Exploiting the network architecture of thiol–ene photo-crosslinked poly(ε-caprolactone) towards tailorable materials for light-based 3D-printing

Abstract

Over the past decade, biodegradable polyesters have played a key role in the field of biomedical engineering. While several of these polymers have been extensively studied and applied in various medical applications, there remains a great need for the development of biodegradable polymers that can be tailored towards specific applications. This study aims to investigate the polymer network architecture of thiol–ene photo-crosslinked poly-ε-caprolactone (PCL) as a tool to fine-tune a range of physical characteristics, including mechanical, thermal and photokinetic properties as well as hydrolytic degradation profiles. Additionally, the study explores the potential of thiol–ene photo-crosslinkable PCL (E-PCL) precursors to be exploited in digital light processing (DLP). The network topology was modified by adjusting the functionality of E-PCL precursors, as well as the functionality of thiolated crosslinkers. The results show that by altering the network architecture, it is possible to fine-tune the physico-chemical properties of the obtained networks. Successful application in DLP has been illustrated as well, exploiting telechelic, tri- and tetra-functional E-PCL precursors combined with a thiolated crosslinker. In conclusion, the reported strategy holds great potential to serve various biomedical applications such as medical devices, drug delivery or tissue engineering.

---

Introduction

In recent decades, there has been an increasing interest in (bio)degradable polymers, particularly in biodegradable polyesters, which have become vital to the future of a plethora of biomedical applications such as drug delivery and tissue engineering (TE).¹ In order to design a polymer exhibiting physical properties that align with a particular envisioned application, the material properties require careful fine-tuning. Bioresorbable synthetic polymers such as poly-ε-caprolactone (PCL), poly(lactic acid) (PLA), poly(glycolic acid) (PGA) and copolymers thereof, have been widely investigated in the context of several biomedical applications.^2,3 These polymers are interesting due to their controllable biodegradability and biocompatibility. Moreover, their use in several biomedical applications is approved by the Food and Drug Administration.⁴ Especially PCL has attracted significant attention as a result of its toughness (tensile strength of 16–24 MPa and elongation at break of approximately 1000%, for thermoplastic PCL) and slow biodegradation rate of 3 to 4 years. Furthermore, due to PCLs excellent visco-elastic properties, it has been frequently applied in injection moulding and extrusion-based 3D-printing.^5–9 It should be noted that the latter encompasses high molar mass thermoplastic PCL (i.e., 80 000 g mol⁻¹). Alternatively to deposition-based 3D-printing, light-based 3D-printing holds great potential to print biodegradable constructs in a patient-specific manner at high resolution. Light-based 3D-printing techniques, such as digital light processing (DLP), characterized by a resolution of 25 μm, employ the principle of spatiotemporally controlled, light-initiated crosslinking.¹⁰ Therefore, in order to process PCL through the aforementioned light-based technique, functionalization with photo-crosslinkable moieties is required. To date, several approaches have already been reported to render PCL photo-crosslinkable. The most frequently reported approach involves functionalization with (meth)acrylate-functionalities via the use of urethane-based coupling chemistry or through reaction with acryloyl chloride.^11–13 Nevertheless, previous research has indicated that (meth)acrylate-based networks result in inferior mechanical properties (i.e., brittle materials with poor toughness), due to the fact that the crosslinking process proceeds through a free-radical chain-growth polymerization, which leads to a more inhomogeneous network topology, as compared to an orthogonal step-growth polymerization process.^14–19 Therefore, step-growth thiol–ene photo-crosslinked networks have been proposed as a promising alternative.^19–21 It should be noted that, in addition to the topology of the polymer network, the number of covalent linkages, formed during crosslinking, should also be considered as an important factor that influences the mechanical properties. Indeed, when considering precursors with an equal molar mass and number of functional groups, twice as many covalent bonds will be formed in case of a chain-growth polymerized network (i.e., acrylate crosslinking) compared to a step-growth polymerized network (i.e., thiol–ene crosslinking), assuming full conversion.²²

Previous studies have already demonstrated that the photo-crosslinking kinetics and mechanical properties of acrylate-crosslinked PCL networks can effectively be modulated by adjusting the molar mass and functionality of the PCL precursors (i.e., crosslink density).¹¹ However, in this study the influence of the polymer network architecture is investigated focussing on thiol–ene photo-crosslinked PCL. The current work builds further on our previous research in which it was demonstrated that the molar mass between crosslinks (M_c) could effectively be used to fine-tune the network's physico-chemical properties.²³ Herein, we further extend the tunability of thiol–ene photo-crosslinked PCL networks by modifying the network topology through both the architecture of the oligomeric precursor (i.e., linear versus star-shaped) and the crosslinker (i.e., tri- versus tetrafunctional thiol). The influence of the network topology on the physico-chemical properties of the crosslinked networks is comprehensively evaluated. Ultimately, the developed materials are exploited in digital light processing (DLP) as well, and their printability is discussed in the context of their network topology. The reported materials have significant potential to serve a broad range of TE applications, such as human nasal cartilage or hyaline cartilage (necessitating a storage modulus range of 234 kPa to 800 kPa), due to their ability to fine-tune the PCL network towards the required mechanical properties and to efficiently generate patient-specific scaffolds.^24,25

Results and discussion

Synthesis and characterization of the alkene-functionalized PCL precursors

In order to create thiol–ene photo-crosslinked PCL networks with varying network topologies, alkene-functionalized PCL (E-PCL) precursors were synthesized with different architectures, according to a protocol reported previously.²³ First, linear and star-shaped (bi-, tri- and tetra-functional) PCL precursors were synthesized, using ethylene glycol, 2-ethyl-2-(hydroxymethyl)-1,3-propanediol (trimethylolpropane) and pentaerythritol as initiator, respectively (Scheme 1). A ring opening polymerization (ROP) approach was chosen, compared to the polycondensation of hydroxycarboxylic acids, since ROP ensures a low polymer dispersity (i.e., 1.1 to 1.6).²⁶ The polymerization reaction continued until all ε-caprolactone monomers had completely reacted, as evidenced by the shift of the H_e proton signal from 4.21 ppm to 4.06 ppm (Fig. S1†). The monomer conversion for each PCL precursor exceeded 99% (Table 1).

Scheme 1 Two-step synthesis of telechelic and star-shaped alkene-functionalized poly-ε-caprolactone (E-PCL(2), E-PCL(3) and E-PCL(4)) according to a protocol previously reported. Telechelic, tri-functional and tetra-functional PCL were obtained through ROP, by using ethylene glycol, trimethylolpropane and pentaerythritol as initiator, respectively.

