3.1. Microstructure of Highly Swollen Homogeneous PGMA Hydrogel
The polyglycerol-based homogeneous, i.e., non-macroporous methacrylate hydrogel was prepared by crosslinking polymerization in bulk while macroscopically clear transparent resin not revealing any microscopic structure by cryo-SEM (
Figure 1a,b) was formed. The gel transparency was maintained in its equilibrium swollen state in water at which the gel contained 2.8 g of water per gram of dry resin. Under closer inspection with the naked eye against a dark background, the water-swollen specimen revealed a very slight hint of scattering of light that disappeared when the swollen gel was even more expanded by swelling in DMSO to ESC = 5.6 g (
Figure 1c,d, respectively).
Cryo-SEM was our first choice to check the water-swollen G0/0.3 sample structure; the sample preparation for cryo-SEM involved sample plunge-freezing in LN
2 and no subsequent freeze-drying. Surprisingly, the cryo-SEM micrographs revealed severe heterogeneity in the form of cellular pores of 5–40 μm distributed over the entire cross-cut sample surface, see images in
Figure 1e,f. Similar heterogeneity but with somewhat larger pores and wider distribution of pore sizes than was showed on cryo-SEM images, was revealed by the HVSEM method where samples had to be Pt-coated after being frozen in the refrigerator and freeze-dried, cf. images g and h in
Figure 1.
The above mentioned minor light scattering from swollen sample makes one think of a possible optical heterogeneity: some initially “white” gels upon swelling may become clear because the refractive indices of light in the gel and in the swelling medium become very similar. The PGMA xerogel resin after polymerization readily swelled in the GMA monomer—the ESC for GMA in polyGMA was 1.7 g/g (RT experiment)—in other words, more of the monomer was absorbed by its polymer homologue. This swelling capacity suggested that the phase separation during polymerization was not in play in this case: otherwise, the polymer would not be compatible with more of its monomer and would not swell in it. Thus, the produced resin is homogeneous similarly to resins of poly(methylmethacrylate). The SEM investigation also showed that structural inhomogeneities on the order of tens of micrometers did not exist in the PGMA xerogel right after polymerization (cf.
Figure 1a–d).
The next source of heterogeneities detected by cryo-SEM and HVSEM could stem from the preparation of sample for cryo-SEM by fast-freeze process. It is commonly assumed that the plunge-freezing of hydrated samples should preserve their structure [
27,
28,
29] or could induce formation of only submicron size pores easily identified as artifacts due to incipient ice formation. Such processes are in detail discussed by Paterson et al. [
12] in the case of poly[HEMA-
co-MeO-PEGMA] hydrogels, that are chemically similar to PGMA and swell more than the neat PHEMA gel, these gels contained even more water during their plunge-freezing in LN
2—around 85 wt% than our PGMA sample (74 wt.% of water) showed no significant change of structure, cf. Figure 3g in reference [
12].
Other reasons of the formation of such a heterogeneous structure like the one seen on
Figure 1e,f considered by us were (a) formation of bubbles during polymerization possibly from dissolved air or nitrogen used for the purging of reactants, and (b) heterogeneity caused by a possible only partial compatibility of the di(ethylene glycol) dimethacrylate crosslinker with the glycerol monomethacrylate monomer (cf. structures in
Scheme 1). To confirm these hypotheses, we synthesized the PGMA gels again under several sets of conditions (described in detail in the
Supplementary information). These conditions included (a) thorough degassing of the polymerization mixture while still DEGDMA was used as the crosslinker, (b) all monomers were freshly distilled prior to polymerization, then degassed and again DEGDMA was used as a crosslinker and (c) monomers were degassed before and after distillation, and the crosslinker was glycerol dimethacrylate, GDMA, as the closest chemical analogue to GMA (cf.
Scheme 1) to ensure good miscibility. Although, such heterogeneous structure as shown in
Figure 1e,f should be detectable in the as prepared polymer system even before its swelling, the resulting resins were always glassy, fully transparent and in cryo-SEM images did not show any signs of such structural elements, cf.
Figure 1a,b.
