Tailoring the Shape-Memory Performance of 2D and 3D Fabricated Semi-Crystalline PCL Networks Via Optimal Crosslinking

Macromolecular Rapid Communications www.mrc-journal.de RESEARCH ARTICLE Tailoring the Shape-Memory Performance of 2D and 3D Fabricated Semi-Crystalline PCL Networks Via Optimal Crosslinking Lorenzo Bonetti1Daniele Natali2Stefano Pandini3Massimo Messori4Maurizio Toselli2 Giulia Scalet1 1Department of Civil Engineering and Architecture, University of Pavia, Pavia, Italy 2Department of Industrial Chemistry "Toso Montanari", University of Bologna, Bologna, Italy 3Department of Mechanical and Industrial Engineering, University of Brescia, Brescia, Italy 4Department of Applied Science and Technology, Politecnico Di Torino, Torino, Italy Correspondence: Lorenzo Bonetti ([email protected]) Received: 1 August 2025 Revised: 15 September 2025 Funding: This research was supported by the European Union ERC CoDe4Bio Grant ID 101039467. Keywords: 3D printing | PCL | photo-crosslinking | semi-crystalline polymer networks | shape-memory polymers ABSTRACT Photo-crosslinking is a fast and efficient approach to obtain chemically crosslinked semi-crystalline networks featuring both one-way and two-way shape-memory effect. However, the effect of photo-crosslinking parameters and fabrication method on the physical, thermo-mechanical, and shape-memory properties of these networks still has to be investigated. This paper aims to fill this gap, specifically focusing on semi-crystalline polycaprolactone (PCL) networks. In detail, the influence of key photocrosslinking parameters -crosslinking temperature and UV light intensityas well as the fabrication method -2D vs. 3Dwere investigated. As a general trend, crosslinking above the melting temperature of PCL and selecting a high UV light intensity yielded structures with superior performance, also displaying stress-free shape-memory behavior. Conversely, crosslinking below the crystallization temperature of PCL and selecting a low UV light intensity led to reduced performance and absence of stressfree actuation. To address this limitation, a post-treatment involving additional UV exposure was introduced, which significantly improved overall performance, particularly enhancing the two-way shape-memory behavior. Interestingly, although the 3D printed samples displayed thermal properties comparable to their 2D counterparts, their shape-memory performance was significantly reduced. Overall, these findings provide practical design guidelines for engineering 2D and 3D PCL-based semi-crystalline structures with tunable physical, thermal, and shape-memory properties. 1 Introduction Shape-memory polymers (SMPs) are a class of smart materials capable of fixing a deformed -temporaryshape and subsequently recovering their original -permanentshape upon exposure to an external stimulus, most commonly heat [1, 2]. Such a capability is commonly referred to as one-way shape-memory effect (SME). However, the reliance on external intervention to reprogram the temporary shape limits the application of these materials in dynamic environments. To address this limitation, two-way SMPs have been developed, which enable reversible switching between two different configurations upon heating and cooling cycles [3–6]. Among the possible classes of SMPs, semi-crystalline polymer networks have attracted growing interest due to their ability to This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Author(s). Macromolecular Rapid Communications published by Wiley-VCH GmbH Macromolecular Rapid Communications, 2025; 0:e00631 https://doi.org/10.1002/marc.202500631 1of14 combine reversible phase transitions with elastic network stability, resulting in efficient and controllable one-way and two-way shape-memory behaviors [7–10]. In this framework, polycaprolactone (PCL) has emerged as a particularly promising candidate for designing thermo-responsive SMPs. Its semi-crystalline nature, low melting temperature (≈50–60◦C), biodegradability, synthetic versatility, and the possibility to blend it with other polymers [11, 12], make it attractive for a wide range of applications, especially in the biomedical field [13–15]. In chemically crosslinked PCL-based networks, the crystalline regions act as thermal switches that fix the temporary shape upon cooling, while the chemical net points ensure effective recovery of the original shape. This dual-phase structure enables an effective SME [16–18]. Particularly, PCL-based semi-crystalline networks have been largely investigated for their thermally-triggered twoway SME under an applied external stress. The two-way SME in PCL was first reported by Hong et al. [19] shortly after the identification of a reversible behavior in a semi-crystalline polymer network based on crosslinked poly(cyclooctene) [20]. In particular, Hong and co-workers showed that a PCL-based shapememory polyurethane -composed of hard