Nonenzymatic Detection of Glucose Using 3D Printed Carbon Electrodes in Human Saliva

One of the most prevalent diseases where point-of-care (POC) diagnostics has focused is diabetes, which impacts hundreds of millions of people globally. Due to the severe negative outcomes including renal failure, nerve damage, and stroke, many POC sensors have been designed to streamline low-cost testing. Recently, the utility of 3D printing for rapidly fabricating housings, electrodes, and sensors for use at the POC has been exploited toward diverse applications. Particularly interesting are 3D printed carbon electrodes (3DpCEs) in POC diagnostics owing to their simplicity, affordability, and mass production capabilities for developing sensors either for direct use or through post-printing surface modifications. Herein, we report a copper modified 3DpCE as a sensitive and selective nonenzymatic biosensor for glucose. Copper deposition, paired with an optimized activation protocol, produced a sensitive and selective sensor for glucose with a larger detection range, enhanced sensitivity, and better reproducibility compared to nonactivated and alkaline immersed 3DpCEs. The sensor displayed excellent linearity between 10–1800 μM and proved to be highly selective over common biologically relevant interferants. The 3D printed sensor successfully determined biologically relevant concentrations of glucose in human saliva which resulted in percent recoveries of 101 ± 8%, 106 ± 6%, and 98 ± 6% for 74, 402, and 652 μM glucose, respectively.

Research into point-of-care (POC) diagnostics has been rapidly advancing with the development of novel sensors, devices, and protocols for the detection and/or monitoring of many common diseases and illnesses impacting health on a global scale. 1,2These efforts strive to provide reliable, selective, cost-efficient, and rapid diagnostic alternatives for patients outside of traditional centralized clinical settings.The use of large and expensive analytical instrumentation like high performance liquid chromatography, gas chromatography, mass spectrometry, etc is unrealistic in many underdeveloped or underfunded areas (e.Native American reservations, low-income urban and rural populations, and developing countries), which motivates researchers to develop diagnostic tools that can be miniaturized. 3,4iabetes mellitus has continually grown as one of the most serious health problems around the world. 5The World Health Organization (WHO) has projected that there will be 366 million diabetics by 2030. 6This chronic disease can lead to extreme complications like renal failure, nerve damage, and even stroke. 7o mitigate these life-threatening outcomes, blood glucose is conventionally monitored in real time ensuring diabetics are within a healthy threshold which prevents hypo/hyperglycemia.In a clinical setting, glucose analyzers rely on bulky and expensive equipment to provide an accurate response. 8However, at-home glucometers have become the primary standard for POC diagnostic devices, allowing diabetics to monitor and maintain a healthy lifestyle without the need for frequent hospital visits.Although glucometers remove the need of hospital-based laboratory testing, a financial and invasive burden is caused by this chronic disease and its continuous monitoring requirements. 9In many cases, these commercial devices require multiple tests throughout the day leading to continuous purchasing of specialized test strips for testing.Along with test strips, sterilized needles are also required for blood collection.4][15] Although enzymatic sensors are selective and sensitive, they tend to lack cost efficiency and longevity due to the enzyme incorporated into the sensor.
Nonenzymatic sensors involve modifying a sensor with metal or complex onto the surface of an electrode which can then be used for sensing. 16These sensors have the benefit of being more cost efficient while maintaining their selective and sensitive properties.At the forefront of POC diagnostics is electrochemistry, owing to its ability not only to be miniaturized but also to permit low-cost, selective, sensitive, and rapid results.Furthermore, electrochemical POC devices are diverse, owing to the versatility of electrode modifications, and only require simple equipment for experimental readout. 17,18In many cases, the modification of an electrode surface is required to provide selective and sensitive responses for specific biomarkers. 19,20Unfortunately, many biosensors are fabricated using complex and multi-step approaches. 21,22Using electrochemical sensors, selective and sensitive determination can also be achieved through the deposition of nanomaterials. 23,24Electrochemical deposition of metals onto electrode surfaces provides inherent electrical and catalytic properties while also producing a large surfaceto-volume ratio to improve its sensitivity and selectivity. 