The influence of structure and morphology on ion permeation in commercial silicone hydrogel contact lenses

The importance of the microstzructure of silicone hydrogels is widely appreciated but is poorly understood and minimally investigated. To ensure comfort and eye health, these materials must simultaneously exhibit both high oxygen and high water permeability. In contrast with most conventional hydrogels, the water content and water structuring within silicone hydrogels cannot be solely used to predict permeability. The materials achieve these opposing requirements based on a composite of nanoscale domains of oxygen-permeable (silicone) and water-permeable hydrophilic components. This study correlated characteristic ion permeation coefficients of a selection of commercially available silicone hydrogel contact lenses with their morphological structure and chemical composition. Differential scanning calorimetry measured the water structuring properties through subdivision of the freezing water component into polymer-associated water (loosely bound to the polymer matrix) and ice-like water (unimpeded with a melting point close to that of pure water). Small-angle x-ray scattering, and environmental scanning electron microscopy techniques were used to investigate the structural morphology of the materials over a range of length scales. Significant, and previously unrecognized, differences in morphology between individual materials at nanometer length scales were determined; this will aid the design and performance of the next generation of ocular biomaterials, capable of maintaining ocular homeostasis.


| INTRODUCTION
The design of a contact lens material should not seek merely to create a bio-inert temporary implant, but preferably it should be capable of maintaining the existing homeostasis of the avascular corneal bed in equilibrium with its external environment. However, currently no single material exhibits the required ocular biocompatibility to enable asymptomatic long-term use and good ocular comfort across a wide patient base. The most common symptoms that are associated with elective discontinuation of wear are discomfort and dryness. [1][2][3] The goal for any new contact lens is that it should allow a suitable balance of oxygen and ion permeability, appropriate on-eye movement, and satisfactory tear exchange. Lens permeability is critical in enabling the free movement of nutrients, the removal of metabolic waste products, and the dynamic distribution of tear film components including water, electrolytes, and proteins to maintain ocular homeostasis. 4,5 Early soft contact lenses were based on poly(2-hydroxyethyl methacrylate) (pHEMA) hydrogels. These, although well hydrated, soft, and relatively comfortable, fell well short of providing the degree of oxygen permeability needed to facilitate corneal cell homeostasis. 6,7 Consequently, there has been a progressive evolution of contact lens materials from conventional hydrogels to more oxygen permeable silicone hydrogel (SiHy) materials. The oxygen permeability of commercial SiHys is known to be high. 8 As contact lens technology has evolved, recognition of the desirability of maintaining a pre-and post-lens tear film that mimics the natural tear film (which has a distinct electrolyte composition) has become clear. [9][10][11][12] The permeation of tear electrolytes is, however, much more variable and at the heart of this study, as reported previously ion permeation does not show a uniformly predictable dependence either on equilibrium water content (EWC) alone or the water structuring within the material. 13 According to Domschke et al., ion permeability within contact lenses is a critical parameter for lens movement on the eye 4 and it is also thought to be a requirement for post-lens tear turnover and metabolic waste removal. 12 Similarly, the electrolyte composition of tears, which is quite distinct from that of serum, can be adversely affected by the presence of a barrier, which disturbs ion transport creating a post-lens tear film. 7 In addition, studies involving more detailed clinical-based analysis have shown that different materials can affect physiochemical properties which in turn can influence the nature and composition of  Table 1 for full monomer names) the post-lens tear. [14][15][16] Differences in the occurrence of corneal infiltrative events have been identified between commercial SiHys 17 ; materials that have similar high oxygen permeabilities but very different ion permeabilities. 13,18 The influence of polymer phase morphology, which can take many forms, on lens movement on the eye has been highlighted. 19 Structure-property-performance relationships are important. Identification of the causative factors that link physicochemical properties, disturbance of corneal homeostasis, and long-term ocular biocompatibility does, however, remain elusive. A more comprehensive understanding of the relationships between physicochemical properties and the underlying SiHy structure needs to be established. All SiHy materials are a subset of a larger group of materials that can be defined as hydrophobically modified hydrogels. The chemical nature of the selected monomers and the ratios in which they are employed can have a profound effect on the nanoscale structure within the material.
This, in turn, plays a huge role in determining the final physical properties displayed by the material. [20][21][22] A range of siloxy monomers and macromonomers is used in the preparation of SiHys, typically combined with other, more hydrophilic, monomers ( Figure 1;  (Figure 1, part 4).
The primary consideration of manufacturers, in selecting combinations and ratios of these monomers, is to ensure that nanoscale separation and self-assembly of the siloxy units does not extend so as to impair the optical clarity of the final water-swollen SiHy network; an obvious requirement for ophthalmic use. The extent of separation and self-assembly of the siloxy groups is limited by the degree of mobility afforded during the polymerization (chain growth, branching, and crosslinking) process. The final nanoscale domain structure is therefore controlled by both composition and polymerization process.
The use of diluents (Table 2) provides another degree of control in the development of haze-free, optically clear contact lens materials by producing an appropriate balance between the various constituents. 28 Despite significant progress in understanding the basic structureproperty relationships of networks, much remains to be learned about how the foundational macromolecular building blocks influence structure from the nano to the macroscopic scale. In this study, four commercial SiHy materials were selected. They enable the study of SiHys with similar water contents but very different ion permeation properties as previously shown by Mann et al. 13 Two of the chosen materials have relatively high quoted EWCs: narafilcon A (44%) and comfilcon A (48%), and two have lower water contents: lotrafilcon B and balafilcon A (both 34%). To investigate factors that may influence the disparate ion permeability properties of each material, the structural morphology at different length scales of each of these four materials was investigated by environmental scanning electron microscopy (ESEM) and small angle x-ray scattering (SAXS) together with specific water structuring properties of each lens material.