segments functioning as crosslinks and soft PCL segments containing both amorphous and crystallizable phasesexhibited noteworthy elongation under load. The mechanism was attributed to the alignment of soft-segment chains along the loading direction, followed by oriented crystallization during cooling under constant stress [19]. Interestingly, such a reversible behavior was soon proposed also in the absence of an external stress applied, when Lendlein’s group reported for the first time bidirectional SMPs capable of switching between two shapes, in stress-free conditions, under heating/cooling cycles [21]. This pioneering work laid the foundation for the development of a plethora of different twoway SMPs capable of self-actuation [4, 5, 22, 23] demonstrating excellent applicability in a wide range of research fields, including biomedical devices [13], soft actuators and robots [3], and smart textiles [24]. As it was possible to understand, the obtainment of semicrystalline networks based on PCL relies on crosslinking, which can be achieved using various techniques, including radiation [25–27], chemical methods [11, 16, 28], and photo-crosslinking [14, 29, 30]. In particular, photo-crosslinked PCL-MA networks are synthesized by modifying PCL chains with methacrylate (MA) groups (e.g., reaction with isocyanatoethyl methacrylate or glycidyl methacrylate) followed by UV-induced radical polymerization in the presence of a photo-initiator [14, 29, 30]. This latter strategy enables rapid and spatially-controlled crosslinking. Moreover, the integration of photo-crosslinkable PCL-MA with additive manufacturing technologies has advanced the field of 4D printing, where PCL has been investigated in different approaches, among which fused filament fabrication (FFF), digital light processing (DLP), and stereolithography (SLA) [13, 14, 31]. In this regard, it is well-known that the shape-memory behavior and the properties of photo-crosslinked PCL-MA networks are significantly influenced by the conditions under which photocrosslinking is performed. Parameters such as UV light intensity, environmental conditions (e.g., temperature), exposure time, photo-initiator type and concentration, can directly affect the crosslinking density, network homogeneity, and crystallinity of the final material, known to govern key shape-memory properties [14, 15]. Despite the critical role of these parameters, systematic studies evaluating the impact of photo-crosslinking conditions on the final shape-memory performance of the obtained structures cannot be found in the literature. Most existing works focus on demonstrating feasibility or showcasing applications, rather than providing a detailed analysis of how processing variables influence material behavior and, most importantly, of how these can be used for tailoring material properties required by the application at hand. In addition, to the best of these authors’ knowledge, no previous works have simultaneously investigated the effect of photo-crosslinking conditions and fabrication approaches. As a result, there is an important need for dedicated investigations that establish clear structure–property relationships in photocrosslinked PCL-MA systems, especially in the context of additive manufacturing where control over printing and curing conditions can vary widely. In this paper, we investigate the effect of two main photocrosslinking parameters, namely the crosslinking temperature and the UV light intensity on the behavior of photo-crosslinked PCL-MA networks. The photo-crosslinking time and the photoinitiator (type and concentration) are kept constant based on previous studies. In addition, we investigate and compare two different fabrication techniques, i.e., compression molding (used to obtain 2D samples) and 3D printing, to assess the influence of the fabrication technique on the performance of the obtained samples. A thorough physical, thermo-mechanical, and shapememory characterization of the specimens is carried out to unveil the effect of the crosslinking parameters and the fabrication technique on the behavior of the obtained samples. We also investigate the possibility to post-treat specimens obtained under low crosslinking conditions as a potential strategy to improve their performance. Overall, this study provides useful guidelines for the optimal design of structures made from PCL-based semi-crystalline networks with tailored physico-mechanical, thermal, and shapememory performance. 