Polymerization of hydroxyethyl methacrylate (HEMA) under rotation to form core-annular hydrogels

Abstract

Hypothesis:

We propose to polymerize high water content hydroxyethyl methacrylate (HEMA) formulations in a rotating cylinder to explore the effect of the rotation on microstructure and critical parameters such as diffusivity of model proteins in porous poly-HEMA gels.

Experiments:

Cylindrical molds were partially filled with water-HEMA-initiator-crosslinker mixtures and exposed to UV light while undergoing rotation to polymerize into a cylindrical tube. The process was repeated multiple times to manufacture a core annular rod with multiple concentric rings, in which at least one ring was porous. The porous gels were imaged by scanning electron microscopy to explore the microstructure. The transport of model proteins bovine serum albumin and human γ-globulin was measured and modeled, in radial and axial directions, to obtain the effective diffusivity and partition coefficient. Also, the true diffusivity of proteins was calculated by accounting for the effects of porosity and tortuosity.

Findings:

The porous gels exhibited diffusion-controlled release of both model proteins. The hydrogels prepared with 55% water in the monomer mixture were porous with non-isotropic structure likely due to axially oriented pores with minimal radial connectivity. The gels with higher water content were isotropic with interconnected pores in both directions. The pore volume increased with water content, but the partition coefficient was relatively constant and less than one likely due to presence of isolated unconnected pores.

Graphical Abstract

1. Introduction

Hydrogels are useful in several biomedical applications including tissue engineering, contact lenses, drug delivery, sensors etc. [1–4]. The pore sizes of the hydrogels impact critical physical properties such as water content, modulus, and diffusion rates of molecules. The hydrogels can be categorized into porous and non-porous depending on the size of the pores. The non-porous gels also have pores but those are typically only a few nanometers in size. The porous gels contain larger pores which scatter visible light resulting in an opaque appearance. The porous hydrogels can be manufactured via many approaches including gas injection, phase separation, electrospinning, freeze-drying, and templating [5–9]. Many researchers have explored applications of porous gels in tissue engineering and delivery of large molecules [10,11]. The microstructure of the pores including size and connectivity impact diffusion coefficient of the biologics, and so novel manufacturing approaches that can yield a desired microstructure are valuable. For example, aligned pores in hydrogels could be useful for applications such as growth of neural tissue [12].

In this work we polymerize HEMA monomer formulations under rotation to explore whether the microstructure is impacted by the rotation. The process can be used to form a hydrogel tube or a core annular structure with the possibility of multiple rings with differing physical properties. These different configurations could be useful in multiple applications. For example, the core–shell structure could be useful for delivery of biologics from the porous annulus, and sustained delivery of small molecules from the non-porous center. These core annular hydrogel structures were obtained by multi-step polymerization of the monomer formulation in a partially filled rotating cylinder. The process was repeated multiple times to form the core annular designs with varying properties in each concentric layer. This approach has some similarities to the novel method developed by Dalton et al. in which a two component monomer mixture was polymerized in a single step in a rotating tube to form concentric tubes due to phase separation during polymerization [13,14]. Dalton et al. focused on a porous central zone and non-porous annulus to serve as nerve guidance channels in the central nervous system, while we are primarily interested in delivery of biologics. To explore whether the structure is isotropic, hydrogels were cut along axial and radial orientations. The hydrogels were imaged along different planes and additionally, protein transport was measured in samples with different orientations. The proteins transport was fitted to the diffusion equation to determine diffusivity and partition coefficient, and these parameters were used to indirectly determine whether the microstructure was isotropic.

2. Methods

2.1. Materials

The monomer 2-hydroxyethyl methacrylate (HEMA) (≥99 %, ≤50 ppm monomethyl ether hydroquinone as inhibitor), crosslinker ethylene glycol dimethacrylate (EGDMA) (98 %, 90–110 ppm monomethyl ether hydroquinone as inhibitor), and photoinitiator Darocur Diphenyl (2,4,6-trimethylbenzoyl)-phosphine oxide (TPO) (97 %) were purchased from Sigma Aldrich. The proteins bovine serum albumin (BSA) (heat shock fraction, pH 7, ≥98 %) and human gamma globulin (hγG) (≥99 % (electrophoresis)), and Dulbecco’s phosphate buffered saline (PBS) (without calcium chloride and magnesium chloride, liquid, sterile-filtered, suitable for cell culture, pH 7.1–7.5) were purchased from Sigma Aldrich. Thermo Scientific Micro BCA assay reagents were purchased from Fisher Scientific.

2.2. Hydrogel polymerization under rotation

The process for polymerizing hydrogel tubes and the core-annular cylindrical rods is illustrated in Fig. 1. A tube spinning apparatus was assembled in the lab to rotate the cylindrical tube which served as the mold for the polymerization (Fig. 2). The spinning apparatus included an18V DC motor, a tail stock, angular contact bearing, and two aluminum cups. The tail stock stabilizes the passive side of the mold while allowing the mold to spin freely with minimal frictional resistance. The tail stock can slide on the 80/20 aluminum framing strip base to load cylindrical molds of any length. The tail stock was attached to an angular contact bearing which fits an aluminum cup to hold the passive side of the mold. On the active side, an 18 V DC motor was mounted on the base. A collar was connected to the motor spindle, and the aluminum cup was attached to the collar using a hex screw. The assembly also included a DC power supply, three UV light bulbs (18 W CFL) and a tachometer to measure the rotation rate.

