Synthesis of the PEG-Based Hydrogels

The synthesis of polymer networks via step-growth reactions require the use of a multi-functional comonomer, bearing at least 3 functionalities (f>2), to create the intermolecular cross-links. However, in the acid-catalyzed thioacetalization mechanism of aldehydes, a dialdehyde comonomer can serve as a tetra-functional cross-linker, forming dithioacetal linkages upon reaction with a dithiol comonomer. Taking into consideration this reaction mechanism, we aimed to synthesize PEG-based networks with dithioacetal moieties at the cross-links. The hydroxyl end-groups of PEG mediate the introduction of diverse functionalities capable of cross-linking; therefore, PEG is a suitable candidate for hydrogel formation. Additionally, PEG-based hydrogels are particularly attractive for biomedical applications owing to the biocompatibility, hydrophilicity, and antifouling properties of the polymer.^[50]

Exploiting the advantages of PEG, dithiol terminated PEG macromonomers, PEG_1.5k-dithiol and PEG_4k-dithiol, were synthesized. The successful functionalization of the macromonomers was confirmed by ATR-FTIR and ¹H NMR spectroscopies (see characterization details in Supporting Information, Figures S1–S3). For the network formation, the PEG-dithiol macromonomer was combined with TPA, which served as the cross-linker and the light-absorbing unit, at a 2:1 PEG_n-dithiol:TPA mole ratio in THF, a good solvent for both comonomers. The step-growth reaction was initiated by the addition of a catalytic amount of sulfamic acid, which is a thermally stable, non-toxic compound that could be easily removed from the hydrogels by washing with water.^[44] The reaction mixture was first heated to 55 °C to remove the excess THF, and then to 80 °C. Gelation of the mixture was observed visually within 15–20 min, while after 2 h of heating, a transparent solid material was obtained (the cross-linked network). Networks prepared using dithiol-terminated PEG of M_n = 1500 g mol⁻¹ (M_w/M_n 1.04) and 4000 g mol⁻¹ (M_w/M_n 1.09), are denoted below as PEG_1.5k hydrogel and PEG_4k hydrogel, respectively. To confirm the formation of the dithioacetal linkages and therefore the proposed gelation mechanism, solid-state ¹H and ¹H-¹³C HSQC NMR spectra of the hydrogels were recorded (Figures 1a and S4–S6) (see characterization details in SI). The ATR-FTIR spectra of the PEG_4k and PEG_1.5k networks were also recorded (Figure S7, blue lines), however, the presence of the C-S vibration bands at 533 cm⁻¹ in the spectra of the PEG-dithiol macromonomers (black lines), prohibited the verification of the presence of the dithioacetal bonds in the polymer network spectra.

The swelling behavior of the prepared networks in water was examined at 20 and 37 °C (Figure 1b). The degrees of swelling of the hydrogels were found to depend on the M_n of the PEG and the solution temperature. The swelling ratio for the PEG_1.5k hydrogel was lower both at 20 and 37 °C, due to the shorter PEG chains and thus the higher cross-link density of the hydrogels. A decrease in swelling was observed at 37 °C compared to 20 °C, which was more pronounced for the longer PEG chains and attributed to the thermo-responsive behavior of PEG in water. The morphology of freeze-dried hydrogels was examined by SEM (Figure 1c,d). Interestingly, the PEG_4k hydrogel exhibited a highly porous structure, with pore sizes ranging from 5 to 40 µm (Figures 1c and S8). On the contrary, the PEG_1.5k hydrogel demonstrated a rather dense structure resembling a wrinkled film (Figure 1d). These results are in good agreement with the lower degree of swelling for the PEG_1.5k hydrogel. It is further noted that both hydrogels in their swollen state were soft and transparent (Figure 1e,f).