Table 1 Chemical characterization of the synthesized E-PCL oligomers with varying architectures, including the number average molar mass obtained via¹H-NMR spectroscopy (M_n), alkene content, degree of substitution, conversion, yield and polydispersity

	M n via ^1H-NMR (g mol^−1)	Ene content (mol g^−1)	Degree of substitution (%)	Conversion (%)	Yield (%)	Đ
E-PCL(2)	8000	2.26 × 10^−4	91	99.5	99	1.15
E-PCL(3)	7900	3.78 × 10^−4	99	99.6	98	1.10
E-PCL(4)	7700	4.89 × 10^−4	94	99.7	99	1.08

---

In a second part of the procedure, the PCL precursor was functionalized with alkene functionalities. To this end, allyl isocyanate was added to the reaction mixture (Scheme 1) and ¹H-NMR spectroscopy was used to monitor the progress. The reaction was finished when a complete shift of the end-standing H_e protons from 3.64 ppm (i.e., next to hydroxyl) to 3.78 ppm (i.e., next to urethane bond) could be observed (Fig. S2†). Ultimately, the E-PCL product was purified through precipitation in cold diethylether, generating E-PCLs at high yield (>98%) (Table 1). To evaluate the agreement between the targeted (8000 g mol⁻¹) and experimentally obtained molar mass, ¹H-NMR spectroscopy was employed (Fig. 1A). The molar mass of telechelic, tri-functional and tetra-functional E-PCL precursors was determined based on the protons next to the urethane functionalities (H_e proton at 3.78 ppm) (eqn (2) and (3): Experimental section). The obtained molar masses were in excellent agreement with the targeted molar mass, as reflected by a minor deviation of 1 to 4% (Table 1). Additionally, the dispersity of the E-PCL precursors, determined through gel permeation chromatography (GPC), ranged between 1.08 to 1.15 (Table 1), demonstrating a narrow molar mass distribution. Furthermore, these results corresponded well with polymer dispersities previously reported in literature for PCL generated through ring opening polymerization of ε-caprolactone exploiting a Sn(Oct)₂ catalyst (i.e., 1.08 to 1.56).²⁷ Additionally, the structure of the synthesized materials was further confirmed through Fourier-Transform Infrared (FTIR) spectroscopy (Fig. 1B). As anticipated, in the FTIR spectra of PCL, a CH₂ stretch, C=O stretch, C–C stretch, C–O stretch and C–O–C stretch could be observed. Since the alkene-functionalized PCL polymers have additional urethane bonds, a N–H stretch was also observed. The precursors with a higher functionality have a higher concentration of urethane bonds, leading to more intense N–H stretch vibrations. Furthermore, in order to quantitatively determine the alkene content and degree of substitution (Table 1), ¹H-NMR spectroscopy with dimethyl terephthalate (DMT) as an internal standard was performed. Comparison of the intensity ratio of the characteristic DMT signal (8.0 ppm) to those of the alkenes (5.15 and 5.85 ppm) allowed to determine the alkene content through eqn (4) and (5) (Experimental section). Furthermore, the degree of substitution for telechelic (E-PCL(2)), tri- (E-PCL(3)) and tetra-functional (E-PCL(4)) PCL precursors equalled 91%, 99% and 94%, respectively. This indicates that the synthetic approach described herein results in alkene-functionalized PCL with excellent control. It should be noted that, in order to generate networks with excellent network connectivity, a high degree of substitution is crucial.

Fig. 1 (A) ¹H-NMR spectra of telechelic, tri-functional and tetra-functional E-PCL with each proton indicated in the spectrum; (B) FTIR overlay spectra of telechelic, tri-functional and tetra-functional E-PCL polymers.

Development and characterization of thiol–ene photo-crosslinked PCL networks

The synthesised E-PCL precursors were subsequently used to create photo-crosslinked networks with varying network topologies (and crosslink-densities) (Fig. 2C). 2D Sheets of the thiol–ene photo-crosslinked networks were obtained by preparing a solution of a telechelic, tri- or tetra-functional E-PCL precursor with pentaerythritol tetrakis(3-mercaptopropionate) (PETA-4SH) or trimethylolpropane tris(3-mercaptopropionate) (PETA-3SH) as thiol crosslinker. Subsequently, ethyl (2,4,6-trimethyl benzoyl) phenyl phosphinate (TPO-L) was added to the mixture, as photo-initiator (0.26 wt% or 1.7 mol% according to E-PCL content). The mixture was irradiated with UV-A light (30 min, 365 nm, 10 mW cm⁻²) in order to obtain photo-crosslinked sheets, which subsequently crystallized, due to cooling to room temperature. As a result of the orthogonal thiol–ene chemistry, used to crosslink the networks, the generated networks were anticipated to be more homogeneous, compared to acrylate crosslinked networks.¹⁸

Fig. 2 (A) Photo-rheology, revealing the gel point and G′, obtained from E-PCL(2)-4SH (green), E-PCL(3)-4SH (orange) and E-PCL(4)-4SH (blue) networks; (B) photo-rheology, revealing the gel point and G′, obtained from E-PCL(3)-4SH (orange) and E-PCL(3)-3SH (purple) networks; (C) schematic depiction of theoretically homogeneous networks formed upon thiol–ene photo-polymerization of E-PCL(2)-4SH vs. E-PCL(3)-4SH vs. E-PCL(4)-4SH and E-PCL(2)-4SH vs. E-PCL(2)-3SH.

To evaluate the photo-crosslinking kinetics of the crosslinked networks, photo-rheology was performed (Fig. 2A and B). The gel point (i.e., crossover between the storage modulus (G′) and loss modulus (G′′)) can be considered the point at which the sample is converted from a liquid to a solid. Interestingly, we found that the gel points were delayed when the functionality of the precursor and crosslinker decreased (Table 2). In order for the networks to gel, a critical number of crosslinks have to be formed, with the respective amount depending on the functionality of the PCL precursor and crosslinker.²⁸ As a result, higher double bond conversions are necessary when PCL oligomers with reduced functionality are considered, therefore delaying gelation.^28,29 Moreover, we evaluated the post-polymerization storage modulus (G′) of each network, as depicted in Table 2. It should be noted that the post-polymerization storage modulus G′ referred to here, is the storage modulus when a plateau was reached, which indicates that the ultimate crosslinking degree was reached. As illustrated, higher G′ values were obtained for E-PCL precursors and crosslinkers with more functionalities. Interestingly, the post-polymerization storage modulus (G′) can be related to the crosslink density, according to the rubber elasticity theory (eqn (1)).³⁰

Table 2 Overview of gel point, experimental storage modulus (G′), theoretical storage modulus (G′), experimental M_c determined via¹H-NMR spectroscopy, theoretical M_c, gel fraction and swelling ratio of E-PCL(2), E-PCL(3) and E-PCL(4) networks crosslinked with PETA-4SH and PETA-3SH

	Gel point (s)	Storage modulus G′ (kPa) (exp.)	Storage modulus G′ (kPa) (theor.)	M c (g mol^−1) (exp.)	M c (g mol^−1) (theor.)	Gel fraction (%)	Swelling ratio
E-PCL(2)-4SH	22 ± 0.3	207 ± 0.06	194	8000	7560	93 ± 1.2	15.3 ± 0.21
E-PCL(3)-4SH	16 ± 0.6	536 ± 0.06	507	2630	2500	97 ± 0.4	7.9 ± 0.13
E-PCL(4)-4SH	14 ± 0.3	741 ± 0.02	812	1925	2110	97 ± 1.1	6.0 ± 0.16
E-PCL(2)-3SH	84 ± 0.8	58 ± 0.02	130	8000	17 990	80 ± 1.2	26.4 ± 1.28
E-PCL(3)-3SH	29 ± 0.8	280 ± 0.01	394	2630	3730	96 ± 1.0	9.8 ± 0.25
E-PCL(4)-3SH	19 ± 1.0	550 ± 0.04	696	1925	2440	96 ± 0.1	7.1 ± 0.01