To track the effect of the plunge-freezing step and treatment under low temperatures at cryo-SEM on the structure of the sample G0/0.3, we also employed methods where no gel freezing was involved: such as the environmental scanning electron microscopy (ESEM). ESEM that does include sample cooling but only to temperatures above the freezing point of water. Indeed, the depression of the freezing point of water in the three-dimensional macromolecular network is involved, too, but discussion of this classical phenomena goes beyond the scope of this work, the reader may start from here [
30]. The gels during ESEM characterization were cooled down to 2 °C. The light microscopy (LM) and finally the laser scanning confocal microscopy (LSCM) using a fluorescently labeled hydrogel were performed with swollen GMA samples at ambient conditions. The results are shown in
Figure 2.
It should be noted that in ESEM, optimization of the chamber pressure is a very important parameter—if the appropriate pressure is not met, the sample surface structure may be flooded by a layer of liquid water (chamber temperature is +2 °C) or the surface structure may collapse by excessive drying (cf.
Supplementary information, Figure S2). Therefore, in this ESEM experiment, the G0/0.3 gel was investigated under a range of pressures 360–660 Pa. Neither the ESEM nor the light microscopy images revealed a heterogeneous structure of this hydrogel (
Figure 2a,b). However, the values of refractive indices of water and the water-swollen PGMA hydrogel were very close (1.333 and 1.366 respectively); hence, possible pores would be hardly observed with LM due to the lack of an optical contrast between the water-filled pores and water-swollen matrix. To resolve the issue of swollen structure existence, the PGMA hydrogel was covalently labeled with a modified fluorescein dye (details of the labeling procedure in the
Supplementary information) and observed using the ambient LSCM. Since after the polymerization, the possible unpolymerized species (including the dye residues) were washed out of the hydrogel, the dye could be present only in the swollen gel phase, but not in water filling the pores (if any). The detected structure was evenly distributed over the entire sample space—see
Figure 2c, where only the darker lines were caused by an imperfect surface of the specimen. In LSCM, the water-swollen G0/0.3 gel was found to be non-porous down to the distinguishable limit of units of micrometers. The evidence that the LSCM method has the capacity to clearly reveal a macroporous structure of fluorescently labeled swollen gels was also acquired and will be discussed in detail and illustrated later (gel H80/1 shown on Figure 5f).
In summary, based on the ESEM, LM and LSCM observations, the swollen G0/0.3 hydrogel did not contain pores like those shown in
Figure 1e,f. Clearly, these structures were secondary pores formed by crystallization of water in the gel and subsequent crystals sublimation under the chamber vacuum. Interestingly, the secondary cellular-like macropores in polyGMA gels were formed with great reproducibility always when water-swollen gels were treated by plunge-freezing and observed with cryo-SEM. We followed the development of secondary pores with observation time, see
Figure 3. Initially, a patterned surface with just a few circular holes was seen (
Figure 3a). The hexagonal pattern appeared within a frozen layer of water. After some time, more holes start forming in the hexagonal shape corners (
Figure 3b). After some more time under the chamber conditions, more cavities appeared on the observed surface, and finally turned into cellular-like structures (
Figure 3c). Similar effects of the prolonged exposure to the reduced pressure, and to the electron beam on the morphology of hydrogels have already been reported [
14,
15].
The structure of a hydrated sample treated by “cryo-fixation” strongly depends on the cooling rate at each volume element of the body that is related to its distance from the surface [
31]. To preserve the swollen structure morphology, ideally, the cooling rate should be high enough to prevent crystallization of water and growth of crystals: water should attain its amorphous form. From the experimental studies with hydrogels, it follows that the plunge-freezing technique (plunging in LN
2) under normal pressure can lead to structure preservation only when samples are very thin (10–40 μm) [
11,
32]. If a thin specimen of gel cannot be prepared, freezing should be performed under external pressure high enough to counteract the ice crystal growth (typically 2100 bar) [
33,
34]. The swelling degree of PGMA gels in water is quite high (EWC 2.8–5.8 g/g depending on gel dilution at polymerization) and the “standard procedure” of cooling by quick immersion of a small piece of the gel in LN
2 when high pressure freeze is not involved results in the development of a new porous structure. This structure was linked to the formation of ice crystals [
34,
35]. Within observation time, these crystals may sublime and leave the empty pores in the material surface. Mechanism of this process in homogenous swollen methacrylate-based hydrogels was qualitatively described in the literature [
28]. Quantitative features of the developing structure such as the average size and total volume of pores depend on the hydrogel swelling degree and dynamics of the freezing cycle.
