Introduction

Self-healing polymers increase the lifetime of products exponentially through both non-reversible healing and reversible healing processes. However, reversible self-healing polymers exhibit poor mechanical stability [1]. In addition, preparation of self-healing polymers involves complex organic chemistry methods, expensive and customized precursors, and specialized equipment. To reduce expenses and equipment demanded for this synthesis, polycaprolactone (PCL), a polymer with similar mechanical properties (i.e., Young’s modulus and elastomeric behavior) was investigated. Similarly, aramid nanofibers (ANFs) were utilized to reinforce PCL for development of biocompatible polymer reinforced composite material.

PCL is a highly versatile biocompatible thermoplastic elastomer. PCL has a relatively low melting point (Tm = 55–60 °C) and an even lower glass transition temperature (Tg = − 60 °C) contributing to its high solubility in organic solvents [2, 3], which make it a highly processable material for sample fabrication. PCL has been used for many biomedical applications including drug delivery materials [4], coatings for implants [5], and scaffolds for tissue engineering [6]. PCL is used in additive manufacturing as a resin to improve end properties such as toughness, flexibility, compression, and tear strength [7].

ANFs are derived from Kevlar®, which is found in ballistic applications [8] and exhibits desirable mechanical properties such as high tensile strength with low fiber elongation [9]. ANFs are relatively non-toxic [10] as compared to other nanoscale materials with similar morphologies such as carbon nanotubes which have been shown to be severely toxic [11].The non-toxic nature of the ANF reinforcement method described in this article lends to broaden applications of the material, extending to use in the biomedical devices and materials or environmental applications. Kevlar® fibers (aramid fibers) possess outstanding mechanical properties due to aligned hydrogen-bonded networks while also having desirable properties of thermostability, electrical insulation, exceptional chemical resistance, and flame retardancy [12]. In turn, one-dimensional aramid nanofibers maintain the qualities of their large-scale molecular chains, but also exhibit nano scale dimensions [13]. ANFs provide a large surface area-to-volume ratio and therefore a high specific surface area [14] to provide good contact with polymer matrix in our composites. In this work, we hypothesized that ANF reinforcement would provide improved strength and toughness to extend the applicable uses of PCL and in turn self-healing polymer applications [15, 16].

To investigate this reinforcement, split-Hopkinson pressure bar (SHPB) testing was performed to characterize axial compression, impact duration, dynamic flow stress and strain rate of PCL, solvent-dissolved PCL, and solvent-dissolved PCL reinforced with one and two weight percent ANFs (1 wt% and 2 wt%). SHPB testing was chosen because of its unique ability to generate complete stress–strain curves at high strain rates from a single test. SHPB also provides data relevant to additive manufacturing of the composite [17]. Herein, we prepared PCL reinforced with ANFs [12] to evaluate dynamic properties of the fortified composite. Results from this work will provide mechanical understanding of ANF reinforcement that will be applied to preparation of ANF-reinforced self-healing polymers that are resource and time-intensive to prepare.

Materials and methods

ANF preparation was accomplished through proton donor-assisted deprotonation of aramid fibers [12]. Poly(p-phenylene terephthalamide) (PPTA) fibers, or aramid fibers, (0.6 g) and potassium hydroxide (KOH) (0.9 g) were added to 300 mL of dimethyl sulfoxide (DMSO). De-ionized (DI) water was then added in a specific volume ratio to DMSO of 1:25, respectively. The ANF/DMSO solution was stirred for 4 h. As the solution stirred, the color of the solution transitioned from clear, showing a slight yellow tint, to a dark red solution (Fig. 1a), providing an indication that deprotonation of the aramid fiber was completed as previously established in literature [12]. The structural restoration of ANFs was then accomplished by utilizing DI water as a proton donor for the system. A volume ratio of DI water to ANF/DMSO solution of 2:1 was utilized. This yielded a uniform ANF/DMSO/H2O solution after being stirred for a minimum of 1 h. Further color transitions occurred, signifying the completion of the structural restoration of the aramid fibers as stated in previous literature [12] as the solutions changed from a dark red, back to a slight yellow tint (Fig. 1b). Such an observation has been established to indicate successful completion of the ANF deprotonation/reformation process.

