1 Introduction

1.1 Bio-based polymers and sustainability

Sustainable bio-based polymers have gained significant interest since the early 2000s, and the scientific literature surrounding their use, improvement, and applications has increased. A meta-analysis of the last 20 years reported that 1874 research papers were published on bio-based plastics and that over half of those papers were published between 2018 and 2021 (approx. 20% in 2021), demonstrating a growing interest in the use of sustainable plastics [1]. According to the same study, the bio-based polymers that garnered the most research interest are poly(lactic acid), poly(hydroxyalkanoate), poly(hydroxy butyrate), and poly(butylene succinate). Lastly, the study specified that the key objective related to the use of poly(lactic acid) (PLA) was its mechanical properties, where food packaging generated the largest demand in interest.

PLA is a polar thermoplastic making it compatible with natural fillers, and it possesses high tensile strength and stiffness, depending on the grade. However, it exhibits a slow crystallization rate, poor thermal resistance, and deformation at temperatures above 50 °C [2]. These drawbacks, along with its higher cost compared to conventional plastics, hinder its use in semi-structural and structural applications. Unlike its fossil-fuel-derived competitors, namely biaxially oriented poly(propylene) (BOPP) and general-purpose poly(styrene) (GPPS), PLA is produced from renewable sources like corn syrup or sugar beet pulp (SBP). A techno-economic analysis reported that with SBP-production methods, the cost of PLA would reach approximately $1.14/kg with the sales of a pectin-rich extract, a byproduct of PLA production, reaching $3–4/kg [3]. In comparison, BOPP and GPPS cost $1.4 and $1.72/kg, respectively.

1.2 Cellulosic nanomaterials

Cellulose nanomaterials (CNs) are widely abundant and naturally occurring around the world. These materials offer high specific strength and stiffness, thermal stability, and chemical functionalization through the abundantly available surface hydroxyl groups (-OH) [4]. However, CNs often require additional treatment to improve hydrophobic matrix compatibility, reduce hydrophilicity, and increase thermal resistance [5]. Cellulose containing only the crystalline components is commonly used for composite reinforcement because of the high axial strength (1.6–6.4 GPa) and stiffness (57–180 GPa) [5].

Cellulose nanocrystals (CNCs) are derived by alkaline or acid hydrolysis from a variety of sources with varying efficiencies [6], and, in suspension, CNCs exhibit two dimensions within the nanoscale. However, after dehydration or compounding, CNCs increase in size by agglomeration, especially with increased weight concentration [7, 8]. This particle size increase occurs when liquid media is removed from the suspension by one of many drying methods [9], where self-interactions are promoted by a multitude of mechanisms [10], including hydrogen bonding [11]. Agglomeration is prevented by surface functionalization or sizing of CNCs, typically by acetylation or with a silane coupling agent for inorganic compatibility, that can increase hydrophobicity or promote covalent bonding for improved matrix interactions [12].

Natural fillers like CNCs work to improve the mechanical properties, but the strength and stiffness are dependent on the filler length. When a nanocellulose filler is compounded in a thermoplastic matrix, a homogeneous distribution of the filler, with good interfacial compatibility, improves the composite properties including strength and stiffness at low (> 1%) filler content [13, 14], but typically at the disadvantage of elongation at break, toughness, and impact strength properties [13]. Improved fiber dispersion and surface interaction dictate what amount of surface area interacts with and improves the properties of the compounded polymer matrix.

1.3 Poly(vinyl alcohol) and cellulose nanocrystals

Poly(vinyl alcohol) (PVA) is a vinyl polymer connected by carbon–carbon bonds with -OH groups and is produced from polyvinyl acetate after hydrolysis, which can be sustainably produced from ethanol [15]. PVA is biodegradable by fungal or bacterial organisms in aerobic or anaerobic environments [16]. PVA is water soluble below the boiling point of water at sea level and atmospheric pressure (100 °C), is non-toxic, and offers biocompatibility and ductility [16, 17]. The hydroxyl functionality presents a major advantage in interfacial compatibility when compounded with CNCs, and these composites can be processed into hydrogels [18] and films [19], wet spun into single fibers [20], and electrospun (ES) into nanofiber mats [21, 22].

The tensile strength and stiffness properties of the various composite CNC forms range from 0.1 to 1000 s of MPa in tensile strength and 0.0001 to 10 s GPa elastic modulus [23]. The composite PVA fiber morphology, especially ES nanofibers, is an attractive option for use as a mechanical reinforcement filler in thermoplastic PLA because of the polar compatibility, high specific tensile strength (100–1000 s MPa), and specific elastic modulus (10–100 s GPa). Furthermore, thermal degradation temperature (225–230 °C) [24] is within the melting range of PLA, especially when combined with CNCs that degrade around 317–330 °C [25]. The utilization of ES composites can address the reduction in agglomeration of CNCs by encapsulation within PVA and the interfacial compatibility of PVA in PLA with respect to surface energy [26, 27].