---

According to the rubber elasticity theory, G′ and the crosslink density are related, under the assumption of full conversion and that the network consists of equally dispersed crosslinks, without entanglements, loops, dangling chain ends or other imperfections.^31–33 This could partly be assumed, since each precursor is characterized by a high degree of substitution (>91%) and the networks are crosslinked by means of a step-growth polymerization. However, it should be kept in mind that thiol–ene chemistry is not completely orthogonal, as there is a possibility for side reactions to occur, such as disulphide formation and unstable carbon radicals forming C–C bonds. These side reactions could potentially introduce some imperfections to the network. Subsequently, the theoretical storage modulus and M_c values of the PCL precursors could be calculated for each network according to eqn (1) (Table 2). As shown for the E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH networks, the theoretical G′ and M_c values closely match to those obtained experimentally (via photorheology and ¹H-NMR spectroscopy), indicating that the topology of the obtained networks can be assumed to be homogeneous with limited presence of network defects. However, when the E-PCL precursors were crosslinked with PETA-3SH, a substantial deviation between the experimental and theoretical G′ and M_c was observed. Moreover, it can be seen that the deviation between both increases as the functionality of E-PCL lowers. This could be attributed to the increased conversions that are required to form an interconnected network when considering building blocks with a lower number of functional groups.^28,29 Thus, it can be expected that the presence of impurities, network defects or disulphide bonds will impact the network formation more severely. Notably, when considering step-growth polymerized networks, the combination of a bi-functional oligomer and a tri-functional crosslinker represent the minimal amount of required functionalities to create an interconnected network. The high theoretical M_c values that were obtained for the networks that were crosslinked with PETA-3SH, result from the low experimental G′ values stemming from the reduced interconnectivity of the networks.

In addition, the gel fraction and swelling ratio of the networks were assessed to further evaluate the network connectivity. The soluble fraction of the crosslinked material should ideally be zero, indicating that no material is able to leach out from the network. The gel fractions of most E-PCL networks exceed 93%, illustrating that highly interconnected networks are formed (Table 2). Nonetheless, the gel fraction of the E-PCL(2)-3SH network displayed a significantly lower value, further confirming imperfect network formation. In addition, a continued decrease of swelling ratio could be observed as the functionality of the E-PCL precursor and thiol crosslinker increased (Table 2). This trend was anticipated since the presence of an increased amount of functionalities, when considering precursors with similar molar mass, implies that M_c is reduced. Hence, a higher crosslink density is obtained which limits the network's ability to swell, according to the Flory–Rehner model.³⁴ To summarize, we found that proper network formation strongly depends on the functionality of the PCL-precursor and thiolated crosslinker.²²

Thermal characterization of thiol–ene photo-crosslinked PCL-based networks

The thermal properties of the photo-crosslinked E-PCL networks were examined by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC). By employing TGA, the thermal degradation temperatures (T_d – 2% mass loss) were found to range between 226 and 266 °C (Table 3). Furthermore, DSC was used to determine the crystallization (T_c) and melting (T_m) temperature for each E-PCL network (Fig. 3A and Table 3). Melting enthalpy (ΔH_m) values of 66.9, 53.5 and 41.1 J g⁻¹, respectively, were obtained for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH networks, as well as 67.6, 54.3 and 42.1 J g⁻¹, respectively, for those crosslinked with PETA-3SH. A clear increase of the melting enthalpy (ΔH_m) upon decreasing functionality could be observed, revealing an increased crystallinity (X_c). These data illustrate a clear correlation between the crosslink density and the degree of crystallization. The higher crystallinities that were observed for networks with lower crosslink densities can be explained by the fact that these networks contain longer, continuous PCL segments exhibiting an increased mobility.

Fig. 3 (A) DSC thermogram obtained from photo-crosslinked E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH networks. An increasing melting temperature and melting enthalpy indicate an increasing crystallinity; (B and C) remaining mass (%)–degradation time (days) plots obtained via accelerated degradation tests with 5 M NaOH, on photo-crosslinked E-PCL networks; (D and E) stress–strain curves for each photo-crosslinked PCL-network, with a close-up to illustrate the Young's moduli.

Table 3 Overview of the degradation temperature at 2% mass loss (T_d), crystallization temperature (T_c), crystallization enthalpy (ΔH_c), melting temperature (T_m) (based on second heating cycle), melting enthalpy (ΔH_m) (based on second heating cycle), crystallinity, Young's modulus, elongation at break and ultimate strength, obtained from telechelic, tri-functional and tetra-functional E-PCL networks crosslinked with PETA-4SH and PETA-3SH

	T d, 98% (°C)	T c (°C)	ΔHc (J g^−1)	T m (°C)	ΔHm (J g^−1)	Crystallinity Xc (%)	Young's modulus (MPa)	Elongation at break (%)	Ultimate strength (MPa)
E-PCL(2)-4SH	266	27.2	56.1	45.4	66.9	49	209.7 ± 34.8	545 ± 92	21.5 ± 1.6
E-PCL(3)-4SH	237	11.4	44.8	34.4	53.5	39	121.7 ± 11.6	348 ± 33	15.4 ± 1.7
E-PCL(4)-4SH	230	1.0	38.0	24.5	41.1	30	93.3 ± 8.8	222 ± 19	12.2 ± 0.8
E-PCL(2)-3SH	238	27.9	56.9	46.1	67.6	50	308.5 ± 6.1	693 ± 99	25.2 ± 3.3
E-PCL(3)-3SH	226	14.8	47.5	36.5	54.3	40	148.8 ± 7.3	445 ± 33	17.2 ± 1.3
E-PCL(4)-3SH	234	4.2	40.0	28.9	42.1	31	105.6 ± 7.3	263 ± 29	12.7 ± 1.0

---

Mechanical characterization of thiol–ene photo-crosslinked PCL-based networks

In order to elucidate the influence of the network architecture on the mechanical properties, tensile testing was performed on each network (Fig. 3D and E). The Young's modulus, elongation at break, and ultimate strength of the designed networks were determined (Table 3). The young's modulus values demonstrated an increasing trend upon decreasing precursor and crosslinker functionality. Interestingly, this trend of the Young's moduli is the opposite of those observed for the storage moduli (G′) obtained via photo-rheology. This can be explained by the fact that the networks were still amorphous during photo-rheological measurements, as they were performed at temperatures above their melting points (i.e., 60 °C). Hence, it is anticipated that the Young's modulus is governed by the crystallinity, given the excellent correlation between both (vide supra). Furthermore, when considering the elongation at break of the networks, a decreasing trend could be observed, when the M_c values of the networks reduced. The strong dependence of the elongation at break on the M_c value is attributed to the shorter inter-crosslink chains, restricting the orientation of the stretched network and decreasing the total elongation at break. Subsequently, the ultimate strengths were obtained for each network as well, demonstrating an increasing ultimate strength upon decreasing E-PCL functionality. The same trend could be observed with decreasing functionality of the thiolated crosslinker. Hence, indicating that the ultimate strength is lower for networks that have an increased crosslink density. Although counterintuitive, the latter can be explained by the higher crystallinity of the more loosely crosslinked networks.