Fig. 1.

A) Step-by-step process to fabricate core-annular hydrogels with at least one porous ring and obtain discs to measure axial transport. A) Polymerization of the outermost ring, which could be porous or nonporous (steps 1–4) and polymerization of the second ring (steps 5–6). The tube was cut along the length and punched to obtain discs for radial transport measurements (steps 7–8). B) Continuation of the fabrication process from step 5 to polymerize a non-porous pHEMA layer inside the porous layer, followed by filling the core with HEMA (no water) to obtain a core-annular pHEMA hydrogel rod (steps 6–9). The rod can be cut into discs (step 10) to measure axial transport. (Created with Biorender.com).

Fig. 2.

A) Lab-assembled apparatus for fabricating core-annular hydrogels. The setup includes the external cover with UV lamps, the spinning apparatus, DC power supply, light switch, and tachometer. B) A closeup of the tube-spinning apparatus showing the glass tube filled with solution loaded into the apparatus.

The spinning RPM was controlled by varying the current supplied to the 18 V DC brushed motor (Digikey) from the power supply (Eventek, 30 V, 5A) to achieve a range of 800–1700 rpm. A box with CFL UV bulbs attached on the inside was placed over the tube spinning assembly to provide radiation for initiating the polymerization reaction. The duration for polymerization was set at ten minutes for the outermost layer, and an additional twenty minutes were provided for each extra layer, to account for the absorption of light by the polymerized layers. In many of the structures explored here, three concentric layers were polymerized, prior to filling the core with water-free HEMA monomer solution and polymerizing vertically to completely fill the core. In most cases, the outermost layer was polymerized with a water/HEMA ratio of 2/3 and the next layer was polymerized with water/HEMA ratio ranging from 11/9 to 4/1 to form the porous annulus (Table 1). The third layer and the central core were polymerized with water-free HEMA monomer formulation. The central core was polymerized in vertical orientation without rotation using MaestroGen transilluminator (LB-16 UltraBright LED Transilluminator, MaestroGen, Hsinchu City, Taiwan) at 345 nm. The fully polymerized rod was removed from the glass vial and submerged in excess water for more than 5 days to hydrate the rod. After equilibrium hydration, the hydrogel rod was cut into thin discs ranging from 2 – 5 mm in thickness and soaked in DI water to extract the unpolymerized monomer, crosslinker and initiator. The DI water was replaced multiple times over 4 or more days to achieve complete extraction of the unreacted components.

Table 1.

Summary water content measurements for all formulations. Columns 1–3 are based on formulation composition and Columns 4–5 are based on measurements of wet and dry weights.

$\frac{VolumeofWater}{VolumeofHEMA}$	$\frac{Massofwater}{Massofformulation}\left(\%\right)$	$\frac{Massofexcesswater}{Massofwater}\left(\%\right)$	EWC(%)	$\frac{Massofwaterinpores}{Massofwater}\left(\%\right)$
2/3	38.3	0	40.7 ± 1.1	–
11/9	53.3	45.2	48.1 ± 2.9	27.7 ± 10.8
3/2	58.3	55.3	51.1 ± 1.0	35.9 ± 2.7
13/7	63.4	63.9	64.9 ± 1.2	49.4 ± 5.2
7/3	68.5	71.3	75.1 ± 3.6	77.7 ± 1.0
4/1	78.8	83.2	82.5 ± 4.7	85.0 ± 4.8

2.3. Measuring water content in pores

The fully hydrated hydrogels were weighed and then allowed to dry at room temperature. The mass of the hydrogel was recorded as a function of time till the weight reached the constant dry weight. The equilibrium water content ($EWC$) was calculated using the following equation

$$EWC=\frac{W_{wet}-W_{dry}}{W_{wet}}\mathrm{*}100$$ (1)

where $W_{\mathit{wer}}$ and $W_{\mathit{dry}}$ are the weights of the hydrated and dry gels. Based on the densities of water (1 g/ml) and HEMA monomer (1.073 g/ml) [15], the volume fraction of water in the formulation was converted to mass fraction (Table 1) for direct comparison with the $EWC$.

Based on literature and our measurements, the poly-HEMA (pHEMA) hydrogel has a saturated water content of 40 % [16], i.e., the mass of water in a pHEMA gel saturated with water is 40 % of the total weight. The mass of water in the gel is equal to the difference between the wet and the dry weight. Thus, the wet and dry weight of a saturated pHEMA gel are related by the following equation

$$\frac{W_{\mathit{wet},\mathit{HEMA}}-W_{\mathit{dry},\mathit{HEMA}}}{W_{\mathit{wet},\mathit{HEMA}}}=0.4$$ (2)

Eq. (2) can be simplified to obtain the following relationship,

$$W_{\mathit{wet},HEMA}=\frac{5}{3}{W}_{\mathit{dry},HEMA}$$ (3)

A pHEMA hydrogel prepared with a monomer formulation that has less water than the saturation content will not have any large pores because the entire water is used is hydrating the polymerized gel. On the other hand, a pHEMA hydrogel polymerized with water in excess of the saturation capacity of 40 % of total weight will result in phase separation between the fully hydrated pHEMA gel and the excess water, which will form large pores [17]. Thus, the mass of water in the large pores ($W_{\mathit{water}\mathit{in}\mathit{pores}}$) in a gel of hydrated weight ($W_{\mathit{wet}}$) and dry weight ($W_{\mathit{dry}}$) can be calculated by subtracting the mass of the fully hydrated pHEMA hydrogel from the total weight of a hydrated porous gel, i.e.,

$$W_{\mathit{waterinpores}}=W_{\mathit{wet}}-\frac{5}{3}{W}_{\mathit{dry}}$$ (4)

where the second term on the RHS represents the mass of the fully hydrated non-porous pHEMA. The mass of water in pores can be divided by the density of water to calculate the volume of pores in any sample. This equation allows calculation of the water volume in pores which is then used to calculate the concentration of proteins in the pores. The same equation can also be used to estimate the expected mass of water in pores based on the formulation composition by replacing the wet weight with the total weight of the monomer mixture and dry weight with the weight of the HEMA monomer.

