Reusable Dual‐Photopolymerized Holographic Glucose Sensors

Diabetes is well established as a widespread, incurable, and fatal disease with glucose monitoring and tight glycaemic control vital for effective illness prevention and management. Hydrogel‐based holographic sensors serve as a low‐cost and label‐free colorimetric sensing platform, directly identifiable by the naked eye and spectroscopy for quantitative monitoring. Herein, a cost‐effective and reusable holographic glucose sensor is developed via single pulse UV‐induced dual‐photopolymerization of boronic acid functionalized hydrogels for point‐of‐care (POC) diagnosis. Computational modeling of holographic sensors response is conducted following Braggs law alongside the study of fabrication parameter optimization and sensor swelling dynamics. Fabrication conditions, responsive and interference hydrogel compositions of holographic sensors are investigated to improve response time, sensitivity in urine (13.03 nm mmol−1 L−1), limit of detection (0.06 mmol L−1), and reusability. Photolithographic patterning of hydrogel‐based holographic sensors permits the inscription of additional information into the sensors for qualitative measurement. Selectivity, reversibility, and continuous monitoring of urine samples are conducted over a physiological glucose concentration range (0.0–9.4 mmol L−1) to demonstrate the viability for diabetic risk identification. The simple incorporation of glucose sensors in a reusable urinary analysis prototype is validated in human urine, showing potential for POC to reduce patient dependency on invasive diabetic monitoring procedures.


Introduction
In modern life, minimal exercise and poor diet linked to sedentary work and processed food are expediting the pandemic growth of diabetes worldwide. [1] It is estimated that 536.6 million people (10.5%) of the global population currently have diabetes with this figure expected to increase to 783.2 million (12.2%) by 2045. [2] Healthcare providers have moved to a proactive approach in the prevention of diabetes launching the national health service Diabetes Prevention Program to reduce extensive costs and detriment to patient health. [3] These schemes utilize novel technologies to encourage patients at risk of diabetes to eat healthier, increase activity, and monitor for biomarkers of prediabetes. [4] Diabetic patients suffer with variations in blood glucose concentration which can exceed normal levels (10 mmol L −1 ), the kidneys are unable to reabsorb all of the glucose at these high levels which leads to excretion via the urine. A urinary glucose concentration greater than 0.8 mmol L −1 is defined as glucosuria and is a key indicator of kidney damage or type 1 and type 2 diabetes. To facilitate tight glycaemic control, patients utilize a small incision to the skin for invasive monitoring, which over time causes significant discomfort and a high risk of infection. [5] The drive toward the non-invasive techniques explore the utilization of readily available biofluids, such as saliva, [6] tear fluid, [7] sweat, [8] and urine [9] removing the requirement for invasive procedures and reducing the burden of repetitive skin incisions. Saliva, tear fluid, and sweat contain significantly lower glucose concentrations and are prone to erroneous readings due to contamination and irritation during sampling. [10] Urinalysis provides the most easily-accessible, accurate biofluid with multiple daily reading opportunities to facilitate continuous monitoring and is regularly utilized by medical professionals to determine diabetic risk.
Commercialized glucose monitoring techniques utilize either electrochemical or colorimetric monitoring for glucose determination in urine. [11] A range of urinary glucose monitoring platforms have been established such as electrochemical cells, fluorescent, and photonic sensors (Table S1, Supporting Information). [12] Electrochemical cells are highly valuable patient monitoring tools with rapid sensitive measurements.