When comparing the thiol–ene photo-crosslinked PCL networks reported herein with previous research on (bifunctional) acrylate-crosslinked PCL networks, it is clear that superior mechanical properties are obtained in the present work. Indeed, acrylate-crosslinked PCLs, based on oligomers with comparable molar mass, have been characterized by a Young's modulus below 15.4 MPa, an elongation at break below 78.5% and an ultimate strength below 2.55 MPa, indicating brittle constructs with poor toughness.^14,15 Herein, superior mechanical properties could be obtained for the thiol–ene crosslinked networks, overcoming the aforementioned brittleness.^15,18 Furthermore, upon alteration of the network topology, the described E-PCL networks could be fine-tuned according to a broad range of mechanical properties.

Accelerated degradation of the thiol–ene photo-crosslinked PCL networks

In order to assess the influence of network topology on the degradation time, accelerated degradation tests were performed in 5 M NaOH. The remaining mass of each network is shown in Fig. 3B and C as a function of time. In contrast to what was anticipated, the data showed a gradual increase of the degradation rate upon increasing crosslink density. Upon crosslinking, thioether functionalities are generated, situated next to an ester group. Previous research has shown that a lower number of methylene groups between the thioether and ester functionality, render the ester linkage more susceptible towards hydrolysis as a result of the more positively charged carbonyl carbon.³⁵ Since only two carbon atoms are spaced between the thioether and ester groups, accelerated degradation could be expected. Therefore, more thioether-ester entities could explain the faster degradation observed herein. On the other hand, networks with a lower degree of crystallinity can be more readily infiltrated by water, increasing the degradation rate as well.^36,37 Thus, these results evidence the tunability of the degradation time, depending on the functionality of the E-PCL precursor as well as the thiolated crosslinker.

3D-printing of thiol–ene crosslinkable PCL via digital light processing

In the field of tissue engineering, 3D-printing has gained increasing attention, because of its ability to create (high-resolution) patient-specific implants. Indeed, patient-specific implants with controlled porosity are crucial to ensure an excellent fit with the tissue defect while ensuring the best possible aesthetic outcome.^38,39 Over the past decades, the application of light-mediated 3D-printing techniques, such as digital light processing (DLP), to manufacture complex tissue-mimicking constructs with high precision in a time-efficient manner, has seen an enormous upsurge. DLP exploits a digital mirror device (DMD) and relies on spatially controlled layer-by-layer illumination to cure a photo-crosslinkable resin. The resulting Computer-Aided Design/Computer-Aided Manufacturing (CAD/CAM) mimicry is significantly improved as compared to deposition-based 3D-printing (e.g., extrusion).^{10,19,40–43} DLP is of particular interest since each layer is printed simultaneously, thereby strongly reducing the manufacturing time and as a consequence, increasing the cost efficiency of the technique. In DLP, the resolution depends on the material properties such as the cure depth and photo-crosslinking kinetics and can be as low as 25 μm.^44–47 Based on photo-reactivity, the developed E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH mixtures were selected and evaluated as photo-crosslinkable resins, for processing through DLP. Notably, resins based on PETA-3SH were not considered herein, given the considerably inferior photo-crosslinking kinetics and subpar network formation as evidenced by the gelation points and mismatch between the experimental and theoretical M_c values. Furthermore, N-methylpyrrolidon (NMP) was selected as a solvent for the photo-crosslinkable resins, due to its high boiling point (i.e., 202 °C), preventing evaporation during printing.^10,48 Furthermore, the viscosity of the resin should be well below 10 Pa s to enable efficient use for DLP. Hence, concentrations of 40, 50 and 50 w/w% E-PCL in NMP were used, exhibiting a viscosity of 1.40, 0.89 and 0.36 Pa s at room temperature, for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH, respectively (Fig. S3†). The concentration of the E-PCL(2)-4SH resin was selected to be lower (i.e., 40 w/w%), due to its faster crystallization. Moreover, to successfully print the selected materials, the photo-initiator (PI) (i.e., TPO-L) and photo-absorber (i.e., quinoline yellow (QY)) concentrations were varied from 1 to 3 wt% and from 0.01 to 0.05 wt%, respectively.¹⁰ An iterative series of experiments was conducted to identify the optimal concentrations. In addition, various intensities (i.e., 7.26–29.52 mW cm⁻²) and irradiation times (i.e., 1–15 s) were considered for each resin as well, in order to determine the optimal printing parameters, resulting in the highest resolution and CAD-CAM mimicry. In a first step, the X–Y resolution was considered and assessed through iterative printing based on a CAD design of a 0.5 cm³ cube. Subsequently, in order to optimize the Z resolution, the penetration depth of the curing light (D_p) and minimal energy required for polymerization (E_c) were determined, by means of the respective working curves for each E-PCL-based resin. Jacob's working curve is defined as the curing thickness (C_d) as a function of the light dose (E) (eqn (8): Experimental section).¹⁰ Through extrapolation of the working curves, D_p and E_c could be identified (Fig. 4B). The critical light dose (E_c) decreased from 230 to 149 mJ cm⁻², when the functionality of the PCL precursor was changed from 2 to 4. A similar trend was already observed through photo-rheology analysis. Indeed, the critical light dose, based on the obtained gelation points, decreased from 649 to 413 mJ cm⁻². The difference in absolute E_c value could be attributed to the different wavelength applied for photo-rheology measurements (i.e., 365 nm), compared to that of the DLP printer (i.e., 405 nm). In addition, it should be noted that photo-rheology was conducted from the melt, while the DLP process proceeded in solution. Furthermore, penetration depths (D_p) of 281.99, 371.18 and 375.42 μm were obtained for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH resins, respectively. It should be noted that the D_p is slightly lower for telechelic E-PCL. This can be attributed to the slightly lower absolute concentration of PI in this resin, due to the fact that a lower oligomer concentration was employed while the PI to oligomer ratio was kept constant. Ultimately, the printed cubes with the best CAD-CAM mimicry were obtained for photo-crosslinkable resins, containing 2.5 wt% TPO-L and 0.0375 wt% QY (according to E-PCL precursor), utilizing an intensity of 29.52 mW cm⁻² and irradiation times of 9.5, 7.5 and 6.0 s for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH resins, respectively. Furthermore, these results indicate that multi-functional E-PCL materials hold the advantage of achieving faster photo-crosslinking kinetics and thus, decreasing the printing time. In addition, said multi-functional E-PCL precursors experience a decreased crystallization rate (vide supra), thereby extending the printable window for these formulations and allowing for E-PCL resins with higher concentrations as well.

Fig. 4 (A) Stress–strain curves, based on tensile test results obtained from DLP-printed, telechelic, tri-functional and tetra-functional E-PCL networks crosslinked with PETA-4SH. A zoom is provided to illustrate the Young's moduli; (B) working curves for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH photo-crosslinkable resins, obtained through DLP with illumination at 405 nm (29.52 mW cm⁻²), determining the critical curing dose (E_c) with high R²-values; (C) the CAD-design and pictures from DLP-printed gyroid structures, based on E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH formulations. An angled-, top- and side-view are shown for each gyroid.

In a final part, more complex porous gyroid scaffolds (1.0 × 1.0 × 0.5 cm³) were 3D-printed. Gyroid structures were selected since their morphology is constituting quadruple junction points with high porosity, which are crucial to enable cell ingrowth, vascularization as well as nutrient and waste diffusion.⁴⁹ Said gyroid structures were successfully printed using light doses of 280, 221 and 177 mJ cm⁻² for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH resins, respectively (Fig. 4C). The excellent agreement with the critical light dose (E_c) obtained via the working curves should be noted in this regard. However, it is important to use slightly higher doses as suggested by the working curve, in order to achieve sufficient inter-layer adhesion. In conclusion, the developed E-PCL oligomers could be effectively exploited to 3D-print complex geometries by DLP. Interestingly, a clear influence of the polymer architecture on the DLP printability was also noted. More specifically, higher E_c values were necessary for E-PCL precursors with a lower functionality, extending the printing process.

