere containing 5% CO₂.

After the incubation period, 100 μL of the liquid from each well was transferred to 96-well black half a rea plates (Corning). Fluorescence intensity was quantified using a Tecan Infinite M200 Pro microplate reader (Tecan Trading, AG) with excitation and emission wavelengths set to 540 and 585 nm, respectively.

To eliminate background fluorescence, two types of controls (blank) were considered: (i) scaffolds without cells and (ii) test solution (DMEM containing 10% Alamar Blue) without contact with cells or scaffolds. Data obtained from the controls were subtracted from the sample signals. All experiments were performed in at least three independent replicates (n ≥ 3) for each condition and time, and the results are reported as mean ± standard deviation (Mean ± SD).

2.7.3. Preparation of Cell-Seeded Scaffolds for SEM Observation

On day 28, the fibroblast-cultured melt electrospun scaffolds were gently washed with PBS (pH 7.4) to remove nonadherent fibroblast and residual culture medium. The samples were then fixed with a freshly prepared 4% paraformaldehyde (PFA, methanol-free, Thermo scientific, USA) solution for 20 min at ambient conditions. After removal of the fixative solution, they were washed three times with PBS to obliterate residual fixative and then dehydrated through a graded ethanol series (30%, 50%, 70%, 90%, and 100%), each for 5 min. The dehydrated samples were subsequently air-dried in a desiccator before observation under a scanning electron microscope (Hitachi TM3030Plus, Japan) at an accelerating voltage of 15 kV.

2.8. Statistical Analysis

Statistical analysis of data was performed using GraphPad Prism software version 10. A two-way ANOVA was used to examine group differences, and post hoc comparisons were performed with Tukey’s test. Results were reported as mean ± standard deviation (Mean ± SD). The class of statistical significance in all analyses was determined as p < 0.05; p < 0.01 and p < 0.001, indicating low, moderate, and very high levels of statistical significance, respectively, and thus increasing the level of confidence in the observed differences. All experiments were performed with at least three independent replicates to ensure reproducibility and reliability of the results.

3. Results and Discussion

3.1. Morphological Properties

The SEM micrographs of binary and ternary blends (PCL/PEG, PCL/PEO, and PCL/PEO/PEG) are presented in Figure (before melt electrospinning, BME), Figure (after melt electrospinning, AME; fiber surfaces), and Figure (AME; fiber cross sections). Note that the scanned morphology is representative of the particular processing technique used in this investigation (i.e., Brabender). While the current results provide a consistent picture under these conditions, the impact of alternative processing methods on the final morphology has been reported elsewhere by the authors. , The morphologies of the blends before melt electrospinning are shown in Figure A–C, while the corresponding microfibers produced by melt electrospinning are depicted in Figure A,C,E. Distributions of dispersed-domain diameters (expressed as pore area fraction) are provided in Figure D–F for the samples before the melt electrospinning and in the Figure B,D,F for the melt electrospun samples. Note that Figure B does not include diameter data for the dispersed domains. In the PCL/PEG (75/25)-AME sample, the PEG, as the dispersed component, does not form spherical domains. In the PCL/PEG-AME sample, the dispersed PEG phase forms fibrils aligned along the longitudinal axis of the melt electrospun fibers. The binary and ternary blends and their fibers do not show pores on the surface or cross-section before soaking the samples in water. However, the immersion of binary and ternary blend samples into water led to the appearance of pores in the specimens (Figures A–C, A,C,E, and A,C,E).

2.

Morphology of (A) PCL/PEG (75/25)-BME, (B) PCL/PEO (75/25)-BME and (C) PCL/PEO/PEG (67/16.5/16.5)-BME samples (After mixing in the Brabender) with (D–F) corresponding area-weighted domain-size distributions quantified from the respective micrographs.

3.

Surface SEM micrographs of melt-electrospun fibers: (A) PCL/PEG (75/25), (C) PCL/PEO (75/25), and (E) PCL/PEO/PEG (67/16.5/16.5). Quantitative pore-size distribution plots extracted from the corresponding images are provided in (B, D, F). Insets in (A, E) show selected regions at higher magnification (30 μm scale) for PCL/PEG and the ternary blend, respectively; main scale bars: 100 μm.

4.