2 Experimental Section 2.1 Materials PCL diol (𝛼,ω-hydroxyl-terminated PCL, Mn≈10 kDa), hereafter denoted as PCL10, 2-isocyanatoethyl methacrylate (2-IEM, 98 %), tetrahydrofuran (THF) and 2-hydroxy-2-methylpropiophenone were purchased from Merck and used without any further purification step. 2.2 Synthesis of Methacrylated PCL The methacrylation reaction of the hydroxyl end groups of PCL10 was carried out following a previously established protocol [30]. Briefly, dried PCL10 and 2-IEM were placed into a glass flask at a molar ratio of 1:2.4 (2-IEM was added in a 20% stoichiometric excess relative to the hydroxyl groups of PCL10 to ensure full conversion). Methacrylation was performed at 100◦C 2of14 Macromolecular Rapid Communications,2025 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License for approximately 3–4 h under stirring in a nitrogen atmosphere. Progress of the reaction was monitored via FT-IR spectroscopy, and considered complete when the intensity ratio of the peaks at 2930 and 2275 cm−1stabilized. Lastly, unreacted 2-IEM was removed under dynamic vacuum, and FT-IR analysis confirmed its absence by the disappearance of the isocyanate absorption band at 2275 cm−1. The obtained methacrylated PCL10 (hereafter denoted as PCL10-MA) was kept in the dark at room temperature (RT) until use. Additional information regarding the synthesis and chemical characterization of the obtained materials was available in reference [30]. 2.3 Sample Preparation 2D and 3D specimens were prepared with two different techniques, i.e., compression molding and 3D printing. To achieve photo-crosslinking, PCL10-MA was melt-mixed (80◦C, 100 rpm, 30 min) with the radical photo-initiator (2-hydroxy-2methylpropiophenone, 0.5 wt. %), then cooled to RT and kept in the dark until use. In the following sections, TXL,t XL,andI XL denote the crosslinking temperature, crosslinking time, and UV light intensity, respectively. 2.3.1 2D Samples The obtained mix was melted and poured between two glass slides (75 x 25 mm) and Teflon spacers (thickness: 0.4 mm) and gently compressed. Photo-crosslinking was achieved by exposure to UV light (Hamamatsu LC8 spot light source) at λ=365 nm for 120 s (i.e., tXL). Two crosslinking parameters were carefully controlled: the temperature (TXL =20◦Cand60 ◦C),towhich the sample was equilibrated before UV exposure, and the UV light intensity (IXL =0.5 and 6 mW/cm2). Please note that the desired UV light intensity was set using a UV power meter (UV Light Checker C9386, Hamamatsu) with a glass slide interposed between the UV source and the meter, thereby accounting for the partial absorption of UV light by the glass. Free films with a thicknesses of about 0.4 mm were obtained by peeling them off from the glass slides. Rectangular specimens (20 x 5 x 0.4 mm, Figure 1) were then cut out from the cured samples and used for the subsequent characterization. Hereafter, specimens obtained via compression molding will be referred to as 2D. Some of the samples obtained at TXL =20◦CandI XL = 0.5 mW/cm2underwent post-treatment (PT) after the first crosslinking step by heating them to TXL =60◦C and irradiating them with high UV light intensity (IXL =6mW/cm 2)foran additional 120 s. Table 1provides details of the post-treatment process. 2.3.2 3D Samples Computer-aided design (CAD) models (30 x 5 x 0.41 mm) were drawn in Autodesk Fusion (v. 2601.1.29) and processed to obtain. stl files, then imported into DNA Studio software (v. 4) for printing. Printing was carried out following a previously FIGURE 1 Representative 2D and 3D fabricated samples used for characterization. TABLE 1 Scheme of the prepared samples. In the sample name, 2D or 3D indicates the fabrication technique (FT), IXL the UV light intensity, and TXL the crosslinking temperature. Accordingly, the label wasexpressedasfollows:FT_I XL_TXL. Sample name FT IXL (mW/cm2)T XL (◦C) tXL (s) 2D_0.5_20 2D 0.5 20 120 2D_6_20 6 20 120 2D_0.5_60 0.5 60 120 2D_6_60 6 60 120 2D_PTa0.5 +620+60 120 +120 3D_0.5_20 3D 0.5 20 120 3D_6_20 6 20 120 aPost-treatment (PT) was performed by heating the sample 2D_0.5_20 to TXL =60◦C and irradiating it with high UV light intensity (IXL =6mW/cm 2)for additional 120 s. established protocol [29]. A pneumatic bioprinter (Cellink BioX6) equipped with a thermoplastic printhead and 22 G (diameter = 0.41 mm) nozzle size was used. The prepared mix was loaded into the print cartridge and heated to 70◦C. Printing was achieved at a pressure of 15 kPa, speed of 10 mm/s, concentric infill (98% density), setting the print bed temperature at 20◦C(i.e.,T XL = 20◦C), and photo-crosslinking was achieved using the integrated UV module (λ=365 nm) of the 3D printer, irradiating the printed layer for 120 s (i.e., tXL) at different UV intensities (IXL =0.5 and 6 mW/cm2) after extrusion. The fabricated samples, consisting of single-layer structures with dimensions of 30 x 5 x 0.41 mm (Figure 1) were used for the subsequent characterization. Hereafter, specimens obtained via 3D printing will be referred to as 3D. For clarity, Table 1provides a summary of all the prepared samples. Macromolecular Rapid Communications,2025 3of14 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2.4 Characterization 2.4.1 Gel Fraction The gel fraction evaluation was carried out by weighting the samples after fabrication (w0), then placing them in