2.4. Transport of model proteins

The transport of model proteins was measured and modeled to determine partition coefficient and diffusivity. It was hypothesized that any preferential alignment of pores will result in non-isotropic transport and so two different geometries were used to measure the transport.

Radial Transport: For testing radial transport, porous hydrogels were polymerized in two different arrangements: 1) the porous hydrogel was polymerized as a single layer to yield a porous tube, and 2) the porous hydrogel was polymerized as the second layer inside a non-porous pHEMA layer. In both cases, the tubular hydrogels were cut along the length of the hydrogel tube (Fig. 3) to yield thin hydrogel sheets which were then used to stamp out circular gels using a ¾“ hole punch. The thickness of these discs was much less than the radius, and so the uptake and release of the proteins could be modeled as 1D diffusion in the axial direction, which corresponded to the radial direction in the original cylindrical geometry. The diffusion equation was used to model transport in both one-layer (punched from the porous tube) and two-layer (punched from the two-layer tubes containing porous gels that were polymerized as the second layer inside the non-porous layer) gels. The boundary conditions were changed to reflect the no-flux boundary condition at the bottom of the two-layer discs and symmetry at the centerline for the one-layer discs.

Fig. 3.

The HEMA hydrogel tube is cut parallel to the central axis to obtain a rectangular hydrogel sheet. The hydrogel sheet is cut to image cross sections of Z and θ-plane by SEM imaging.

Axial Transport: For measuring axial transport, the porous annulus was sandwiched in between an outermost non-porous pHEMA ring and a non-porous core. The hydrogel rod was cut into discs such that the thickness was much smaller than the radius, and thus the uptake and release of the proteins could be modeled as 1D diffusion in the axial direction, which corresponds to the axial direction of the original hydrogel. The non-porous outermost layer and the center prevented radial transport from the curved surface.

BSA (~65 kDa) and hγG (~120 kDa) were used to investigate transport in the porous pHEMA hydrogels as these proteins have similar size and molecular weights to therapeutic proteins. Either BSA or hγG was loaded into the gels by soaking the gels in 3 mL of 2 mg/mL protein solution at 4 °C for 14 days or longer, which was sufficiently long to achieve equilibrium. After equilibrium, the protein-loaded gels were removed from the uptake solution and placed in 8 mL of PBS to measure the release. The protein concentration in release medium was determined by collecting 300 μ L samples at each timepoint and measuring the concentration by the Micro BCA assay. The absorbances of microplate wells were read on a plate reader (Synergy H1 microplate reader, BioTek, Winooski, VT) at 562 nm.

The porous hydrogels were characterized by determining the partition coefficient, K, and diffusivity, ${\mathit{D}}_{\mathit{eff}}$, of both proteins. The partition coefficient is the ratio of the concentration of the protein in the pores of the hydrogel and the concentration of the protein in solution at equilibrium. The mass of protein loaded into the hydrogel can be determined by measuring the initial ($C_{w,0}$) and the final ($C_{w,f}$) concentration of the protein in the loading solution, which can then be divided by the volume of water in the pores ($V_{\mathit{pores}}$ based on Eq. (4) to determine the protein concentration in the pores. It is noted that the protein concentration could also be reported based on the total gel weight, but it is more appropriate to report based on the pore volume because the proteins are not expected to diffuse into the non-porous regions. The protein concentration in the pores of the gel can then be divided by the final protein concentration in the loading solution to obtain the partition coefficient ($K$), i.e.,

$$K=\frac{C_{g,f}}{C_{w,f}}=\frac{V_w\left(C_{w,0}-C_{w,f}\right)}{V_{\mathit{pores}}{C}_{w,f}}$$ (5)

where $V_w$ is the volume of the loading solution. The mass of protein diffusing into the gel is much less than the mass of protein in the solution, and thus the decrease in concentration in the loading solution is very small, which can lead to errors in the calculated value of the partition coefficient by using Eq. (5). To address this issue, the mass of protein in the gel is calculated based on the total mass of protein that is released from the gel during the release study ($M_{r,f}$), i.e.,

$$K=\frac{M_{r,f}}{V_{\mathit{pores}}{C}_{w,f}}$$ (6)

The transport of protein in the pHEMA gel with large aspect ratio can be described by the following one-dimensional diffusion equation

$$\frac{\partial C}{\partial t}=-\frac{\partial }{\partial y}j$$ (7)

where $C$ is the average protein concentration in the gel and $j$ is diffusive flux. The radial transport can be neglected because the diameter of the gels is much larger than the thickness.