Rheological Characterization

The linear viscoelastic properties of the hydrogels were determined by monitoring the evolution of the storage (G’) and loss (G’’) moduli as a function of the applied stimulus. Dynamic frequency sweep (DFS) measurements were conducted in the linear viscoelastic regime at an angular frequency range 100–0.1 rad s⁻¹, at a constant stress amplitude of 2 Pa, and at RT (298K). To validate the measurements, the DFS experiments were repeated three times for each type of hydrogel using three different batches (Figure S9). The PEG_4k hydrogels exhibited a G’ plateau value of about 2.5 ± 0.38 kPa (Figure S9b), while for the PEG_1.5k hydrogels, the DFS experiments showed a higher G’ of about 6.8 ± 2.1 kPa (Figure S9b), which correlated with the higher cross-link density of the latter sample. The measured G’ values for both samples are in good agreement with those reported earlier for similar PEG-based hydrogels.^{[36, 37, 51, 52]} The small differences of the G’’ values were attributed to slight differences in the thickness of the hydrogels. The cross-link densities of the hydrogels were calculated using the Flory–Rehner and the rubber elasticity theory^[53] for both ideal and non-ideal networks, and the calculated values are summarized in Table 1. The cross-link density refers to the number of cross-links per unit volume in a polymer network. As expected, the cross-link density of the PEG_4k network was found to be approximately 3.5 times lower than the PEG_1.5k network.

The photodegradation of the hydrogels was monitored by shear rheology using a quartz lower plate, which allowed the irradiation of the sample with UV light (λ = 254 nm, 0.063 mW cm⁻¹), while the variation of G’ was monitored as a function of the irradiation time. It is worth mentioning that the photodegradation of the hydrogels is a process that occurs mainly on the surface of the hydrogel by photolyzing layer-by-layer the photosensitive moieties. So, the thickness of the gel, as well as the density of the photoactive sites, play a key role in the photodegradation rate of the system.^[54] To study the viscoelastic changes of the hydrogels upon UV exposure, first, a PEG_4k hydrogel with a thickness 350 ± 8.7 µm was placed between the plates of the rheometer and irradiated. Dynamic time sweep (DTS) experiments were conducted at a constant frequency of 10 rad s⁻¹ and stress amplitude of 2 Pa, revealing a fast decay in G’ within the first 10 min of exposure, denoting the softening of the hydrogel because of the photocleavage of the labile dithioacetal bonds (Figure 2a, black squares). Indeed, G’ decreased from ∼2.5 kPa to ∼0.2 kPa after 10 min of irradiation, and then reached 0.01 kPa after 30 min of exposure. Interestingly, after about 40 min of irradiation, a crossover between G’ and G’’ was observed, indicating the gel-to-sol transition, and the disruption of the coherence of the network structure. To test the effect of hydrogel thickness, a PEG_4k hydrogel with nearly half thickness ∼150 µm was prepared and irradiated under the same conditions. The DTS experiment revealed a faster gel-to-sol transition compared to the thicker (∼350 µm) hydrogel with the crossover at only ∼7 min of exposure (Figure 2a, blue rhombus). In the following, all measurements were carried out with a constant sample thickness of about 350 ± 8.7 µm. Similar changes in the shear modulus curve were observed for the PEG_1.5k network upon photoirradiation (Figure S10). However, the irradiation time required for the disintegration of this hydrogel was substantially longer (∼8 h). This observation was associated with three critical considerations. The first is the higher cross-link density of the PEG_1.5k hydrogel, which required a larger number of photons to photolyze the labile bonds.

The second, is related to the partial absorption of the incident light by the aromatic photo-products which remain in the sample holder,^[55] and the third is associated to smaller conformational entropy of the shorter chain, as proposed by Kowolik et al.^[56] As shown in Figure S10, even after 8 h of irradiation, the crossover of the storage modulus G‘ and loss modulus G’ was not observed, signifying an incomplete transformation of the gel to the sol state, which was verified visually by the presence of a thin hydrogel layer upon lifting the rheometer's top plate.

