Cross-linked nanofilms for tunable permeability control in a composite microdomain system

Abstract

Fabrication of microcapsule based composite materials with precise control over diffusion is desirable for applications in anti-corrosive agent release, drug delivery, and biosensing. Self-assembled layer-by-layer (LbL) nanofilms may be used to control analyte flux in these advanced materials because they allow accurate manipulation of interface properties on the nanoscale. The transport of a model analyte was evaluated across glutaraldehyde cross-linked PAH [poly(allylamine hydrochloride)]/PSS [poly(sodium 4-styrenesulfonate)] nanofilm constructs to determine the potential of these bilayers to precisely control small-molecule diffusion. Measurements of glucose permeation rates across nanofilms deposited on planar porous substrates revealed that glutaraldehyde-mediated cross-linking drastically decreased transport across PAH-containing bilayers. Additionally, we found that analyte permeation rates can be finely tuned by controlling the degree of intralayer and interlayer PAH cross-linking. To realize the practical application of these nanofilms in micro-scale flux-based devices, the planar multilayer scheme was used to line glucose-sensing microdomains entrapped in alginate matrices. These glucose sensing nanocomposite hydrogels exhibited sensitivity and analytical range that are adjustable depending upon the characteristics of the multilayers; cross-linking of the nanofilm lining the microdomains to limit glucose diffusion resulted in an extension of the analytical range by ≈249% and a decrease in the corresponding sensitivity by ≈85%. This demonstration of control over small-molecule diffusion in microdomain-populated hydrogel materials opens the possibility to use these devices for multifunctional purposes, including biosensing and controlled release of encapsulated species such as drugs, coloring/flavoring agents, anti-corrosives, and other active molecules.

---

Introduction

Microcapsule technology plays a critical role in numerous applications for biotechnology, food, agriculture, cosmetic, and pharmaceutical industries.^1–3 A capsule shell acts as a barrier between the payload material and the external microenvironment; functioning to modulate specific surface interactions and to selectively control permeability.^4–6 An abundance of technology exists for the encapsulation of materials via entrapment, adsorption, or diffusional loading,^7–12 where the preferred method for a given application is chosen based on the properties of the cargo being encapsulated and the desired function of the capsule. The interaction of the colloid-cargo composite with its environment may then be further modified by tuning the physical and the chemical properties of the capsule exterior.

Layer-by-layer (LbL) self-assembly provides a versatile route to modulate the surface properties of an assortment of template materials in a precise manner using a very simple process that may nevertheless result in a highly complex multifunctional coating.¹³ The deposition of these nanofilms is driven by a combination of chemical and physical interactions such as electrostatics, covalent bonding, hydrogen bonding, or bioaffinity;¹⁴ these may be adjusted by engineering the deposition conditions and the multilayer composition itself. Fortuitously, a large set of molecules are available for use in constructing these materials. The ability to achieve nanoscale architectural precision makes LbL uniquely suited to construct coatings that selectively control molecule permeation through the capsule wall for applications such as drug delivery,¹⁵ self-healing smart coatings,¹⁶ or biosensing.¹⁷ The nanofilm coating can be used to match burst/sustained release profiles of active agents (drugs, anticorrosive substances) or to gate diffusion into the capsule interior for selective biomolecular sensing. For the special case of enzymatic bioreactors, the nanofilm coating can be used as a diffusion-limiting barrier for finely tuning substrate flux and hence substrate consumption. In enzymatic biosensors, this controls the reaction-kinetics of the sensor chemistry which manifests as changes in the sensor analytical range and sensitivity.¹⁸

LbL has been employed to make nanofilms for gas barrier coatings and separation membranes by the inclusion of high-aspect-ratio materials such as inorganic clay platelets (e.g. LAPONITE®, vermiculite, or montmorrilonite) and graphene.^19–22 These platelets are assembled to align parallel to each other within the nanofilm layers (normal to the direction of analyte flux) to create a highly tortuous path for molecules diffusing through the layers. However, the use of platelet LbL to regulate molecular permeation rates has been more effective on planar films than on colloidal templates. There have been a few examples of capsules fabricated with platelet-based nanofilm layers^23–26 but the reduction in observed diffusion was substantially lower compared to the planar equivalent. An innovative approach to reduce permeability of capsule walls involves the introduction of preformed nanoparticles or in situ nucleation and growth of nanoparticles within the polymer shell.^27–29 These methods produce structurally stable microcapsules with stimuli responsive burst release capabilities; however, they still face a challenge in achieving reproducible and controlled reduction of permeability.²⁷

In addition to the compositional route, permeability through soft material nanofilms may also be controlled by changing the number of layers or the thickness and structure of these layers using different deposition conditions. Electrostatic deposition of polyelectrolytes can be controlled with a high degree of reproducibility and precision by controlling pH and salt concentration.^30–32 These methods have been shown to be effective on planar films^33,34 but face challenges for replication in capsule formation due to tedious process flow and quality control issues ascribed to colloidal destabilization.³⁵

UV, thermal, or chemical cross-linking of polyelectrolyte multilayer (PEM) films have also been demonstrated as an effective means to affect film permeability while also varying structural properties. UV cross-linked diazoresin/poly(styrene sulfonate) microcapsules that are extremely stable to solvents and osmotic pressure fluctuations, have exhibited reduced permeability to enzymes.^36,37 Carbodiimide-activated or thermally-induced cross-linking between amine and carboxylic acid groups present in PEMs have also been reported to produce capsules with enhanced physical stability and decreased permeability.^38–41 Another promising approach to fabricate high diffusion barrier nanofilms is to cross-link amine containing polyelectrolytes using glutaraldehyde. Grunlan's group reported that glutaraldehyde cross-linking of an amine-containing polyelectrolyte polyethylenimine (PEI) could reduce the diffusion of oxygen across planar nanofilms below detection limits.^42,43 It has been established that glutaraldehyde cross-linked planar nanofilms can limit the diffusion of monovalent and divalent cations;⁴⁴ glutaraldehyde cross-linked polymeric microcapsules have also been shown to possess better chemical, thermal, and mechanical stability, with reduced permeability to larger macromolecules (460–2000 kDa FITC–dextran) compared to non-cross-linked microcapsules made from similar amine-containing PEMs.^45–47