Substituting diffusive flux from the Fick’s law $\left(-D_{\mathit{eff}}\frac{\partial C}{\partial y}\right)$, yield the following diffusion equation,

$$\frac{\partial C}{\partial t}=-\frac{\partial }{\partial y}\left(D_{eff}\frac{\partial C}{\partial y}\right).$$ (8)

For the samples without nonporous backing, the governing equation is subjected to the following boundary conditions.

$$\frac{\partial C}{\partial y}(t,y=0)=0$$ (9)

$$C(t,y=H)=0$$ (10)

where $H$ is half thickness of the gel. The first boundary condition (Eq. (9) corresponds to symmetry about the middle plane, and the second boundary condition (Eq. (10) is based on the assumption that the release is occurring under sink conditions.

The rate of the mass of protein released into the external medium can be calculated by calculating the rate of protein released from the gel, i.e.,

$$\frac{d{M}_r}{dt}=-2D_{eff}A\frac{\partial C}{\partial y}(y=H)$$ (11)

where $2A$ is the total area of the hydrogel exposed to the solution.

The initial conditions, i.e., the concentration in the loading solution and the gel at the beginning of the release phase are given by the following equations,

$$C(y,t=0)=C_i$$ (12)

$$C_w(t=0)=0$$ (13)

The differential equation can be solved analytically to obtain the following expression for the fraction of protein released ($F(t)$) into the solution [18],

$$F(t)=1-\sum _{n=0}^{\mathrm{\infty }}\frac{8}{(2n+1{)}^2{\pi }^2}•\mathrm{e}\mathrm{x}\mathrm{p}\left(-(2n+1{)}^2{\pi }^2\frac{D_{eff}t}{4H^2}\right)$$ (14)

where H is the half thickness of the hydrogel. The fractional release $F(t)$ is experimentally determined as the mass of protein released at any time $t$ divided by the cumulative amount released after infinite time.

2.5. SEM imaging

The porous hydrogels were imaged by scanning electron microscopy (SEM) (JEOL 7000 FESEM Akishima, Tokyo, Japan) to obtain the microstructure and to observe if there is a preferential pore alignment in any direction. The hydrogels were cross-sectioned along the Z-plane and the θ-plane for the SEM imaging. The hydrogel samples were placed on an aluminum stub with carbon tape, air dried, and coated with gold for 30 s, prior to imaging. The samples were imaged using an accelerating voltage of 20 kV.

3. Results

3.1. Manufacturing pHEMA tubes and rods with porous annulus

The approach of stepwise polymerization of rings in a rotating glass or polypropylene vials was successfully used to fabricate multiple core-annular structures with at least one porous ring (Figs. 4 and 5). Fig. 4 includes the photograph of the polymerized porous hydrogel inside the glass vial (Fig. 4A) and the photograph of the gel after removal from the vial (Fig. 4B). The photograph in Fig. 4B includes the top view (left) and the front view (right). The photographs in Fig. 4C show the punched-out gels that were used for measuring water content and protein transport in the radial direction.

Fig. 4.

Polymerization and preparation of single layer porous pHEMA hydrogels for measuring transport and water content. A) Top view of pHEMA hydrogel made with 80% water in formulation. B) Top view (left) and side view (right) of the hydrogel tube after removing from the vial. C) Discs punched from the hydrogel. These samples were specifically used for measuring bidirectional, radial transport.

Fig. 5.

Photographs of the gels at different stages of the process for manufacturing core annular hydrogels. A) a clear non-porous pHEMA later is polymerized at the bottom to form a no-flux surface, B) 1st (outer) layer of non-porous concentric pHEMA, and C) 2nd layer of porous concentric pHEMA are polymerized. D) The pHEMA tube with a non-porous outer ring and a porous inner ring is removed from vial, cut along cylindrical axis and arch punched. E) The core-annular tube is filled with non-porous pHEMA to make the pHEMA rod with a clear center and porous annulus. The hydrogel rod is removed from the vial and cut into discs.

Fig. 5 includes photographs at various stages of the process for obtaining gels that were used for measuring the radial transport with no-flux at the bottom surface, and for measuring the axial transport. Fig. 5A–C include the front view on the left and the top view on the right. The Fig. 5B photo shows the result after the first step in which a nonporous outermost ring is polymerized, which is followed by polymerization of the porous ring (Fig. 5C). The photo in Fig. 5D shows the two-ring cylindrical tube after removal from the mold (left) and half of the tube after cutting along an axial plane passing through the center. The photo on the right was obtained after punching a circular disc from the half-tube shown in the middle. The disc on the right of Fig. 5D was used to measure the axial transport with no-flux condition on the bottom surface. Fig. 5E (left) shows the photo of the rod after the tube shown in Fig. 5D was filled with HEMA monomer formulation and polymerized to yield a transparent core. The rod was cut (middle) into thin discs (right) which were used to measure the axial transport. Fig. 5A shows that a thin layer of non-porous pHEMA can be polymerized at the bottom of the vial prior to polymerizing the concentric cylindrical rings. In a few cases, the disc shown in Fig. 5E are cut such that the top surface of the disc included the concentric rings while the bottom surface of the disc was cut through the non-porous layer at the bottom, yielding discs where the diffusion of the biologics occurred only from the top. Thus, we obtained discs for measuring radial transport with protein diffusion from both top and bottom (Fig. 4C) and protein diffusion only from the top (Fig. 5D). We also obtained discs for measuring axial transport from both top and bottom (Fig. 5E) and only from the top (discs cut from the bottom of the rod). These different disc samples are shown in Fig. 6 along with schematics illustrating the direction of transport. The samples with a nonporous backing (unidirectional diffusion) could be useful in applications requiring preferential release in only one direction.