Compared to hydrolytic or enzymatic cleavage, which are the main mechanisms employed for hydrogel degradation, photodegradation allows external spatiotemporal control of the applied light source, providing greater control over the hydrogel properties.^[57-59] As a consequence, photodegradable hydrogels can be advantageous for various applications, including photoremovable wound dressings, photopatterning, regulation of the cellular functions (migration, proliferation, and differentiation), and controlled release of encapsulated cells or therapeutic agents.^{[16, 26, 37]} We investigated the temporally controlled photodegradation of the PEG_4k hydrogel (Figure 2b). Two irradiation cycles of 20 min duration each were applied to the hydrogel under shear. A rapid decrease in G’ from ∼2.5 to ∼0.16 kPa was found during the first irradiation interval (UV ON region), followed by a constant value of G’ for a period of 80 min when the irradiation ceased (UV OFF region). Upon switching on again the UV light (second UV ON region), the degradation recommenced and a fast drop in G’ to 0.025 kPa was observed, which again remained constant, when the light source was switched off (second UV OFF region). These results confirmed the ability to temporally control the cross-link density, hence the viscoelasticity of the hydrogel, and clearly show that the photolabile dithioacetal linkages within the polymer network can be easily cleaved upon light irradiation.

Owning to the high reactivity of thiols towards various functional groups (e.g., carbonyl, vinyl) under mild conditions, as well as their ability to undergo fast exchange reactions (dynamic exchange), we investigated the ability of the polymer networks to reform following photodegradation. Since the formation of the dithioacetal bonds required an elevated temperature, the reformation of the hydrogels after photocleavage was examined at 80 °C. The intrinsically low storage modulus of the PEG_4k hydrogel in combination with its fast photodegradation rate and transformation into a sol, rendered this sample suitable to study the hydrogel reformation process and the recovery of its mechanical properties. Rheological analysis using DTS experiments was performed at 10 rad s⁻¹ and 2 Pa on the PEG_4k hydrogel with a thickness 350 ± 8.7 µm. The hydrogel was first irradiated with UV light for 70 min, when G’ gradually decreased from ∼2.2 to ∼0.07 kPa, and the crossover of G’ and G’’ was observed after ∼40 min of irradiation (Figure 2c). Then, the UV lamp was switched off, and the temperature in the rheometer was set to 80 °C. As shown in Figure 2c, the elastic modulus increased from ∼0.03 to ∼1 kPa within a period of ∼2.5 h and then it remained constant for ∼2 h. The increase of G’ and the moduli crossover indicated the sol-to-gel transition of the material and the formation of a polymer network with mechanical properties similar to those of the initial network, before photodegradation. Interestingly, upon completion of the test the reformed hydrogel appeared transparent, malleable and soft (Figure 2c, inset), similar to the original hydrogel, verifying the reformation of the network upon heating after photolysis. Based on these results, we suggest that the dithioacetal-based networks can easily photodegrade to a viscoelastic liquid state, while the degradation products can react again upon heating to reform the polymeric network. The photodegradation and the subsequent heating processes of the highly cross-linked PEG_1.5k hydrogel revealed similar changes of the G’, G’’ curves to those found for the PEG_4k hydrogel. However, as mentioned above, an extended irradiation time ∼8 h was necessary for photodegradation (Figure S10). Moreover, while G’ decreased from ∼6.5 to 0.016 kPa upon light exposure, its recovery during heating was only partial and reached ∼0.82 kPa after 3 h, while it remained constant for the next 8 h.

The reversibility of the dithioacetal network formation was examined in a 2-cycle irradiation-heating DTS experiment on the PEG_4k hydrogel. Upon UV irradiation (30 min), G’ steeply decreased from ∼3.5 to ∼0.27 kPa and then increased to ∼3.0 kPa (85% recovery) with heating for ∼125 min (Figure 3d). Next, the sample was irradiated again for ∼10 min, which led to a rapid decay of G’ to ∼0.3 kPa, while upon heating (∼125 min), G’ reached a constant value of ∼2.4 kPa, signifying the reformation of the network (68% recovery) (Table S1). The slight decrease in modulus observed after each exposure cycle to the two stimuli was attributed to the formation of a small fraction of disulfide byproducts, which hindered the quantitative reformation of the polymer network (see “Insights into the photodegradation mechanism” section below). To prevent dehydration of the samples and minimize local fluctuations in water content during heating, the hydrogels were fully immersed in water. Any evaporated water was replenished in the rheometer bath surrounding the measuring fixture, ensuring uniform sample hydration.