However, the reported electrochemical urinary glucose sensors are dominated by the use of enzymatic binding, which can be costly and prone to denaturing in harsh environments. [11] In addition, electrochemical cells require a power source and are prone to signal drift over time, requiring patients to recharge or recalibrate their device to maintain accuracy, research is currently ongoing to mitigate these effects. [13] Hence there is a desire to move toward optical systems for their power-free monitoring and minimal requirement for recalibration. [14] Fluorescent sensors have been applied in biosensing, [15] pharmacology, [16] and environmental monitoring. [17] However, fluorescent sensors are susceptible to photobleaching, irreversible measurements, require complex fabrication techniques, and UV excitation to stimulate emission, limiting their potential in POC devices. [18] The current gold standard of urinary glucose analysis employs single-use dipsticks, which rely on enzymatic detection of glucose. [19] Dipsticks are single use and provide a semi-quantitative colorimetric result, comparable against a reference chart. Photonic sensors such as holographic sensors (HSs) provide a reusable platform for quantitative colorimetric analysis and are a focal point of research due to the simplicity of fabrication, [20] readout, and removal of the requirement for complex analytical equipment. [21] HSs, comprise of a subwavelength variation in refractive index (RI) produces a tuneable Bragg mirror structure reflecting specific wavelengths dependent on the internal fringe spacing. [22] They can be developed into a reusable photonic sensing platform providing qualitative and quantitative analysis of urine glucose concentration. Due to their cost-effective, reversible, chemical functionalization, resistance to electromagnetic interference, and real-time monitoring, [23] HSs have been demonstrated for quantification of pressure, [24] ions, [25] and drug molecules. [26] Conventional HSs are based on inorganic nanoparticles are suffered from the toxicity of chemicals and tedious fabrication in POC biosensing applications. [27] Compared with nanoparticle-based systems, dual photopolymerized (DP) HSs are comprised entirely of biocompatible materials presenting a beneficial advantage to the development of POC devices. Nanoparticle systems rely upon continuous wave lasers were employed to record holographic gratings, requiring extended exposure times and potentially vibration diminishing equipment to ensure reliable recording. The use of a pulsed single flash polymerization simplifies the previously complex fabrication process into a simple 4-step procedure without the requirement for complex exposure setups.
Herein, a single-flash UV dual photopolymerization was developed to produce a stable, nanoparticle-free, and reusable hydrogel-based holographic glucose sensor that can continuously, and quantitatively monitor glucose levels in urine. Computational modeling was applied to inform the optimization of sensor recording and provide an approximation of sensor swelling characteristics. DP HS consists of periodically alternating layers of two polymeric hydrogel materials with different RIs, named as the responsive matrix (RM) and interference layer (IL). The formulations of the RM hydrogel with low RI and the IL hydrogel with high RI were optimized to improve the peak stability and device sensitivity. Experiments investigate the fabrication parameters that influence sensor performance, to expand the current understanding of doubly UV photopolymerized HS characteristics. Demonstrations of sensor photopatterning with text define sensors applicability to label-free sensing. Sensor properties such as angular dependency, response time, reversibility, and sensitivity were characterized. Glucose sensors are validated for POC analysis in artificial urine (AU) over a physiological glucose range, and a simple device is demonstrated in human urine for a reusable urinalysis apparatus.

Fabrication of Holographic Glucose Sensor
In this work, RM is a low cross-linking density hydrogel prepared by photo cross-linking of acrylamide and glucose-specific 3-acrylamidophenylboronic acid (3-AAPB) with 1.0 mol% methylenebisacrylamide (MBA) as a cross-linker (Figure 1a-i). The IL contains a high cross-linking density hydrogel formed by polymerization of acrylamide and 29 mol% 1,4-bis(acryloyl) piperazine (BAP) (Figure 1a-ii). By increasing cross-linking density in IL, the doubly polymerized fringe has a higher RI than RM, leading to the production of a Bragg reflector. Herein, the cross-linker is varied between the RM (MBA) and IL (BAP) were selected to permit higher cross-linking densities in the IL monomer solution, due to the low solubility of MBA. 3-Acrylamidophenyl boronic acid (3-AAPB) was used to functionalize the RM hydrogel, of which the boronic acid group permits covalent binding of glucose reversibly. [28] Boronic acids function as a Lewis acid and have been well documented across a range of glucose sensing platforms. [29] 3-AAPB is trigonal and can react with water to form an anionic tetrahedral boronate ( Figure 1b). [30] 1,2 or 1,3 cis-diols of carbohydrates act as Lewis bases which can bind with boronic acid to form a 5 or 6 membered cyclic boronate ester. [31] Boronic acids can bind through the trigonal and tetrahedral form, determined through several factors such as pH, temperature, and concentration. [30] DP HSs were generated by a 4-step fabrication process ( Figure 1c). 1) A hot washed RM film was saturated with aerated IL monomer solutions and allowed to soak 5 min before 2) being dried for 10 min and 3) exposed to a coherent UV laser pulse, and finally 4) washed to remove unpolymerized material. HSs were recorded utilizing a single pulse from an Nd:YAG laser (355 nm, 5 ns, 30 µs delay). Unreacted monomers were removed, and the hydrogel thickness pre-exposure was reduced via drying, upon hydration sensors expand producing a replay wavelength in the visible region. Hydrogels are positioned at an angle in exposure (5⁰) to separate holographic signal from specular reflection interference. The pulsed laser was used in replacement of a continuous wave laser that can reduce the exposure time to nanoseconds, removing requirements for vibration-diminishing equipment. Post-exposure immediately washing removes unpolymerized material (deionized water, 25 °C, >2 h), producing the two distinct hydrogel regions of the sensor. The developed sensors vary volumetric expansion relative to glucose concentration, due to glucose association and dissociation with 3-AAPB forming a polymerized ionic charge. The simplification of fabrication offers potential for mass production. However, the high cross-linking density in IL www.afm-journal.de www.advancedsciencenews.com is anticipated to stifle the expansion (Figure 1d). The diffraction of the DP hologram corresponds to Braggs law, where d corresponds to the grating period, n is an integer, λ is the reflected wavelength and θ is the angle of illumination from the normal (Figure 1e). The boronic acid-binding of glucose initiates a volumetric change of the hydrogel film, expanding the d spacing of the hologram and thus changing the reflection wavelength.