Fig. 6.

Design and photographs of single layer and two-layer hydrogels to measure transport in radial (A) and axial (B) directions.

3.2. Water content and drying dynamics

The ratio of water and HEMA in formulations explored here are listed in Table 1 on volume basis in column 1 and mass basis in column 2. The excess water, that is the mass of water more than the saturation capacity of the pHEMA polymer, is listed in Column 3. Table 1 also includes the measurements for a single layer of porous gels obtained by punching out disc shaped samples from a porous tube. The calculated equilibrium water content obtained from the measured dry and wet weights (Eq. (1) is listed in Column 4. The wet and the dry weights were also used to calculate the volume of pores by using Eq. (4). The ratio of the mass of pores and the total mass of water is listed in Column 5. The EWC values are in reasonable agreement with the percentage of water in the formulation for the high-water content gels, which implies that the entire mass of water in the formulation was trapped in the gel, either as water hydrating the polymer or as water in pores. However, the EWC values are less than the expected values for the gels with 11/9 and 3/2 water: HEMA (v/v) ratio in the formulation. The excess water is expected to form pores so the calculated ratio in Column 3 is expected to match the ratio in Column 5. The measured mass fraction of water in pores is less than the excess water in the formulation particularly for the lower water content gels which suggests that a fraction of the water was not trapped in the gels during polymerization.

The rate of water loss from the porous gels is expected to be faster due to the presence of the large pores. The rate of water loss can thus be used to indirectly assess the presence and connectivity of the pores. The data in Fig. 7 shows that the time required for about 90 % loss of water due to drying is much shorter for the porous gels. The drying rate becomes independent of the water content in the formulations for 55 % or less. Based on the water content data, we expect the gels with 11/9 and 3/2 ratio of water/HEMA in the formulation to contain pores, but the drying data overlaps with the non-porous gels likely because the pores are not connected, and thus even the water in pores must diffuse through the nonporous regions (Fig. 7A). The trends are similar for the two-layer gels in Fig. 7B.

Fig. 7.

Drying dynamics for A) single layer hydrogels of 1.5–1.6 mm thickness and 19.05 mm diameter and multi-concentric layer hydrogels of B) 2.5 and C) 5 mm length. Water/HEMA (v/v) = 2/3 (Pink), 11/9 (black), 3/2 (yellow), 13/7 (green), 7/3 (blue), and 4/1 (orange). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. Transport of proteins

The single layer tube of the porous gels with thickness of approximately 1.5–1.6 mm was punched to a diameter of 19.05 mm to obtain discs that were utilized to measure radial transport. The concentric layer rods were cut into discs of 2.5 mm and 5 mm lengths to measure the axial transport. The gels were loaded with the model proteins and then soaked in fresh buffer to measure the release profiles. The BSA and hγG release profiles are shown in Fig. 8 and Fig. 9. The dynamic concentration in the release medium was fitted to the sink model to determine diffusivities of the two model proteins in the porous hydrogels. Additionally, the total mass of the protein released from the gels was utilized to determine the concentration of the proteins in the pores, which was divided by the protein concentration in the loading solution to determine the partition coefficient. The $D_{eff}$ and K values from both radial and axial samples are listed in Table 2 and plotted in Fig. 10 as a function of the water/HEMA ratio in the formulation. The release studies were conducted with gels of two different thicknesses to ensure that the transport was diffusion controlled.

Fig. 8.

Fig. 8. BSA and hγG release profiles to measure axial transport. Samples were prepared with two different thicknesses. The solid lines are model fits based on diffusion control sink model. Error bars are standard deviations (n = 3). Water/HEMA (v/v) = 11/9 (red), 3/2 (black), 13/7 (orange), 7/3 (blue), and 4/1 (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 9.

Fig. 9. BSA and hγG release profiles to measure radial transport. Samples were prepared with two different thicknesses. The solid lines are model fits based on diffusion control sink model. Error bars are standard deviations (n = 3). Water/HEMA (v/v) = 11/9 (red), 3/2 (black), 13/7 (orange), 7/3 (blue), and 4/1 (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2.

Partition coefficients, K, and diffusivities, ${\mathrm{D}}_{\mathrm{eff}}$, were obtained for porous hydrogel formulations by fitting the release data to the sink-release model.

Protein	Hydrogel Pore Direction	Water/HEMA (v/v)	Partition Coefficient, K	Effective Diffusivity, $D_{eff}=D\frac{\epsilon }{\tau }$ (cm2/s)
BSA	axial	11/9	0.184 ± 0.052	3.21 ± 1.79
		3/2	0.310 ± 0.039	1.53 ± 0.54
		13/7	0.410 ± 0.031	2.6 ± 1.89
		7/3	0.465 ± 0.091	4.27 ± 2.08
		4/1	0.823 ± 0.136	3.51 ± 0.92
	radial	11/9	0.025 ± 0.006	–
		3/2	0.336 ± 0.055	3.15 ± 0.67
		13/7	0.375 ± 0.133	3.42 ± 0.78
		7/3	0.685 ± 0.341	0.73 ± 0.53
		4/1	0.831 ± 0.136	0.83 ± 0.48
hyG	axial	11/9	0.331 ± 0.091	3.41 ± 0.62
		3/2	0.650 ± 0.144	4.28 ± 0.07
		13/7	0.693 ± 0.109	4.44 ± 0.42
		7/3	0.808 ± 0.186	4.32 ± 0.22
		4/1	0.917 ± 0.149	4.96 ± 0.79
	radial	11/9	0.149 ± 0.046	–
		3/2	0.211 ± 0.066	3.18 ± 0.02
		13/7	0.360 ± 0.087	2.69 ± 0.97
		7/3	0.740 ± 0.158	1.99 ± 0.71
		4/1	0.843 ± 0.152	4.75 ± 0.34

Fig. 10.