Although there has been considerable progress in LbL technology to forge capsules with enhanced stability and stimuli-sensitive switchable permeability characteristics,^48,49 advancement related to finely tuning diffusion of analytes in/out of polymeric capsules has been scarce and limited to methods involving irreversible enzymatic degradation of the microcapsule shell.^50,51 This is somewhat surprising given that capsule permeability control is especially vital in payload release and biosensing applications that demand precise diffusion control to achieve desired functionality.^17,52

In this study we explored the use of glutaraldehyde cross-linked PAH [poly(allylamine hydrochloride)]/PSS [poly(sodium 4-styrenesulfonate)] bilayers to provide accurate and reproducible control over the diffusion of a model analyte glucose. The controlled reduction in glucose permeation rate is examined by varying the number of cross-linked PAH/PSS bilayers deposited on planar Anopore membranes. The cross-linked nanofilm constructs were translated onto sacrificial calcium carbonate templates to construct microcapsule-based enzymatic reactors for glucose oxidation, which were also functionalized with a phosphor for optical readout. These micron-sized glucose sensors were ultimately dispersed in an alginate hydrogel matrix. Flux-based enzymatic microcapsule sensors immobilized in a matrix act in an ensemble fashion, thus any change to substrate permeation properties of individual microcapsules manifests as substantial shifts in bulk sensor properties. This provides a powerful tool to probe substrate diffusion across the microcapsule wall towards the engineering of a tunable biosensing hydrogel device, which may be used as a fully implantable soft-material enzymatic sensor to monitor physiologically relevant biomarkers.

Experimental section

Chemicals

Sodium carbonate (Na₂CO₃), calcium chloride (CaCl₂), poly(sodium 4-styrenesulfonate) (PSS, average M_w 70 000 Da), poly(diallyldimethylammonium chloride) (PDADMAC, average M_w 100 000–200 000 Da), poly(allylamine hydrochloride) (PAH, average M_w 15 000 Da), glutaraldehyde solution (grade II, 25% in H₂O), alginic acid sodium salt from brown algae (100–300 cP, 2% at 25 °C), and buffer salts (NaHCO₃, MES, C₂H₃NaO₂, and TRIS) were obtained from Sigma and were used as received without further purification. Glucose oxidase (GOx) from Aspergillus niger (257 U mg⁻¹, BBI solutions) and Pd-meso-tetra (4-carboxyphenyl) porphine (PdTCPP, Frontier Scientific) suspended in DMSO (10 mM) solution were used in all experiments. Glucose used for all sensor response studies was obtained from Macron Fine Chemicals.

Layer-by-layer assembly on planar substrate

Nanofilms were deposited on Whatman Reserved (R) Anopore inorganic aluminum oxide membrane filters (25 mm diameter, 0.02 μm pore size) placed in an open-face filter holder (Pall Co.). The open face of the filter membrane was exposed to oppositely charged polyelectrolyte solutions 20 mg mL⁻¹ PDADMAC (pH 8), 20 mg mL⁻¹ PAH (pH 8), 20 mg mL⁻¹ PSS (pH 7.2) alternately with wash steps (5 mM NaHCO₃) between each polyelectrolyte exposure step. A primer coating consisting of [PSS]–[PDADMAC/PSS]₅ was deposited to achieve complete surface coverage⁵³ before depositing the desired number of PAH/PSS bilayers (Scheme 1A). After depositing the target number of PAH/PSS bilayers, the nanofilms were exposed to 0.1 M glutaraldehyde solution for 30 minutes to cross-link the amine groups on PAH (Scheme 3A). Excess glutaraldehyde was removed by washing the nanofilms with 5 mM NaHCO₃ (pH 7.2). All polyelectrolyte solutions were prepared in 5 mM NaHCO₃.

Scheme 1 (A) [PAH/PSS]_n nanofilms and the primer coating ([PSS]–[PDADMAC/PSS]₅) deposited on Anopore membrane, (B) [PSS/PAH/PSS/PDADMAC]_n−1–[PAH/PSS] nanofilms and the primer coating ([PSS]–[PDADMAC/PSS]₅) deposited on Anopore membrane.

To fabricate interspersed cross-linked PAH layers (Scheme 1B), a PSS/PDADMAC layer was deposited between successive PAH/PSS bilayers. Cross-linking of the interspersed layers was performed using the same protocol to cross-link non-interspersed PAH/PSS bilayers. It should be noted that when depositing PAH/PSS bilayers wash steps were performed using 5 mM NaHCO₃ (pH 7.2), and while depositing PDADMAC/PSS bilayers 5 mM NaHCO₃ (pH 8) was used for the washing steps, to ensure that the polyelectrolytes were sufficiently ionized while deposition.

Permeability measurements

Nanofilms fabricated on Anopore membrane filters, were placed between the feed and the permeate chambers of a side-by-side diffusion cell (Permegear Inc.). The feed chamber was filled with 7 mL of 5 mM NaHCO₃ (pH 7.2) containing 1 g L⁻¹ glucose and the permeate chamber was filled with 7 mL of 5 mM NaHCO₃ (pH 7.2). Samples were collected from both the feed and the permeate sides at regular time intervals, and the glucose concentration of the samples were measured using a YSI biochemistry analyzer (2700 Select). The rate of increase in concentration over time in the permeate chamber (dC/dt) was calculated by linear regression for the different nanofilm formulations.