| MATERIALS AND METHODS
The properties and components of four commercial SiHys namely narafilcon A, comfilcon A, lotrafilcon B, and balafilcon A which were selected based on their water content and ion permeation properties 13 are shown in Table 2.  Using this information and dividing by the heat of fusion of pure water (79.72 cal/g = 333.55 J/g) then multiplying by 100 allows the percentage of free or freezing water to be calculated. The total area under the endotherm peak corresponds to the total freezing water content (FWC), which is the sum of the ice-like water content (ILWC) and the polymer-associated water content (PAWC), illustrated previously. 13 The non-freezing water content (NFWC) was calculated by subtraction of total freezing water from the measured EWC. All values are expressed as percentages of the total mass of the hydrated lens.

| Environmental scanning electron microscopy (ESEM)
The topography and heterogeneities of the SiHy contact lenses were observed with a Philips XL30 FEG (Thermo Fisher Scientific, Waltham, MA) ESEM equipped with a cold stage. Each hydrated lens was prepared for mounting as shown in Figure 2 to image the cross-sectional surface. The lenses were prepared using a colloidal carbon mount.
They were rapidly immersed at −180 C for 1 min and sublimated at −95 C for 5 min. Samples were then transferred into a PolarPrep 2000 cryo-stage system chamber enabling rapidly frozen samples to be analyzed at −140 C. The accelerating voltage was set at 5 kV, with a working distance range of 3 to 5 mm. The samples were analyzed at ×35,000 magnification. The system was set to detect secondary electrons used for sample surface topography.
The heterogeneity of each image suggests discontinuities in the physical topographical structure of each material. Using five individual lens matrix section images-which are consistently characteristic of each material-the percentage of what we have termed "apparent porosity" (black vs. white) of the materials was determined. In principle, the lower the mean atomic number of a sample under a beam, the fewer electrons are generated/detected and the resultant pixel for that region will consequently be dark. Thus, a pore or void, which is atomically light, will always appear dark if it is adjacent to an atomically heavier polymer region. Each image was assessed by using the Java-based image processing program ImageJ. 29 After adjusting the initial threshold, the ImageJ plugin for particle analysis was used.  In SEM, the returned signal is provided by the amount of material in the path length of the electron beam. In our samples, where there are deep crevasses or voids, the returned signal is less intense (less material in the beam path) and therefore appears dark. In contrast, the areas where there is ample polymer in the beam path appear brighter (more material = higher intensity). In short, these allow the topology of the material to be mapped out and give an impression of the surface porosity.