Cross-sectional SEM morphology of melt-electrospun fibers: (A) PCL/PEG (75/25), (C) PCL/PEO (75/25), and (E) PCL/PEO/PEG (67/16.5/16.5). The corresponding area-weighted pore-size distribution plots are shown in (B, D, F). The inset in (A) (10 μm scale) provides a closer view of the fiber cross-section to emphasize internal morphology; main scale bars: 100 μm.

In PCL/PEG-BME and PCL/PEO-BME blends, PEG and PEO form a Drop-in-Matrix structure in the samples (Figure A,B). Considering the pore-size distribution plots (Figure D,E), the domains of PEO (∼4.2 ± 1.6 μm) compared to PEG (∼2.8 ± 0.8 μm) are slightly larger, which may reflect differences in molecular weight and crystallization kinetics. PEO is expected to exhibit a high melt viscosity and elasticity, which intrinsically restricts its chain mobility because of its very high molecular weight (∼600,000 g/mol). Hence, it tends to remain poorly dispersed in the PCL, creating large, well-defined domains.

Although the mixing temperature (70 °C) was close to the melting transition of PEO, no evidence of persistent unmelted PEO particles was observed after compounding. Because a DSC melting peak represents the maximum melting rate rather than necessarily the complete disappearance of the crystalline phase, the transient presence of residual PEO crystallites during compounding cannot be excluded. Upon addition to molten PCL, the PEO particles were expected to undergo progressive surface-to-core melting, assisted by conductive heat transfer, shear-induced fragmentation, and viscous heat generation. − Thus, most of the PEO was likely molten during the later stages of mixing, although complete melting could not be directly verified. Any partially molten PEO could have resisted deformation and breakup, thereby contributing to larger or more irregular PEO-rich domains and a broader domain-size distribution. In addition, owing to its high molecular weight, molten PEO remained highly viscous near its melting temperature, which may itself have limited domain refinement.

The ternary blend-BME sample consists of two different sizes of dispersed phases (Figure C). The large pores, with a mean diameter of 60 ± 0.1 μm, could be attributed to the PEO-rich domains, and the tiny pores, with a mean diameter of 0.3 ± 0.0 μm, could be the footprints of PEG-rich domains. The reason for the distinct domain sizes in the ternary blend compared to the binary blends is probably the assumption that the different molecular weights of PEO and PEG influence the complex interplay of phase separation and crystallization processes. The impact of molecular weight on the liquid–liquid phase separation and miscibility of different polymers was reported before. , PEO and PEG, as the polymer’s high and low molecular weights, introduce a complex interaction to the system. The high molecular weight PEO tends to form larger domains due to its slower diffusion. Simultaneously, the low molecular weight PEG, being more mobile, might be squeezed into smaller regions, resulting in smaller domains.

The average fiber diameter of PCL fibers decreases from 43.6 ± 0.5 μm to 39.4 ± 1 μm when PEG is added to the sample (Figure A). This could be because adding PEG decreases the melt viscosity due to its low molecular weight, which acts like a plasticizer in the system. On the other hand, the addition of PEO to the system increases markedly the average fiber diameter to 116.4 ± 24.8 μm (Figure C). This observation emphasizes the importance of molecular weight and its impact on the sample’s viscosity, which can significantly change the fiber diameter. The same statement for different systems was reported before. Because of its low molecular weight, PEG can be stretched easily during electrospinning and form a fibril shape, all aligned in the direction of the applied stretching force caused by the electrical charge difference applied between the needle and the collector. In order to illustrate the traces of PEG “fibrils” after dissolving in water, a high-magnification inset is used in the left corner of Figure A (scale bars: main 100 μm; inset 30 μm). It illustrates the homogeneous creation of fibrils of the PEG in the PCL matrix. The imposed stretching triggers the PEG-rich domains to be thinner. The tiny pores on the cross-section of the fibers (Figure A) show the cross-section of PEG fibrils trapped in the PCL matrix. The lack of pore on the fiber’s surface confirms the alignment of the PEG fibrils in the direction of imposed tension (Figure A). From the perspective of multiphase fluid mechanics, the fibrillar morphology observed in the PCL/PEG-AME sample is likely the result of sequential deformation mechanisms acting during melt electrospinning. The dispersed PEG droplets may undergo deformation within the elongational flow field as the molten blend passes through the needle. , Upon exiting the nozzle, the electrically driven stretching of the polymer jet is expected to further amplify this deformation before rapid solidification fixes the final morphology. Although quantitative verification would require determination of the capillary number and its comparison with the critical value for droplet deformation, the morphological observations are consistent with this sequence of events. −