Nanofilm coated microparticles with encapsulated sensing chemistry

PdTCPP and GOx containing calcium carbonate (CaCO₃) microparticles were synthesized using the co-precipitation method,⁵⁴ with minor modifications. Briefly, 200 μL of 10 mM PdTCPP solution was added to 8 mL of 0.2 M Na₂CO₃ containing 64 mg of GOx under continuous stirring (800 RPM). After 5 min, 8 mL of 0.2 M CaCl₂ was added rapidly and the reaction was allowed to continue for 10 min. Nanofilms were deposited on the PdTCPP and GOx containing microparticles, by alternately exposing the particles to polyelectrolyte solutions (20 mg mL⁻¹ PDADMAC (pH 8), 20 mg mL⁻¹ PAH (pH 8), 20 mg mL⁻¹ PSS (pH 7.2)) with intermediate wash steps. The wash solutions used were the same as describe above for making nanofilms on planar substrates. After depositing the desired number of nanofilms, 3.3 mg of nanofilm-coated microparticles was suspended in 10 mL, 3 M glutaraldehyde solution for 30 min. Excess glutaraldehyde was removed by washing the microparticles with 5 mM NaHCO₃ (pH 7.2). The amount of glutaraldehyde used for the microcapsules was based on the ratio of [nanofilm surface area] : [mass of glutaraldehyde]. Hollow microcapsules were made from the PEM lined, payload containing microparticles, by exposing them to 0.2 M sodium acetate buffer at pH 5.1.

Microporous alginate composite (MPAC) hydrogels

MPAC hydrogels were made using a modified version of the protocol described by Roberts et al.⁵⁵ Briefly, microcapsules fabricated using 3.3 mg of PEM coated CaCO₃ microparticles were suspended in 75 μL of deionized water, non-coated CaCO₃ microparticles (25 μL of 33.3 mg mL⁻¹), 3% alginate solution (200 μL), and MES (100 μL, 0.5 M, pH 6.1) were mixed to make a slow-gelling hydrogel precursor. The precursor was then poured between two glass slides separated by a 1.5 mm Teflon spacer, and allowed to gel for 24 hours.

Characterization

Confocal fluorescence and differential interference contrast (DIC) microscopy images were captured using an inverted laser spinning-disk confocal microscope (Olympus IX81, Yokogawa CSU-X1). Hydrogel samples excited at 488 nm were viewed through 40× and 100× oil immersion objectives. Images were analyzed using ImageJ 1.48 v software.

SEM images of nanofilm coated microparticles, microcapsules and MPAC hydrogels were captured using a JEOL 7500 scanning electron microscope. A diluted sample of either nanofilm coated microparticles or microcapsules was placed on a silica wafer and was allowed to dry overnight. To prepare a hydrogel sample for SEM imaging, a 5 mm × 5 mm hydrogel was placed on a silica wafer and dried overnight. All samples were sputter-coated with 2.5 nm of palladium/platinum before imaging. SEM images were analyzed using ImageJ 1.48 v software.

Sensor response testing

Hydrogel discs (4 mm radius) were excised from the hydrogel slab using a biopsy punch. Each sample was placed in a liquid flow cell (Scheme 2), and changes in lifetime with varied glucose and oxygen concentrations were recorded using a custom time-domain lifetime measurement system as described elsewhere.⁵⁶

Scheme 2 Schematic diagram of (A) enzyme/dye containing microdomain bound by nanofilm coating, (B) a section of the microdomain containing hydrogel and (C) the experimental setup consisting of the flow through cell and the time-domain lifetime measurement system used to evaluate sensor response.

The response to oxygen was evaluated by flowing buffer having varied dissolved oxygen concentrations (0–206.8 μM). The dissolved oxygen concentration of 10 mM TRIS (pH 7.2) containing 10 mM CaCl₂ was changed by purging air and nitrogen with mass flow controllers (type 1179A, MKS).

To determine the response to glucose, solutions containing different concentrations of glucose (0–400 mg dL⁻¹) dissolved in 10 mM TRIS (pH 7.2) with 10 mM CaCl₂ were flowed over the hydrogel samples. The response parameters were calculated from each of the obtained response curves. The limit of detection (LOD) was estimated by calculating the glucose concentration at which the response was 10% higher than the response at 0 mg dL⁻¹ glucose concentration. Similarly, the maximum differentiable glucose concentration (MDGC) was estimated by calculating the glucose concentration at which the response was 10% lower than response at 400 mg dL⁻¹ glucose concentration. The range of the sensor was defined as R = MDGC − LOD, while the sensitivity was defined as the percent change in the maximum and minimum response observed per unit range of the sensor.

Results and discussion

The effect of glutaraldehyde cross-linking of PAH/PSS bilayers on the diffusion of glucose was evaluated by measuring the rate of glucose diffusion across PAH/PSS nanofilm constructs deposited on Anopore membranes. PAH/PSS bilayers were deposited on the primer coating (PSS–[PDADMAC/PSS]₅) to fabricate PSS–[PDADMAC/PSS]₅–[PAH/PSS]_n multilayers, where n was varied from 1 to 10. The glucose diffusion across different nanofilm formulations was evaluated by calculating the linear slope of the glucose concentration change dC/dt (where C is the concentration of glucose (g L⁻¹) and t is time (hours)) on the permeate side of the diffusion cell. The data presented in Fig. 1 shows the decrease in dC/dt for both the cross-linked and non-cross-linked PAH/PSS bilayers as the number of layers is increased. It is quite clear that the decrease in dC/dt is much more pronounced in the case of the cross-linked PAH/PSS bilayers. Specifically, the glucose permeation rate through non-cross-linked PAH/PSS bilayers decreases by ≈46% when n is increased from 3 to 9, whereas the dC/dt of cross-linked PAH/PSS bilayers decreases by ≈98% for the same number of bilayers. It is evident from this that the cross-linked films more effectively prohibit the free diffusion of glucose compared to the native nanofilm constructs. For the same number of bilayers, cross-linking significantly decreases the dC/dt across the multilayer constructs. Comparing the glucose permeation rates through cross-linked and non-cross-linked PEMs when n = 3, 5, and 9, the dC/dt of glucose through the cross-linked PEMs was found to be less than the corresponding non-cross-linked PEMs by ≈71%, ≈88% and ≈99%, respectively. Extrapolating fitted data (ESI, Fig. S1†), revealed that 124 non-cross-linked PAH/PSS bilayers would be required to achieve the glucose permeation rate obtained when using 9 bilayers of cross-linked PAH/PSS. The first five cross-linked bilayers decrease the glucose permeation rate drastically; however, further increase in the number of cross-linked bilayers has a proportionally lower effect. The dC/dt values for glucose through the cross-linked PEMs change by ≈39% when comparing diffusion rates between n = 1 and n = 2, whereas the decrease was only ≈15% when comparing n = 5 and n = 6. This is believed to simply be a function of the percentage of total film thickness represented by the additional layers; in the first case, doubling the thickness of cross-linked layers is a more substantial increase in thickness than adding one layer to five already deposited (15–20% increase).