Electrospinning is a technique that manipulates liquid media with electric fields and low amounts of current through an air gap. The ES method gained popularity in the early 2000s using a single capillary metallic needle (SNE) connected to a high voltage electric potential (+ / −), which generates an electric field. The solution is mechanically pumped by a syringe and propelled by the high voltage to produce nanoscale fibers that accumulate into a random nonwoven mat. Furthermore, a rotating collector can be used to align the fibers [28]. Production with a single needle range from 0.01–0.1 g/h of spun nanofibers, but with advancements in free surface [29] and shear-aided [30] electrospinning, the productivity of spun fibers can increase to 1–10 s g/h, depending on the dissolved polymer viscosity and conductivity. This increased productivity can also occur without reducing the continuous fiber quality or increasing the fiber diameter [31].

Over the past two decades, the interest in ES has increased from hundreds of publications in the early 2000s to thousands a year in the past 5 years, with about one-third of those studies relating to composites [32]. Electrospun fibers exhibit an internal alignment of polymer chains from the elongation process that improve strength, elongation and toughness properties, promoting their use as fiber reinforcement in polymer resins [33]. Various composites have been produced by incorporating organic and/or inorganic fillers into the dissolved polymer solutions to spin composite fibers into mats, forming hybrid composite mats by co-spinning multiple polymer solutions, utilizing the neat polymer or composite mats as continuous reinforcement in solvent casted films and epoxy resins, and by compounding continuous fiber mats between thermoplastic sheets by thermal compression molding [34,35,36].

1.4 Research objective

Recent literature reviews have expressed a need for further investigation into short fiber reinforcement in polymer resins [32, 37]. Few processes offer both an ability of surface modification and production of nanofibers with the option of controlling the fiber lengths as a melt compounding additive. Producing fibers with controlled aspect ratios allow for further analysis of the critical aspect ratio and lengths associated with optimization of the composite stiffness [23]. Additionally, natural fiber PLA composites are brittle and often lack toughness once reinforced. Investigation into ductility and impact strength using fiber reinforcement presents an opportunity to broaden the applications of PLA. Therefore, this research is focused on using well known and compatible components like PVA and CNCs, implementing the facile production of composite ESNFs by electrospinning and evaluating them as cut fiber reinforcement in a thermoplastic (PLA) matrix within the industrially scalable convention of melt compounding. The work done in this study investigated the challenges of producing nanometer diameter composite fibers from mixing PVA solution and CNC suspension, the processing steps related to producing discontinuous fibers from nonwoven mats, compounding the fibers, and characterization of the composite material and mechanical properties.

2 Materials and methods

2.1 Electrospinning solution

Electrospinning solutions were prepared by dissolving a 99% hydrolyzed PVA with a Mw of 130,000 g/mol that was purchased from Sigma-Aldrich (St. Louis, MO, USA) in water and mixing it with a 10.3 wt.% CNC suspension produced by the Forest Products Laboratory (FPL, Madison, WI, USA). Surface tension of the PVA and PVA/CNC mixtures was reduced by addition of 0.5% (v/v) of Triton X-100 with a Mw of 80,000 that was purchased from Sigma-Aldrich (St. Louis, MO, USA). Solution preparation is detailed in previous work [22]. The ESNFs processing conditions were selected with respect to consistency of fiber diameter and alignment, while maximizing CNCs wt.% content during electrospinning at the specified parameters listed in Table 1.

Table 1 Electrospinning and spraying settings used for production of composite filler by type
Full size table

2.2 Sample preparation

The reinforcement fibers were first prepared by electrospinning PVA and CNC composite nanofibers in 4:1 (4:1-nf) and 3:2 (3:2-nf) ratios and PVA nanofibers as control reinforcement for the two types of composite nanofibers, at room temperature (24 ± 1 °C) and humidity (50 ± 10, %). The CNCs, detailed in previous work [38], were electrosprayed under the same environmental conditions, accumulated on a planar collector, removed by a straight edge, and collected in a beaker.

The ESNFs were prepared for compounding by first cutting 1–5-mm strips of the spun mat perpendicular to the alignment direction with an IDEAL 1071 (Krug & Prister GMbH & Co. KG, Balingen, Germany), 710-mm guillotine blade, and woven mat cutting board. Afterwards, the strips were cut along the orientation direction into squared mat sections. The sections were soaked for a minimum of 3 min in liquid nitrogen that was contained in a steel “Cryo-Cup” (Research Products International Corp, Mt. Prospect, IL USA) and plastic housing to increase the fiber stiffness prior to blending, at sub ambient temperatures, with a coffee grinder (Mr. Coffee: model#: IDS77RB; Cleveland, OH USA), which was equipped with a stainless-steel/plastic cup and twin steel blades for two 10-s periods. This process separated the fibers from their as spun state to produce fibrous bundles as shown in Fig. 1.

Fig. 1
figure 1

Processing steps including a strips cut from the continuous fiber mat, b liquid nitrogen soaked and c blended fiber bundles, and d the melt compounded bundles in a PLA matrix

Full size image

2.3 Thermoplastic compounding and specimen preparation

The sample identifiers (ID) are named by filler type. The IDs are differentiated by the ratio of PVA:CNC-nf as 4:1-nf and PVA:CNC2-nf as 3:2-nf. The PVA nanofibers, PVA-nf and PVA2-nf, correspond as references to the PVA wt.% content with respect to the two composite fillers. The weight percent (wt.%) of each fiber type is listed in Table 2 along with the ratio of PVA:CNC and wt.% of PLA matrix in the composite. The electrosprayed CNC nanoparticles (CNC-np) were included as a reference for both composite nanofiber groups with equal CNC weight contents as composite ESNFs [38]. The fillers were conditioned at 60 °C for 4 h prior to melt compounding.