THF at RT for 24 h using an immersion ratio of 0.5 g PCL to 15 mL THF. Following immersion, the swollen specimens were removed from the solvent and dried at RT for 24 h until a constant weight (wd) was reached, representing the residual weight after extraction. The gel content (Gf(%)) was calculated using the following equation (Equation (1)) [29]: 𝐺𝑓(%)=𝑤𝑑 𝑤0 ×100 (1) 2.4.2 Thermal Characterization The thermal properties of the samples were examined using Differential Scanning Calorimetry (DSC 250, TA Instruments), conducted over a temperature range of −20◦Cto100 ◦Cata heating/cooling rate of 10◦C/min (heating/cooling/heating scans were carried out). The melting temperature (Tm) and crystallization temperature (Tc) were identified at the peak maximum of the endothermic transition and at the peak minimum of the exothermic transition, respectively. The degree of crystallinity (𝜒𝑐(%)) was calculated using the following equation (Equation (2)) [29]: 𝜒𝑐(%)=Δ𝐻𝑚 Δ𝐻100 𝑚 ×100 (2) where Δ𝐻𝑚represents the melting enthalpy of the specimen in the second heating scan, Δ𝐻100 𝑚was the theoretical melting enthalpy for 100% crystalline PCL (139.5 J/g [32]). 2.4.3 Mechanical Characterization The mechanical behavior of the samples was evaluated using a Dynamic Mechanical Analyzer (DMA Q850, TA Instruments) configured for tensile testing [29]. Quasi-static tensile tests were conducted at 80◦C, i.e., above the melting temperature of PCL (T >Tm). The preload was set to 0.001 N, then a load ramp of1 N/min was applied up to 18 N. The Young’s modulus (E) was determined from the slope of the stress-strain curves in the initial linear range (ε=0-5%, R2>0.9) and the crosslinking density (ν) was calculated by applying the statistical Gaussian rubber theory following Equation (3)[16, 29]: 𝐸=3𝜈𝑅𝑇 (3) where E was the Young’s modulus, R was the universal gas constant, and T was the absolute temperature. 2.4.4 Shape-Memory Characterization The shape-memory behavior of the samples was examined by evaluating the shape-memory effect (SME), including both oneway (1W) response and two-way (2W) response, the latter assessed under either stress-driven or stress-free conditions [12, 29]. 2.4.4.1 One-Way Shape-Memory Effect. Before testing, the samples were first preconditioned at T =80◦C (i.e., above the melting temperature of PCL) for 10 min. The preload was set to 0.001 N, then a load ramp of 1 N/min was applied keeping the T =80◦C until a nominal strain εappl =20% was achieved. Subsequently, keeping the strain constant (εappl =20%), the samples were cooled down to −20◦C(T<Tc)at2 ◦C/min. Once the target temperature was reached, the load was removed (F = 0.001 N), and the specimens were heated up to 80◦C at a constant heating rate of 2◦C/min under quasi-stress-free conditions. The material’s ability to retain a temporary shape was quantified by calculating the strain fixity ratio (𝑅𝑓), as defined in the following equation (Equation (4)): 𝑅𝑓(%)=𝜀𝑢𝑛𝑙𝑜𝑎𝑑 𝜀𝑎𝑝𝑝𝑙 ×100 (4) where 𝜀𝑎𝑝𝑝𝑙 was the applied nominal strain and 𝜀𝑢𝑛𝑙𝑜𝑎𝑑 was the strain after load removal. The material’s ability to recover its original (permanent) shape following the quasi-stress-free heating step was assessed by calculating the strain recovery ratio (𝑅𝑟) according to the following equation (Equation (5)): 𝑅𝑟(%)= 𝜀𝑎𝑝𝑝𝑙 −𝜀𝑟𝑒𝑐 𝜀𝑎𝑝𝑝𝑙 ×100 (5) where 𝜀𝑎𝑝𝑝𝑙 was the applied nominal strain and 𝜀𝑟𝑒𝑐 was the residual strain measured after the heating ramp. 2.4.4.2 Stress-Driven Two-Way Shape-Memory Effect. Before testing, the samples were first preconditioned at T =80◦C (i.e., above the melting temperature of PCL) for 10 min. The preload was set to 0.001 N, then a load ramp of 1 N/min was applied keeping the T =80◦C until a nominal strain εappl =30% was achieved. (A higher strain value, εappl =30%, was applied in the test to generate higher stresses in the specimens, which was necessary to observe a significant actuation magnitude). Then, keeping the force fixed (corresponding to εappl =30%), a coolingheating cycle between −20 and 80◦C(2 ◦C/min rate) was carried out. To quantitatively characterize the two-way shape-memory behavior, three parameters were determined: the actuation magnitude (AM), the recovery magnitude (RM), and the the stress-driven reversible deformation (Δ𝜀𝑟𝑒𝑣(𝑠𝑡𝑟𝑒𝑠𝑠−𝑑𝑟𝑖𝑣𝑒𝑛)) as defined in Equations (6)–(8) respectively. 