Fig. 1 The glucose permeation rate (dC/dt) through PAH/PSS bilayers composed of cross-linked PSS–[PDADMAC/PSS]₅–[PAH/PSS]_n (blue ◊), cross-linked PSS–[PDADMAC/PSS]₅–[PSS/PAH/PSS/PDADMAC]_n−1–[PAH/PSS] (green □), non-cross-linked PSS–[PDADMAC/PSS]₅–[PAH/PSS]_n (red ○), and the primer coating PSS–[PDADMAC/PSS]₅ where n = 0 (purple Δ). Error bars represent 95% confidence intervals for three separate nanofilm constructs.

The drastic decrease in permeability to glucose when the –NH₂ groups of the PAH layer are cross-linked (Scheme 3A) in the presence of glutaraldehyde may be attributed to the decrease in free volume present in the PEMs.^42,43 Apart from cross-linking the –NH₂ groups of PAH in an individual PAH layer, the possibility also exists to have cross-linked –NH₂ groups present in successive PAH layers due to the interpenetrating nature of LbL assembled PEMs.⁵⁷ This led us to question the extent of interlayer and intralayer cross-linking and the corresponding influence on glucose diffusion. To investigate this, nanofilms were designed with a spacer bilayer PSS/PDADMAC introduced between successive PAH/PSS bilayers (Scheme 3B). The spacer containing PEMs fabricated are represented by PSS–[PDADMAC/PSS]₅–[PSS/PAH/PSS/PDADMAC]_n−1–[PAH/PSS]. The spacer containing cross-linked PAH/PSS bilayers were found to limit glucose diffusion to a greater extent than non-cross-linked nanofilms but less than cross-linked PAH/PSS nanofilms without spacer bilayers (Fig. 1). Introduction of the PSS/PDADMAC spacer bilayer allowed glucose to diffuse through the nanofilm coatings more freely as compared to glucose diffusion across non-spacer containing successively cross-linked films with the same total number of PAH layers. For n = 3, 5, and 9 the dC/dt of cross-linked P PSS–[PDADMAC/PSS]₅–[PSS/PAH/PSS/PDADMAC]_n−1–[PAH/PSS] was 2.5, 4.5, and 81 times greater, respectively, than their cross-linked counterparts without the spacers (PSS–[PDADMAC/PSS]₅–[PAH/PSS]_n). It is important to recognize that the metrics used are of glucose permeation rate and are not normalized by film thickness. Thus, even though the cross-linked spacer-containing PEMs contain more layers and are overall thicker, the total glucose diffusion barrier is less than the cross-linked PEMs without the spacer bilayers. This increase in dC/dt is ascribed to the reduced interlayer cross-linking by the introduced spacer bilayer that decreases the interpenetration of neighboring PAH layers.

Scheme 3 (A) Cross-linking of poly(allylamine hydrochloride) (PAH) by glutaraldehyde. (B) Glutaraldehyde cross-linked PAH/PSS nanofilm constructed without and with a PDADMAC/PSS spacer bilayer.

Once the glucose permeation rate effects were determined with the planar multilayers, the nanofilm architectures were translated to microparticle template coatings as a way to fabricate microcapsule based optical glucose sensors. The expectation was that the varying glucose permeation rate of the different nanofilms would result in correspondingly shifted glucose sensor behavior (sensitivity and response range). In these enzymatic glucose sensors, GOx catalyzes the oxidation of glucose in the presence of molecular oxygen, ultimately producing gluconic acid (glucose + O₂ + glucose oxidase + H₂O → gluconic acid + H₂O₂). If the enzyme is in excess, and the reaction is not limited by oxygen supply then the decrease in molecular oxygen is proportional to the amount of glucose oxidized. Thus measuring the decrease in molecular oxygen using an oxygen sensitive phosphorescent dye (PdTCPP) enables the indirect measurement of glucose concentrations. The design described, places severe demands on transport control. In a generic enzymatic biosensor, a diffusion-limiting coating is used to restrict the amount of substrate entering the enzyme module per unit time, effectively making the system substrate-transport limited rather than reaction-kinetics limited.⁵⁸ It is important to appreciate that the overall range and the sensitivity of an enzymatic biosensor can be tailored by controlling the rate of substrate diffusion.⁵⁹