Table 2 Composition details for the studied materials
Full size table

Thermal compounding was done in 100 g quantities using general grade 4043D Ingeo Biopolymer PLA pellets (Minnetonka, MN) in a batch mixer (C.W. Brabender Instruments Hackensack, NJ) at 180 °C for 3 min at 70 RPM. Fillers were then added and mixed for another 5 min. The contents were cooled at ambient conditions and reduced to ≤ 3-mm pieces by a granulator (Hellweg MDS 120/150, Hackensack, NJ, USA).

2.4 Scanning electron microscopy (SEM)

Scanning electron micrographs were taken with a Zeiss NVision 40 (Oberkochen, Germany) electron microscope. Samples were observed under a 3 kV electron beam intensity and a 12 ± 1 mm working distance after sputter coating with a 4-nm-thick layer of gold palladium. Fracture surfaces of the impact testing bars were observed at low magnification (× 50) and high magnification (× 5.0 k +).

2.5 Atomic force microscopy (AFM)

Thermoplastic composites were cut with an Ultra-AFM Diamond Knife 35° 3.0 mm knife into 1 × 2 mm area and 700-nm thick ribbons with a Leica EM UC6 (Leica Mikrosysteme GmbH, Vienna) for observation on the MFP3D (Oxford Instruments Oxon, UK) AFM. Images were collected with AC160TS-R3 probes with ~ 20 (nN/nm) spring constants. Alternating current (AC) mode produced images with a constant free air (1.5 V) amplitude to set point (0.5 V) ratio of 0.33.

2.6 Differential scanning calorimetry (DSC)

The composite melting temperatures were measured with a DSC 2500 (TA Instruments New Castle, DE USA) using T-zero aluminum pans and a heat-cool-heat cycle with an isothermal hold after heating and cooling, a ramp rate and temperature range of 5 min, 10 °C/min, and 20–200 °C, respectively. Sample measurements consisted of approximately 8 mg of sample and were repeated in triplicate.

2.7 Density

Specific gravity measurements were taken following D792-20 protocol at 23 °C.

2.8 Mechanical properties—tensile, flexural, and impact measurements

Tensile, flexural, and impact samples were injection molded using a Mini-Jector Model #50 (Miniature Plastic Molding, Solon, OH, USA) at 180 °C with an injection pressure of 17 MPa in accordance with ASTM D638-14 and D790-10. Tensile and flexural tests were conducted using a universal testing machine (Instron 5966) with a 10kN loadcell (Norwood, MA, USA) at 23 ± 2 °C and relative humidity (RH) of 50 ± 10%. Type I tensile specimens were tested at a 5 mm/min rate using a 50-mm gauge length and flexural specimens of 125 × 12.7 × 3.2-mm dimensions were tested with a 16:1 span to depth ratio at 1.4 mm/min. The flexural test specimens were used to measure impact resistance according to D256-10 on a Ceast pendulum impact tester—Model Resil 50B (Ceast USA Inc. Charlotte, NC). Toughness was determined by integrating the area under the tensile stress vs. strain curve until the sample failed.

2.9 Statistical analysis

Statistical analysis was conducted using JMP® Software Version 16 (SAS Institute Inc., Cary, NC). Mechanical properties were analyzed using Tukey’s Honestly Significant Difference (HSD) mean comparison of each group with the equation:

$${Y}_{ijk}={A}_{i}+{B}_{j}+{C}_{k}+{\in }_{ijk}$$

where Yijk represents each mechanical property; the letters A, B, and C represent the fixed effect of each reinforcement at varying levels of wt.%; and \(\in\) term represents the error. This test was chosen as a conservative analysis of variance with respect to the unequal sample sizes between groups.

3 Results and discussion

3.1 Scanning electron microscopy

The SEM images at × 50 magnification in Fig. 2 show a change in roughness in the PLA matrix with increased ESNF filler, specifically the 15 wt.% 4:1-nf (d) and 12 wt.% PVA-nf (g). However, a threshold exists where the increased roughness only occurred in composites with filler content above 4–6 wt.% ESNF filler, like that observed in 20 wt.% CNCs and poly(ethylene brassylate) composites [39]. In this study, the composite break surface roughness was attributed to the fiber distribution, the close proximity of fibers with hydroxyl (-OH) functionalities to the PLA matrix that promoted hydrogen bonding [40], and the increased complexity of stress transfers throughout the matrix that was attributed to the long fiber reinforcement. High magnification (≥ 10 k ×) images of neat PLA and the composite test coupons are shown in Fig. 3 where evenly distributed fiber holes or exposed ends are visible on the fracture surface. Gavande et al. modeled ESNF nylon-reinforced epoxy resins and bridging mechanisms through fracture analysis. The authors attributed exfoliation, partial exfoliation, and complete exfoliation to the nanofiber matrix interfacial strength [41]. Partial exfoliation or the separation of the fiber from the matrix resulted in pores with diameters exceeding the fiber diameters and the presence of exposed fibers, indicating weaker interfacial compatibility. In this study, exfoliation was most visible in the 3:2-nf PLA composite fracture surface, shown from this in Fig. 3c while some nanofiber ends were visible in 4:1-nf and PVA-nf PLA filled PLA. With an increase in ESNF content, the surface roughness of the fracture surface and frequency of holes or exposed fibers increased. Additionally, with the increased CNC content in the ESNFs, a variable morphology was observed in the fiber. Visible characteristics included a smooth surface break in the 4:1-nf and neat PVA nanofibers, while a rough surface was observed in the 3:2-nf.