Computational Modelling
Computational modelling is a vital component of HS research, permitting rationalization, and visualization of the funda-mental mechanisms within HS. Previous techniques utilized silver nanoparticle photopatterning to produce a RI change, TEM imaging allowed visualization of the internal swelling response of fringes. [32] However, DP HS layers are difficult to be observed experimentally as RI variation is achieved by the hydrogel composition and cross-linking density variation. Computational models applied here have been designed to represent the general DP HS system, however, RI variation between the IL has been elevated to correspond to expected RI variation in the acrylamide glucose system. Herein, a study of the fabrication procedures affected on sensor response is conducted, examining the correlation between the layer number and the hydrogel thickness at the point of recording. Alongside this  analysis of increasing layer number producing a reduction in sensor response is conducted to provide an approximation of the correlation between layer number and sensor performance.
The DP HSs were recorded through the constructive and destructive interference between incident and reflected beams of a coherent laser light. Initial replay wavelength was determined by the constructive interference regions initiating polymerization nodes correlating to the recording wavelength. If the photosensitive hydrogel is swollen when exposed to the laser pulse, no expansion occurs upon hydration producing an initial replay wavelength comparable to the recording wavelength (355 nm), outside of the visible spectrum. Pre-exposure swelling was determined by IL solvents, humidity, and temperature. It is hypothesized that when the hydrogel thickness increases, more ILs and cross-linking density across the hydrogel produce, and therefore reducing the sensitivity. The developed modeling examines the variation of hydrogel thickness upon exposure, experimentally associated with pre-exposure drying. The holographic model employed was adapted from the COMSOL "single bit hologram." The geometry demonstrated in Figure 2a-i was designed to investigate the effect of photosensitive hydrogel thickness on the IL number utilizing COMSOL. The external RI was set to 1.33, and the hydrogel matrix was set to 1.46 to correspond to RM. Counter propagating incident and reflected beams propagate from the two ports, with a physics-controlled trigonal meshing applied. The layer number was determined through the effective RI variation along the 2D cut line across the hydrogel. Thickness of layers at 6, 9, 12, 15, and 18 µm were examined with layer numbers at 51, 76, 101, 128, and 151 respectively ( Figure 2b; Figure S1, Supporting Information), with the electric field density plots highlighted in (Figure 2c; Figure S2, Supporting Information). Figure 2d depicts the COMSOL computational model utilized to study fringe spacing variation corresponding to optimized sensor response. [33] The model consists of 3 separate materials, the RM layer (RI = 1.40), IL region (RM = 1.60), and the exterior medium to represent the aqueous medium the hologram is situated in (RI = 1. 33  Graph of layer number recorded as a function of hydrogel thickness; c) Electric field intensity plots for i) 18 µm and ii) 6 µm of film thicknesses in exposure; d) Overview of the i) geometry designed to model the expansion and response of HS, defining the specific regions such as RI and IL, and ii) demonstration of the trigonal meshing applied, with the magnified section demonstrating meshing geometry in more detail; e) Simulated reflection wavelength maxima obtained for hydrogel sensors with layer numbers of 5, 10, 15, 20, and 25 with expansion iterations of 5, 4, 3, 2, and 1% relative to the initial wavelengths; f) Simulated reflection spectra obtained from HS with 15 layers and an initial layer spacing of RM = 290, IL = 145. of the sensor under the broadband light, an input port is situated on the left of the model, and a receiving output port on the right. Parameter sweeps were conducted to examine the effect of layer thickness variation on the reflected wavelength. It is hypothesized that increasing the layer number recorded produces both a reduction in initial replay wavelength and diminishing sensitivity, experimentally correlated to film thickness prior to exposure. An investigation was designed