During the observation in a cryo-SEM chamber, the frozen specimen was subjected to the electron beam irradiation under low vacuum (around 100 Pa), and the etching of the specimen occurred via the sublimation of ice crystals together with freeze drying of the hydrogel. The resulting voids could not be refilled with the network material, since the dehydrated gel under the observation conditions was a glassy material. When reswollen, the pores in the gel might partly recover.
3.2. PGMA Hydrogels Prepared in the Presence of Water as a Diluent
When gelation in the macromolecular system proceeds in a diluent, the resulting material is a swollen gel. If the swelling capacity of the gel macromolecular network is exceeded already during polymerization, the network will expel the additional liquid, i.e., it will phase separate and a heterogeneous material will be formed showing a typical microstructure: for example 2-hydroxyethyl methacrylate polymerized in water presence above 50 vol.% attains a well-known fused-sphere-like microstructure [
3] with hydrogel spheres connected throughout the whole sample volume. In PHEMA polymerizing in 80 vol.% water, the size of a sphere is approximately between 4 and 8 μm. The thorough description of the phase separation phenomenon in forming gels and examples of resulting morphologies in methacrylate hydrogels can be found in [
19,
24,
36,
37] and an example of a PHEMA sphere-like hydrogel formed by reaction induced phase separation, H80/1, will be given in Part 3.3 below.
It was found earlier by Refojo [
26] that in lightly crosslinked poly(glycerol monomethacrylate) systems (containing below 0.5 wt% of a bi-vinyl crosslinker) and diluted up to approx. 90 vol.% water in the monomer solution, the gels still absorbed more water after polymerization—a clear indication that PGMA gels did not phase separate during their formation. Indeed, the presence of a diluent during network formation alters the hydrogel macromolecular network topology—with more dilution, more elastically inactive loops are formed on the polymer backbone and less chains contribute to network elasticity, thus the final network is more loose, more swellable and it exerts lower modulus of elasticity than a network formed in the absence of a diluent [
38]. The dilution of the GMA-based system in our experiment was 40 vol.% so the water level was far below the phase separation critical dilution and the gel was macroscopically homogeneous.
The equilibrium swelling of G40/0.3 hydrogel at room temperature corresponded approx. to 85 wt.% water (EWC 5.8 g/g) while for G0/0.3 the water content was 74 wt% (EWC 2.8 g/g), cf. swelling values of all tested gels in
Table 1.
This difference in swelling well marks the difference of the macromolecular networks in G40/0.3 vs. G0/0.3 caused merely by network formation in the presence of various amounts of diluent. The G40/0.3 hydrogel was optically transparent in all states: as-prepared, equilibrium-swollen—expanded and dried at elevated temperature. The cryo-SEM images of the water-swollen G40/0.3 gel revealed porous cellular-like structure shown in
Figure 4a. This structure is very similar to that of gel G0/0.3, cf.
Figure 1e,f. In the sample G40/0.3, the porous structure revealed by cryo-SEM was attributed to the formation and sublimation of ice crystals following the plunge-freezing (in LN
2). However, the observed morphology of G40/0.3 was somewhat different, revealing the number of small round-shape voids in the walls of cellular cavities. When the swollen gel G40/0.3 was chemically labeled by FITC and investigated by LSCM, there was no structure on the level of tens of micrometers seen, cf.
Figure 4a vs.
Figure 4b. The major difference between the G0/0.3 and G40/0.3 gels was thus their swollen macromolecular network topology resulting in a considerably different swelling degree and equilibrium modulus of elasticity, 250 kPa vs. 30 kPa (
Table 1) respectively. The change of morphology of these hydrated gels when exposed to freezing conditions was qualitatively still similar, compare
Figure 1e,f and
Figure 4a,b.
3.3. PHEMA Hydrogels
Covalently crosslinked macromolecular hydrogels based on poly(2-hydroxyethyl methacrylate), PHEMA, swell in water significantly less compared with the gels based on PGMA, cf.
Table 1. For example, PHEMA hydrogel prepared without diluent absorbed 0.6 g of water per g of dry matter in equilibrium while EWC of G0/0.3 was 2.8. Although PHEMA is chemically rather similar to PGMA, its chains are less hydrophilic: they attain a special state of equilibrium conformation in water achieving balance between hydrophilic and hydrophobic interactions as described in [
24]. We prepared PHEMA hydrogels in the absence as well as in the presence of a water diluent: sample H0/0.3 and sample H40/0.3. Both hydrogels as prepared and swollen to equilibrium volume in water were expectedly transparent. In their dry state, they revealed homogeneous morphology by ESEM, LM and cryo-SEM—similarly as for PGMA hydrogels (
Figure 1a,b; the images for dry PHEMA not shown).