Fig. 2
figure 2

SEM micrographs of each filler type (wt.%) PLA composite impact test specimen’s break surface at × 50 magnification

Full size image
Fig. 3
figure 3

SEM micrographs of ESNF-reinforced PLA impact test coupon break surfaces high magnification

Full size image

3.2 Atomic force microscopy (AFM)

Images produced by AFM provided a means to qualitatively observe the material property changes generated by the interactions between PLA/PVA and PVA/CNC, as shown in Fig. 4. The topography, shown in Fig. 5a, c, and e, of the section of composite surface was produced from signal feedback within the closed loop (set point amplitude) from the sample z-piezo [42]. In the topographic images, the white represents the highest point (~ 40 nm) and the black the lowest point (− 40 nm), with increased heights occurring by tip interactions with stiffer materials like cellulose in comparison to PLA. The AFM amplitude modulation technique simultaneously generates phase contrast imagery in a second output to sense qualitative material differences. These differences are produced from changes in the tip’s deflection upon a brief interaction resulting from polymer viscoelastic properties, stiffness, and increased forces within the closed loop [43, 44].

Fig. 4
figure 4

Topographic (nm), phase contrast (°), and surface roughness (nm) output in 5-µm2 cross sectional areas of neat PLA (a, b), PVA-nf PLA (c, d), and PVA:CNC-nf PLA (e, f); root mean square, RMS

Full size image
Fig. 5
figure 5

FTIR spectra of 4:1-nf (a), PVA-nf (b), and CNC-np filled PLA (c), including the fillers without matrix (d), broken into three sections: 3600–1200 (a–d), 1300- (a1–d1) and 980–900 cm−1 (a2–d2)

Full size image

The circular fiber morphologies observed in the height (black and white) are represented by the relative phase change and assigned an arbitrary color scale, shown in Fig. 4b, d, and f. The fiber cross-section dimensions, represented by blue, green, and yellow, are consistent in dimension with the sub-micrometer diameters observed in SEM images (Fig. 3a and b). However, variable surface roughness from cutting the neat PLA before observation with AFM, as shown in Fig. 4a, may not accurately represent the relative stiffness with respect to phase changes on the surface of the neat polymer because of the drastic height differences. Color changes in ESNF composites more strongly suggest material stiffness differences attributed to phase changes. The greater change in phase is attributed to the higher stiffness of the randomly oriented CNCs along the inside the ES fiber (blue) [45]. The phase change is less in the interphase region surrounding the CNCs or in ES fibers with a lower concentration of CNC ends within the cross section (green/yellow), and the lowest phase change is attributed to the matrix (yellow/brown). Further examination by force curve analysis or force modulation is needed to verify stiffness changes based on the initial phase contrast observations, but initial observations indicate that fiber surfaces are engaged with the matrix without voids forming near the fiber interface after cutting the specimen surface. A transcrystalline layer at the interface of the fiber and the PLA matrix may explain the phase change from green to yellow. The interactions of the -OH groups within the fiber and between the fiber and the matrix interface are observed via FTIR analysis.

3.3 FTIR

FTIR spectra were measured to qualify molecular interactions between the neat and composite ESNF filler and PLA matrix. The CNC-np was also measured for comparison. Intermolecular bonding of hydroxyl groups is indicated by the broad 3550–3200 cm−1 region. Rojas et al., in their characterization of composite ESNFs with PVA and CNC, reported two characteristic peaks that indicated free and bound -OH groups within this region [46]. In the spectra for this study, shown in Fig. 5, the former appeared around 3340 cm−1 and the latter around 3295 cm−1. In observation of the fillers by type, shown in Fig. 5c, the free peak was most prominent in the CNC-np, and with increasing CNCs in the composite ESNFs, the free peak intensity shifted towards the bound wavenumber, appearing more like neat PVA-nf. The C-H stretching regions were observed between 2840 and 3000 cm−1 with PVA peaks (annotated with black text in Fig. 5C1) appearing at 2939, 2909, and 2863 cm−1. The CNC peaks appeared at 2943, 2900, and 2854 cm−1, and the composite ESNFs appeared in line with the PVA-nf frequencies.