Furthermore, the mechanical properties of DLP-printed structures were evaluated as well and compared to those of the earlier described 2D-films (vide supra). Stress–strain curves were obtained using DLP-printed dog bones (0.5 mm thickness), based on previously stated photo-crosslinkable resins (i.e., E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH). As shown in Table 4 and Fig. 4A, the Young's modulus, total elongation at break and ultimate strength values exhibit the same relative trends as described earlier for the thin casted films. However, their absolute values decreased. These results are in line with previous reports in literature. More specifically, Thijssen et al. described that the observed reduction of mechanical properties could be attributed to the reduced polymer concentration which was used for DLP-printing due to viscosity restrictions, compared to when photo-crosslinking from melt. In addition, a decrease of crystallinity and the presence of solvent traces might as well have decreased stated mechanical properties.¹⁰ Indeed, a reduction of crystallinity could be observed for the DLP-printed structures (Table 4). However, it is hypothesized that controlled recrystallization after solvent removal from the DLP-printed dog bones might enable to overcome the observed discrepancy.

Table 4 Overview of the Young's modulus, elongation at break, ultimate strength, crystallization enthalpy (ΔH_c), melting enthalpy (ΔH_m) (based on second heating cycle) and crystallinity (%), obtained for DLP-printed, telechelic, tri-functional and tetra-functional E-PCL networks crosslinked with PETA-4SH

	Young's modulus (MPa)	Elongation at break (%)	Ultimate strength (MPa)	ΔHc (J g^−1)	ΔHm (J g^−1)	Crystallinity Xc (%)
E-PCL(2)-4SH DLP	148.2 ± 24.4	483 ± 118	11.1 ± 2.8	49.0	55.3	40
E-PCL(3)-4SH DLP	77.8 ± 14.9	238 ± 68	6.9 ± 0.9	45.26	46.7	34
E-PCL(4)-4SH DLP	46.3 ± 17.7	76 ± 36	4.6 ± 1.4	40.3	34.8	25

---

Finally, the shelf life stability of the photo-crosslinkable E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH resins was assessed by monitoring the remaining alkene content over 7 days at different temperatures (i.e., 4, 20 and 40 °C), via¹H-NMR spectroscopy using DMT as internal standard. Results indicated that the formulations remained relatively stable up to three days (at 4 °C), with alkene contents of 99.6%, 98.9% and 91.8%, respectively (Table S3†). These reductions were considered acceptable, indicating no significant issue for the printing of larger structures. Additionally, storage of the formulations at lower temperatures has also shown to slightly extend the obtained shelf lives (Table S3†). While the formulations exhibit relative stability up to three days under appropriate storage conditions (i.e., low temperature, under argon and in the dark), forthcoming research will encompass the elaboration of strategies to improve resin stability to ensure extended shelf lives. One potential approach involves developing the complete mixture and incorporating the crosslinker (i.e., thiol) prior to the printing process. Another prospective strategy involves introducing a pot-life stabilizer (e.g., aluminum salt of N-nitroso-N-phenylhydroxylamine (ANPHA)^50,51) into the formulations, thereby augmenting their stability at elevated temperatures over time.

Conclusions and future perspectives

In the present work, thiol–ene chemistry was employed for the formation of photo-crosslinked PCL-based networks. Alkene-functionalized PCL precursors were synthesized and crosslinked using a thiolated crosslinker in the presence of a suitable photo-initiator. The influence of the network architecture on several physico-chemical properties (mechanical, thermal, photo-crosslinking kinetics and relative hydrolytic degradation profile) was herein thoroughly investigated through variation of the functionality of E-PCL precursors and thiolated crosslinkers. A significant impact of the network topology could be observed. More specifically, upon alteration of the network architecture, a range of mechanical properties (elongations at break ranging from 222 ± 19% up to 693 ± 99%, Young's moduli from 93.3 ± 8.8 MPa up to 308.5 ± 6.1 MPa and ultimate strengths from 12.2 ± 0.8 MPa up to 25.2 ± 3.3 MPa), a range of thermal properties (melting enthalpy from 41.1 J g⁻¹ up to 67.6 J g⁻¹), a range of photo-rheological properties (storage moduli from 58 ± 0.02 kPa up to 741 ± 0.02 kPa) and relative hydrolytic degradation profiles could be observed. These results evidence the tunability of each property influenced by the network topology. Depending on the required application, the properties could be fine-tuned accordingly. In addition, successful application in digital light processing for telechelic, tri- and tetra-functional E-PCL resins with PETA-4SH crosslinkers, has been illustrated as well. Furthermore, the use of multi-functional E-PCL precursors offers the opportunity towards post polymerization modification (PPM) strategies (e.g., functionalization with specific (biologically active) moieties onto the available ‘ene’ functional groups). Said moieties could potentially alter the physico-chemical and/or biological properties of the network.

Hence, this approach of altering the network topology through the functionality of both precursor and crosslinker is very encouraging in the context of tissue engineering applications in which mimicry of the physical and morphological properties of the targeted tissue is of utmost importance. In the future, our strategy will be further elaborated towards one specific tissue engineering application, i.e. hyaline cartilage.

Experimental section

Materials and methods

Materials. All chemicals were used as received, unless stated otherwise. ε-Caprolactone (>99%), supplied by Tokyo Chemical Industry (TCI), was dried over calcium hydride (CaH₂) and retrieved through vacuum distillation at 120 °C. Allyl isocyanate (98%), ethylene glycol (anhydrous, 99.8%), pentaerythritol (99%), 2-ethyl-2-(hydroxymethyl)-1,3-propanediol (>99%, for synthesis), Tin(II) 2-ethylhexanoate (92.5–100%), dimethyl terephthalate (99.93%), pentaerythritol tetrakis(3) (>95%), trimethylolpropane tris(3mercaptopropionate) (>95%) and NaOH (>97%) were supplied by Sigma-Aldrich (Diegem, Belgium). Ethyl(2,4,6-trimethylbenzoyl) phenyl phosphinate (Speedcure TPO-L, 94.5%) was obtained from Lambson Ltd HQ (West Yorkshire, UK). Toluene (>99%), chloroform (stabilised with amylene, >99%) and diethylether (stabilised with 5–7 ppm BHT, >99%) were supplied by Chem-lab NV (Zedelgem, Belgium). Toluene was dried over Na and kept dry with molecular sieves (4 Å). Deuterated chloroform (stabilised with silver foils +0.03% TMS, 99.8%) was purchased from Eurisotop.