The cargo containing microparticles and capsules were first characterized by optical and electron microscopy to confirm that the desired products were produced in the fabrication process. SEM images of cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ coated microparticles (Fig. 2A) revealed the spherical morphology of the CaCO₃ microparticles bound by fuzzy PEMs. The average diameter of the coated particles is estimated to be about 3.6 μm. Elemental analysis carried out using an EDS system attached to the SEM (ESI, Fig. S2†) confirmed that CaCO₃ was no longer present in the microcapsules after dissolution of the CaCO₃ core. The hollow microcapsules (Fig. 2B) appear as collapsed structures, which occurs when specimens are dried during sample preparation for SEM. The SEM image of the cross-linked PAH/PSS coated microparticles are similar to those reported by Volodkin et al.³ However, the cross-linked PAH/PSS microcapsules appear to have capsule walls that collapse less than the cross-linked PAH/PSS microcapsules fabricated by Tong et al.⁴⁵ and Wang et al.⁴⁷ This minor difference is attributed to the greater number of cross-linked bilayers and increase in overall coverage by the PAH/PSS bilayers compared to prior studies. Images of the microcapsules entrapped in MPAC hydrogels (Fig. 2C) reveal a wrinkled microporated morphology for the hydrogel when dried, which is characteristic of microcapsule-containing MPAC gels.⁵⁵

Fig. 2 SEM images of sputter-coated (A) cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ coated microparticle (B) cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ bound microcapsule (C) MPAC containing cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ bound microdomains. Scale bars correspond to 1 μm.

Differential interference contrast (DIC) micrographs (Fig. 3A and C) revealed that the cross-linked microcapsules containing GOx and PdTCPP were uniformly distributed when immobilized in MPAC hydrogels. The average diameter of the microcapsules entangled in the MPAC gels was estimated using ImageJ software to be 4.0 ± 0.4 μm, which is comparable to the size of the corresponding nanofilm-coated microparticle templates viewed by SEM. The slightly larger estimated size of the microcapsules viewed by DIC images may be attributed to the hydrated state of the samples relative to the dried samples used for SEM. In the confocal micrographs (Fig. 3B and D) the phosphorescent dye PdTCPP is clearly localized in the spherical microdomains, which is expected as the dye is confined within the nanofilm boundary. The MPAC hydrogels were imaged under ambient (Fig. 3D) and reduced oxygen (Fig. 3E) conditions, while keeping the camera settings constant to obtain a ratio image (Fig. 3F). An intensity ratio (reduced oxygen : ambient oxygen) corresponding to >1 depicts that the PdTCPP intensity has increased under reduced oxygen. While not absolutely quantitative, these images show that the dye remains localized within the capsules and retains its response to oxygen after glutaraldehyde cross-linking and immobilization in the hydrogel.

Fig. 3 DIC images of MPAC hydrogels containing cross-linked microcapsules ([PDADMAC/PSS]₅–[PAH/PSS]₉) at (A) 40× magnification and (C) 100× magnification. Fluorescence microscopy images of microcapsule containing MPAC gels at (B) 40× magnification, (D) 100× magnification (ambient oxygen), (E) 100× magnification (reduced oxygen) and the intensity ratio of the MPACs under reduced oxygen to ambient oxygen (F). Scale bars correspond to 10 μm. Color coded scale corresponds to intensity ratios of PdTCPP containing microdomains (reduced oxygen : ambient oxygen).

Alginate matrices containing GOx/PdTCPP microdomains can potentially function as enzymatic biosensors. GOx contained in these microdomains oxidizes glucose, reducing local oxygen concentrations proportional to the glucose permeation rate. We envisioned to tune the analytical range and sensitivity of the hydrogel based sensor by decreasing glucose permeation rate into the microdomains, eventually making it a truly glucose-diffusion limited system. However, it is imperative to understand that glucose-limited behavior is achieved only if the influx of oxygen is much higher than or equivalent to the influx of glucose; requiring the nanofilms to preferentially decrease glucose diffusion rates over oxygen diffusion rates.

The response of the MPAC hydrogels containing PdTCPP and GOx loaded microdomains to changing oxygen concentrations was evaluated to ascertain whether cross-linking of PAH/PSS bilayers affects oxygen diffusion. As a control, the oxygen sensor response of MPAC hydrogels containing non-cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ microcapsules was also determined. Fig. 4 represents the Stern–Volmer normalized lifetime (lifetime at zero oxygen concentration divided by lifetime at given oxygen level) against varying oxygen concentrations (non-normalized quenching curves of lifetime versus oxygen concentration may be seen in ESI, Fig. S3†). K_SV values were calculated using the Stern–Volmer equation τ₀/τ = 1 + K_SV[O₂], and found to be 0.030 ± 0.002 μM⁻¹ on an average. All the hydrogel samples having different nanofilm compositions showed a high sensitivity to oxygen at levels less than 100 μM, and a decreased sensitivity at higher oxygen concentrations, characteristic of oxygen-sensitive palladium porphyrin dyes.^60–62 The similar oxygen response characteristics show that cross-linking of the nanofilms in the hydrogel does not affect the kinetics of oxygen diffusion.

Fig. 4 Stern–Volmer quenching curves. Lifetime (normalized to the lifetime at zero oxygen concentration) against varying oxygen concentrations for MPAC hydrogels containing microdomains bound by different nanofilm architectures. The cross-linked nanofilm architectures are represented by [PDADMAC/PSS]₅–[PAH/PSS]_n where n = 3 (red □), n = 5 (purple Δ), n = 7 (green ○), n = 9 (blue ◊) and non-cross-linked nanofilm architecture [PDADMAC/PSS]₅–[PAH/PSS]₉ (black ×). Error bars represent 95% confidence intervals for three separate MPAC hydrogels. The dashed lines are provided only as a guide to the eyes.