| Small angle x-ray scattering (SAXS)
SAXS, a nondestructive x-ray technique, was used to investigate particle size, shape distribution, and morphology at the nanoscale level.
Measurements were conducted on a laboratory SAXS instrument (Xeuss 2.0, Xenocs, France) equipped with a liquid gallium MetalJet Xray source (Excillum, Sweden), wavelength (λ) = 0.134 nm, beam size approximately 2 mm 2 , camera length = 2.4 m (calibrated using a silver behenate standard) and Dectris Pilatus 1 M pixel detector. Samples (ca. 100 μm thickness) were placed in a cell (1.5 mm depth) with deionized water and sealed with Kapton®. Scattering data were collected for 900 s using collimating slits of 0.5 × 0.6 mm ("high flux" mode). The scattered x-rays were adjusted for transmission, background (water and Kapton®), sample thickness, and acquisition time before the intensity data were placed on an absolute scale (cm −1 ) using scattering from a standard sample of glassy carbon. 30 Data reduction from a 2D scattering pattern to 1D intensities as a function of q (scattering vector, nm −1 ) was performed using the instrumentspecific Foxtrot software and custom LabView code. Spacing between domains was calculated using the following equation: d = 2π/q, where the q values of each lens were taken from the first major feature of the corresponding radially integrated scattering pattern.

| Statistical methods
Descriptive statistics in terms of mean and standard deviation are provided, Figure 4 where n ≥ 3 and in Figure 6 they denote the variation for five separate images on each lens material.   Table 3 summarizes the hydration properties of the lenses after a 7day hydration in deionized water before ion permeation. One representative DSC trace showing the water structuring profile for each of the four materials is provided in Figure 3. The difference between FWC (= PAWC + ILWC) and total EWC indicates that all contact lenses have approximately 15 to 20% nonfreezing water of the total lens weight. It is the distribution of freezing water between ILWC and PAWC (Section 2.2) that is of interest here, since this is critical for the interaction of hydrated ions with water within the polymer matrix.

| Water structuring
Water structuring values for lenses in packing solution, DI water, and post ion permeation showed slight differences for each condition but these lay within standard deviations (e.g., FWC for comfilcon A in packing solution, DI water, and post NaCl permeation were 27.9 ± 0.6, 26.9 ± 1.7, and 26.1 ± 1.9%). Water structuring results discussed here and thereafter refer to those measured in DI water to represent the ion permeation analysis conditions. Narafilcon A material was observed to contain uniquely high values of polymer-associated water (23.7 ± 2.1%) compared to a very low ice-like water content (4.4 ± 0.4%). In contrast and at the other extreme, the level of ice-like water within comfilcon A was higher (16.6 ± 1.6%) than that of polymer-associated water (10.4 ± 0.7%). Balafilcon A also shows a higher ratio of ice-like (13.9 ± 0.6%) compared to polymer-associated water (5.6 ± 0.3%), although in balafilcon A the total water content is significantly lower than that of comfilcon A (34.2% compared to 47.5%). Lotrafilcon B has the same EWC as balafilcon A, a similar amount of polymer-associated water (5.2 ± 0.8%), but less ice-like water (10.3 ± 0.9%).
Unlike conventional hydrogels, 31 Table 2). It was therefore logical to investigate differences in physical

| ESEM imaging of the matrix of the SiHy contact lens materials
Characteristic structures were observed for all samples (Figure 5), suggesting a degree of heterogeneity more marked in comfilcon A and balafilcon A than in narafilcon A and lotrafilcon B. The "apparent porosity" was analyzed using the ImageJ software, and calculated as a percentage of the total image in each case at a magnification of ×35,000 (n = 5), as illustrated in the insert in Figure 6. The least ion permeable material, narafilcon A, showed the lowest percentage of "apparent porosity" while the most ion permeable material, balafilcon A, exhibited the highest percentage ( Figure 6). Cluster size population averages for the lens materials are summarized in Table 4. Visual inspection of porosity and cluster size by the ImageJ suggest that narafilcon A, the least ion permeable material, showed the lowest porosity and higher polydispersity of sizes, which has three different populations. The more porous materials, comfilcon A and balafilcon A, show larger and more uniform domain populations (Table 4).