The photodegradation kinetics of the system were determined by calculating the relaxation times (τ) from the normalized modulus (G’_t/G’₀) as a function of exposure time (Figure S11). The hydrogels exhibited two distinct τ values, τ_1–4k = 71.4 ± 6.5 s and τ_2–4k = 323.7 ± 15.5 s for the PEG_4k hydrogel, and τ_1–1.5k = 1554.4 ± 33.6 s and τ_2–1.5k = 32 956.9 ± 9357.8 s for the PEG_1.5k hydrogel (Table 1). These values signify that the photodegradation of the dithioacetal hydrogels is governed by two kinetic processes, one faster, which takes place during the initial exposure time, and a second at longer irradiation times. Dithioacetal bonds connected to stretched PEG strands require less energy to undergo photocleavage, hence these bonds are cleaved first, with photodegradation rate constants k_PD 14 × 10⁻³ s⁻¹ for the PEG_4k hydrogel and 6.4 × 10⁻⁴ s⁻¹ for the PEG_1.5k hydrogel, while the rest of the cross-links, which are connected to more relaxed polymer strands, are cleaved at a second stage with k_PD 3.08 × 10⁻³ s⁻¹ and 0.3 × 10⁻⁴ s⁻¹ for the PEG_4k hydrogels and PEG_1.5k hydrogels, respectively. The k_PD values clearly show that the degradation kinetics of the PEG_1.5k hydrogel is one order of magnitude slower compared to the PEG_4k hydrogel. Interestingly, the G’_t/G’₀ of the thinner PEG_4k hydrogel revealed one relaxation time (τ = 103.3 ± 3.1 s), signifying that the photodegradation process is governed by a single kinetic process. This result was attributed to the fewer structural inhomogeneities (e.g., loops) present in the thinner hydrogel sample, which resulted in fewer stretched PEG strands, and thus a single photo-degradation rate. It is noteworthy to mention that the dithioacetal networks presented herein undergo photocleavage with rates of the same order of magnitude as the extensively studied o-NB ester or coumarin-ester based photodegradable hydrogels.^{[27, 28, 36]} However, the photodegradation of a material is directly related to the intensity of the light source, which in our case was significantly lower (0.063 mW cm⁻¹) compared to that used to cleave the ester-based hydrogels (4–40 mW cm⁻²), revealing the high sensitivity of the S─O─S bond to light irradiation,^{[27, 31, 36]} and therefore rendering these photosensitive materials promising candidates for photopatterning applications, network recycling and drug delivery technologies.

Photo-Controlled Release

As indicated by the rheological measurements, the dithioacetal hydrogels can be effectively photodegraded at very low exposure intensities, whereas the photodegradation process can be remotely controlled by the use of the light source. These characteristics render our hydrogels particularly attractive for applications in 3D photopatterning, as smart photoremovable wound dressings or photodegradable and recyclable plastics. In all these applications, the encapsulation of active compounds such as antimicrobial agents, drugs or nanoparticles for catalysis (e.g., TiO₂ for water purification) is highly desirable. To study the loading and releasing capacity of the dithioacetal hydrogels, a hydrophobic dye (Sudan Red, SR) was incorporated as a model compound within the porous structure during hydrogel formation and its release profile was investigated under light irradiation. As shown in Figure 3a,b, the photocleavage of the dithioacetal cross-links, upon irradiation, resulted in the disruption of the porous hydrogel structure and the gradual release of the dye. For both samples, the released dye exhibited an almost linear dependance with exposure time up to 80% release. The release profile of the dye molecules for the PEG_4k hydrogel was significantly higher compared to the PEG_1.5k hydrogel due to the faster photodegradation of the network. Specifically, 44% of the dye was released after 20 min of exposure of the PEG_4k hydrogel (Figure 3c, red dots) while only ∼8% of the dye was released from the PEG_1.5k hydrogel at the same irradiation time (Figure 3c, black squares). Notably, in contrast to the rheological study, the photo-decomposition of the PEG_1.5k hydrogel was quantitative in this experiment due to the replacement of the aqueous medium by fresh water after each irradiation cycle, which also removed the aromatic photoproducts that absorb light and could possibly hinder the photodegradation process. It is important to note that the dye release rate from the hydrogels is influenced not only by the hydrogel's photodegradation rate but also by several other factors. These include the solubility of the dye molecules in water (with more hydrophilic molecules leaching out faster than less hydrophilic ones), the polymer-dye interactions, such as hydrogen bonding, π-π stacking, electrostatic forces, and others, and therefore a quantitative comparison of the photodegradation rate determined by rheology with the dye release rate would not be credible.