The glucose sensing characteristics of MPACs containing non-cross-linked [PDADMAC/PSS]₅–[PAH/PSS]_n nanofilm bounded microdomains were examined to establish that cross-linking of PAH/PSS bilayers was necessary to alter the sensor characteristics significantly. The phosphorescence lifetime of MPAC hydrogels containing PdTCPP and GOx loaded microdomains was recorded as the materials were exposed to buffer solutions containing varied concentrations of glucose (0–400 mg dL⁻¹). The normalized sensor response curves for MPACs containing non-cross-linked [PDADMAC/PSS]₅–[PAH/PSS]_n nanofilm bounded microdomains are illustrated in Fig. 5A, where the change in lifetime is calculated relative to the lifetime obtained at maximum glucose concentration (refer to ESI, Fig. S4† for non-normalized lifetime response curves). A coherent trend was observed in terms of sensitivity and range of the sensors as the number of bilayers was increased. This was anticipated since altering the transport properties of the microcapsule directly influences the sensor characteristics.⁶³

Fig. 5 (A) Sensor response curves of MPACS containing non-cross-linked [PDADMAC/PSS]₅–[PAH/PSS]_n nanofilm bounded microdomains when n = 3 (red □), n = 5 (purple Δ), n = 7 (green ○) or n = 9 (blue ◊). (B) Sensor response curves of MPACs containing cross-linked [PDADMAC/PSS]₅–[PAH/PSS]_n nanofilm bounded microdomains when n = 3 (red □), n = 5 (purple Δ), n = 7 (green ○) or n = 9 (blue ◊). Error bars represent 95% confidence intervals for three separate MPAC hydrogels. The dashed lines are provided only as a guide to the eyes.

With an increase in the number of PAH/PSS bilayers from n = 3 to n = 9, the analytical range increases by ≈106% while the sensitivity over the same range decreases by ≈59%. This inverse relationship between range and sensitivity is characteristic of flux-based sensors.⁵⁹ Table 1 summarizes the sensor parameters for non-cross-linked microcapsule-containing hydrogels. The decrease in the flux of glucose diffusing into the microdomains as the number of bilayers are increased accounts for the changed sensor response characteristics. Although the analytical range increases as the number of bilayers are increased, the analytical range achieved for the materials using non-cross-linked nanofilms still does not encompass the usual operational range for glucose sensors (0–400 mg mL⁻¹).⁶⁴ All of the sensor formulations made using non-cross-linked PEMs were highly sensitive to glucose changes in the hypoglycemic range (<70 mg dL⁻¹),⁶⁵ but they failed to detect glucose concentration changes above 98 mg dL⁻¹. This suggests that the glucose flux into the microdomains is too high, which either overwhelms the enzyme or consumes oxygen too fast. These findings indicate that the diffusion of glucose into the discrete microdomain sensors should be decreased further, to make the composite hydrogel system glucose-diffusion limited.

Table 1 Calculated sensor parameters for MPACs containing non-cross-linked and cross-linked [PDADMAC/PSS]₅–[PAH/PSS]_n nanofilm-bounded microdomains^a

	LOD^b (mg dL^−1)	MDGC^c (mg dL^−1)	Range^d (mg dL^−1)	Sensitivity/range (% per mg dL^−1)
a In each case, data from three separate MPAC hydrogels were used to calculate mean values (95% confidence).b LOD, limit of detection.c MDGC, maximum differentiable glucose concentration.d Range, MDGC − LOD.
Non-cross-linked
[PAH/PSS]3	12.0 ± 6.8	54.4 ± 3.2	40.7 ± 8.8	13.0 ± 4.5
[PAH/PSS]5	15.6 ± 0.1	54.6 ± 4.9	39.8 ± 4.5	13.2 ± 0.6
[PAH/PSS]7	11.5 ± 2.3	62.5 ± 2.1	51.0 ± 4.3	9.4 ± 1.6
[PAH/PSS]9	14.4 ± 2.0	98.2 ± 7.4	83.8 ± 5.5	5.3 ± 0.5
Cross-linked
[PAH/PSS]3	14.3 ± 5.0	65.4 ± 7.3	52.2 ± 11.1	12.5 ± 3.4
[PAH/PSS]5	24.3 ± 4.8	170.8 ± 23.4	168.0 ± 13.4	2.0 ± 0.4
[PAH/PSS]7	32.9 ± 3.7	296.4 ± 28.9	271.4 ± 23.0	0.9 ± 0.1
[PAH/PSS]9	33.2 ± 9.7	321.2 ± 8.2	292.7 ± 9.5	0.8 ± 0.1

---

Having demonstrated that non-cross-linked nanofilm containing sensor formulations are not effective in controlling sensor dynamics considerably, the sensor response of formulations containing glutaraldehyde cross-linked microdomains were evaluated (refer to ESI, Fig. S5† for non-normalized response curves). As expected, with an increase in the number of cross-linked PAH/PSS bilayers in the nanocomposite hydrogels, the analytical range of the sensors increases and the sensitivity decreases (Fig. 5B). The analytical range increases by ≈461% while the sensitivity over the range decreases by ≈94%, as n is increased from 3 to 9. The analytical range and sensitivity of MPAC hydrogels containing cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉ was found to be ≈227% more and ≈85% less respectively than MPAC hydrogels containing non-cross-linked [PDADMAC/PSS]₅–[PAH/PSS]₉. Thus, cross-linking of PAH/PSS bilayers is crucial to significantly decrease dC/dt of glucose into the sensor and alter sensor response parameters considerably. However, as discussed previously, the rate of change of dC/dt decreases with the increase in the number of cross-linked bilayers that consequently affects how the sensor characteristics change as number of bilayers are increased. The analytic range increases by ≈8% and the sensitivity over the range decreases by ≈8% when the number of cross-linked bilayers is increased from 7 to 9. This change is insignificant compared to the change in response between sensors fabricated from 3 cross-linked bilayers and 5 cross-linked bilayers. Sensor figures of merit for cross-linked microcapsule-containing hydrogels are summarized in Table 1.