Fig. 1
figure 1

Images showing the ANF preparation. a Deprotonation of aramid fibers, b structural reformation of ANF, c scanning electron micrograph (SEM) of ANF using a Hitachi S-4500 SEM, and d ImageJ analysis of SEM shown in (c)

Full size image

To utilize the high-quality fibers formed through the reformation process, ANFs needed to be removed from the DMSO/H2O solution and transferred to new solvents. Similarly, because the reformation process is uncontrolled, larger fibers tend to form, requiring the solution to be filtered. Filtering was done using Nalgene rapid flow filters with pore sizes of 0.2 µm attached to a vacuum. This filtered solution was analyzed using scanning electron microscopy (SEM) with a Hitachi S-4500 and analyzed by ImageJ software to determine ANF reformation.

Next, incorporation of ANFs in PCL required using a solvent to suspend ANFs in random orientation while also ensuring even distribution within PCL. A vacuum oven was used to remove DMSO and H2O in a two-step evaporation process. First, the ANF/DMSO/H2O solution was placed into the vacuum oven directly after filtration. The oven was then heated to 85 °C and a vacuum of approximately 510 mmHg was used to reduce the evaporation point of water. As the water evaporated from the solution, the pressure would periodically drop, causing the evaporation point of water to increase. In turn, this process was monitored and consistently performed to keep the vacuum at appropriate levels for evaporation. After the solution reduced in volume to approximately 50 mL, indicating DMSO was the only solvent remaining in the system before the solution was transferred to multiple 40 mL scintillation vials in equal parts to evaporate the remaining solution.

To evaporate the remaining DMSO from the solution, the oven was set to 150 °C with a slowly induced vacuum of 510 mmHg. After the DMSO was evaporated, a green mass dried to the walls of the scintillation vial and was dispersed into 20 mL of trifluoroethanol (TFE). Ultra-sonication was then used to distribute ANF in TFE, leaving a milky white solution. These solutions were re-analyzed through SEM and ImageJ to obtain the concentration and weight percent of the ANF/TFE solutions. A precise volume of ANF/TFE solution was placed on an SEM stub to evaporate TFE to confirm the presence and morphology of ANF in SEM imaging (Fig. 1c) in the new solvent. Obtaining SEM micrographs of ANF shows high-contrast fibers that was used for ImageJ analysis to estimate the area percentage of the fibers using a minimum of eight micrographs. This allowed for the calculation of fibers that were retained in the new solvent solution by setting a contrasting threshold of approximately 5% to define fibers from the background (Fig. 1d). Concentration of ANF/TFE solutions was calculated using average area of fiber present, density of aramid fibers, and volume of the initial solution.

Samples for dynamic impact testing were prepared by forming a polymer solution or composite, depositing the solution or composite onto wax paper to form beads, and then letting the solvent evaporate before beads were melted into a mold for SHPB tests. A PCL-only sample was prepared by melting as-received polymer beads into the mold. Other than the PCL-only sample, samples were prepared by dissolving PCL in TFE at 10 wt% PCL. For the solvent-dissolved PCL samples, a syringe was used to place this solution in beads onto wax paper for solvent evaporation. For the solvent-dissolved PCL samples containing ANFs, sonication was used to promote even distribution of the nanofibers throughout the solution prior to bead deposition and solvent evaporation. The PCL-ANF composite was placed into a heated sonication bath. Sonication of samples was performed for 20 min before sample beads were deposited on the wax paper. Sample beads were allowed to dry for a minimum of 48 h to ensure that excess TFE had evaporated. Sample beads could then easily be taken from the wax paper and melted into sample molds for SHPB testing. Weight percent of ANFs was chosen based on previous studies that utilize ANF composites that exhibited desired mechanical properties [10, 18].

ANFs were analyzed utilizing diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) IRTracer-100 to obtain infrared spectra (IR) of fibers after reformation. IR spectra for ANFs were compared to aramid fiber counterparts. DRIFTS measurements for ANFs vs aramid fibers are shown in Fig. 2.