to simulate this theory, whereby sensors with 5, 10, 15, 20, and 25 layers were given initial thicknesses at 310, 300, 290, 280, and 270 nm for RM, and 155, 150, 145, 140, and 135 nm for IL, respectively. Expansion iterations for both RM and IL were set to incremental increases of 5%, 4%, 3%, 2%, and 1% for both RM and IL relative to the initial thicknesses ( Figure 2e; Table S2, Supporting Information). The thickness ratio of RM:IL was set to 2:1 correlating with previously simulated DP HS modeling. [33] The simulation conducted aims to simulate the restricted effect predicted to occur with a greater layer number. Layer numbers chosen here were to facilitate simple running of the computational model, values above 30 layers produced impractical simulation times. Simulated reflection spectra obtained permit an approximation of relative wavelength change expected with different degrees of expansion. Figure 2f shows the spectra obtained from 15 layers with a 3% expansion rate per iteration. As expected, the lower the layer number the greater relative wavelength change observed, the simulation can provide an approximation of experimental sensor swelling from the relative wavelength shift.

Chemical Optimization of Hydrogel Formulations
A water bath optical setup integrated with a thermocouple as the temperature control was used for HS analysis (Figure 3a). Sensors were placed into a cuvette with a consistent analyte volume (1.5 mL) and illuminated with a broadband light source. Glucose concentration was altered via successive analyte removal and injection of a glucose solution (150 µL, 20.0 mmol L −1 ) producing glucose concentrations of 0.0, 2.0, 3.8, 5.4, 6.9, 8.2, and 9.4 mmol L −1 , and optical response was collected via fiber optics. The RM monomer solution was comprised of AM (82 mol%), 3-AAPB (15 mol%), MBA (1.0 mol%), and HMPP (1.3 mol%) with a dilution of 0.67 g mL −1 (dimethyl sulfoxide (DMSO):H 2 O, 1:1, v/v). IL monomer solutions were comprised of AM (61 mol%), BAP (29 mol%), and HMPP (10 mol%) with a dilution of 0.173 g mL −1 (DMSO:H 2 O, 1:1, v/v). A temperature range of 25-40 °C was set to demonstrate sensor applicability to cover the physiological temperature. The rate of response was expedited as temperature increased from 25 to 40 °C (Figure 3b). As the temperature increased, the initial replay wavelength increased marginally due to thermal expansion of the hydrogel, while the total wavelength shift remained constant in response to 2.0 mmol L −1 of glucose. Figure 3c highlights wavelength changes after 20 min, with 40 °C demonstrating the fastest response rate at 0.47 nm min −1 . 37 °C as the physiological temperature in human body was employed for biosensors throughout the following experiments.
3-AAPB acts as a functional co-monomer in hydrogel materials, binding glucose to produce boronate anions, thus increasing Donnan osmotic pressure and stimulating swelling of hydrogels. [34] The hydrophilic/hydrophobic balance of the hydrogel can affect the swelling of HS. 3-AAPB contains a phenyl ring which increases the hydrophobicity of the hydrogel. Therefore, there is expected to be a point of optimum 3-AAPB concentration, where the number of ionizable groups and the hydrophilic/hydrophobic balance are achieved. To optimize the sensitivity to glucose, HSs were prepared by varying 3-AAPB concentrations (10,15,20,25, and 30 mol%). Figure 3d-i demonstrates normalized Bragg reflection spectra for an optimized sensor with a 3-AAPB concentration at 20 mol% to physiological glucose concentrations (0.0-9.4 mmol L −1 ). Figure 3d-ii demonstrates the optical response observed through a smartphone camera via the bifurcated receiving fiber to avoid interference from external light sources. The observed structural color changes improved qualitative identification of glucose concentrations. Figure 3e demonstrates the relative wavelength changes observed for each sensor as the glucose concentration varied from 0.0 to 9.4 mmol L −1 . The total shift was 36.0 nm for 20 mol% of 3-AAPB concentration, where the maximum ionizable functional groups were incorporated while maintaining the hydrogel hydrophilicity (Figure 3f). Sensor shifting range has been significantly improved with 20 mol% sensors with a 36.0 nm wavelength shift in response to a glucose concentration of 9.4 mmol L −1 , demonstrating a 44% increase compared to 24.9 nm for 15 mol%.