25Copper and nickel are two common metals deposited onto electrodes for the detection of glucose. 26,27Copper plated sensors have been shown to be a promising alternative to enzymatic glucose sensors due to their ease of fabrication, stability, biocompatibility, and high sensitivity. 28mperative in the success of POC diagnostic devices is simplicity, not only in device use-permitting use by untrained individuals-but also in device fabrication.Carbon-based sensors (e.g., conventional (glassy carbon), fibers (microfibers), and nanomaterials (graphene)) have been shown to be compatible with diverse modification strategies and therefore have become crucial in the development of electrochemical POC devices and sensors. 16,293D printing has become a widely utilized technique in analytical chemistry over the past decade. 30,31With the ability to print metals, polymers, ceramics, and biomaterial, 3D printing provides a customizable, low-cost, prototypic, and rapid approach for fabricating innovative and functional materials with diverse applications. 32Such advancements make accessible healthcare more attainable than ever where 3D printing has proven to be an asset in biomedical diagnostics for POC applications. 33,34While 3D printing can provide many important characteristics in the ability to fabricate unique designs, the materials used also provide an adaptable platform for numerous applications. 35n the development of electrochemical sensors, many researchers have begun to manipulate and optimize 3D printable materials.7][38] Recently, commercial carbon-infused thermoplastic filament has been propelled into the field of electrochemical sensing utilizing fused deposition modeling (FDM) 3D printing.FDM printing fabricates 3D structures out of thermoplastic polymer filaments such as polylactic acid (PLA) and acrylonitrile butadiene styrene (ABS). 39,40By incorporating carbon fiber, carbon black, carbon nanotubes, etc, this filament can be used for the fabrication of functional sensors.Other research has shown that 3D printed sensors are a powerful and low-cost alternative to conventional electrodes to selectively detect electroactive biomarkers and pharmaceuticals. 41,42hrough the combination of alkaline assisted electrochemical activation (AAEA) and copper deposition a selective nonenzymatic glucose sensor for noninvasive monitoring for diabetics can be achieved using 3D printed carbon electrodes (3DpCEs).Herein, a 3D printed nonenzymatic glucose sensor was created for noninvasive detection in human saliva.To highlight the novelty of this sensor (i) a comparative study of AAEA 3DpCEs compared to nonactivated 3DpCEs and alkaline immersed 3DpCEs, (ii) characterization of the sensitivity and selectivity of glucose detection with a 3DpCE, modified with copper nanomaterial, using amperometry, and (iii) recovery analysis of AAEA 3DpCEs in human saliva was performed.
Fabrication of 3DpCEs.-Acomputer aided design (CAD) file was created using CAD software (Fusion 360) for the 3DpCEs.A 0.425 cm 2 working electrode with a 2 mm wide connector (Fig. S1) was printed using the Creative Pro FDM 3D printer using the Conductive PLA filament from Protopasta.Once printed, electrodes were stored in airtight containers until use.
Activation of the 3DpCE surface.-Alkalineimmersion.-3DpCEswere activated using 5 M NaOH, and the working surface was fully immersed for 60 min.After the immersion process, each 3DpCE was rinsed thoroughly with DI water and air dried prior to use.

Alkaline assisted electrochemical activation (AAEA).-3DpCEs
were electrochemically activated using a model set of parameters unless otherwise specified.The model parameters include a scan rate of 0.1 V s −1 , 5 M NaOH solution, a voltage range from −1.5 V to 1.5 V, and 7 cycles of activation.All experiments used a Ag/AgCl (1 M KCl) reference electrode and carbon fiber rod counter electrode.After activation, 3DpCEs were rinsed with DI water and air dried prior to use.
Electrochemical deposition of copper onto 3DpCEs.-Copperdeposition was completed on nonactivated, alkaline immersion, and AAEA 3DpCEs using chronoamperometry.Each type of 3DpCE was placed in 10 mM CuCl 2 with 100 mM KCl and a CV was obtained to determine the deposition potential.A reduction potential of −0.65 V was chosen for the deposition of Cu.Copper was deposited on 3DpCEs in 10 mM CuCl 2 with 100 mM KCl where the reduction potential was applied for 600 s.After deposition, each Cu-3DpCE was rinsed with DI water and air dried before use.
Electrochemical impedance spectroscopy (EIS) of 3DpCEs.-Acomparison of nonactivated, alkaline immersion, and AAEA 3DpCEs was completed using EIS.Each 3DpCE was used to collect the open circuit potential through CV in 2 mM ferri/ferrocyanide in 100 mM KCl and 2 mM glucose in 0.1 M NaOH, respectively.EIS was then collected at both potentials for each 3DpCE to determine the ohmic resistance.EIS parameters: High Frequency-1 MHz, Low Frequency-0.1 Hz, and Amplitude-0.005V.