Partition coefficient, K, and effective diffusivity, $\mathrm{D}\text{eff}$, of pHEMA hydrogels made from formulations with water/HEMA ranging from 11/9 v/v to 4/1 v/v.

The fits between the diffusion control sink model (solid lines in Figs. 8 and 9) and the experimental data are good which suggests that the release is controlled by diffusion of protein through the pores. The mass of the protein released increases with an increase in the water fraction in the formulation due to an increase in the volume of the pores in the polymerized gels. The diffusivities of both BSA and hγG are lower in the radial direction compared to the axial direction. The diffusivities change with water fraction in the formulations, but the trends are not clear partly due to the large error bars. The partition coefficients are higher in the radial gels compared to axial gels, and additionally the partition coefficients increase with an increase in the water fraction in the formulation.

3.4. SEM imaging

The cross sections of the hydrogels made from formulations with water/HEMA ranging from 11/9 to 4/1 were imaged using SEM (Fig. 11). The Z-plane image of the 55 % water formulation (Water/HEMA ratio = 11/9) includes small pores a few microns in size but only towards the inner surface of the tubular hydrogel. The region further away, i.e., towards the periphery does not contain any pores. The pores are only visible in the higher magnification image due to the small size. This again suggests that the centrifugal forces lead to heterogeneities in the local composition, and pores only form where the local water content exceeds the water content of the saturated pHEMA hydrogel. Additionally, the presence of the air–liquid interface at the inner radius may have an impact on the pore formation as well. The θ-plane of the 55 % formulation appears to contain less pores which could be due to alignment of the pores in the z-direction. The formulations with 60 % (Water/HEMA ratio = 3/2) or more water appear to be more uniformly porous though the 60 % formulation appears to include a more porous band somewhere in between the inner and the outer radius. The 65 % (Water/HEMA ratio = 13/7) water formulation shows that the gel has macropores across the entire radial length and the pores appear more uniform in size particularly in the region closer to the inner surface. A few larger pores are present closer to the outer surface. The formulations with 70 (Water/HEMA ratio = 7/3) and 80 % (Water/HEMA ratio = 4/1) water contain very large, interconnected pores, and additionally smaller pores within the pHEMA matrix.

Fig. 11.

SEM images of porous pHEMA hydrogels. The radial direction extends downward in the images. The left columns show cross sections of the z-plane, and the right columns show cross sections of the θ-plane. The lower magnification image is included to show the entire gel, while the higher magnification images are included to present the detailed features.

4. Discussion

4.1. Effect of centrifugal forces on morphology evolution

The rotation of the fluid may impact the morphology of the polymerizing layer by pushing the microgels that form during polymerization radially outwards. This will result in gradients in water content within the porous layer, with the possibility of higher polymer density and lower water content closer to the periphery. To assess this possibility, we consider that the polymerization has led of formation of microgels of radius ${\mathrm{R}}_{\mathrm{m}}$ and density ${\mathrm{\rho }}_{\mathrm{m}}$. The microgels eventually merge to cause gelation [19], but prior to that the microgels could move radially outwards due to the centrifugal forces. The terminal velocity of the microgels in the radial direction can be approximated by equating the net force on the microgel with the viscous drag. Assuming a spherical shape for the microgels, the following equation can be used to calculate the terminal velocity U.

$$U=\frac{2}{9}\frac{R_m^2\left({\rho }_m-{\rho }_f\right)r{\omega }^2}{\mu }$$ (15)

where ${\mathrm{\rho }}_{\mathrm{f}}$ and $\mu$ are the density and viscosity of the fluid surrounding the microgel, respectively, $r$ is the radial location of the microgel, and $\omega$ is the angular velocity of the rotating cylinder.

The density of hydrated pHEMA is reported to range from 1.15–1.34 g/mL [20] and the viscosity of the solution can be approximated by viscosity of water which is 1 cP. The size of the microgels will vary considerably depending on the time of polymerization, and the value of $r$ ranges from zero at the center to about 1 cm at the edge of the cylinder. As an approximation, we estimate the terminal velocity at a radius of 1 cm for a 1 μm radius microgel particle subjected to rotation at 1800 RPM. Using these values, we obtain a radial velocity of about 20 μm/s. Considering the polymerization time of 40 min, the distance traveled by the microgel due to the centrifugal forces could be comparable to the thickness of the gel. It is noted though that the polymerization process is not linear in time, and the fluid gels soon after the microgels form, through aggregation and growth of the microgels. Additionally, the viscosity of the solution near gelation would be significantly higher than the viscosity of water. Inclusion of these factors would reduce the expected radial displacement of the microgels. While the model proposed above is simple, it suggests that the centrifugal forces could impact the microstructure, which is supported by the SEM images and by diffusivity measurements.