In the composite ESNF-reinforced PLA, the peak intensity between 3200 and 3500 cm−1 increased with a large wt.% filler for both ratios and PVA-nf, which were relatively lower in the 3:2-nf filler (not shown). Further indications of -OH interactions (bending and stretching) were observed in the 1330–1420 and 1150–1000 cm−1 regions with peaks around 1080 cm−1. Interactions between PVA and PLA is reported to appear around 1750 and 1185 cm−1 by the carbonyl (C = O stretching) and the ester (C-O stretching) [47]. However, in the results shown in Fig. 5a and b, the respective peaks were observed at 1745 and 1180 cm−1. A similar effect was observed in CNC-np filled PLA in Fig. 5c. Both the amorphous and α crystalline bands of PLA were visible at 955 cm−1 and 921 cm−1, respectively, indicating that the crystallinity of PLA is somewhat affected by larger quantities of PVA-nf content, but this effect requires further investigation by differential scanning calorimetry (DSC) and X-ray diffraction (XRD) analysis.

3.4 Thermal properties

Thermal properties obtained by DSC analysis, shown in Table 3, include the glass transition (Tg) and melting (Tm) temperatures (°C), enthalpy of cold crystallization (Hcc) and melting (Hm), and percentage of crystallinity (χc). Glass transition in the first heating cycle was a few degrees higher in each sample group (not listed) but reduced after annealing for 5 min at 200 °C. Addition of the ENSF did not appear to hinder PLA chain mobility with respect to a negligible change in Tg from PLA. Melting temperature also showed little change after the first heating cycle and annealing at 200 °C for 5 min, apart from the 15% 4:1-nf group, which showed a 4.2 °C increase. Despite previous studies, which indicated PVA up to 8% CNCs in composite films did not improve thermal resistance [17], the 4:1-nf at 15% (with 12% PVA and 3% CNCs) imparted some thermal resistance in PLA with the 153.2 ± 0.2 °C melting temperature in comparison with the 149 ± 0.1 °C PLA, which is currently not well understood and requires further investigation using thermogravimetric analysis. However, a secondary melt peak was observed in the PVA-nf 12% and 4:1-nf 15% groups DSC graph shown in Fig. 6 and suggests that the crystallization effect that is observed between PVA and CNCs, reported by Niinivaara et al. [48] studying thick films, affects the crystallinity in the PLA ESNF composite. Often, nucleation is detected by DSC analysis when measuring heat flow of composites containing small particles in semicrystalline polymer matrix [49]. An increase in crystallinity was observed to varying degrees and is listed in Table 3.

Table 3 Second heating cycle DSC and crystallinity percent calculation results of ESNF and CNC nano particle-reinforced composite PLA
Full size table
Fig. 6
figure 6

Graph of second DSC heating run of neat PLA and composite specimens

Full size image

The PLA selected for this study is an amorphous grade and contains low crystallinity, reportedly from high D-lactide stereoisomer [50]. Addition of cellulose containing filler improved the Xc to 3.5% with 1.3 wt.% 3:2-nf, and with increased fiber content, it decreased to 2.2 and 2.4% Xc, which was consistent among most ESNF filled PLA. However, thermal conductivity has been shown to improve in cellulose filled PVA ESNFs and could be attributed to a crystallization window at the 10 °C/min cooling rate used in this study [51]. Inconsistencies may be attributed to the distribution of fillers during melt mixing and potentially resolved by compounding in a twin-screw extrusion system.

3.5 Density

The calculated density was subtracted by the measured density, performed based on buoyancy (ASTM D792) that is listed in Table 4, and divided by the ideal density to give void fraction (Vf, %). Void content calculations were performed based on the weight fraction of the contents with respect to the theoretical volume based on the sum of constituent material density.

$${V}_{f}=100*\frac{{\rho }_{c}-{\rho }_{exp}}{{\rho }_{c}}$$

where \({\rho }_{c}\) is the theoretical density and \({\rho }_{exp}\) is the experimental density. The theoretical density was calculated using the equation

$$\frac{1}{{\rho }_{c}}=\sum_{i=1}^{n}\frac{{W}_{i}}{{\rho }_{i}}$$

where 1 is divided by the sum of the respective weight fraction of the components \({W}_{i}\) divided by the constituent materials density \({\rho }_{i}\).

Table 4 Calculated density with mean experimental density, standard deviation, and percentage difference from PLA
Full size table

The void calculation results for all samples were negative values and may indicate the lack of accuracy in determining the experimental density of the composite. Additionally, improvements in the crystallinity of the composites were attributed to the change in experimental density which resulted in ostensibly void free composites. Literature regarding this subject notes that a 1% change in void calculation can occur with 1% error in the theoretical density [52]. It is well known that compounds containing small particles create nucleation in semicrystalline polymer matrices.