The DP HSs are reliant on the high cross-linking density in IL to achieve the efficient RI variation in the Bragg reflector. However, due to an inability to obtain a monitorable grating with lower cross-linking densities in the IL, an investigation of the effect of IL monomer concentration was conducted to determine the correlation between IL dilution and sensitivity. The IL monomer solution contained AM (61 mol%), BAP (29 mol%), and HMPP (10 mol%). Increasing the cross-linking density in the film could reduce sensitivity, and reduction of the crosslinking density in the monomer solutions did not produce viable holograms for data collection. Therefore, IL monomer concentration (0.50, 0.60, 0.75, 1.00, and 1.50 mmol L −1 ) was varied to observe the effect on glucose sensor performance. Figure 3g highlights the normalized Bragg reflection spectra for optimal IL monomer solution concentration (0.60 mmol L −1 ) to glucose concentration variation (0.0-9.4 mmol L −1 ).
The total wavelength shifts increase as IL concentration decreases (Figure 3h). The optimized IL concentration was determined to be 0.60 mmol L −1 . Total wavelength shift has been extended to 77.4 nm (±1.5 nm) for 0.0-9.4 mmol L −1 , with an initial response of 23.4 nm to a 2.0 mmol L −1 of glucose concentration. The sensors of 0.50 mmol L −1 demonstrated a greater glucose sensitivity (98.1 nm, [glucose]: 0.0-9.4 mmol L −1 ). However, the error margin for measurements ±5.2 nm rendered a relative low sensor reliability. The optical response range has been extended from green (≈558 nm) to red (≈636 nm), permitting a more easily distinguishable response amenable to quantitative monitoring.

Physical Optimization of Sensor Fabrication
Film thickness has been proved to effectively alter response time and diffraction efficiency of DP HS. [35] Controlling the www.advancedsciencenews.com volume of RM monomer solution from 25, 50, 75, 100 to 125 µL effectively could vary film thickness from 6.0, 7.5, 11.0, 13.0 to 18.5 µm (Figure 4a). The optical changes of HS were examined at 1-min intervals after immersing in glucose solution  indicates that the rate-determining step is glucose binding to the boronic acid moiety. An unexpected effect of decreasing the sensor thickness was an increase in sensitivity, with films of 6 µm demonstrating a 21% increase wavelength shift (29.3 nm) compared to 18.5 µm (24.1 nm) to a 2 mmol L −1 glucose concentration. The sensitivity of the 6 µm sensor increased, with a 90 nm wavelength shift in response to a 9.4 mmol L −1 glucose concentration (Figure 4b-inset). Theoretically, the sensitivity is not affected by the thickness change when all variables, such as monomer solutions, drying time, and pulsed laser were maintained. It was hypothesized that the drying process of IL monomer solutions in RM was more effective in thinner films when exposing to the hot-air flow at the same time ( Figure S3, Supporting Information).
The IL monomers were dissolved in the mixture of H 2 O and DMSO, with boiling points of 100 and 189 °C respectively, selected to ensure monomers solvation. To achieve superior drying control, the RM soaked by IL monomer solutions was transferred to an oven at 110 °C for the major removal of water. Sensors were prepared by producing a 6-µm thick film with the oven drying time varied from 2 to 10 min to examine the sensitivity by injection of the glucose solution at 2.0 mmol L −1 (Figure 4d). As the extent of drying increased, the initial replay wavelength shifted to a longer wavelength from blue to red alongside an increase of total wavelength shift (Figure 4e; Figure S1, Supporting Information). Longer drying time leads to the removal of more solvent, reducing film thickness before pulsed laser exposure. Polymerization was induced by high intensity constructive interference between recording incident and reflected beams. The spacing of these regions is independent of the film thickness, therefore the number of polymerized fringes is increased for thicker films. Consequently, the more ILs recorded within the sensor increase overall crosslinking density. Initial replay wavelength is set by the degree of hydrogel expansion from recording thickness, with initial recorded fringe spacing considered to be ≈λ/2, as the hydrogel expands the RM fringe spacing increases producing a red shifting of hologram reflection. Sensor initial replay wavelength can be directly correlated to sensitivity. A potential explanation could be due to the number of IL fringes recorded upon exposure in agreement with computational modelling previously demonstrated (Figure 2a-c). The phenomenon was seldomly reported, hence the detailed investigations on the drying effect strengthen the understanding of DP HS and offer a simple way to adjust the sensor performance. An optimum 8-min drying time was selected due to the comparable sensitivity to 10 min, however improved visual differentiation between 0.0-2.0 mmol L −1 permits qualitative measurement of glucose levels correlated with glycosuria (Figure 4f). Data collected experimentally here confirmed the computational modelling that by increasing the layer number recorded, an overall decrease in sensitivity was observed. Further examination of this effect should be conducted in future to determine the exact level of restriction caused by IL polymerization.