Insights into the Photodegradation Mechanism

To further elucidate the reversible viscoelastic properties of the dithioacetal hydrogels, insights into the photodegradation/reformation mechanism were gained by studying the response to light irradiation of a small dithioacetal analogue. A 4-arm star polymer comprising a monofunctional thiolated PEG of molecular weight M_n = 2000 g mol⁻¹ linked to the TPA molecule via dithioacetal bonds was synthesized and used for the photodegradation study (see supporting information and Figures S12–S14). The synthesis and the photodegradation kinetics of the star polymer were monitored by ¹H NMR spectroscopy. D₂O was selected as the solvent for the degradation study to provide similar environmental conditions to those used for the degradation of the swollen hydrogels. As shown in Figure S15, the signal at δ 5.22 ppm, attributed to the dithioacetal protons (He), decreased significantly after 20 min of exposure to light irradiation, and disappeared after 60 min of irradiation, indicating the photocleavage of the dithioacetal bonds. Concomitantly, the peak at δ 4.26 ppm, which corresponds to the methylene protons (Hb) next to the ester bonds of mPEG_2k, was gradually shifted to 4.30 ppm (Hb’) denoting the cleavage of the mPEG chains from the star structure and the formation of the thiolated mPEG. Moreover, the signals at δ 2.60 and 3.03 ppm, which correspond to the methylene protons of the mercaptopropionate group, Hd and Hc, respectively, gradually decreased and four new peaks emerged in the δ 2.60 to 3.03 ppm range, signifying the formation of two photoproducts, the mPEG-thiol macromonomer and a small fraction of mPEG-S-S-mPEG dimer (Figure S16, inset). The latter explains the partial recovery of networks’ mechanical properties after photodegradation, since the dimers consume the thiol end groups of mPEG, preventing their reaction with the TPA molecules. ¹H NMR spectroscopy revealed that the protons of the TPA molecule also underwent changes upon light irradiation. The peak at δ 7.53 ppm (Hg), attributed to the aromatic protons of the TPA ring, decreased with irradiation time, while several new peaks in the δ 7.4 and 8.5 ppm range appeared (Figures S15 and S16), denoting the cleavage of the dithioacetal bonds and the release of the aromatic molecules. However, due to the reduced solubility of the aromatic compounds in D₂O, the verification of their structure was not possible. To confirm the formation of the aromatic photoproducts, the photodegradation experiment was studied in CDCl₃ which solubilizes both comonomers (Figure 4). The decrease of the peak at δ 4.97 ppm attributed to the dithioacetal protons, confirmed the photocleavage of the labile cross-links. Irradiation induced the cleavage of the dithioacetal bonds and the release of the precursor mPEG_2k-thiol polymer, as well as the mPEG-S-S-PEGm dimers. The appearance of a peak at δ 10.15 ppm, corresponding to the aldehyde protons of TPA (H3), confirmed the formation of the initial TPA comonomer.

Based on the above results, the proposed mechanism of photodegradation and re-formation of the dithioacetal cross-links of the polymer networks is illustrated in Scheme 1. Upon irradiation at very low intensity, the aromatic moieties of the TPA comonomer absorb the incident photons and, via an energy transfer process, cleave the labile C─S bonds located next to the aromatic groups, which have a very low bond energy (272 kJ mol⁻¹), even lower than that of the C─O bonds (358 kJ mol⁻¹) of the photosensitive polyacetals studied earlier by us.^[60] Following the cleavage of the dithioacetal bonds in the presence of H₂O molecules, the initial TPA comonomer and the thiol-terminated PEG chains are obtained as the main photoproducts, justifying the thermo-reversibility of the network structure, as well as a small fraction of PEG-S-S-PEG dimers, which have a minimum effect on the network reformation, as revealed by the rheological data.