3.6 Mechanical properties

Melt compounded ESNFs generally improved the tensile yield strength (TYS) of PLA as shown in Fig. 7 with values listed in Table 5. The 2.5 wt.% composite 4:1-nf initially increased TYS by 17%, but with increased loading levels to 15 wt.%, the TYS improved 21% to 68.3 ± 1.2 MPa. In comparison, 3:2-nf improved tensile strength by 8 and 15% in the 1.3 and 2.6 wt.% groups, but by 3% with 7.5 wt.% filler in PLA. Previous work determined a 1:1 PVA:CNC fiber mat stiffness was 2.6 × that of 4:1 ratio mat and may translate to the 3:2-nf [22]. Additionally, the 15% coefficient of variance (COV) in the 7.5 wt.% 3:2-nf was attributed to potential inconsistency in the fiber reinforcement performance, sample defects, or inconsistent distribution of fibers within the PLA matrix from less PVA content in the fiber. ESNFs may have also entangled into fiber bundles and remain undispersed during compounding; PLA composites containing neat PVA fillers, between both control groups, improved TYS 15–20%, from 64.9 ± 0.1 with 1.6 wt.% PVA2-nf and to 67.8 ± 1.8 with 4 wt.% PVA-nf. The CNCs showed improvements in 3% and 5 wt.% specimens. Previous work, using a TSE to compound CNC-np at the same weight percentage showed a 10% improvement in strength [38].

Fig. 7
figure 7

Tensile yield strength (TYS, left) and tensile modulus of elasticity (TMOE, right) percentage change by filler (wt.%) and organized by filler type

Full size image
Table 5 Mean (µ) tensile and flexural properties of composite PLA groups including the standard deviations ( ±) and percent change (Δ,%) by filler (wt.%)
Full size table

Statistical analysis using Tukey’s HSD mean comparison and letter difference among filler wt.% listed in Table 6 determined changes in TYS which were insignificant, apart from the 15 wt.% 4:1-nf. Strength improvements were attributed to interactions of hydroxyl groups in PVA with the ester bonds in thermoplastic PLA [47], as indicated by FTIR analysis shown in Fig. 5. The shift in unbound -OH groups improves the even distribution of the reinforcement fibers and increased the discontinuity in stress transfer across the sample. In effect, the surface roughness of broken samples from SEM images correlates with the statistical analysis letter difference of significant changes in TYS of the 15 wt.% 4:1-nf filler listed in Table 6. Figure 3a shows the even distribution of fibers and lack of fiber pull out that is attributed to the TYS improvement.

Table 6 Statistical analysis on mechanical properties by Tukey’s HSD mean comparison of filler types
Full size table

TYS tensile yield strength, TMOE tensile modulus of elasticity, TSb tensile strain at break, FYS flexural yield strength, FMOE flexural modulus of elasticity.

An apparent trend exists between 4:1-nf (wt.%) and tensile modulus of elasticity (TMOE) results, shown in Fig. 7 and listed in Table 5. Samples compounded with increasing wt.% of 4:1-nf showed increased TMOE of up to 4.74 ± 0.34 GPa, a 30% increase. PLA compounded with neat PVA-nf increased at most 10.2% to 4.0 ± 0.1 GPa with 4 wt.% filler and reduced TMOE as little as 3.1% in 0.8 wt.% PVA2-nf. PLA composites containing 3:2-nf exhibited reduced TMOE at 1.3 wt.% and increased up to 11% to 4.04 ± 0.45 GPa with 2.6 wt.%. Statistical analysis attributed letter differences, listed in Table 6, was based on significance of mean comparisons among variations in filler wt.% in 4:1-nf and the CNC-np, but not among the neat PVA-nf and PVA2-nf or the 3:2-nf filled PLA groups. The observed improvements attributed to the 4:1-nf relate to the interfacial compatibility starting with the PVA encapsulated CNC fiber and translating to PVA shell interactions with the PLA matrix. Despite having more CNCs within the fiber, the 3:2-nf interacting with the PLA matrix, like the 4:1-nf, provided less intra fiber resistance to stress acting on the composite.

In support of this notion, Park et al. attributed hydrogen bonding and interaction of the ester linkages in the CNCs’ anhydroglucose units with PVA hydroxyl groups (-OH) to the increased rigidity observed in the solvent cast films containing 3 and 5 wt.% CNCs [51]. Additionally, alignment of CNCs was observed by solvent etching ESNF poly(methyl-methacrylate) (PMMA), observed with SEM, where Dong and co-workers concluded that alignment occurred during the polymer elongation phase (spinning) of sub-micron fiber formation [53]. In the same work, the author indicated that the fibers’ modulus increased with the addition of CNCs up to 17 wt.%, as measured with nanoindentation, and that increased wt.% of CNCs were counterproductive to improving the elastic modulus. Moreover, Niinivaara et al. measured a 13-fold improvement in modulus with 10% CNCs that the author attributed to the CNCs forming a percolated network in thick PVA films, nucleating the PVA, and compatibility between the filler and matrix [48]. Considering the fiber content between the two ratios used in this study, the 4:1-nf falls within the increasing modulus effect, while the 3:2 ratio may distribute throughout the matrix; by reducing the PVA content available for surface interactions with PLA, the interfacial bond may provide less resistance to tensile stress during testing.