Conclusion

This work demonstrates the utility of glutaraldehyde cross-linked LbL nanofilms for precise diffusion control and the application of this technology for the development of glucose sensors. The diffusion control of cross-linked films was first investigated using planar films to develop a method capable of providing a large diffusion barrier, while also considering the design constraints needed for eventual use in engineering microcapsules. Cross-linking just a few bilayers of nanofilms reduced glucose permeation rate to levels, which could only be theoretically achieved using hundreds of non-cross-linked bilayers. The effect of cross-linking was further probed by the introduction of spacer bilayers as it was expected that the nanofilm cross-linking was a combination of interlayer and intralayer cross-linking due to the fuzzy nature of the interpenetrating PEMs. Spacer bilayers lessen the degree of diffusion restriction, which could be useful for achieving more specific permeation rates or for the diffusion modulation of larger molecules. The cross-linking scheme was adapted for colloidal templates using glucose as a model analyte and microcapsules containing encapsulated glucose sensor chemistry. Cross-linked nanofilms preferentially restricted glucose movement over oxygen transport across the microcapsule walls, which manifested as an extension in the analytical range of the hydrogel based composite biosensors.

We anticipate that similar results will be observed by cross-linking other amine containing components (e.g. peptides, enzymes, polysaccharides) which are commonly used to fabricate LbL nanofilms. This technology can be used to tailor the functionality of flux-based biosensors for a variety of small-molecule analytes (i.e. urea, lactate, etc.) and we expect it can also play a role in controlled-release systems for applications such as drug delivery and self-healing coatings. Having the ability to differentially control the release or intake of molecules, opens the possibility to formulate multiplexed and multifunctional nanocomposite devices. Future work will focus on the development of these unique microcapsule enabled hydrogel-based devices for multi-analyte sensing and theranostic applications.

Acknowledgements

This material is based upon work supported by the National Science Foundation under Grant No. 1066928 (CBET), 1258696 (CMMI), and 1403002 (CBET). We also thank Swayoma Banerjee for her insight and expertise in confocal imaging. We are grateful to Dr Luis Rene Garcia and Dr Wayne Versaw for allowing us access to the Olympus IX81. Use of the TAMU Materials Characterization Facility is also acknowledged.