𝐴𝑀 (%)=𝜀𝑙𝑜𝑤 −𝜀𝑎𝑝𝑝𝑙 (6) 𝑅𝑀 (%)= 𝜀𝑙𝑜𝑤 −𝜀ℎ𝑖𝑔ℎ 𝜀𝑙𝑜𝑤 −𝜀𝑎𝑝𝑝𝑙 ×100 (7) Δ𝜀𝑟𝑒𝑣(𝑠𝑡𝑟𝑒𝑠𝑠−𝑑𝑟𝑖𝑣𝑒𝑛)(%)=𝜀𝑙𝑜𝑤 −𝜀ℎ𝑖𝑔ℎ (8) 4of14 Macromolecular Rapid Communications,2025 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License where 𝜀𝑎𝑝𝑝𝑙 represents the strain applied at 80◦C (30% nominal), 𝜀𝑙𝑜𝑤 the strain after cooling under load at −20◦C, and 𝜀ℎ𝑖𝑔ℎ the recovered strain at the end of the following heating under load at 80◦C. 2.4.4.3 Stress-Free Two-Way Shape-Memory Effect. Before testing, the samples were first preconditioned at T =80◦C (i.e., above the melting temperature of PCL) for 10 min. The preload was set to 0.001 N, then a load ramp of 1 N/min was applied keeping the T =80◦C until a nominal strain εappl =30% was achieved. (A higher strain value (εappl =30 vs. 20%) was applied in the test to generate higher stresses in the specimens, which was necessary to observe a significant reversible actuation.) Then, keeping the force fixed (corresponding to εappl =30%), a cooling ramp at 2◦C/min was applied down to −20◦C. Subsequently, the sample was unloaded (F =0.001 N) and heated under quasi-stress-free conditions (2◦C/min) up to a selected actuation temperature (Tact). Then thesamplewascooledto−20◦C(2 ◦C/min), and a second heating/cooling cycle, achieving a second Tact, was performed before the final heating to 80◦C. Please, note that preliminary tests were carried out to select the actuation temperatures (data not shown). These tests followed the same procedure as the stress-free two-way tests, except for the final heating step. Instead of heating up to the Tact, the specimens were heated up to 80◦C, undergoing full recovery. From each strain-temperature curve, in the recovery step, two Tact were selected after the beginning of the recovery of ε. To quantitatively characterize the stress-free two-way shapememory behavior, the stress-free reversible deformation (Δ𝜀𝑟𝑒𝑣) was calculated as defined in Equation (9): Δ𝜀𝑟𝑒𝑣(𝑠𝑡𝑟𝑒𝑠𝑠−𝑓𝑟𝑒𝑒)(%)=𝜀𝑚𝑎𝑥 −𝜀𝑚𝑖𝑛 (9) where 𝜀𝑚𝑎𝑥 and 𝜀𝑚𝑖𝑛 represent the maximum and minimum strains achieved during the reversible cooling/heating cycle. 3 Results and Discussion The aim of this work was to explore how selected crosslinking parameters and different fabrication techniques influence the thermo-mechanical, physical, and shape-memory properties of semi-crystalline PCL-based networks. In details, the explored crosslinking parameters were: i) the crosslinking temperature (TXL =20 or 60◦C), below the crystallization temperature (Tc)or above the melting temperature (Tm) of PCL, to explore the effect of the material’s state (in the presence or absence of a crystalline phase) during crosslinking. ii) The UV light intensity (IXL =0.5 or 6mW/cm 2), the two values being selected as corresponding to the lowest and highest intensity achievable in the operational range of the printing machine (Cellink BioX6). Crosslinking times were kept constant (tXL =120 s) based on a previous study [29]. Specimens fabricated via 2D or 3D technique were then compared, to investigate the effect of the printing process on the behavior of the obtained samples. Please note that a TXL =60◦C was not investigated with 3D printed specimens, as shape fidelity could not be maintained at this temperature. For further details, the reader is referred to [29]. 3.1 Physical and Thermal Properties of the Samples The obtained specimens were characterized in terms of gel fraction by immersion in THF. Thermal parameters were determined through DSC analysis. The resulting data are presented in Table 2 and Figure 2. As it is possible to observe by comparing the 2D specimens, the crosslinking parameters, i.e., TXL and IXL, strongly influenced the thermal and physical properties of the obtained samples. Crosslinking at TXL =20◦C led to high values of Tm,T c,and crystallinity content (χc). Instead, crosslinking at TXL =60◦C, when the sample is fully melted, led to a decrease in the Tmand Tc of the crosslinked specimen, as well as in their χc. In particular, the Tmwas reduced of about 10◦C(Figure2B), while the Tcof about 15◦C(Figure2A). Accordingly, a decrease in χcof about 8% was observed (Figure 2C) in the samples crosslinked at TXL =60◦C. The decrease of the degree of crystallinity can be ascribed to a hindered crystallization process, due to the restricted chain mobility in the presence of a larger amount of chemical net points. Moreover, the physical state of the sample -in the presence or absence of a crystalline phasepossibly influences the spatial distribution of crosslinking points within the material. The associated decrease in the melting/crystallization temperatures can be ascribed to stricter bonds between chains, resulting in a less perfected structure of the crystals [29, 33]. In terms of gel fraction, most of the specimens displayed high gel fraction values of ≈95% (Figure 2D). Such results are in line with previous works, where Gfvalues in the 90–94% range were found for UV-crosslinked PCL10-PEG semi-crystalline networks [12] and PCL10 crosslinked by sol–gel chemistry [16]. Conversely, sample 2D_0.5_20 displayed a lower Gfcompared with other specimens (≈80 vs. ≈95%) due to the lower crosslinking extent achieved in this specimen. The results thus suggest that by either increasing IXL or increasing TXL above melting temperature promotes an increase in the crosslinking efficiency. Interestingly, the impact of the post-treatment (PT) on the thermal properties of the specimens was found to be noteworthy. Specifically, this treatment led to an important decrease of Tm,T c, and