Tensile strain at break (TSB) of PLA increased when compounded with all wt.% of neat PVA-nf and at the lower wt.% of 4:1 ratio composite fiber; the values are listed in Table 5. Up to a 23.5% increase in strain was observed in 12 wt.% PVA-nf composite PLA, and less filler had a varied effect on TSB with improvements of 5–8% between PVA and PVA2-nf filled PLA. The 4:1-nf PLA composites showed improvements with 2.5 and 5 wt.% filler of 11.1 and 10.3%, respectively, but the 15 wt.% and all of the 3:2-nf reduced PLA TSB. Noteworthy, however, statistical analysis determined that the TSB results of fiber and particle filled PLA composites were not significantly different from neat PLA by mean comparison (Table 6).

Flexural yield strength (FYS), illustrated in Fig. 8, exhibited ± 2% changes when filled with ESNFs, apart from the 6–9% improvements observed in PVA2-nf. The filled materials within each ESNF type were not considered significantly different from neat PLA by statistical analysis, listed in Table 5. The CNC-np showed some increased FYS values, as cellulose nanomaterials in polymeric matrix are well known to improve flexural strength and stiffness [54, 55].

Fig. 8
figure 8

Flexural yield strength (FYS, left) and flexural modulus of elasticity (FMOE, right) by filler (wt.%) and organized by filler type

Full size image

The addition of ESNFs generally produced a consistent reduction in flexural modulus of elasticity (FMOE), apart from the 6 and 4% increase observed in 15 wt.% 4:1-nf and 3:2-nf, respectively. The CNC-np showed a 1% reduction in FMOE, which is attributed to the low loading level (low volume) and short aspect ratio of CNCs [56]. Addition of CNCs was documented to reduce mechanical properties after reaching a percolation threshold [57].

3.6.1 Impact strength

Impact strength improvements, shown in Fig. 9, are typically attributed to the interfacial interaction of the fiber filler, the reduction in crack propagation, and increased complexity of stress transfer deviations in the matrix between the reinforcement fibers [58]. Williams et al. elaborate further, explaining that the key factors to controlling fracture energy relate to the fiber dimensions, mechanical strength, volume fraction, and coefficient of friction during pull-out [59]. In this study, PLA compounded with PVA-nf and 4:1-nf improved the impact strength 32 and 54% with 12 and 15 wt.% ESNF reinforcement, listed in Table 7. The 7.5 w.t% PVA:CNC2-nf improved the IS of PLA by 23.5%, and the 4.5 wt.% PVA2-nf control showed a 27.6% improvement. The expectation of increasing impact strength with increasing PVA-nf in PLA was unsupported by the results of 2 and 4 wt.% sample groups. This implies an inconsistent dispersion of ESNFs among melt compounding of each formulation, presumably attributed to fiber agglomeration. In this study, statistical analysis of ESNF filled PLA by mean comparison of filler wt.% within each filler type determined that all groups at the highest wt.% ESNFs were statistically significant from neat PLA. At lower levels, some overlap occurred, as indicated by letter differences that are shown in Table 6.

Fig. 9
figure 9

Impact strength (IS, left) and modulus of toughness (T, right) by filler (wt.%) and organized by filler type

Full size image
Table 7 The mean (µ), toughness (T), and impact strength (IS) of composite PLA groups including the standard deviations ( ±) and percent change (Δ, %) by filler (wt.%) and type
Full size table

3.6.2 Toughness

Integration of the stress strain curves divided by density produced toughness (T, J/g) results [60]. The results are listed in Table 7 and illustrated in Fig. 9. The bar graph shows how composite ESNFs of both ratios, along with CNC-np, reduced the composite toughness, but the PVA-nf increased toughness by 19% with a 12 wt.% content in PLA. Contrary to the 4 wt.% PVA-nf filler’s 13% improvement to 1.19 ± 0.15 (J/g), the 4.5% PVA2-nf, as the control group for the 3:2 ratio composite, decreased the 1.05 ± 0.06 J/g PLA toughness by 21.9%, which would indicate poor mixing when compared to the former. Moreover, addition of CNC-np initially reduced the toughness by 56%; but with increasing levels, toughness decreased less by 12.4% and may indicate improved distribution with increased filler content. Liu and coworkers reported an increase in tensile yield strength with the addition of PVA in ESNF PLA mats but measured a decrease in toughness with 20 wt.% levels of PVA. A similar effect is observed in this study, where TYS values were improved in composite ESNF filled PLA, but the elongation decreased with increased filler amounts.

4 Discussion

4.1 Fiber effect and mechanical properties

In a review of natural fiber-reinforced composites [23], discussion surrounding aspect ratio expounds the effect of cellulose aspect ratio on the mechanical properties of an polymer matrix. Therein, small aspect ratio fibers improve the stiffness but reduce the strength from the fiber end effect [61], while high aspect ratio fibers improve both the strength and stiffness because of more surface interactions along the fibers length, an improved interfacial shear strength. The reduction of particle size to the nanoscale may contribute little to reinforcing a polymer matrix if the interfacial interactions with the matrix are poor. With increased surface interactions, the critical aspect ratio decreases, and smaller lengths are required for reinforcement. When compounding a polymer matrix with unsized CNC particles, the particles reportedly re-aggregate within the composite during extrusion [26]. With ESNF composites, the CNCs are encapsulated in the fiber before composite production. An added benefit of electrospinning is the increased specific surface area attributed to the nanoscale ESNF diameter. Compatibility between the polymer matrix and fiber is address via the material selection process, as opposed to a surface treatment or additive for chemical compatibility. Post processing ESNF mats by cutting fiber lengths of aligned mats allows for some consistency in the aspect ratio of cut fibers. This provides some consistency in the control of mechanical properties with respect to the fiber interfacial adhesion to the matrix and orientation parallel to the stressed axis [33, 62].