Fig. 2
figure 2

IR spectra collected via IRTracer-100 for ANF and aramid fibers. ANF1, ANF2, and ANF3 define repeats of the deprotonation/reformation process to ensure consistency in samples. Functional groups are signified by red dotted lines

Full size image

Key peaks that define the functionalities of aramid fibers are located at approximately 3300 cm −1 corresponding to NH stretching vibrations, 1510 cm−1 for C–C stretching of the aromatic ring, and 1246 cm−1 and 823 cm−1 for the out-of-plane vibrations of the phenyl group, [10, 19]. These matching peaks for the functional properties can be confirmed through the comparison spectra in Fig. 2, providing confirmation that the process of ANF synthesis was reliable and repeatable for polymer composite development and reinforcement. Similarly, matching functional groups indicate ANFs retain properties critical to mechanical performance of aramid fibers.

For SHPB testing, an aluminum bar was used for characterization of dynamic flow stress and strain rate for the complete range of samples. An aluminum bar set-up was chosen due to its close, but higher, mechanical impedance to the samples. Furthermore, copper pulse shapers were used to obtain prolonged loading by shaping the striker-bar pulse to achieve improved stress distribution within the samples. High-speed camera (HSC) recordings aided in characterizing the impact deformations, impact durations, and a distinct dynamic deformation extent of the samples. Samples were tested at three strain rates to observe the variance in dynamic flow stress to fully describe the complete range of samples under dynamic loading.

Trial runs were conducted prior to testing PCL samples to ensure the appropriate adaptation of the testing procedure and data analysis. The end goal of the testing was to obtain dynamic flow stress over a specific range of strain rates in the samples and how this response varies with increasing ANF loading. Samples were prepared with copper tubing molds that were approximately 8 mm in diameter and 6 mm in height (Fig. 3b). The sample material was heated to approximately 90 ˚C and placed on Teflon® films to allow for easy release. Sample beads were added in small amounts for proper heating to melt and mold the sample. Beads were compacted between every additional layer to remove potential voids within the sample. After the full volume of the sample mold was filled, the sample and sample mold were inserted into a vulcanizer and compressed with approximately 400 psi to ensure parallel, flat faces and to increase sample uniformity.

Fig. 3
figure 3

a Compaction percentages for the complete range of PCL samples at high strain rates. Error bars signify standard deviation of samples. b SHPB sample pre-test to clarify initial dimension of samples prior to first cycle compaction seen in (c). c HSC of SHPB compaction testing of PCL + TFE at three strain rates capturing first compression cycle demonstrating energy absorption and deformation of samples during testing

Full size image

Results

Samples were tested at four different velocities (n = 3) to observe the varying strain rate grouping of the material. The strain rate grouping based on the complete range of PCL samples was 1500 s−1 ~ 2500 s−1, 2500 s−1 ~ 3500 s−1, and 3500 s−1 ~ 4500 s−1. All testing was supplemented by HSC to account for differences between bar and sample stiffness that caused inequal face forces within the sample. HSC recorded sample responses at 210 kfps allowing for precise measurement of impact duration and compaction for the overall axial deformation of the samples. HSC images were interpreted using ImageJ software to understand the unique dynamic loading and deformation extent of the samples. Samples further aided results by initial plastic deformation during testing aligning any imperfection in sample faces, increasing wave propagation and peak dynamic loading of the samples. Compaction differences between sample types are shown in Fig. 3a.

Compared to all other samples, the highest compaction at the lower strain rates (1500 s−1 ~ 2500 s−1) was observed when PCL was reinforced with 2 wt% ANFs. At mid strain rates (2500 s−1 ~ 3500 s−1), PCL at 1 wt% ANF loading was much stiffer and deformed significantly less than other samples. At high strain rates (3500 s−1 ~ 4500 s−1), all samples performed similarly.