4.2. SEM imaging

The SEM images show that all gels have the expected porous microstructure in the annulus because the water content of the formulation is higher than the saturation water content of a pHEMA gel. The pores are a few microns in size for the higher water content (>65 % water content). The pores are smaller and appear to be axially aligned in the low water content formulations, while the pores appear to be interconnected for > 65 % water content. There is evidence of a hierarchy of pores for the high-water content systems including larger interconnected pores and smaller disconnected pores in the polymer matrix.

4.3. Water content and drying dynamics

The water content of the polymerized films is either equal to or less than the water content in the initial formulation. The water content after hydration could be less than the starting value if a fraction of the water in the formulation was not trapped into the hydrogel structure. The total water content can be easily separated into water in the pores and the water in the non-porous regions by assuming that the nonporous region has water content equal to that in a saturated pHEMA gel.

The water loss dynamics show that the rate of water loss increases with increasing water content. The water loss is expected to include an initial phase in which the porous annulus loses water rapidly, followed by a second phase with slower loss from the non-porous regions. The relative amount of water loss in the two phases is related to the amount of water in the large, connected pores (rapid loss) and the nanosized pores in the nonporous region (slow loss). The data shows the expected trends though the two phases of rapid and slower water loss are not clearly separated, possibly due to presence of a fraction of the water water in isolated pores.

4.4. Transport of BSA and hγG

The transport of proteins could be anisotropic if the pores have a preferred alignment. Thus, we decided to measure the axial and radial diffusivity in separate experiments. The axial transport was measured for two different disc thicknesses to ensure that the process was controlled by diffusion through the hydrogel disc. The release duration increased for the thicker samples which suggests that diffusion through the gel is the rate controlling mechanism. The release profiles for both axial and radial transport can be fitted to the diffusion equation which also implies that diffusion in the porous hydrogel is the rate limiting step. The diffusivity and partition coefficient are comparable for both axial and radial samples for the higher water content formulations. However, for the lowest water content formulation, the K value for the radial sample is significantly less than that for the axial sample. We hypothesize that the pores in this sample are axially oriented and thus the pores are not accessible to the proteins in the sample prepared for measuring radial transport. These conclusions agree with the SEM images which also suggest axial orientation for the lower water content sample and interconnected pores without preferred orientation for the high-water content samples. The partition coefficient increases with an increase in the water fraction in the formulation. This increase cannot be due to higher pore volume because the partition coefficient is based on the actual pore volume. The increase in K suggests that some of the pores are isolated in the low water content gels and thus not accessible to the proteins, but still accessible to water. The lower than one partition coefficient also supports this hypothesis because the concentration of the protein in the water in the pores is expected to be equal to that in the loading solution, which would imply a partition coefficient of unity.

The diffusivity values are lower than the free diffusivity of albumin (~6.32 × 10⁻⁷ cm²/s) and free diffusivity of γG (~3.6 × 10⁻⁷ cm²/s) which is expected because the pore size is significantly larger than the molecular size of BSA (~66.5 kDa, ~14 nm) and hγG (~1193 kDa, ~25 nm) [21–23]. The fitted value of the diffusivity is the effective diffusivity which is lower than the true diffusivity in the pores of the gel because of tortuosity $\tau$ and porosity $\epsilon$. The effective diffusivity can be related to the corrected diffusivity in the pores by the following relationship,

$$D_{eff}=D\frac{\epsilon }{\tau }$$ (16)

There are many equations relating porosity, $\mathrm{\epsilon }$, and tortuosity factor, $\mathrm{\tau }$ [24]. We consider the following two relationships to estimate the tortuosity factor and porosity:

$$\tau =\frac{1}{\sqrt{\epsilon }}$$ (17)

$$\tau =2-\epsilon$$ (18)

It is noted that the Bruggeman relation is valid only in situations where the non-aqueous phase is present in a low volume fraction, and it can be represented by random, isotropic spheres [25]. The porous hydrogels appear to be isotropic, at least for the higher water content, but the microstructure is more complex than a packed bed of spherical particles. Thus, the Bruggeman relation is not strictly valid, and so we consider an alternative relationship given by Eq. (18). The porosity was calculated as the ratio of the mass of water in the pores divided by the total gel weight. Based on the calculated porosity and the tortuosity given by Eq. (17) and Eq. (18), we calculated two different values of the true diffusivity. These calculated diffusivities are included in Table 3. The diffusivities D₁ and D₂ are calculated based on using Eq. (17) and Eq. (18), respectively, for the tortuosity factor. Additionally, diffusivity D₃ is calculated based on assuming a tortuosity factor of unity. Only the mean values of the effective diffusivities are used in these calculations.

Table 3.

Diffusivity of proteins in pores of the hydrogel. The porosity was calculated as the ratio of mass of water in pores and the total weight of the porous hydrogel. The diffusivities D₁ and D₂ were calculated by using Eq. (17) and Eq. (18), respectively, for calculating the tortuosity factor. Additionally, diffusivity D3 was calculated by assuming a tortuosity factor of unity.