4.2 Surface characteristics and melt compounding

Colijn et al. notes that the polar and dispersive characteristics of the filler and polymer matrix should be similar for spontaneous mixing [27]. This applies to the PVA fiber and PLA matrix and to the interaction of PVA and CNC during ES production of the composite nanofibers. The dispersive surface energy of PLA at 25 °C (298 K) was determined to be 33.3 mJ/m2 [63]. Chen et al. determined that the dispersive energies of PVA, between 110 and 150 °C, were around 45 mJ/m2 but could not extrapolate the value to room temperature [64]. Peng and Gardner measured dispersive energies of multiple varieties of commercial microcrystalline cellulose (MCC) and found the results ranged between 55 and 75 mJ/m2, depending on the manufacturing process [65]. Encapsulating the CNC in PVA may allow for more spontaneous mixing in molten PLA as the PVA coated CNCs are more closely related in surface energy compatibility, as opposed to CNCs self-affinity when compounded directly in PLA. Additionally, the PVA and CNC fiber specific strength improves with reduced dimensions, likely attributable to a closer proximity of fiber to matrix surface interaction [40], and offers the advantage of producing a hydrogen bonding network between each material in the composite.

4.3 ESNF composite fiber and PLA

The hydrogen bonding network, coupled with the high aspect ratio of the ESNFs, produced a noteworthy improvement in the toughness and impact properties in PLA composites containing PVA-nf and 4:1-nf, respectively. Control of the fiber lengths allows for greater utility of the mechanical properties imparted by the fiber [66]. In previous studies, elastomeric PLA blends greatly improved the strain at break and impact strength to 104 ± 53.2% and 59 ± 2.1 (kJ/m2) while reducing the tensile property performance [67]. Another study, that filled PLA with 20% acetyl tributyl citrate and 4 wt.% nano fibrillated cellulose (NFC), saw similar increases to elongation without as much tensile property loss [68]. By increasing the composite elongation and chain mobility, the tensile property performance cannot be maintained. However, coupling the ESNF reinforcement with a plasticizer and the use of long fibers with polymer specific surface interaction to provide interfacial compatibility could promote strength and stiffness improvements.

Over filling of CNCs in the 3:2-nf hindered the hydrogen bonding network between PVA and CNC, as indicated by FTIR (Fig. 5) and IS properties (Fig. 9), from performing as reinforcement on the level observed by the 4:1-nf reinforcement. Shalom et al. varied the ratios of CNC to PVA and encountered morphological changes that enhanced tensile strain and toughness in 1:5 PVA:CNC composite films by 570% and 202 MJ/m3, regardless of the crosslinker additive [19]. However, in this case, the improvements to toughness and strain failed to translate with thermally compounded PLA. One disadvantage in this study involved the small batch size in a bowl mixing unit. Between the PVA-nf and PVA2-nf filled PLA, mechanical property values were inconsistent, which can be attributed to the distribution of fibers in the matrix. Adopting a masterbatch methodology was proven to increase dispersion of reinforcing agents in thermoplastics with a twin screw extrusion unit [69]. However, at the ESNF production levels used in this study, accumulating sufficient filler is prohibitively time consuming. However, high throughput techniques like free surface [70] or shear-aided [30] electrospinning could be utilized to address this setback.

5 Conclusions

Electrospun nanofibers were produced in quantities required to melt compound with PLA after mechanical cutting to a 1–5-mm length. The ESNFs were observed by SEM and AFM in the PLA matrix as partially exfoliated fibers protruding from the fracture surface after melt compounding. Long fiber length, nano-scale dimensions, and hydrogen bonding interactions were attributed to the tensile property improvements. A synergistic effect was observed by the 4:1 composite ESNF that lacked in the 3:2-nf PLA. The FTIR analysis indicated smaller peaks in the unbound -OH frequencies of ESNF fillers compared to CNC-np. At 15 wt.%, the 4:1-nf filled PLA exhibited the highest impact strength of improvement of a statistically significant 54%, maintained an increased Tm after 5 min of annealing at 200 °C, and reduced the tensile strain at break by 7% with a 21% and 30% improvement in tensile strength and modulus. The PVA-nf PLA exhibited improved strain at break and tensile strength by 19% and 24%, respectively. Furthermore, the toughness was improved in PVA-nf PLA composites. The flexural strength was statistically equal to neat PLA results with ESNF fillers. The bending modulus showed up to 6% and 4% improvement with 4:1 and 3:2 ratio composite ESNFs and decreased with all others by 5–10%. Table 8 summarizes the 4:1 ratio composite ESNF properties.

Table 8 Summarized mechanical property results with the mean (σ), standard deviation ( ±), and % change from neat PLA with 15 wt% 4:1 PVA:CNC-nf reinforcement
Full size table