Unlike PCL/PEG-AME, in the PCL/PEO-AME, PEO domains in the PCL (Figures C and C) create a sea–island phase morphology. The sea–island morphology is analogous to the droplet-matrix morphology, though there is a critical difference in the scale and distribution. The “islands” describe the dispersed PEO phase embedded within the “sea,” PCL, the continuous matrix phase. The islands are usually more extensive and less uniformly shaped compared to the droplets in droplet-matrix morphology. This is why large pores are observed in PEO-containing binary (Figures C and C) and ternary mixtures (Figures E and E). The formation of sea–island morphology for the PCL/PEO system was reported before where the sea–island morphology was defined as when the PEO spherulite exclusively develops near the PCL phase borders during the cooling step, like water streaming around pebbles and engulfing them inside. There are signs of the simultaneous presence of PEO and PEG on the ternary blend melt electrospun samples (Figures E and E). In the fiber structure of the ternary blend, fibrils are observed that are slightly shorter than those formed in the PCL/PEG-AME sample and are scattered on the surface and interior of the fibers. The authors believe that this feature reflects the role of PEG in the system. In addition, the presence of large, connected, and open pores on the surface and cross-section of the fibers, according to the authors, indicates the effect of PEO in creating structural porosity in this blend. These observations were also confirmed by hot-stage microscopic analysis (Movies S1–S4, Supporting Information; Movie S1 for PCL/PEO (75/25), Movie S2 for PCL/PEG (75/25), Movie S3 for PCL/PEO/PEG (67/16.5/16.5), and Movie S4 for PEO/PEG (75/25)). The hot-stage microscopy observations revealed that PEG is initially dispersed as discrete droplets within the PCL matrix (Figure S1, Supporting Information). Upon heating, these droplets undergo progressive coarsening, resulting in the formation of larger PEG-rich domains (Figure S3, Supporting Information). In contrast, under identical thermal conditions, PEO exhibits substantially more pronounced coarsening, leading to the development of distinct morphological features. Also, the examination of the PEO/PEG binary sample clearly showed that under the applied mixing conditions, these two polymers form distinct phases (Movie S4, Supporting Information). This behavior is also repeated in the ternary system, indicating the simultaneous presence of both types of morphologies, such that PEO is observed as island structures and PEG as dispersed droplets within the PCL matrix (Movie S3, Supporting Information). The creation of two separate domains of PEO and PEG in a blend was previously reported by Soo et al. They explained that this phase separation is due to the significant differences in molecular weight between PEO and PEG, which disrupt uniform crystallization and result in distinct crystallite domains, even though the two polymers share similar chemical structures. However, it is important to note that the miscibility of PEO and PEG strongly depends on factors such as their initial concentrations, the ratio of PEO to PEG in the system, the difference between their molecular weights, and the employed blending method. ,,

One limitation of the current work is that the morphology is characterized for specimens fabricated through Brabender. Given that polymer blend morphology can be sensitive to processing parameters, the authors have reported elsewhere further studies investigating whether and how various blending approaches might affect phase distribution and domain formation. ,

3.2. Structural and Chemical Characteristics

Figure A presents the FTIR spectra of all the samples before and after melt electrospinning. Measurements were conducted at ambient conditions with no additional thermal treatment, allowing a direct comparison of chemical fingerprints between the pre-electrospinning blends and the as-spun fibers. The samples before melt electrospinning show higher intensities than the melt electrospun fibers. This is owing to the higher thickness of samples before melt electrospinning than the fine and distinct melt electrospun fibers. Besides, it is worth mentioning that the characteristic peaks that emerged for specimens before and after melt electrospinning are entirely duplicates. Peaks of neat polymers can be distinguished in the blend samples, with no evidence of additional peaks or shifts, confirming no presence of intermolecular interactions. This result confirmed also that there is no chemical interaction between PCL and PEO or PEG nor degradation. The study by Nguyen Tri et al. in the PCL/PEG (50/50) system also reported no chemical interactions between the blend components. Yet, minor physical interactions between the phases in that system were suggested. Nevertheless, in the current study, no such changes were observed, which could be attributed to differences in process conditions or the sensitivity of the test methods. , The absence of new peaks or significant shifts in the FTIR spectra does not necessarily rule out the existence of weak physical interactions between the system’s components, such as dipole–dipole interactions or weak hydrogen bonds. Such interactions may only cause very minor changes in the functional groups, which are not detectable owing to the overlap of absorption bands and the sensitivity limitations of the FTIR method in multicomponent polymer systems. However, the absence of appreciable spectral changes, together with the preservation of the two-phase morphology of the samples, indicates that, if present, these interactions were not sufficiently intense to produce detectable chemical changes. The FTIR spectra of PEO and PEG are identical as they are the same polymers with different molecular weights.