Synthesis of E-PCL(2), E-PCL(3) and E-PCL(4) 8000 g mol⁻¹ precursors. Telechelic poly-ε-caprolactone diol (8000 g mol⁻¹) was synthesized using the following procedure. A Schlenk equipped with a magnetic stirrer was flame-dried prior to the start of the experiment. ε-Caprolactone (20 g, 0.175 mol, 1 eq., MM = 114.14 g mol⁻¹), Sn(Oct)₂ catalyst (0.1 g, 0.247 mmol, 0.5 wt% with respect to ε-caprolactone monomer, 405.122 g mol⁻¹), ethylene glycol initiator (0.14 mL, 2.5 mmol, 1 : 70 stoichiometric ratio initiator to monomer, 62.07 g mol⁻¹) and anhydrous toluene (67.96 mL, 4 mol L⁻¹, 92.14 g mol⁻¹) were added to the Schlenk while being exposed to argon atmosphere. Three freeze–pump–thaw cycles were performed by freezing the solution in liquid N₂. After freezing the solution, vacuum was applied, causing the headspace above the frozen solution to be removed. The reaction proceeded for 24 h at 100 °C while stirring under argon atmosphere. The reaction continued until ¹H-NMR spectroscopy enabled verification of complete conversion of ε-caprolactone into poly-ε-caprolactone. Subsequently, the obtained PCL diol was modified to photo-crosslinkable alkene-functionalized PCL (E-PCL). The exact molar mass (8000 g mol⁻¹) was calculated with ¹H-NMR spectroscopy and based on this value, an excess of 1.2 eq. allyl isocyanate (0.83 g, 4.97 mmol, 2 eq., MM = 83.09 g mol⁻¹) was added with respect to the hydroxyl groups. The solution was stirred for 1 h at 60 °C. ¹H-NMR spectroscopy was used to monitor the progress of the transition of the hydroxyl functionalities to alkene end groups. Ultimately, the E-PCL product was purified through precipitation in cold diethylether, while fast stirring and filtration. Tri-functional E-PCL was obtained similarly, with a different initiator. More specifically, 2-ethyl-2-(hydroxymethyl)-1,3-propaandiol (trimethylolpropane) initiator (0.341 g, 2.54 mmol, 1 : 69 stoichiometric ratio initiator to monomer, 134.17 g mol⁻¹) was added to the mixture. The exact molar mass (7900 g mol⁻¹) was calculated with ¹H-NMR spectroscopy and based on this value, an excess of 1.2 eq. allyl isocyanate (1.25 g, 7.55 mmol, 1.2 eq., MM = 83.09 g mol⁻¹) was added with respect to the hydroxyl groups. Tetra-functional E-PCL was obtained through a similar procedure as well. More specifically, pentaerythritol initiator (0.346 g, 2.534 mmol, 1 : 69 stoichiometric molar ratio initiator to monomer, 136.15 g mol⁻¹) was added to the mixture. The exact molar mass (7700 g mol⁻¹) was calculated using ¹H-NMR spectroscopy and based on this value, an excess of 1.2 eq. allyl isocyanate (1.25 g, 7.55 mmol, 2 eq., MM = 83.09 g mol⁻¹) was added with respect to the hydroxyl groups.

Photo-curing. A mixture of telechelic E-PCL (4 g, 8050 g mol⁻¹, 2.26 × 10⁻⁴ mol g⁻¹ alkene content), TPO-L as photo-initiator (4.5 mg, 0.26 wt% or 1.7 mol% according to E-PCL precursor) and thiolated PETA-4SH crosslinker (0.110 g, 0.226 mmol, 488.66 g mol⁻¹) was dissolved in chloroform to homogenize the solution. The PETA-crosslinker was added according to an equimolar thiol to alkene ratio. Subsequently, chloroform was evaporated completely and the remaining solution was poured between two glass plates separated by a silicone spacer (0.5 mm thickness). Finally, the mould was placed between UV lamps and was crosslinked into sheets upon UV-A irradiation (365 nm, 10 mW cm⁻²) for 30 minutes. Upon preparation of the sheets based on tri- and tetra-functional E-PCL, the amount of PETA-4SH crosslinker was altered according to the alkene content, ensuring an equimolar thiol-to-alkene ratio. Sheets with PETA-3SH (398.56 g mol⁻¹) were developed similarly.

Characterization

Determination of the molar mass. Employing ¹H-NMR spectroscopy, the molar mass of telechelic, tri-functional and tetra-functional E-PCL precursors was determined based on the end-standing protons next to the urethane functionalities (H_e at 3.78 ppm). First, the integration of these protons was set at a value of four, six and eight, respectively. Subsequently, the molar masses of the respective polymers were calculated based on the value of the H_e protons (4.1 ppm) in the repeating backbone units according to eqn (2) and (3) (Table S1†).

MM = # repeating units × MM_monomer + MM_I (3)

Determination of the alkene content. The alkene content (mol g⁻¹) was determined by adding DMT (10 mg, 0.05 mmol, 194.18 g mol⁻¹) as an internal standard to 10 mg E-PCL. Analysing the obtained spectra from ¹H-NMR spectroscopy in Mestrenova software, the alkene content could be determined as follows (Table S2†):

With ‘I’ referring to the integrated value of a specific signal. Both alkene proton signals I(δ5.2) and I(δ5.8) were compared to an integration value of 1000 for I(δ8.0) for the aromatic protons of DMT. ‘N’ represents the number of protons of the functional group, ‘m’ corresponds with the exact mass of the product (around 10 mg) and ‘MM’ represents the molar mass. Subsequently, based on the alkene content and previously determined molar mass of the synthesized E-PCLs, the degree of substitution could be determined as well (Table S2†).

Rubber elasticity theory. Exploiting the rubber elasticity theory, the average molar mass between crosslinks (M_c) could be calculated. According to the rubber elasticity theory, G′ and the crosslink density are correlated under the assumption that full conversion takes place. The calculated molar mass between the crosslinks (M_c) could be defined based on the crosslink functionality (f) and the storage modulus (G′), according to eqn (1).³⁰

With M_c referring to the molar mass between crosslinks, ρ the network density (1.13 g mL⁻¹), R the universal gas constant (8.314 J mol⁻¹ K⁻¹), T the temperature (333.15K), G′ the experimental storage modulus and f referring to the crosslink functionality.

¹H-NMR spectroscopy. ¹H-NMR spectroscopy was performed on a Bruker spectrometer (400 MHz). Deuterated chloroform (CDCl₃) was used as solvent with DMT as an internal standard to obtain quantitative results. The obtained spectra were analysed using Mestrenova software, under full automatic baseline correction (Whittmaker Smoother).

Gel permeation chromatography. GPC was performed on a Waters 2695 Alliance Separation Module and RI detector Waters 2414. Calibration occurred with Polystyrene (PS) standards. Solutions of 5 mg E-PCL and 1 mL HPLC graded chloroform were prepared and possible solid particles were filtered.

Fourier-transform infrared spectroscopy. FTIR spectra were recorded on a PerkinElmer Frontier FTIR spectrometer combined with a MK2 golden gate set-up equipped with a diamond crystal from Specac. The spectra were obtained at room temperature, recorded between 600 and 4000 cm⁻¹ at a resolution of 4 cm⁻¹.