Holographic Patterning and Angle Dependency
To facilitate simple understanding of the HS application, photopatterning of secondary information can be recorded onto the hologram such as lettering or images, with a measured diffraction efficiency of 31% ( Figure S5, Supporting Information). Figure 5a highlights the photolithographic patterning at a nanoscale using a mirrored macroscale "Glu" characters, communicating sensor purpose to patients. Shaping the mirrored section limits the areas capable of reflecting incident laser light, and therefore the regions are capable to produce the constructive and destructive interference required for holo-graphic recording. For sensors to be multiplexed in a device, several holograms functionalized to alternative biomarkers are required in close proximity, inscription of simple information will reduce the risk of human error. By controlling drying time, initial replay wavelength can be tuned to achieve a qualitative or quantitative sensor dependent on the specific requirements of treatment. Respective sensors were immersed in glucose free PBS solutions, after equilibration the analyte was changed to a 10 mmol L −1 glucose solution to observe the response. As can be seen from Figure 5a (Movies S1 and S2, Supporting Information) sensors demonstrate a clear colorimetric change to the presence of glucose, with the sensor dried for a shorter time (2 min) presenting a green to orange change and the sensor dried to a longer time (6 min) demonstrating a yellow to deep red color change. Therefore, by varying the drying time the replay wavelength can be tuned, however reducing the drying time reduces the shifting range, therefore a balance needs to be made between quantitative and qualitative sensitivity of the device.
As previously described, the HS was recorded at an angle relative to the laser interference planes to facilitate the separation of the Bragg reflection from specular reflection. As the sensor swells, the angle of the reflection can vary requiring a reorientation of the receiving fiber. Therefore, to examine the extent of angle change, an automated goniometer optical set-up was used to determine the angle of reflection variations (Figure 5b). Glucose concentrations were varied from 0.0 to 9.4 mmol L −1 , achieved via successive glucose additions, produce an overall angle change of 5° (Figure 5c; Figure S6, Supporting Information). All concentrations present a homogenous peak with a singular maximum and minimal hysteresis. Minimal variation in reflection angle has permitted receiving fibers to be fixed throughout experimental monitoring. Fixation of the viewing angle validates sensors applicability to devices, where user expertise to align the collection angle is not required. The results demonstrate the ability to tune sensor characteristics, such as replay wavelength, recorded information, and collection angle to facilitate device development.