5.

(A) FTIR spectrum for samples before and after melt electrospinning and (B) XRD pattern of neat polymers and their binary and ternary blends.

Given that PEG is susceptible to thermo-oxidative degradation under elevated-temperature processing conditions in the presence of oxygen, , it is important to note that, although melt electrospinning in the present study was carried out under ambient atmosphere, the processing conditions were such that direct exposure of the molten polymer to ambient air occurred primarily after exiting the needle tip. Nevertheless, the possibility of a limited degree of thermo-oxidative degradation or oxidative chain scission cannot be completely excluded. If such degradation occurs, it may alter the molecular architecture of the polymer and consequently affect the rheological behavior of the melt, the fiber formation process, and ultimately the chemical and mechanical properties of the fibers. ,

However, the FTIR spectra of the processed samples did not show any new absorption bands or noticeable shifts in the characteristic polymer peaks, indicating that significant chemical degradation did not occur under the processing conditions used in this study. Therefore, although a limited degree of oxidative chain scission cannot be completely ruled out, the available experimental evidence suggests that, if present, its extent was insufficient to lead to detectable changes in the chemical structure of the processed samples. Since the mechanical properties of the fibers were not evaluated in the present study, no direct conclusions can be drawn regarding the possible impact of such limited degradation on their mechanical performance.

Figure B depicts the XRD patterns recorded under ambient conditions for neat PCL-AME, PEO-BME, PEG-BME, and their binary-AME and ternary-AME blends. Considering the fact that PEO and PEG are identical polymers with different molecular weights, their characteristic peaks are the same and determined at 2θ of 18.9° and 23.3°. , Signals correlated with PCL orthorhombic crystal construction, peaks (110) and (200) emerged at 21.2° and 23.3°, respectively, and the latter overlaid with the PEO or PEG. −

The comparison of XRD patterns between melt electrospun fibers of PCL and PCL/PEG reveals the disappearance of PEG’s characteristic peaks in the binary blend, suggesting that PEG either remained in an amorphous phase or formed crystalline domains too small or disordered to be detected by XRD.

Although no detectable PEG crystalline reflections were observed in the XRD patterns of the as-spun samples, indicating that any PEG crystallinity at room temperature is either below the detection limit or represents only a minor fraction of the PEG phase, the physical state of the PEG domains may nevertheless significantly influence scaffold performance. Specifically, amorphous PEG is expected to dissolve more readily upon water immersion than crystalline or highly ordered PEG, , due to the removal of sacrificial components, potentially affecting the evolution of porosity and, consequently, the mechanical integrity of the final scaffold. However, this behavior is also governed by the morphology and spatial distribution of the PEG domains, and establishing a direct quantitative relationship would require dedicated time-dependent mechanical characterization, which is beyond the scope of the present study.

In the PCL/PEO sample (Figure B), it is indicative of the ability of PEO domains to crystallize and form independent crystal structures. Due to the high molecular weight of PEO, the melt viscosity increases, limiting the ability to stretch and flow the chains during melt electrospinning; as a result, under the same process conditions, fibers with larger diameters are formed (Figures A and A). This increase in fiber diameter reduces the cooling rate, providing sufficient time for PEO chains to organize and form crystals. As a result, the presence of PEO crystal peaks in the XRD pattern of the PCL/PEO binary sample is expected and justified. Also, the difference in crystallization behavior between PEO and PEG has been confirmed by hot-stage microscopic observations (Movies S1–S4, Supporting Information).