Gel fractions and swelling degrees. These were determined by punching out 4 mm diameter discs in triplicate, which were immersed in chloroform (in excess, 5 mL) for 2 days at room temperature. The mass of the discs was determined in dry state before swelling (M_d1), in swollen state (M_s) and in dry state after swelling (M_d2). In order to determine the swelling ratio and gel fraction, the following equations were used:

Thermal characterization. The degradation temperature (T_d, at 2% mass loss) was determined through Thermogravimetric Analysis (TGA), using the TGA Q50 (TA instruments). A sample of approximately 10 mg was placed in a platinum pan under a nitrogen flow of 60 mL min⁻¹. The sample was heated from 35 °C to 600 °C at a rate of 10 °C min⁻¹. The Q series software was used to analyse the TGA thermograms. The melting temperature (T_m), crystallization temperature (T_c), melting enthalpy (ΔH_m), crystallization enthalpy (ΔH_c) and crystallinity (X_c) were determined via Differential Scanning Calorimetry (DSC). The measurements were performed using a DSC Q2000 (TA instruments) with RSC 500 cooler (Zellik, Belgium). 5 mg sample was measured in an aluminum Tzero pan under nitrogen flow, with an empty pan being used as reference. The sample was equilibrated at 20 °C. Then, during the first heating cycle the sample was heated at 10 °C min⁻¹ up to 100 °C. Subsequently, during the first cooling cycle the sample was cooled at 10 °C min⁻¹ down to −80 °C. Finally, during the second heating cycle the sample was heated at 10 °C min⁻¹ up to 100 °C. The melting point data were collected from the second heating cycle. The Q series software was used to analyse the DSC thermograms. The degree of crystallinity (X_c) was calculated by dividing the melting enthalpy (ΔH_m) with the enthalpy of fusion of completely crystalline PCL (139.3 J g⁻¹ according to literature).³⁶

Mechanical characterization. The total elongation at break, Young's modulus and ultimate strength were determined with tensile experiments. The measurements were performed using an electromechanical universal 5ST tensile machine (Tinius Olsen) with a load cell of 500 N. Dog bones were punched out from the crosslinked sheets in triplicate. A crosshead speed of 10 mm min⁻¹ and a preload tension of 0.1 N were applied, until fracture of the dog bones occurred.

Degradation assays. Relative degradation assays were obtained by punching out 3 mm discs from a crosslinked sheet, in triplicate and with a thickness of 0.5 mm. Every day, during twelve days, three discs were taken out from the 5 M NaOH solution, rinsed with water, dried and weighed.

Photorheology. Photo-crosslinking kinetics were recorded, using the Physica MCR 301 Rheometer (Anton Paar, Sint-Martens-Latem, Belgium) combined with a PP25 spindle (25 mm diameter) by applying UVA irradiation at a wavelength of 365 nm and an intensity of 10 mW cm⁻² (EXFO Novacure 2000 UV irradiation source), at 60 °C. A 1 mL solution of E-PCL precursor, TPO-L photo-initiator (0.26 wt% or 1.7 mol% according to E-PCL content) and thiolated PETA-crosslinker (equimolar thiol-to-alkene ratio) was prepared from melt. Approximately 300 μL of this solution was placed in the parallel plate set-up, between the supporting plate and the PP25 spindle, with the gap set to 150 μm. The measurements were performed at a strain of 0.1%, an oscillation frequency of 1 Hz and a normal force of 1 N. After 15 s, the UV-A light of the rheometer was switched on, exposing the sample to UV-A light during 550 s, after which the measurement was terminated.

Digital light processing. A LumenX digital light processing (DLP)-printer from Cellink (405 nm LED and DMD) has been used to print 3D-constructs. An intensity of 75% (29.52 mW cm⁻²) was applied. A solution of 40, 50 and 50 w/w% of the respective E-PCL(2), E-PCL(3) and E-PCL(4) precursors in N-methylpyrrolidon (NMP) was prepared, containing thiolated PETA-4SH crosslinker (according to a 1 : 1 thiol-to-alkene ratio), TPO-L photo-initiator (2.5 wt% or 16.3 mol% according to E-PCL content) and Quinoline Yellow photo-absorber (0.0375 wt% according to E-PCL content). The viscosity of each photo-crosslinkable resin was determined through rheology, using an Anton Paar Physica MCR 301 rheometer. A solution of 300 μL was placed in the parallel plate set-up, between the supporting plate and the PP25 spindle, with the gap set to 150 μm. The measurements were performed at a constant shear rate of 0.01 s⁻¹ and linearly increasing temperatures from 20 °C up to 50 °C. The working curves were obtained through the development of 1 cm² squares (50 μm thickness) on glass slides, using varying cure doses (75% intensity and irradiation times from 8 to 15 s), in triplicate. Subsequently, the cure depth (thickness) was determined through the use of an optical microscope (Zeiss Axioteck). The working curves are described as the relation between the applied dose (E) and cured thickness (C_d), according to eqn (8).

Gyroid structures were developed, with the use of a 1.0 × 1.0 × 0.5 cm³ CAD design. The layers (50 μm) were illuminated at an intensity of 29.52 mW cm⁻² for 9.5, 7.5 and 6.0 s, for telechelic, tri- and tetra-functional E-PCL-based photo-crosslinkable resins, respectively. In a final step, the printed 3D-constructs were washed and sonicated three times within acetone. In addition, dog bones were DLP-printed with the same printing parameters as described above (vide supra). The dog bones were printed with a thickness of 0.5 mm (50 μm per layer). Tensile testing of the DLP-printed dog bones was performed similar as for those obtained from the thin casted films. The measurements were performed using an electromechanical universal 5ST tensile machine (Tinius Olsen) with a load cell of 500 N. A crosshead speed of 10 mm min⁻¹ and a preload tension of 0.1 N were applied, until fracture of the dog bones occurred. Finally, stability tests were performed as well, for E-PCL(2)-4SH, E-PCL(3)-4SH and E-PCL(4)-4SH resins. The resins were prepared identically as described above and stored in the dark and under argon, at 4 °C, 20 °C and 40 °C in parallel. During seven days, the alkene content was determined through ¹H-NMR spectroscopy, using DMT as internal standard. 10 mg DMT was added to the resin (containing 10 mg E-PCL). The NMR spectra resulting from these mixtures were analysed as previously described and ‘ene’ quantification occurred through eqn (4).

Conflicts of interest

The authors declare no conflicts of interest.

Acknowledgements

The authors would like to acknowledge funding from FWO (FWO-SB fellow – 1SA2323N). The 400 MHz NMR, used in this work, is owned by the NMR Expertise Centre at UGent and was funded through an FFEU-ZWAP grand (FWO) awarded to Prof. Martins J. (UGent). The authors would like to acknowledge that high resolution pictures of the printed structures were taken and edited by Van Damme L., MD (UGent).