The DP HS Characterization for Urinalysis
Urine samples contain an array of compounds which can cause interference in glucose monitoring. To determine sensor selectivity to common interference compounds, glucose (2.00 mmol L −1 ), citrate (2.45 mmol L −1 ), creatinine (7.79 mmol L −1 ), fructose (0.04 mmol L −1 ), lactate (0.25 mmol L −1 ), sucrose (2.00 mmol L −1 ), urea (249.75 mmol L −1 ), and uric acid (1.49 mmol L −1 ), in PBS solutions were prepared respectively, and the pH value was adjusted to 7.6 ( Figure 6a and Table S3, Supporting Information). Fructose and lactate demonstrate minor shifts, 3.0 and 0.9 nm respectively, while glucose shows a shift of 27.6 nm. Although fructose and lactate bare cis-diol groups capable of binding with the boronic acid groups, their low concentrations in urine produce a minor response. Sucrose is a disaccharide of glucose and fructose where the cis diols have been covalently bonded, hence the minor wavelength shift was observed (0.93 nm). A negligible shift observed for uric acid and creatinine are waste products from respiration excreted through the urine in low concentrations. Urea as a Lewis base is the most concentrated interference compound present in urine, a wavelength shift of 13.47 nm is observed. The attraction between Lewis acid and base could produce a dative covalent bond between the boronic acid and urea, [36] forming a bound ionic charge within the hydrogel and therefore inducing Donnan osmotic pressure swelling. Citrate shows a wavelength shift of 13.5 nm at 2.45 mmol L −1 , possibly due to the interaction between citrate and boronic acids. [37] The two carboxylic acid groups paired with the tertiary alcohol group present a potential binding site. Glucose response in the presence of each contaminate was investigated as interfering compounds could be competing with glucose to bind within the hydrogel. HSs were equilibrated in a PBS solution and secondary analyte solutions were prepared containing physiological concentrations of interference compounds and glucose (2.0 mmol L −1 , pH 7.6) (Figure 6b). A comparable trend of sensor sensitivity is observed with signal shift increasing by the previous interference compounds respective shifts. The developed sensors show suitable selectivity for the determination of urinary glucose concentration, and further improvements could be made via the use of diboronic acid co-monomers. [38] The advantages of the HSs over other sensing platforms for urinalysis is their ability to determine glucose concentrations reversibly and accurately over an extended period. To demonstrate this reliability and reversibility, sensors were successively immersed between PBS solutions of glucose concentrations of 0.0 and 2.0 mmol L −1 at 37 °C (Figure 6c). After 15 cycles sensors showed minimal hysteresis between initial and final values. Accurate determination of concentrations greater than 2.0 mmol L −1 is imperative for patient urinary monitoring. Sensor response was monitored at set intervals (2 min) to hypothetical daily urinary glucose variations (Figure 6d). The HSs demonstrate a quick response time irrespective of glucose concentration (10-12 min). The response is slower (30 min) when moving to glucose-free conditions, with minimal hysteresis observed at the baseline value (± 2.06 nm). The reduced equilibration time is a minor issue for urinary monitoring as intervals between measurements will often exceed 30 min. This high reversibility demonstrates the applicability of sensors to POC, where devices can be reused multiple times daily by patients, reducing both waste and costs incurred through single-use techniques.
To establish sensor applicability to urinary glucose monitoring in patients, artificial urine (AU) was prepared according to the composition described by Sarigul et al. (Table S4, Supporting Information) with pH corrected to 7.6. [39] Optimized sensors were immersed in glucose-free AU and allowed to equilibrate (1.5 mL, 1 h). The glucose concentration was varied through successive removals of glucose-free AU (150 µL), followed by the addition of glucose-containing AU (150 µL, 20 mmol L −1 ), maintaining analyte volume at 1.5 mL. The sensors displayed a wavelength shift of 95.4 nm for 0-9.4 mmol L −1 of glucose, comparable to PBS testing (Figure 6e). Although the sensitivity showed no hysteresis between AU and PBS, a red shift was observable (0.0 mmol L −1 , AU = 628.2 nm, PBS = 601.1 nm) (Figure 6e). Figure 6f demonstrated the dynamic linear calibration curves for HS response in AU. A linear relationship is achieved between glucose concentration and wavelength change over the range of 0.0-5.4 mmol L −1 with a sensitivity value of 13.03 nm mmol −1 L −1 , with sensor limit of detection determined to be 0.06 mmol L −1 . The red shifting visible in the color changes (Figure 6f-inset) could be attributed to the presence of interference compounds such as citrate and urea. Concurrently, the AU and PBS standard solutions utilized possess different ionic strength values, 300 and 162 mmol L −1 respectively. HSs volumetric change is directly correlated with the Donnan osmotic pressure inducing a net movement of ions across the hydrogel membrane, therefore this increase in ionic strength could correlate with the red shift observed. [40] A simple POC device was developed, whereby a HS was fitted with a small amount of epoxy adhesive to a suitably sized urine receptacle fitted with a colorimetric reference chart and illuminated with a broadband light source (Figure 6gi,ii). Glucose presence within urine identifies patients with diabetes or prediabetes to medical professionals. Figure 6g-iii demonstrates the devices response to glucose-free urine was donated from