The characteristic peak of PEO (2θ = 18.9°) is observed in the ternary melt electrospun fibers. The shorter peak of PEO in the ternary blend fibers, PCL/PEO/PEG-AME, compared with the neat PEO-BME, can be attributed to less PEO in the ternary blend fiber. The total amount of PEO and PEG in the ternary blend, 33 wt % (16.5 wt % PEO and 16.5 wt % PEG) is higher than the amount of PEO, 25 wt %, in PCL/PEO (75/25)-AME and the amount of PEG, 25 wt %, in PCL/PEG (75/25)-AME. To clarify, a semiquantitative integration of the diffraction peak centered at approximately 2θ = 18.9° was performed using identical baseline correction and peak integration procedures for both samples. The integrated peak area decreased from 4824.1 for PCL/PEO (75/25)-AME to 3928.0 for PCL/PEO/PEG (67/16.5/16.5)-AME; if the PEO and PEG were miscible with the same peaks, they would have shown a sharper peak in PCL/PEO/PEG (67/16.5/16.5) fibers. In line with the SEM results, in the PCL/PEO/PEG-AME sample (Figures E and E), the simultaneous presence of features attributed to PEG and PEO is clearly discernible. PEG, by creating surface grooves and fine pores in the fiber cross-section, and PEO, by forming large, interconnected pores, a sea–island morphology in the cross-section, and surface porosity, all play a role in the formation of the final structure. This distinct behavior is attributed to the formation of distinct domains for PEO and PEG, and to differences in their behavior during cooling, as confirmed by hot-stage microscopic observations (Movie S3, Supporting Information). Based on hot-stage microscopy, the crystallization sequence in the ternary PCL/PEO/PEG-AME system is PEO first, then PCL with a slight delay, while PEG is in the final stage and remains mainly in the amorphous state. This behavior is intensified under the rapid-cooling conditions of the melt electrospinning process, leading to the dominance of the amorphous phase within the PEG domains.

3.3. Thermal Properties

Table summarizes the thermal indices obtained from DSC results of the first heating cycle and cooling run of the samples before (BME) and after (AME) the melt-electrospinning process. In Table , the changes in crystallization enthalpy (ΔH _C), melting temperature (T _m), crystallization temperature (T _c), and crystallinity percentage (X _C) are reported for each polymer component, allowing for the evaluation of the effect of composition and electrospinning process on the thermal behavior. All the samples show one T _g (Table and Figure S3 of Supporting Information). On the condition that a sample shows one single glass transition, the components are miscible in the molten state. However, in this study, it could be because of the closeness of the components’ glass transition temperatures (Table ). So, distinguishing their T _g would be impossible using DSC.

2. Thermal Properties of the Samples before (BME) and after (AME) the Melt Electrospinning.

property	component	PCL/PEO-BME	PCL/PEG-BME	PCL/PEO/PEG-BME	PCL/PEO-AME	PCL/PEG-AME	PCL/PEO/PEG-AME
T C [°C]	PCL	31.3 ± 0.7	29.4 ± 0.4	32.3 ± 1.5	35.8 ± 1.8	30.5 ± 0.9	29.9 ± 0.4
	PEO	45.8 ± 0.8		44.8 ± 1.4	42.4 ± 1.8		44.8 ± 0.6
	PEG		0.9 ± 0.1, 44.4 ± 1.5	–0.7 ± 0.2		–1.3 ± 0.1	1.3 ± 0.1
ΔH C [J/g]	PCL	33.0 ± 0.9	53.4 ± 0.5	34.1 ± 0.9	32.7 ± 0.5	32.9 ± 0.7	29.5 ± 1.1
	PEO	20.7 ± 1.4		20.8 ± 0.8	6.0 ± 0.0		25.3 ± 1.0
	PEG		9.5 ± 0.0, 8.8 ± 0.1	1.0 ± 0.0		19.9 ± 0.2	3.0 ± 0.0
T m [°C]	PCL	60.9 ± 0.1	61.6 ± 1.1	62.1 ± 1.3	59.9 ± 0.2	60 ± 0.8	58.3 ± 0.4
	PEO	66.1 ± 0.0		66.9 ± 1.8	63.3 ± 0.0		64.3 ± 1.0
	PEG		65.4 ± 1.3			63.1 ± 0.1
X C , [%]	PCL	33.3 ± 1.3	50.1 ± 0.5	54.1 ± 1.0	63.8 ± 0.9	70.7 ± 0.6	50.1 ± 1.2
	PEO	35.3 ± 0.8		48.0 ± 1.0	44.1 ± 1.1		40.0 ± 2.0
	PEG		31.5 ± 0.6
T g [°C]	Blend	–50.2 ± 1.3	–49.3 ± 1.5	–50.9 ± 1.5	–50.2 ± 3.4	–51.7 ± 1.9	–52.9 ± 1.6

^aData is obtained from the cooling cycle.