Contrary to G0/0.3, the swollen bulk-prepared H0/0.3 hydrogel revealed a homogeneous, non-macroporous structure when treated by plunge-freezing in the same way as G0/0.3 for the cryo-SEM investigation, cf.
Figure 5a. We explained this different morphology by the much lower content of water in PHEMA gel (EWC
H0/0.3 = 0.6 g/g compared with EWC
G0/0.3 = 2.8 g/g). Clearly, the water phase in the PHEMA gel was interacting with the macromolecular network to a much higher extent and thus water did not freeze out under the plunge-freeze conditions. Interestingly, the H40/0.3 swollen gel also attained a low amount of water (EWC
H40/0.3 = 0.7). This gel revealed a uniform non-porous structure (in given resolution range) by HVSEM, see
Figure 5b but incipient porous morphology was revealed by the cryo-SEM of H40/0.3, see
Figure 5c. However, the number of pores over the H40/0.3 frozen surface observed at the same exposition time was significantly lower than that in PGMA hydrogels, compare
Figure 4a and
Figure 5c.
The PHEMA macroporous hydrogels were produced by phase separation occurring above the critical water content that is approx. 45–50 vol.% for lightly crosslinked PHEMA. Thus, the gel H80/1 made at 80 vol.% of water falls in the range where the reaction-induced phase separation occurs for thermodynamic reasons as well described in the past [
19,
20]. The forming PHEMA gel matrix formed the typical fused-sphere-morphology while the voids between the spheres formed a continuous porous space as revealed by SEM and LM and shown in
Figure 5d,e,f. The equilibrium water content in H80/1 was 4.4 g/g, which was relatively high compared with the swelling of H0 and H40 gels (cf.
Table 1). The EWC
H80/1 corresponded to the dilution by water during reaction, which was 4.3 g of the water diluent per g of monomers. This means, that the overall volume of the formed structure is almost unchanged while the gel matrix in forming H80/1 forms the gel spheres expelling some water into the interstitial space. In the swelling equilibrium of H80/1, certain portion of water is located in the space between the gel spheres while some water swells the spheres. Thus, the swelling of the hydrogel phase in H80/1 must be significantly less than EWC = 4.4 g/g. The swelling of spheres was previously found to be close to swelling of H40/0.3, EWC
Hspheres = 0.7 [
24]. This suggests that the gel in the spheres was not changed under the conditions of plunge-freezing and cryo-SEM and behaved similarly as the gel H40/1. Moreover, comparison of the morphological structure of H80/1 by cryo-SEM with that by LSCM and even by HVSEM showed that the morphology of H80/1 seen by cryo-SEM was not significantly distorted, see
Figure 5d,e,f. Clearly, the spatial arrangement of the gel spheres and their flexible connections could facilitate accommodation of the possible solid ice formed upon plunge-freezing in the LN
2.
3.4. IPN Hydrogels
We have also investigated much stiffer swollen hydrogels made by reinforcing the primary hydrogel network (network 1) with the second, chemically similar interpenetrating network 2. The IPN hydrogels were made by sequential polymerization of network 2 components swollen in network 1. As the first networks, either homogeneous, non-porous (G40/3, H0/1) or heterogeneous, phase-separated (H80/1) matrices were chosen. The preparation of such gels and their morphologies are schematically shown in
Scheme 2, details of the preparation were elaborated by Sadakbayeva et al. [
39]. The morphological microstructure of the water-swollen IPN gels was investigated by cryo-SEM. Again, the images reflected the capacity of water to form crystals in the non-porous gels at plunge-freezing in LN
2 and following low temperature treatment in the microscope. Let us consider the example of two IPN hydrogels with macroporous network significantly differing in water content: 1/IPN hydrogel H80/1-H0/0.3 that contains 38 vol.% of water and 2/IPN hydrogel H80/1-G0/0.3 that contains 70 vol.% of water (cf.
Table 1).