HSC footage provided frame-by-frame insight into the deformation of the samples as well as first cycle compression. Figure 3c displays SHPB of PCL/TFE specimens at the three strain rates, demonstrating conical compaction in post-compaction as well as deformation extents. Utilizing ImageJ, the first cycle compaction was measured, and the deformation pattern characterized. Samples under SHPB tests exhibited conical deformation as opposed to cylindrical deformation, due to face force imbalances [20], which may be due to stiffness differences between the samples and the bar as well as void compaction occurring in ANF-reinforced material that resulted in increased compaction ranges [21]. SHPB responses were used to calculate dynamic flow stress for the low, mid, and high strain rates per sample (Fig. 4). All samples showed relative uniformity for dynamic flow stress across all strain rates tested.

Fig. 4
figure 4

Dynamic flow stress average for the complete range of PCL samples at high strain rates exhibiting unique loading and stress distribution systems. Error bars signify standard deviation of samples

Full size image

The most notable difference observed from the dynamic flow stress of the samples occurred at mid strain rates. While all samples experienced strain rate strengthening, the solvent-dissolved PCL samples at mid strain rates also experienced the most pronounced adiabatic thermal softening. Low and high strain rates samples appeared to maintain dynamic loading while showing similar percentage compression. Furthermore, the most uniform response across sample testing is noted at high strain rates for both compression percentage and dynamic flow stress, suggesting that the low glass transition temperature and melting point of PCL create unique stress distribution and loading properties in dynamic high-energy environments.

Conclusions

SHPB samples performed similarly across strain rate groups, lacking statistical differences in mean value performance for dynamic flow stress. The inherent plasticity of PCL caused dynamic flow stresses to remain relatively consistent through strain rate groupings, while indicating softening from dissolution by TFE had occurred through decreased dynamic flow stress. HSC provided data on the extent of deformation and strengthening that occurred in PCL matrix reinforced with ANFs. Samples reinforced with 1 wt% ANF compacted less but sustained high dynamic flow stress perhaps indicating a more effective fiber to matrix ratio for energy absorption through a stiffer material. Samples reinforced with 2 wt% ANFs showed high compaction ranges to all samples at 1500 s−1 ~ 2500 s−1 strain rate. This increased deformation shows the toughness of the material by deforming over a larger range while experiencing similar dynamic flow stress capacities. At 2500 s−1 ~ 3500 s−1 and 3500 s−1 ~ 4500 s−1 strain rates, native PCL and PCL matrices struggled to support dynamic loading due to the high-energy of the test. At 2500 s−1 ~ 3500 s−1 strain rate grouping, samples dissolved by TFE, experienced a unique phenomenon known as adiabatic thermal softening [22] creating a temporary fluid-like matrix due to the low melting point and glass transition temperature of PCL. Samples dissolved by TFE experienced crystallinity alignment losses after being re-solubilized [23]. These crystallinity losses increased the thermal sensitivity of the composite, causing the high-energy test to affect the material differently that native PCL which sustained the highest dynamic flows stress for the 2500 s−1 ~ 3500 s−1 strain rate. Compaction decreased severely for solvent-dissolved samples for 2500 s−1 ~ 3500 s−1 strain rate. Like dynamic flow stress, losses in crystallinity and molecular alignment creating thermal sensitivity cause samples dissolved by TFE to enter a fluid-like state where the compression of these fluidic materials changes in comparison to native PCL. These phenomena are more clearly seen for 3500 s−1 ~ 4500 s−1 where energy of the test is so high, all samples enter a temporary fluid-like state. Causing a homogeneous response from all samples in both dynamic flow stress and compression analysis where the dynamic flow stress of a temporary fluid matrix would remain constant because crystallinity and molecular alignment are fully but temporarily lost. Similarly, the temporary fluid matrix cannot experience differences in compression due to the inherent properties of fluids. In future studies, observing polymer composite response accuracy would increase from a polycarbonate bar, or utilizing a stiffer polymer matrix. Furthermore, developing a methodology for ANF dispersant without the use of a solvent would benefit the performance of polymer composite materials. The end properties of the composite would be less affected without being dissolved by solvent that causes crystallinity and molecular alignment losses.