χcvalues, that reached values closer to those of the specimens treated at TXL =60◦C. The fabrication technique (i.e., 2D vs. 3D) did not appear to have an impact on the thermal properties of the samples, but rather to influence the physical ones. In particular, the gel fraction values of the 3D printed specimens resulted significantly lower than those of 2D ones (Figure 2D). Similar observations were made in our previous work [29], where lower gel fraction values (≈85%) were obtained for 3D printed specimens, in line with the Gfof sample 3D_6_20. These outcomes can be ascribed to differences in the degree of crosslinking achieved during 3D printing, most Macromolecular Rapid Communications,2025 5of14 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TABLE 2 Thermal and physical properties of semi-crystalline PCL networks prepared. Sample name Tc(◦C) Tm(◦C) ΔHm(J/g) χc(%) Gf(%) 2D_0.5_20 28.9 ±0.3 52.5 ±0.0 51.5 ±0.0 36.9 ±0.0 80.9 ±0.8 2D_6_20 26.3 ±2.3 51.4 ±2.1 48.4 ±2.4 34.7 ±1.7 95.7 ±0.5 2D_0.5_60 12.5 ±0.4 42.2 ±0.1 40.5 ±0.0 29.1 ±0.0 95.3 ±0.4 2D_6_60 12.3 ±0.3 42.6 ±0.8 39.5 ±2.4 28.3 ±1.7 95.9 ±0.2 2D_PT 16.4 ±0.2 45.0 ±0.1 42.2 ±0.7 30.3 ±0.5 94.5 ±1.6 3D_0.5_20 26.1 ±0.1 52.7 ±0.2 51.4 ±0.4 36.9 ±0.3 67.5 ±3.2 3D_6_20 24.8 ±0.1 51.9 ±0.2 48.2 ±0.2 34.6 ±0.2 83.9 ±0.4 FIGURE 2 Thermal and physical properties of the semi-crystalline PCL networks prepared: (A) crystallization temperature, Tc;(B)melting temperature, Tm; (C) degree of crystallinity, χc; (D) gel fraction, Gf.Bluedots:T XL =20◦C, red squares: TXL =60◦C. Empty symbol: IXL =0.5 mW/cm2, full symbol: IXL =6 mW/cm2. Black triangle =PT. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.). likely arising from the inhibitory effect of oxygen on crosslinking. Oxygen present in the air has a well-documented inhibitory effect on the photopolymerization of acrylate and methacrylate groups by forming a superficial inhibited layer, usually referred to as the OIL (oxygen inhibition layer). At the surface exposed to air, oxygen diffuses easily and reacts with the radicals generated by light, forming less reactive peroxide radicals. This “consumes” the active radicals, slowing down or blocking the polymerization reaction [34]. Notably, this effect is more pronounced in 3D specimens, as 2D specimens are processed between two glass slides, which markedly limits OIL formation. 3.2 Mechanical Properties of the Samples Quasi-static tensile tests were carried out on the samples at 80◦C(T>Tm). Data obtained from these analyses are reported in Tables 2, 3,andFigure3. It is worth clarifying that the specimens did not break within the test window, but all reached TABLE 3 Mechanical properties derived from QS tensile tests on the semi-crystalline polymer networks prepared. Sample name E (MPa) Crosslinking density, ν (10−4mol cm−3) 2D_0.5_20 0.8 ±0.1 0.9 ±0.1 2D_6_20 2.0 ±0.6 2.3 ±0.7 2D_0.5_60 4.2 ±0.4 4.8 ±0.4 2D_6_60 4.6 ±0.2 5.3 ±0.2 2D_PT 3.7 ±0.3 4.2 ±0.3 3D_0.5_20 0.6 ±0.0 0.7 ±0.0 3D_6_20 1.3 ±0.0 1.5 ±0.0 the displacement limit of the machine (depending on their initial length). 6of14 Macromolecular Rapid Communications,2025 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License FIGURE 3 Mechanical characterization of semi-crystalline PCL networks prepared. (A) Representative σ/εcurves and (B) Young’s moduli obtained from QS tensile tests at T =80◦C. Blue dots: TXL =20◦C, red squares: TXL =60◦C. Empty symbol: IXL =0.5 mW/cm2, full symbol: IXL =6 mW/cm2. Black triangle =PT. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.). As expected, stress-strain curves (Figure 3A) were highly influenced by the crosslinking parameters. Considering 2D specimens, an increase in the Young’s modulus (Figure 3B) was observed by increasing the UV power intensity, particularly for specimens crosslinked at low temperatures (E =0.8 ±0.1 vs. 2.0 ±0.6 MPa for 2D_0.5_20 and 2D_6_20, respectively). The influence of the crosslinking temperature was even more noticeable, leading to a 5-fold increase in the E values comparing specimens irradiated at the same UV power intensity (E =0.8 ±0.1 vs. 4.2 ±0.4 MPa for 2D_0.5_20 and 2D_0.5_60, respectively). The E values obtained in this work are consistent with our previous findings on PCL10 semi-crystalline networks photo-crosslinked via UV light. Specifically, previous samples crosslinked at TXL =60◦CandI=6mW/cm 2exhibited E =4.55 ±0.16 MPa [35], matching that of the 2D_6_60 samples (E =4.6 ±0.2 MPa, Table 3). The same observations done for the Young’s modulus can be made for the crosslinking density (ν,Table3). Also in terms of mechanical performance, the impact of the posttreatment on the specimens was noteworthy. Specifically, this treatment resulted in a significant increase in the E values (E =3.7 ±0.3 for 2D_PT), approaching those observed under high crosslinking conditions. Accordingly, the crosslinking density increased, likely because the post-treatment