References

1. I. Manavitehrani, A. Fathi, H. Badr, S. Daly, A. N. Shirazi and F. Dehghani, Polymer, 2016, 8, 20.
2. M. M. Stevens, Mater. Today, 2008, 11, 18–25.
3. B. D. Ulery, L. S. Nair and C. T. Laurencin, J. Polym. Sci., Part B: Polym. Phys., 2011, 49, 832–864.
4. C. Z. Liu and J. T. Czernuszka, Mater. Sci. Technol., 2007, 23, 379–392.
5. M. A. Woodruff and D. W. Hutmacher, Prog. Polym. Sci., 2010, 35, 1217–1256.
6. I. Engelberg and J. Kohn, Biomaterials, 1991, 12, 292–304.
7. R. M. Mohamed and K. Yusoh, Adv. Mater. Res., 2016, 1134, 249–255.
8. N. Herrera, P. Olsén and L. A. Berglund, ACS Sustainable Chem. Eng., 2020, 8, 11977–11985.
9. S. F. Abdellah Ali, IOP Conf. Ser.: Mater. Sci. Eng., 2016, 137, 012035.
10. Q. Thijssen, L. Parmentier, E. Augustyniak, P. A. Mouthuy and S. Van Vlierberghe, Adv. Funct. Mater., 2022, 32, 2108869.
11. B. J. Green, K. S. Worthington, J. R. Thompson, S. J. Bunn, M. Rethwisch, E. E. Kaalberg, C. Jiao, L. A. Wiley, R. F. Mullins, E. M. Stone, E. H. Sohn, B. A. Tucker and C. A. Guymon, Biomacromolecules, 2018, 19, 3682–3692.
12. T. Kuhnt, R. M. García, S. Camarero-Espinosa, A. Dias, A. T. ten Cate, C. A. van Blitterswijk, L. Moroni and M. B. Baker, Biomater. Sci., 2019, 7, 4984.
13. J. Tzeng, Y. Hsiao, Y. Wu, H. Chen, S. Lee and Y. Lin, BioMed. Res. Int., 2018, 2018, 2314–6133.
14. L. Elomaa, S. Teixeira, R. Hakala, H. Korhonen, D. W. Grijpma and J. V. Seppälä, Acta Biomater., 2011, 7, 3850–3856.
15. O. D. Mcnair, B. J. Sparks, A. P. Janisse, D. P. Brent, D. L. Patton and D. A. Savin, Macromolecules, 2013, 46, 5614–5621.
16. J. Borrello, P. Nasser, J. C. Iatridis and K. D. Costa, Addit. Manuf., 2018, 23, 374–380.
17. H. Quan, T. Zhang, H. Xu, S. Luo, J. Nie and X. Zhu, Bioact. Mater., 2020, 5, 110–115.
18. M. J. Kade, D. J. Burke, C. J. Hawker and D. J. Burke, J. Polym. Sci., Part A: Polym. Chem., 2010, 48, 743–750.
19. C. Yu, J. Schimelman, P. Wang, K. L. Miller, X. Ma, S. You, J. Guan, B. Sun, W. Zhu and S. Chen, Chem. Rev., 2020, 120, 10695–10743.
20. C. E. Hoyle, T. A. I. Y. Lee and T. Roper, J. Polym. Sci., Part A: Polym. Chem., 2004, 42, 5301–5338.
21. C. E. Hoyle and C. N. Bowman, Angew. Chem., Int. Ed., 2010, 49, 1540–1573.
22. J. J. Karnes, T. H. Weisgraber, C. C. Cook, D. N. Wang, J. C. Crowhurst, C. A. Fox, B. S. Harris, J. S. Oakdale, R. Faller and M. Shusteff, Macromolecules, 2023, 56, 2225–2233.
23. Q. Thijssen, A. Quaak, J. Toombs, E. De Vlieghere, L. Parmentier, H. Taylor and S. Van Vlierberghe, Adv. Mater., 2023, 24, e2210136.
24. E. Hoch, C. Schuh, T. Hirth, G. E. M. Tovar and K. Borchers, J. Mater. Sci. Mater. Med., 2012, 23, 2607–2617.
25. J. Van Hoorick, P. Gruber, M. Markovic, M. Tromayer, J. Van Erps, H. Thienpont, R. Liska, A. Ovsianikov, P. Dubruel and S. Van Vlierberghe, Biomacromolecules, 2017, 18, 3260–3272.
26. N. Gökalp, C. U. Turan, Y. A. Guvenilir, N. Gokalp, C. Ulker and Y. Avcibasi Guvenilir, J. Polym. Mater., 2016, 33, 87–100.
27. M. Degirmenci, G. Hizal and Y. Yagci, Macromolecules, 2002, 35, 8265–8270.
28. P. J. Flory, J. Am. Chem. Soc., 1941, 63, 3083–3090.
29. A. Torres-Knoop, I. Kryven, V. Schamboeck and P. D. Iedema, Soft Matter, 2018, 14, 3404–3414.
30. S. S. Sheiko and A. V. Dobrynin, Macromolecules, 2019, 52, 7531–7546.
31. M. Quesada-Pérez, J. A. Maroto-Centeno, J. Forcada and R. Hidalgo-Alvarez, Soft Matter, 2011, 7, 10536–10547.
32. M. Sharifi, K. A. Ghorpade, V. I. Raman and G. R. Palmese, ACS Omega, 2020, 5, 31011–31018.
33. N. A. Neuburger and B. E. Eichinger, Macromolecules, 1988, 21, 3060–3070.
34. P. T. M. Albers, L. G. J. Van Der Ven, R. A. T. M. Van Benthem, A. C. C. Esteves and G. De With, Macromolecules, 2020, 53, 862–874.
35. R. G. Schoenmakers, P. Van De Wetering, D. L. Elbert and J. A. Hubbell, J. Controlled Release, 2004, 95, 291–300.
36. M. J. Jenkins and K. L. Harrison, Polym. Adv. Technol., 2008, 19, 1901–1906.
37. W. Limsukon, R. Auras and T. Smith, ACS Appl. Polym. Mater., 2021, 3, 5920–5931.
38. R. N. Maniar and T. Singhi, Curr. Rev. Musculoskelet. Med., 2014, 7, 125–130.
39. D. A. Zopf, C. L. Flanagan, A. G. Mitsak, J. R. Brennan and S. J. Hollister, Int. J. Pediatr. Otorhinolaryngol., 2018, 114, 170–174.
40. B. Grigoryan, D. W. Sazer, A. Avila, J. L. Albritton, A. Padhye, A. H. Ta, P. T. Greenfield, D. L. Gibbons and J. S. Miller, Sci. Rep., 2021, 11, 1–13.
41. Z. Wang, R. Abdulla, B. Parker, R. Samanipour, S. Ghosh and K. Kim, Biofabrication, 2015, 7, 045009.
42. J. Field, J. W. Haycock, F. M. Boissonade and F. Claeyssens, Molecules, 2021, 26, 1199.
43. M. T. Sultan, O. J. Lee, J. S. Lee and C. H. Park, Biomedicines, 2022, 10, 3224.
44. E. J. Carrasco-correa, E. Francisco, S.-A. J. M. Herrero-Martínez and M. Miro, TrAC, Trends Anal. Chem., 2021, 136, 116177.
45. G. Zhou, Q. Han, J. Tai, B. Liu, J. Zhang, K. Wang, X. Ni, S. Wang, X. Li and J. Zhang, J. Bioact. Compat. Polym., 2017, 32, 429–442.
46. S. L. Sherman, O. Kadioglu, G. F. Currier, J. P. Kierl and J. Li, Am. J. Orthod. Dentofacial Orthop., 2020, 157, 422–428.
47. A. Kessler, R. Hickel and M. Reymus, Oper. Dent., 2020, 45, 30–40.
48. J. Delaey, L. Parmentier, L. Pyl, J. Brancart, P. Adriaensens, A. Dobos, P. Dubruel and S. Van Vlierberghe, Macromol. Rapid Commun., 2023, 44, 2200955.
49. L. Germain, C. A. Fuentes, A. W. van Vuure, A. des Rieux and C. Dupont-Gillain, Mater. Des., 2018, 151, 113–122.
50. J. J. Schwartz, D. H. Porcincula, C. C. Cook, E. J. Fong and M. Shusteff, Polym. Chem., 2022, 13, 1813.
51. Z. Belbakra, Z. M. Cherkaoui and X. Allonas, Polym. Degrad. Stab., 2014, 110, 298–307.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3py00381g

---