a healthy 27-year-old male, validated utilizing Roche dipstick testing. The pH value of the glucose-free urine was adjusted to 7.6, and spiked with glucose to validate sensor response to urinary glucose concentrations, Figure 6g-iv shows color changes for healthy (glucose free), and patient at-risk urinary glucose concentrations (0.0, 2.0, and 5.4 mmol L −1 , 37 °C) after a 15 min incubation time with sensors washed between measurements in deionized water. Glucose concentrations remain constant within urine after collection, therefore permitting temperature is maintained response time and shifting will be consistent. Herein, temperature was maintained within a water bath to ensure a consistent response time however, in future POC devices an insulated receptacle could ensure temperature is maintained throughout the equilibration time. A qualitative red shifting of the signal is observable to the user; however, a minor interference can be observed due to the yellow color of the sample. The human eye is less sensitive to color variation in the red region of the visible spectrum. Quantitative analysis of the diffracted light wavelength could be achieved via a smartphone app with a background subtraction to ensure reliable analysis. Between individual measurements HS were submerged in deionized water to maintain a level of hydration within the sensor. To demonstrate HS reusability between measurements, sensors were monitored in human urine over a 5-day period Figure 6. Establishment of optimized glucose sensor selectivity, performance in biomimetic solutions, and the construction of a urinalysis diagnostic device. Testing of the sensor wavelength shift to the addition of a) common interference compound in PBS solutions (pH 7.6) at physiological concentrations (Table S3, Supporting Information); b) both interference compounds and 2.0 mmol L −1 of glucose; c) Cycling of HS between 0.0 mmol L −1 and 2.0 mmol L −1 glucose concentrations for sensor reliability over 15 cycles; d) The sensor response to glucose at 2 min intervals, green and red shaded areas highlight normal and at risk glucose concentrations respectively; e) Comparisons of the HS in response to glucose between AU and PBS (Figure 4f), inset spectra obtained from monitoring glucose variation in AU over a range of 0.0-9.4 mmol L −1 ; f) Biphasic calibration curves of HS response to glucose in AU over a range of 0.0-5.4 mmol L −1 and 5.4-9.4 mmol L −1 ; g) i) Schematic to demonstrate the preparation technique and analysis method for a simple HS, (20 min, 37 °C). ii) Images display the HS fabricated and the components of the device, iii) Demonstration of HS performance in human urine, iv) Smartphone images of reusable device performance in human urine samples of varying glucose concentration.
with minimal hysteresis ( Figure S7, Supporting Information). Here, the HSs have been proven to be viable in biological equivalent solution and a simple device has been demonstrated in a POC application.

Conclusion
In this work, comprehension of DP holographic fabrication techniques has been expanded with development of viable, colorimetric, reusable, nanoparticle-free, glucose sensitive devices with improved sensitivity and response stability. This approach reduces the cost incurred from raw materials and complex syntheses but also minimizes waste produced in urinary glucose monitoring. Sensors provide patients with reliable quantification of their urinary glucose concentration and identify their risk of diabetes. Simulation of HS swelling provides an approximation of the reduction of device sensitivity with increasing hydrogel thickness when the direct experimental measurement is not applicable at the nanoscale. The developed HSs are efficient in determining glucose concentration over a physiological range (0.0-9.4 mmol L −1 ) via both qualitative and quantitative colorimetric measurement. Optimization of both the boronic acid functionalized RM and the high RI IL monomer solutions alongside the fabrication techniques permits a significant expansion of the DP HS for complex analytes. Optimized sensors achieve a 13.03 nm mmol −1 L −1 sensitivity, a 397% increase compared to initially prepared HS and a limit of detection of 0.06 mmol L −1 . Investigation of the effect of drying time on sensor performance has expanded current understanding of the fundamental fabrication principles of the DP HS, informing future research of parameters that determine device sensitivity. Further investigations in the structure-performance relationships could fundamentally bring DP HS to the forefront as an optical sensing platform. Photolithographic patterning of text permits users to qualitatively assign sensor purpose in future multiplexed devices minimizing human error. Sensor angular dependence has been quantified with minor variations over sensor range, future works using spherical mirrors or rough surface mirrors as a recording substrate may assist in the improvement of this factor. Sensor performance has been studied in biomimetic and human urine validating sensor viability with minimal interference from common biological contaminates. Moreover, the functionalization with more specialized boronic acid moieties could facilitate a more selective sensor for monitoring more complex biological analytes such as blood. [41] Alongside this, the integration of a smartphone app to directly monitor the structural colorimetric changes would facilitate the POC urinary analysis. [42] This work demonstrates the progression of simple, adjustable and biocompatible DP HS is highly promising to produce a wide and powerful bioanalytical platform for advanced healthcare diagnostics.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.