| Ion permeation behavior and water structure
The differences in the behavior of a series of commercial SiHy materials of apparently similar chemical structure ( Figure 1; Table 2) were examined using hydrated cations as molecular-level probes. Despite the many apparent similarities (e.g., optical clarity, oxygen permeability, and tensile modulus), the materials differ dramatically in other ways, particularly ion permeability ( Table 2). Ion permeability in conventional hydrogels is known to be governed simply by water content and water structure, the picture for SiHys is more complex, a point underlined by the huge differences in ion permeabilities found with SiHys of similar water content. 13 In seeking an explanation for the unusual ion transport behavior of SiHys, it is logical to explore the way that differences in the self-assembly characteristics of siloxy moieties leads to complex morphologies in SiHys. These are not found in their conventional hydrogel counterparts and may well reveal unexplored characteristics of this group of materials.
Balafilcon A and comfilcon A have been shown to exhibit the highest ion permeation rates, followed by lotrafilcon A, and then narafilcon A, by far the least ion-permeable material of this group. Potassium permeation values were higher than those for sodium, which in turn were higher than calcium. 13 Relative permeation values of the individual ions were observed to agree with the Stokes' (or hydrodynamic) radii and the relative ionic mobilities, which increase in the order K + <Na + <Ca 2+ (Table 5). Table 5 also shows the hydrodynamic radius of the chloride counter-anion which, importantly, does not have a rate-determining role. 31 The differences observed in the initial profile, shape, and slope of the curves shown previously 13 reflect differences in the way that the various material structures and charges interact with the aqueous salt solutions and with water in affecting transport through the matrix. From inspection of Table 3 and Figure 4, the broad correlations between the ion permeation values and EWC or FWC found in simpler hydrogel systems 31  Disruption and "ordering" of hydrogen bonded water networks is greater near a hydrophobic solute. Water hydrates the hydrophilic portion of amphiphiles but excludes the hydrophobic regions to yield (e.g., ) micelles and lipid bilayers, thereby reducing entropy in thermodynamically driven processes. Extending these principles to optically clear hydrophobically modified hydrogels, the nature of "polymerassociated" water becomes clearer. Disruption and reordering of the tetrahedral structure of water is induced by even modestly hydrophobic polymer chains with consequent deviation from perfect ice networks on freezing, which has an inevitable effect on melting behavior.
If the polymer chains are clustered to form pores, the proportion of "ice-like" to "polymer-associated" water increases. This can be seen here in the higher proportions of ice-like water in the more porous comfilcon A and balafilcon A matrices.
The free mobility of hydrated ions through an aqueous matrix inevitably requires the ability of the hydrated ion to exchange water molecules with the surrounding medium. It is logical to expect that hydration phenomena related to this will influence ion permeation behavior in SiHys. It was observed that narafilcon A was found to have both low permeability characteristics and much lower ice-like water than the other three materials studied. In simplistic terms this means that there is very little available water with the ability to hydrate cations and sustain their freedom of movement within the polymer network. It is clear, however, that results from water freezing analysis and the ratios of bound and free water alone do not provide a complete explanation of the ion permeation properties demonstrated by these materials, unlike conventional pHEMA-based contact lenses. 13 The phase separated nanostructure in SiHys, reflected in the SAXS results, and the mesostructures as observed by ESEM will also affect their ion permeation behavior.