References

1. Y. Ma, W. F. Dong, M. A. Hempenius, H. Möhwald and G. J. Vancso, Angew. Chem., Int. Ed., 2007, 46, 1702–1705.
2. B. G. Stubbe, K. Gevaert, S. Derveaux, K. Braeckmans, B. G. De Geest, M. Goethals, J. Vandekerckhove, J. Demeester and S. C. De Smedt, Adv. Funct. Mater., 2008, 18, 1624–1631.
3. D. V. Volodkin, A. I. Petrov, M. Prevot and G. B. Sukhorukov, Langmuir, 2004, 20, 3398–3406.
4. C. E. Mora-Huertas, H. Fessi and A. Elaissari, Int. J. Pharm., 2010, 385, 113–142.
5. M. Samadzadeh, S. H. Boura, M. Peikari, S. M. Kasiriha and A. Ashrafi, Prog. Org. Coat., 2010, 68, 159–164.
6. A. A. Antipov and G. B. Sukhorukov, Adv. Colloid Interface Sci., 2004, 111, 49–61.
7. M. Yang, J. Ma, C. Zhang, Z. Yang and Y. Lu, Angew. Chem., 2005, 117, 6885–6888.
8. D. Chen and M. Jiang, Acc. Chem. Res., 2005, 38, 494–502.
9. A. Sunder, M. Krämer, R. Hanselmann, R. Mülhaupt and H. Frey, Angew. Chem., Int. Ed., 1999, 38, 3552–3555.
10. A. Manna, T. Imae, K. Aoi, M. Okada and T. Yogo, Chem. Mater., 2001, 13, 1674–1681.
11. J. Ruths, F. Essler, G. Decher and H. Riegler, Langmuir, 2000, 16, 8871–8878.
12. J. Guo, W. Yang, Y. Deng, C. Wang and S. Fu, Small, 2005, 1, 737–743.
13. G. B. Sukhorukov, E. Donath, H. Lichtenfeld, E. Knippel, M. Knippel, A. Budde and H. Möhwald, Colloids Surf., A, 1998, 137, 253–266.
14. X. Zhang, H. Chen and H. Zhang, Chem. Commun., 2007, 1395–1405.
15. K. Ariga, Y. M. Lvov, K. Kawakami, Q. Ji and J. P. Hill, Adv. Drug Delivery Rev., 2011, 63, 762–771.
16. D. G. Shchukin and H. Mohwald, Chem. Commun., 2011, 47, 8730–8739.
17. M. McShane and D. Ritter, J. Mater. Chem., 2010, 20, 8189–8193.
18. E. W. Stein, P. S. Grant, H. Zhu and M. J. McShane, Anal. Chem., 2007, 79, 1339–1348.
19. F. Xiang, P. Tzeng, J. S. Sawyer, O. Regev and J. C. Grunlan, ACS Appl. Mater. Interfaces, 2014, 6, 6040–6048.
20. T. Guin, M. Krecker, D. A. Hagen and J. C. Grunlan, Langmuir, 2014, 30, 7057–7060.
21. W.-S. Jang, I. Rawson and J. C. Grunlan, Thin Solid Films, 2008, 516, 4819–4825.
22. A. A. Gokhale, J. Lu, N. J. Parker, A. P. Izbicki, O. Sanyal and I. Lee, J. Colloid Interface Sci., 2013, 409, 219–226.
23. J. Hong, K. Char and B.-S. Kim, J. Phys. Chem. Lett., 2010, 1, 3442–3445.
24. R. Kurapati and A. M. Raichur, Chem. Commun., 2012, 48, 6013–6015.
25. J. Hickey, N. A. D. Burke and H. D. H. Stöver, J. Membr. Sci., 2011, 369, 68–76.
26. R. A. Caruso, A. Susha and F. Caruso, Chem. Mater., 2001, 13, 400–409.
27. P. R. Gil, L. L. del Mercato, P. del_Pino, A. Muñoz_Javier and W. J. Parak, Nano Today, 2008, 3, 12–21.
28. K. Zhang, Q. Wang, H. Meng, M. Wang, W. Wu and J. Chen, Particuology, 2014, 14, 12–18.
29. B. Sun, S. A. Mutch, R. M. Lorenz and D. T. Chiu, Langmuir, 2005, 21, 10763–10769.
30. J. D. Mendelsohn, C. J. Barrett, V. V. Chan, A. J. Pal, A. M. Mayes and M. F. Rubner, Langmuir, 2000, 16, 5017–5023.
31. L. Chang, X. Kong, F. Wang, L. Wang and J. Shen, Thin Solid Films, 2008, 516, 2125–2129.
32. R. A. McAloney, M. Sinyor, V. Dudnik and M. C. Goh, Langmuir, 2001, 17, 6655–6663.
33. X. Liu and M. L. Bruening, Chem. Mater., 2004, 16, 351–357.
34. A. M. Balachandra, J. Dai and M. L. Bruening, Macromolecules, 2002, 35, 3171–3178.
35. H. Gao, D. Wen and G. B. Sukhorukov, J. Mater. Chem. B, 2015, 3, 1888–1897.
36. H. Zhu and M. J. McShane, Langmuir, 2005, 21, 424–430.
37. I. Pastoriza-Santos, B. Schöler and F. Caruso, Adv. Funct. Mater., 2001, 11, 122–128.
38. P. Schuetz and F. Caruso, Adv. Funct. Mater., 2003, 13, 929–937.
39. V. Kozlovskaya and S. A. Sukhishvili, Macromolecules, 2006, 39, 5569–5572.
40. H. Lee, Y. Jeong and T. G. Park, Biomacromolecules, 2007, 8, 3705–3711.
41. W. Tong and C. Gao, Polym. Adv. Technol., 2005, 16, 827–833.
42. Y.-H. Yang, M. Haile, Y. T. Park, F. A. Malek and J. C. Grunlan, Macromolecules, 2011, 44, 1450–1459.
43. Y.-H. Yang, L. Bolling, M. Haile and J. C. Grunlan, RSC Adv., 2012, 2, 12355–12363.
44. K. L. Cho, A. J. Hill, F. Caruso and S. E. Kentish, Adv. Mater., 2015, 27, 2791–2796.
45. W. Tong, C. Gao and H. Möhwald, Chem. Mater., 2005, 17, 4610–4616.
46. W. Tong, C. Gao and H. Möhwald, Macromolecules, 2006, 39, 335–340.
47. F. Wang, J. Feng and C. Gao, Colloid Polym. Sci., 2008, 286, 951–957.
48. A. A. Antipov, G. B. Sukhorukov and H. Möhwald, Langmuir, 2003, 19, 2444–2448.
49. A. A. Antipov, G. B. Sukhorukov, S. Leporatti, I. L. Radtchenko, E. Donath and H. Möhwald, Colloids Surf., A, 2002, 198–200, 535–541.
50. M. V. Lomova, A. I. Brichkina, M. V. Kiryukhin, E. N. Vasina, A. M. Pavlov, D. A. Gorin, G. B. Sukhorukov and M. N. Antipina, ACS Appl. Mater. Interfaces, 2015, 7, 11732–11740.
51. I. Marchenko, A. Yashchenok, T. Borodina, T. Bukreeva, M. Konrad, H. Möhwald and A. Skirtach, J. Controlled Release, 2012, 162, 599–605.
52. A. A. Antipov, G. B. Sukhorukov, E. Donath and H. Möhwald, J. Phys. Chem. B, 2001, 105, 2281–2284.
53. J. Jeon, V. Panchagnula, J. Pan and A. V. Dobrynin, Langmuir, 2006, 22, 4629–4637.
54. A. I. Petrov, D. V. Volodkin and G. B. Sukhorukov, Biotechnol. Prog., 2005, 21, 918–925.
55. J. R. Roberts, D. W. Ritter and M. J. McShane, J. Mater. Chem. B, 2013, 1, 3195–3201.
56. B. B. Collier and M. J. McShane, Proc IEEE SENSORS, 2011, 943–946; B. B. Collier and M. J. McShane, Anal. Chem., 2012, 4725–4731.
57. G. Decher, Science, 1997, 277, 1232–1237.
58. G. S. Wilson and Y. Hu, Chem. Rev., 2000, 100, 2693–2704.
59. D. A. Gough, J. Y. Lucisano and P. H. S. Tse, Anal. Chem., 1985, 57, 2351–2357.
60. D. B. Papkovsky, G. V. Ponomarev, W. Trettnak and P. O'Leary, Anal. Chem., 1995, 67, 4112–4117.
61. C.-S. Chu and Y.-L. Lo, Sens. Actuators, B, 2008, 134, 711–717.
62. A. Biswas, A. T. Nagaraja and M. J. McShane, ACS Appl. Mater. Interfaces, 2014, 6, 21193–21201.
63. H. Zhu, R. Srivastava, J. Q. Brown and M. J. McShane, Bioconjugate Chem., 2005, 16, 1451–1458.
64. H. Koschwanez, F. Yap, B. Klitzman and W. Reichert, J. Biomed. Mater. Res., Part A, 2008, 87, 792–807.
65. E. R. Seaquist, J. Anderson, B. Childs, P. Cryer, S. Dagogo-Jack, L. Fish, S. R. Heller, H. Rodriguez, J. Rosenzweig and R. Vigersky, Diabetes Care, 2013, 36, 1384–1395.

---

Footnotes

† Electronic supplementary information (ESI) available: Energy dispersive X-ray spectroscopy (EDS) spectra, non-normalized lifetime data, and additional data for glucose permeation rates when using non-cross-linked bilayers. See DOI: 10.1039/c6ra13507b

‡ The manuscript was written with contributions of all authors. All authors have given approval to the final version of the manuscript.

---