^bData is obtained from the first heating cycle.

^cCrystallinity percentage, X _C, is calculated using eq .

DSC shows one melting point (Table and Figure S3 of Supporting Information) for PCL/PEG-AME, PCL’s melting point for PCL/PEG fiber. This observation could be because of the remaining PEG in the amorphous phase due to rapid solidification during the melt electrospinning. However, it is worth noting that a subtle deviation from the main melting peak of PCL is detected in the first heating cycle (Figure S3, Supporting Information), such that a slight shoulder arises on the typical melting peak of PCL. Although this feature is not well-defined as an independent melting peak, it may suggest the existence of a slight portion of PEG domains trapped within the PCL matrix, which could have experienced limited crystallization during the cooling step. This hypothesis is further supported by the distinct crystallization signal observed at approximately 1 °C during the cooling scan (Table and Figure S3 of Supporting Information). This phenomenon has also been reported in several studies. − For instance, Luo et al. reported a similar low-temperature crystallization peak in PCL/PEG systems, which they attributed to the construction of microcrystalline domains by PEG chains that remained confined within the PCL matrix. Their analysis confirmed that these crystals are remarkably small, consistent with limited phase separation of PEG during solidification while the larger portion of PEG, phase separated from PCL. These findings further support the hypothesis that the minor crystallization peak detected at ∼1 °C (Table and Figure S3 of Supporting Information) may originate from trapped PEG chains within the PCL phase. Furthermore, it implies that the larger portion of PEG forming fibrils in the system remains in the amorphous phase, as indicated by the comparison between the original amount of PEG in the system and the minor enthalpy of this peak. This interpretation is further supported by the absence of any characteristic PEG diffraction peaks in the XRD pattern (Figure B) in addition to the lack of a distinct melting point for the PEG in the first heating cycle of DSC (Table and Figure S3 of Supporting Information).

On the other hand, the PCL/PEG-BME sample exhibits two melting points (Table and Figure S3 of Supporting Information). This observation suggests that the slower cooling rate experienced by the samples after removal from the Brabender provides sufficient time for the PEG-rich domains to crystallize. In contrast, the rapid solidification of the microfibers during melt electrospinning significantly restricts chain mobility, thereby suppressing the crystallization of the PEG-rich domains. This difference in thermal history and crystallization kinetics explains why PCL/PEG-AME exhibits a single melting peak, whereas PCL/PEG-BME displays two distinct melting peaks.The DSC curve of the ternary blend exhibits two melting points for both samples before and after melt electrospinning (Table and Figure S3 of Supporting Information). To elucidate the behavior of the ternary blend, it is helpful to understand the crystallization characteristics of the samples by studying the cooling cycle of DSC analysis. The neat polymers indicate well-defined and distinct crystallization patterns (Table ), mirroring their inherent thermal properties. For instance, PEO and PEG crystallize at the same temperature (Table ); nonetheless, with its lower molecular weight, PEG demonstrates a much sharper crystallinity peak. This is explained by the shorter PEG polymer chains, which facilitate quicker and more effective crystallization. In contrast, with its higher molecular weight, PEO has longer, more entangled chains, resulting in slower crystallization and a broader peak (Table and Figure S3 of Supporting Information).

The cooling curve for the PCL/PEO-AME blend (Table and Figure S3 of Supporting Information) suggests that melt electrospinning does not disrupt the construction of crystalline networks for PEO and PCL. This could be attributed to the high molecular weight of PEO, which may lead to the melt exhibiting significantly high viscosity, thereby restricting the chain mobility during processing. As a result, PEO lacks sufficient stretchability under the applied shear conditions, resulting in the formation of discrete, island-like domains within the system (Figure C, and Figure C). Given the relatively large size and PEO-rich nature of these islands, they readily crystallize upon cooling, as evidenced by the distinct crystallization behavior observed in the DSC cooling curve.