targeted domains that were crystalline at T <Tcand therefore inaccessible to crosslinking [36]. Upon melting (T >Tm), these regions became amorphous and accessible to additional crosslinking. Interestingly, the fabrication technique was also disclosed here to influence the mechanical properties of the specimens, especially for samples crosslinked at high UV power intensity (6 mW/cm2). In fact, by comparing 2D_6_20 and 3D_6_20 specimens, a slight decrease in the Young’s modulus was observed (E =2.0 ±0.6 vs. 1.3 ±0.0 MPa for 2D_6_20 and 3D_6_20, respectively). Such observations can be related to inter-filament bonding issues. This latter issue has been highlighted in numerous studies unveiling how issues in mechanical continuity, particularly related to interfilament and inter-layer bonding, are intrinsic to extrusion-based 3D printing techniques. These issues stem mainly from the nature of the extrusion-based printing process, which can result in structural defects and anisotropic mechanical properties in printed parts [37, 38]. 3.3 Shape-Memory Properties of the Samples 3.3.1 One-Way SME The one-way shape-memory behavior of the samples was evaluated using a defined thermo-mechanical cycle via DMA (Figure 4A). The high Rfvalues (Rf≥95%, Figure 4B,Table4) obtained for all the tested specimens demonstrate that the temporary shape was effectively retained after unloading, indicating efficient crystallization during cooling. Similarly, high Rrvalue (Rr≥97%, Figure 4C,Table4) confirmed nearly complete recovery of the permanent shape upon reheating above Tm[39]. This analysis indicates that, although slightly, high TXL and IXL values positively influenced the one-way shape-memory performance. In particular, a high TXL led to a slight (≈1–2%) increase in the Rf(96.5 vs. 97.5% for 2D_6_20 and 2D_6_60, respectively) and Rr(97.6 vs. 99.1% for 2D_6_20 and 2D_6_60, respectively) compared to a low TXL. Interestingly, the fabrication technique was not found to have a significant influence on the one-way shape-memory behavior of the specimens. These results align well with our previous studies on PCL-based SMPs fabricated both through traditional fabrication techniques [16, 39] and additive manufacturing [29], both displaying Rfand Rrvalues ≥95%. Therefore, it can be concluded that the printing process does not impair the material’s excellent one-way shape-memory performance. 3.3.2 Stress-Driven Two-Way SME The two-way SME was then assessed to evaluate the potential of the obtained semi-crystalline PCL networks for reversible shape transformation. First, reversibility was assessed under mechanical load (i.e., stress-driven conditions) by applying a thermal cooling/heating cycle. Similar to the one-way SME, the two-way SME results from the interplay between the macromolecular architecture of the polymer network and its thermo-mechanical history [3, 29]. Macromolecular Rapid Communications,2025 7of14 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License FIGURE 4 One-way shape-memory effect of the semi-crystalline PCL networks prepared. (A) Representative one-way shape-memory test: (1) the specimen is stretched to εappl =20% at 80◦C, (2) cooled to −20◦C while maintaining constant strain (εappl =20%), (3)unloaded(F=0.001 N), and finally (4) reheated to 80◦C under quasi stress-free conditions (F =0.001 N). (B) Strain fixity ratio, Rf, and (C) strain recovery ratio, Rr, obtained from one-way shape-memory tests. Blue dots: TXL =20◦C, red squares: TXL =60◦C. Empty symbol: IXL =0.5 mW/cm2, full symbol: IXL =6 mW/cm2. Black triangle =PT. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.). TABLE 4 One-way shape-memory properties of the semicrystalline PCL networks prepared. Sample name εappl (%)aRf(%) Rr(%) 2D_0.5_20 20 95.1 96.7 2D_6_20 96.5 97.6 2D_0.5_60 97.6 99.7 2D_6_60 97.5 99.1 2D_PT 97.0 97.9 3D_0.5_20 20 95.2 97.5 3D_6_20 95.6 96.7 a=nominal strain. In particular, to achieve two-way SME in semi-crystalline polymer networks, both a crystallizable phase and physical or chemical net points must be present. Under a constant tensile load, the polymer network elongates upon cooling due to two distinct mechanisms, namely the entropy elasticity effect in the rubbery region and the crystallization-induced effect close to Tc [5]. Conversely, heating the material above its Tmcauses contraction, which is attributed to the melting of previously oriented crystalline domains. Therefore, it can be concluded that the twoway SME is based on a crystallization-induced elongation (CIE) process during cooling and on a melting-induced contraction (MIC) process during heating [3, 5, 40]. The stress-driven two-way SME of the samples was evaluated using a defined thermo-mechanical cycle via DMA (Figure 5A). This thermo-mechanical cycle allowed the evaluation of the reversible strain changes resulting from CIE during cooling and MIC during heating (Table 5). The actuation magnitude (AM) and recovery magnitude (RM) were calculated using Equations (6)and(7)(Figure5and Table 5). The AM values (Figure 5B) were strongly influenced by the crosslinking parameters, with the crosslinking temperatures playing the most critical role. Indeed, when the UV intensity was kept constant (IXL =0.5 mW/cm2), more than a 2-fold increase in the AM values was obtained at the two crosslinking temperatures (AM =28.7 and 12.3% for 2D_0.5_60 and 2D_0.5_20, respectively). Such an effect was less noticeable at higher UV light intensity (AM =28.1 and 23.8% for 2D_6_60 and 2D_6_20, respectively), possibly because, at high temperature, the crosslinking plateau is already reached, whereas, at low temperature, a higher UV intensity is required to achieve the same level of crosslinking. Consistent with this observation, the effect of the UV intensity on the extent of the AM was more noticeable at low TXL,where almost a 2-fold increase in the AM was obtained at the two UV light intensities explored (AM =23.8 and 12.3% for 2D_6_20 and 2D_0.5_20, respectively). It is important to evidence a correlation between the AM and the applied stress. Since the tested samples exhibited different crosslinking degrees, different values of stress were required to reach the target strain of 30% (Table 5). It is well known in the literature that a higher applied stress generally results in increased AM [22], a trend that is also reflected in the data presented in this work. However, this is likely not the only factor influencing the AM. In fact, when comparing samples 2D_6_60 and 2D_PT, which experienced similar stress levels, 8of14 Macromolecular Rapid Communications,2025 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License FIGURE 5 Two-way shape-memory effect under an applied load (i.e., stress-driven) of the semi-crystalline PCL networks prepared. (A) Representative two-way shape-memory test under load: (1) a controlled load ramp of 1 N/min is applied at 80◦C(i.e.,T>Tm) until a strain level of 30% is reached. Maintaining the force constant, corresponding to the strain obtained at the end of this step, a (2) cooling and (3) heating cycle between −20◦C and 80◦C is then executed at a controlled rate of 2◦C/min. (B) Actuation magnitude, AM, and (C) recovery magnitude, RM, obtained from twoway shape-memory tests under load. Blue dots: TXL =20◦C, red squares: TXL =60◦C. Empty symbol: IXL =0.5 mW/cm2, full symbol: IXL =6 mW/cm2. Black triangle =PT. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article). TABLE 5 Two-way shape-memory properties under an applied load (i.e., stress-driven) of the semi-crystalline PCL networks prepared. Sample name εappl (%)aσ(kPa) AM (%) 𝚫𝜺𝒓𝒆𝒗(𝒔𝒕𝒓𝒆𝒔𝒔−𝒅𝒓𝒊𝒗𝒆𝒏) (%) RM (%) TCIE (◦C)bTMIC (◦C)b 2D_0.5_20 30 284.4 12.3 10.5 86.0 44.6 65.6 2D_6_20 833.2 23.8 21.5 90.3 40.0 61.1 2D_0.5_60 1075.5 28.7 27.5 94.5 27.9 56.9 2D_6_60 1056.9 28.1 26.4 93.9 26.0 59.2 2D_PT 924.8 21.0 19.1 90.9 32.3 60.2 3D_0.5_20 30 246.2 10.5 8.0 76.1 43.6 59.1 3D_6_20 361.2 13.5 11.0 81.0 40.5 57.7 anominal strain; bTCIE was calculated from the onset point of the steepest elongation process in the εvs. T curve, while TMIC was calculated from the onset point of the steepest contraction process in the εvs. T curve. marked differences in AM were observed. This difference may be attributed to variations in the three-dimensional network architecture. Specifically, the different crosslinking conditions likely influenced the spatial distribution of the crosslinking points within the polymer network, ultimately impacting on the AM values. In terms of RM values (Figure 5C), less pronounced differences were observed among the 2D specimens, with RM values moving from 86% for less crosslinked samples (2D_0.5_20) to 93.9% for the more crosslinked ones (2D_6_60). These AM and RM values obtained for high crosslinking conditions (AM =28.1%, RM =93.9% for 2D_6_60) are consistent with earlier findings on PCL-based shape-memory polymers obtained by both sol-gel crosslinking (AM =26%, RM =94%, εappl =40%) [16] and UV crosslinking (AM =25%, RM =97%, εappl =20%) [35]. The post-treatment resulted in an overall enhancement of the two-way shape-memory performance of the specimens. Notably, it had a positive impact on the AM values, that reached values (AM =21.0% for 2D_PT) midway between the low (AM =12.3% for 2D_0.5_20) and high (AM =28.1% for 2D_6_60) crosslinking conditions. As already stated, these observations can be related to a combination of factors: i) the higher crosslinking degree achieved through the post-treatment process, which leads to increased applied stress (at 30% strain), and ii) a different spatial distribution of the crosslinking points. Macromolecular Rapid Communications,2025 9of14 15213927, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/marc.202500631 by CochraneItalia, Wiley Online Library on [20/12/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License