The used macroporous matrix was the same in both cases, yet its state during the polymerization of network 2 was somewhat different due to different swelling in the GMA monomer (ESC
H80/1 for HEMA was 10.2 g/g and for GMA 13.0 g/g). High swelling of network 1 in both monomers suggested that its polymer phase was evenly penetrated with the monomer of network 2 like depicted in the
Scheme 2. The EWC of the resulting structured IPNs was different (2.2 g/g for H80/1-G0/0.3 and 0.6 g/g for H80/1-H0/0.3). Thus the possibility of “secondary” formation of pores during cryo-SEM was more likely for the system H80/1-G0/0.3. Indeed, in the H80/1-G0/0.3 macroporous morphology was found by cryo-SEM: the boundaries of spherical particles of the first gel were partially preserved and embedded in second network, but numerous holes in both phases were clearly visible (
Figure 6a). The H80/1-H0/0.3 sample provided a clearly distinguished two-phase area with visible gel fused spheres and material filling the space around the spheres, both phases showing no porosity (
Figure 6b). The mechanical failure at the cryo-SEM experiment of the gel spheres in the H0/1-G0/0.3 hydrogel was investigated in detail by the experiment (see
Supplementary data, Figure S3).
Finally, the morphology of the microstructured IPN (MIPN) hydrogels in their native swollen state was elucidated using LSCM. This method was already applied for various IPN systems [
33,
40,
41,
42]. To distinguish the two networks in the IPN hydrogel, we introduced the two different modified dyes (the methacryloylated fluorescein—green color and DY-677—red color) in both monomer mixtures of H80/1-G0/0.3 at polymerization. Different light absorption and emission properties of these dyes allowed distinct observation of the corresponding hydrogel phases, see
Figure 6c. The image shows that the H80/1-G0/0.3 hydrogel consists of fused spherical regions of mixed color: the gel spheres should contain both PHEMA (labeled by methacryloylated fluorescein, green color) and PGMA (labeled by DY 977, red color) separated by the regions containing only PGMA (network 2 labeled by red color); no polymer-free voids were identified with LSCM in this IPN hydrogel. Thus, the porous morphology evident in
Figure 6 developed during cryo-SEM as a secondary structure and is considered artifactual.
3.5. Effect of Plunge-Freezing at Cryo-SEM on the Hydrogel Microstructure
As it was shown with the H40/0.3 homogenous sample, the swelling capacity alone cannot serve as a single predictive parameter for assessing the possibility of “secondary porosity” formation induced by the freezing cycle. For example, the H40/0.3 sample revealed the artificial porous structure despite its low equilibrium swelling in water (EWCH40/0.3 = 0.7 g/g). We constructed an empirical map of the cryo-SEM behavior of hydrogels studied here in order to understand the limits of this popular technique and to set semi-predictive criteria of the applicability of plunge-freezing with cryo-SEM for correct visualization of the swollen gels morphology without artifacts.
In the schematic map on
Figure 7, the hydrogels were arranged according to their tensile modulus (
E) as a mechanical parameter along the ordinate and their equilibrium water content along the abscissa. Four segments (A–D) in the map are related to the gels with relatively high or low values of
E and EWC, which indicate the reliability borders of the plunge-freezing cryo-SEM method. According to these schematics, the cryo-SEM observation gave relevant results exclusively for the samples from segment B: those with low EWC and high
E. When the EWC was too high (segments C and D), the formed ice crystals led to the appearance of secondary porosity irrespectively of the gel mechanical strength. When the tensile modulus of the gel was too low (segment A), the matrix could not withstand the formation of ice crystals even at relatively low swelling capacity.
The pore size of PGMA-based hydrogels revealed by means of cryo-SEM observation was little dependent on the position of the sample in the map. Moreover, the pore size distribution was quite wide (
Table 1). Finally, the observed pore size may be a function of observation time (
Figure 3). The pores appearing in the H40/0.3 gel were significantly smaller as compared to the PGMA gels, due to the lower equilibrium water content of the PHEMA gel.
In conclusion, it should be noted that the empirical two-variable criterion could be applied to the multiphase samples (such as MIPNs). As shown for the macroporous H80/1 sample (for the structure see
Figure 5f), no secondary pores were found by the cryo-SEM images in the gel spheres, even though its overall EWC was relatively high (4.4 g/g) and its apparent Young’s modulus was very low (4 kPa). In the case of such heterogeneous systems, rather the hydrogel phase properties should be considered but these may not be easily experimentally accessible.