In both PCL/PEO/PEG-BME and PCL/PEO/PEG-AME samples, two melting peaks are observed during the first DSC heating cycle (Table and Figure S3 of Supporting Information). While one of these peaks is attributed to the melting of the PCL phase, the nature of the second peak requires further investigation. Based on the hot-stage microscopy results, this peak is attributed to the PEO melting peak, not to the PEG or PEO/PEG melting peaks. The hot-stage microscopy shows that PEO and PEG form distinct regions and that upon cooling, crystallization first begins in the PEO regions, followed by PCL crystallization, while the PEG regions, which appear as spherical domains and are much smaller than the PEO domains, remain largely amorphous until the end of the test, which is the ambient condition.

Similar and completely consistent behavior is observed in the cooling curves of both PCL/PEO/PEG-BME and PCL/PEO/PEG-AME samples (Table and Figure S3 of Supporting Information). In both systems, three crystallization peaks can be distinguished. Based on the provided discussion above, the crystallization peak at about −1 °C for PCL/PEO/PEG-BME and about 1°C for PCL/PEO/PEG-AME can be attributed to the crystallization of a limited portion of PEG domains that are trapped within the PCL matrix. At the same time, the bulk of PEG seems that remain amorphous until the end of cooling. On the other hand, considering that in the PCL/PEO-BME and PCL/PEO-AME systems, no evidence of entrapment of PEO domains in the PCL matrix is observed, it does not seem reasonable to attribute this low temperature peak to PEO crystallization. Two other crystallization peaks, observed at around 30 °C and 44°C, are assigned to the crystallization of the PCL and PEO phases, respectively (Table and Figure S3 of Supporting Information). It implies that, despite the chemical similarity of PEO and PEG, the significant difference in their molecular weights leads to the formation of distinct domains and different crystallization behaviors in this system.

3.4. Cellular Metabolic Activity and Viability Assessment

To evaluate the initial cellular compatibility of melt electrospun scaffolds, the Alamar Blue assay was performed to quantify the metabolic activity of cultured fibroblasts on different samples (Figure A). The SEM of cell-seeded PCL and ternary blend fibers after 28 days of culturing is shown in Figure B,C. The Alamar Blue assay indicates the initial response to the samples’ surface and morphology. The results show that, across all groups, fibroblast cell metabolic activity increases in a time-dependent manner, indicating overall compatibility with the scaffold and the absence of toxic effects. However, the signal intensity in the scaffolds coming from binary and ternary blends is significantly higher than that of the neat PCL sample (Figure A). This increase cannot be attributed solely to increased cell numbers; rather, it most likely reflects improvements in physiological status, spatial penetration, and the dynamic interactions of cells with the open scaffold network. The results of relative metabolic activity, obtained by dividing the Alamar Blue test data by the corresponding PCL values at each time interval, are presented in Figure S4 of Supporting Information.

6.

(A) Metabolic activity of fibroblasts on different scaffolds over a period of 28 days. Cellular metabolic activity was measured using the Alamar Blue assay on days 1, 3, 7, 14, 21, and 28 after culture. Data is presented as mean ± SD. Two-way ANOVA performed statistical analysis with Tukey’s post hoc test. Statistical symbols are displayed as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ***p < 0.0001, and ns indicates nonsignificance. To avoid cluttering the form and maintaining readability, only two key families of comparisons are shown in the graph: (1) ternary combination (PCL/PEO/PEG) versus PCL as the baseline reference, and (2) PCL/PEO versus PCL/PEG. All other comparisons (including all pairs between formulations at each time point) are presented in full in the Supporting Information (Table S1). (B) SEM image of fibroblast cells on PCL scaffold at day 28 showing scattered cell distribution. (C) SEM image of PCL/PEO/PEG scaffold on the same day showing formation of a dense network of cells and extracellular matrix. Scale bar: 50 μm.