| Composition, morphology, and internal nanostructure
It has been postulated that the existence of two separate phases in SiHy materials, rather than a single homogeneous phase, may be advantageous, promoting both the diffusion of ions through the hydrophilic regions of the polymer (so-called hydration channels), and increased diffusion for oxygen through the silicone-rich regions. 24 Although the hydrophilic regions of the material provide good diffusion of ions, they provide a relatively high barrier to oxygen diffusion. 24 An excellent discussion of the evolution of contact lens morphologies and the concept of independent control of oxygen and ion transport by means of co-continuous morphologies reflects aspects of the commercial knowledge that has been amassed in recent years. 19 The ESEM images provide good quality imaging of hydrated sam- between ice-like water contents and the permeation of salts through these SiHys was observed, whereas there seems to be no correlation between the EWC (which takes into account bound water) and ion permeation. It is interesting to note that balafilcon A has less ice-like water than comfilcon A, but a higher permeation coefficient for each salt studied in this work. However, balaficon A has a higher percentage of "apparent porosity" ( Figure 6) and greater visually assessed interconnectivity between the heterogeneities observed in Figure 5.
This appears to allow increased diffusion of salts through the material.
The samples with lower "apparent porosity": narafilcon A and lotrafilcon B, show much smaller nanoscale cluster sizes (as measured by SAXS), and have the lowest values of permeation coefficient for all salts studied here.
The data suggest that materials with a relatively large interconnected mesoporous structure lead to higher ion permeabilities.
SAXS was used to probe smaller length scales than ESEM, that is, the    Tables 1 and 2) with the resultant differences in morphology; this in part is due to the lack of precise compositional information.
The patent literature, together with regulatory submissions, has enabled the compiled information shown here to be assembled. This provides a useful basis for some general observations and conclusions.
In contrast, the "apparent heterogeneities" observed here occur at much shorter length scales than those influencing optical clarity (ca 500 nm) and require sophisticated investigative instrumentation-in the domain of molecular self-assembly-driven by the fundamental incompatibility of silicone moieties and water. 8,21 The driving principles in the design of SiHys with optical clarity have been to combine siloxy groups (e.g., TRIS, Figure 1, structure 2b) with hydrophilic monomers, compatibilizing macromers, and diluents designed to maintain compatibility during the lens processing stage.
The material with the least "apparent heterogeneity" is narafilcon A.
The SiHPMA structure (Figure 1, structure 2e) has a hydroxyl ester moiety in close proximity to a relatively short siloxy chain containing five silicon atoms. Of all the SiHy material structures, this presents the least opportunity for self-assembled siloxy units that exclude covalently bonded hydrophilic groups. This material clearly has the lowest degree of "apparent porosity" (Figures 5 and 6). The description of the design considerations that accompanied the development of lotrafilcon A is contained in an extensive patent application. 24 This places great stress on the development of appropriate macromer structures, which dominate the formulation, typified by Figure 1, structure 4c. This is characterized by relatively short linear siloxy segments interspersed with fluoroether units. As with narafilcon A, the hydrophilic "solvent-monomers" are N,N-dimethyl acrylamide (NNDMA) and HEMA (Table 1; Figure 1). Without comparative data, no definitive conclusion can be drawn but it is relevant to observe that NNDMA is an exceptionally good solvent and may well contribute to the low heterogeneity of these two materials. Comfilcon A and balafilcon A exhibit appreciably greater heterogeneity than their two counterparts and their respective macromer structures (Figure 1, structures 4a and 4b) appear to offer greater potential for hydrophobic clustering. Comfilcon A is characterized using both N-methyl-Nvinyl acetamide and NNDMA which contrasts with balafilcon A which relies almost completely on the solvent power of N-vinyl pyrrolidone.
These factors may well contribute to the difference in morphologywhich is striking. The various process-related factors that govern cure rates and kinetics of polymerization will also influence network formation and morphology but the nature and extent of these effects is not in the public domain. It is clear that differences in chemistry and morphology of these materials are interlinked and that further systematic studies will be necessary to resolve the detailed effects of structure on self-assembly.
The use of a series of techniques that probe different aspects of morphology and heterogeneity demonstrates that heterogeneity can lead to higher levels of ion permeability than are achieved with homogeneous hydrogels. This is a consequence of the way in which the aqueous domains are organized. It is through this investigation and understanding of the length scale of morphological self-assembly that we have been able to explain why the bulk picture of water content and water structure, which predicts ion and oxygen permeability in homogeneous hydrogels does not apply to the SiHys exemplified here. Our results suggest that "morphological heterogeneity" of the polymer phase explains what cannot be determined directly-the "morphological organization" of the aqueous phase.  19 The limited amount of public domain information underlines the fact that the relationships between structure and performance that exist between the varied and versatile structural components that characterize this group of biomaterials are currently unexplored and unexploited for other applications. However, that fact provides a logical impetus to study structure-property relationships of purpose-synthesized silicone-based copolymers. The ability to manipulate and predict the structure and physical properties of a polymer network by changing specific variables (i.e., polymer molecular weight and concentration, chemical groups, initiator, cross-linker and polymerization method) is fundamental to the successful development of any process.