		$D_{\mathit{eff}}$ (× 10−7) (cm2/s)	ε	τ 1	τ 2	D1(× 10−7) (cm2/s)	D2(× 10−7) (cm2/s)	D3(× 10−7) (cm2/s)
BSA	axial	3.2	0.25	2.01	1.75	25.83	22.55	12.88
		1.5	0.33	1.74	1.67	7.84	7.54	4.52
		2.6	0.42	1.55	1.58	9.71	9.92	6.26
		4.3	0.50	1.42	1.50	12.20	12.93	8.62
		3.5	0.67	1.23	1.33	6.44	7.01	5.26
	radial	–	0.25	2.01	1.75
		3.2	0.33	1.74	1.67	16.73	16.08	9.64
		3.4	0.42	1.55	1.58	12.69	12.97	8.18
		0.7	0.50	1.42	1.50	1.99	2.11	1.40
		0.8	0.67	1.23	1.33	1.47	1.60	1.20
hyG	axial	3.4	0.25	2.01	1.75	27.45	23.96	13.68
		4.3	0.33	1.74	1.67	22.48	21.60	12.95
		4.4	0.42	1.55	1.58	16.43	16.78	10.59
		4.3	0.50	1.42	1.50	12.20	12.93	8.62
		5.0	0.67	1.23	1.33	9.20	10.02	7.51
	radial	–	0.25	2.01	1.75
		3.2	0.33	1.74	1.67	16.73	16.08	9.64
		2.7	0.42	1.55	1.58	10.08	10.30	6.50
		2.0	0.50	1.42	1.50	5.67	6.02	4.01
		4.8	0.67	1.23	1.33	8.83	9.61	7.21

The calculated values of diffusivities are in reasonable agreement with the free diffusivity of albumin (~6.32 × 10⁻⁷ cm²/s) for axial transport except for the 55 % water content gels. There is no significant dependence on the relationship used for estimating the tortuosity factor. However, the diffusivity value calculated by assuming a tortuosity factor of unity is closer to the solution diffusivity, which is consistent with the hypothesis that the axial pores are probably aligned for the formulation with the lowest water content. The calculated diffusivities in the radial direction for BSA are much lower than the free diffusivity which suggests that the centrifugal forces may lead to formation of a low porosity barrier layer near the outer radius. The trends are similar for hγG, though the calculated values are higher than the free diffusivity value of 3.6 × 10⁻⁷ cm²/s. The protein diffusivity in pores cannot be higher than that in solution. The estimation of diffusivity is based on Eqs. 16–18 which involve several assumptions that may not be valid for the porous pHEMA hydrogels.

The partition coefficients are less than unity, which implies that the concentration of the proteins in pores is less than the concentration in the loading solution. This unexpected result suggests that a fraction of the pores may be unconnected to the porous network and these pores contain water but do not contain any proteins. The average concentration of proteins in the pores includes the entire volume of water in both connected and unconnected pores, and so the concentration could be an underestimate, which could be the reason for the lower than one values of the partition coefficients.

5. Conclusion

The proposed method of layer-by-layer polymerization in a rotating cylindrical vial was successfully used to fabricate multiple core-annular structures with at least one porous ring, which could be useful for many applications including delivery of biologics from the porous regions and delivery of small molecules from the non-porous regions. The layers adhered well to prevent any delamination and the pHEMA discs exhibited sufficient structural integrity for the eventual use of these materials in biomedical applications. This method of polymerizing hydrogels in a rotating cylinder could be useful in manufacturing a wide range of structures with rings of varying pore size and chemical properties. The microstructure is isotropic in most cases, although the rotation appears to allow formation of aligned pores under a small window of operating parameters which could be useful in some applications. The SEM images provide some evidence for anisotropy, but clearer evidence was obtained indirectly by measuring protein partition coefficient and diffusivity along radial and axial directions. The dynamics of protein transport from samples cut along different planes was fitted to the diffusion equation to determine diffusivity. The effective diffusivity obtained from the fits, the measured porosity and estimated tortuosity factor were used to estimate the true diffusivity of proteins in the pores. The partition coefficient increased with an increase in water fraction in the formulation suggesting better pore connectivity for higher water fraction. The axial diffusivity values calculated by accounting for porosity and tortuosity factor were comparable to the solution diffusivity for BSA, but higher than the solution diffusivity for hγG. The radial diffusivity values were lower than the axial diffusivity, possibly due to formation of a low porosity barrier layer due to the centrifugal forces. The SEM images also suggest variation in pore structure along the radial direction likely due to centrifugal forces. A simple model based on calculation of the terminal velocity supports the hypothesis that the centrifugal forces could squeeze out water resulting in variation in microstructure and hence physical properties in the radial direction. The effect of centrifugation would be enhanced at higher rotation speed which could potentially be a useful design variable if the heterogenous structure is desirable for some applications. Results from this study provide another approach in controlling microstructure of pHEMA monoliths which could be useful in many applications including drug delivery. The pore directionality could be useful in applications involving growth of neural tissue [12]. However, the degree of pore alignment may not be adequate, and additional studies are necessary to obtain a full three-dimensional mapping of the structure and to increase alignment by further exploring the parameter space. Additionally, biocompatibility of the materials must be established by cell-based assays and in vivo studies to explore potential in drug delivery applications.

HIGHLIGHTS.

• A novel approach is developed for manufacturing core annular hydrogels with a porous annulus.

• The diffusivities of model proteins are measured in radial and axial directions.

• The partition coefficients of model proteins are measured.

• The porous hydrogels are imaged along different planes.

• The SEM images and diffusivities and partition coefficients are used to assess pore connectivity and homogeneity of microstructure.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jcis.2024.07.105.

Data availability

Data will be made available on request.