Since the scaffolds were placed in water for a week before cell seeding, the PEO and PEG phases are mostly dissolved (Figures and ), yielding a structure with interconnected intrafibrous porosity. SEM observations (Figures and ) clearly show that after this stage, the fiber surface transitioned from a dense to a porous state. Such a morphological transformation, similar to that reported for multiphase electrospun scaffolds, increases the effective contact area, facilitates nutrient and oxygen exchange, and forms three-dimensional microenvironments for cell adhesion and survival. −

This finding aligns with the report by Giacaman et al. on PCL-b-PEG/PCL mats, which identified enhanced porosity and permeability as critical factors for fibroblast migration and metabolic stability. Also, Xia et al., in their study of PCL/nHA/PEG hybrid scaffolds, provided a multiscale porous environment that led to a significant increase in cell density and activation of growth-related metabolic pathways. These results are consistent with the present observation of a gradual increase in Alamar Blue signal in ternary blend samples (Figure A).

From the perspective of cell physiology, in such substrates, cells not only grow on the scaffold surface but also penetrate the voids created by the removal of soluble phases, gradually forming a three-dimensional network (Figure C). The increase in metabolic signaling in this situation could be because of enhanced intercellular communication, activation of mitochondrial respiratory pathways, and improved gas exchange, rather than merely an increase in cell number. A similar observation has been reported for studies of electrospun blends with connected porosity, in which the increase in metabolic capacity was attributed to upregulation of ECM-related genes and integrins.

In contrast, the neat PCL scaffold, due to its smooth surface, likely limited cell penetration into the network and cell accumulation on its surface (Figure B). Under such conditions, nutrient exchange in the lower layers might be reduced, and it could be the reason why the metabolic signal intensity remained lower than in the open-pore groups.

Overall, the results of metabolic activity (Figure A), along with imaging evidence (Figure B,C), demonstrate that transforming the PEO and PEG phases into porous intrafibrous structures after a preimmersion step in water creates a microenvironment with enhanced permeability, three-dimensional accessibility, and a cellular metabolic response. This feature, coupled with the morphological stability provided by the PCL matrix, suggests a beneficial equilibrium between structural integrity and biological activity, a concept previously proposed as a key indicator for scaffold design in bone tissue engineering. However, since this evaluation is only a preliminary fibroblast-based test, larger in vitro studies with primary osteoblast cells are necessary to investigate the biological behavior of the scaffolds for bone tissue engineering application, which is the subject of our future work.

4. Conclusion

In this work, the behavior of a ternary blend of PCL/PEO/PEG under the melt electrospinning process was studied. The SEM images reveal the different morphology of the samples before and after melt electrospinning because of the effects of the high temperature of the process, heating time, which acts like annealing, fast solidification, and stretching that is due to the imposed electrical field to the system. A comparison between the behavior of PEG and PEO mixing with PCL was also made. The FTIR spectra of melt electrospun fibers indicate no reaction or chemical interaction between PCL, PEO, and PEG. It also alludes to no difference between the FTIR spectra of samples before and after melt electrospinning. PEO can form large, PEO-rich domains during the solidification of melt electrospun fibers made from PCL/PEO/PEG-BME. These larger domains develop a crystalline structure as they solidify. In contrast, the XRD spectra of samples vividly indicate that the bulk of PEG remains in the amorphous phase in the ternary blend fibers, which DSC confirms. However, the minor fraction of PEG trapped in the PCL could form the microcrystalline structure. Metabolic activity results indicate that the use of PEO and PEG as sacrificial components is a practical approach to improve the biological properties of scaffolds. However, further studies with osteoblast cells are required to evaluate the actual potential of these scaffolds for bone tissue engineering. Besides, the results explained here are specific to the selected processing technique (i.e, Brabender). To further elucidate the potential effects of different compounding techniques on the blend microstructure, a comprehensive investigation has been conducted and is reported elsewhere.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c05235.

• PCL/PEO (75/25) (Movie S1) (MP4)

• PCL/PEG (75/25) (Movie S2) (MP4)

• PCL/PEO/PEG (67/16.5/16.5) (Movie S3) (MP4)

• PEO/PEG (75/25) (Movie S4) (MP4)

• Design of experiments (JMP), rheological isotherm/time-sweep plots, hot-stage optical microscopy images of the PCL/PEG blend, DSC heating and cooling curves, normalized fibroblast metabolic activity data, and statistical comparisons (ANOVA with Tukey post hoc analysis) (PDF)

The authors declare no competing financial interest.