Kinetic study of 3DpCEs.-Acomparative study of alkaline immersion and AAEA 3DpCEs was completed to study the kinetic behavior of each electrode.CVs were obtained at scan rates of 0.025, 0.05, 0.075, 0.1, 0.15, 0.25, and 0.5 V s −1 for each 3DpCE.Each 3DpCE was tested in 2 mM ferri/ferrocyanide in 100 mM KCl and 2 mM glucose in 0.1 M NaOH, respectively, and the peak oxidation currents were plotted against the square root of the scan rate.Each CV was taken in triplicate, and the average oxidation peak current was reported.
Glucose detection using Cu-3DpCEs.-Nonactivated,alkaline immersion, and AAEA Cu-3DpCEs were placed in 5 mM glucose (Glu) with 0.1 M NaOH to obtain a CV of the oxidation of glucose to gluconolactone.Linear calibrations were then obtained for each Cu-3DpCE using chronoamperometry.For alkaline immersion and AAEA Cu-3DpCEs an applied voltage of 0.6 V was used, and nonactivated Cu-3DpCEs used an applied voltage of 0.8 V. Serial addition calibrations were collected through the addition of Glu into a background of 0.1 M NaOH.Selectivity was also analyzed for AAEA Cu-3DpCEs where successive additions of Glu and potential interferents (thiocyanate, caffeine, urea, uric acid, and ascorbic acid) were completed using chronoamperometry to monitor the change in current after each addition.
Scanning electron microscopy imaging of Cu-3DpCEs.-Electronmicrographs were taken using an Apreo VolumeScope™ scanning electron microscope (SEM), and energy dispersive X-ray analysis (EDAX) was conducted using an integrated TEAM™ Pegasus energy dispersive spectroscopy-electron backscatter diffraction (EDS-EBSD) system consisting of an Octane Elect Energy Dispersive Spectroscopy (EDS) detector and a Velocity™ Electron Backscatter Diffraction (EBSD) detector.
Recovery analysis of glucose in human saliva using Cu-3DpCEs.-Aspiked recovery analysis was performed in a human saliva 1:10 dilution in 0.1 M NaOH solution.3DpCEs were activated and deposited with copper prior to use in this experiment.Each 3DpCE was calibrated in the human saliva solution using chronoamperometry.The background current of the saliva/NaOH solution was monitored using chronoamperometry under stirred conditions (300 rpm) prior to additions of 30, 120, 300, 500, and 800 μM Glu (50, 150, 300, 400, and 600 μl of 5 mM Glu).After calibration, the 3DpCEs were placed in a fresh diluted saliva solution and a low (74 μM), medium (402 μM), and high (652 μM) unknown was spiked into the background and recorded.

Results and Discussion
Figure 1 shows a schematic representation for fabricating 3DpCEs using 3D printing, which were then activated using AAEA and electrochemically deposited copper for the detection of glucose.Initially, 3DpCEs are printed using FDM 3D printing (Fig. 1a).As printed, 3DpCEs are not functional enough to produce meaningful current responses for diagnostic use.This issue has been previously discussed and is related to the much higher percentage of non-conductive PLA support material in the 3D printable filament. 43hrough the process of electrochemical activation and saponification, active carbon was made uniformly accessible on the surface and charged lactate caps were produced which improved the conductivity of the 3DpCEs. 44Once an optimized activation process is performed for 3DpCEs, an electrochemical deposition of copper was completed to facilitate the detection of glucose using a nonenzymatic process (Fig. 1c).
Along with the nonactive surface, the size of the electrode stem (SI Fig. 1) also affects the current response.SI Fig. 2 shows the resistance of 3DpCEs at different distances from the face of the electrode.This information shows that 3DpCEs do not maintain a constant resistance like other electrode materials.Due to the PLA working as an insulator, the resistance of 3DpCEs linearly increases with distance away from the electrode face.This additive resistance also deteriorates the response of 3DpCEs which requires precise connection for reproducible experiments (SI Fig. 1).Using the electrode configuration shown in SI Fig. 1, the ideal connection was determined to be between 5-8 mm from the electrode surface.This connection offers easy experimental use and provides the most reliable results while mitigating as much resistivity from the PLA support material.After determining these physical parameters of 3DpCEs, we further investigated the alkaline-assisted electrochemical activation protocol.
AAEA provides a rapid and simple approach for modifying 3DpCEs, however, many parameters can be optimized for a specific application.An optimized AAEA protocol was determined for copper deposition by exploring the concentration of NaOH, the number of potential cycles, and the scan rate to determine the most responsive 3DpCE following activation.SI Fig. 3 shows a CV overlay of 2 mM ferri/ferrocyanide using AAEA 3DpCEs in different concentrations of NaOH during activation while maintaining model conditions (SI Fig. 3a and a CV overlay of 2 mM ferri/ ferrocyanide for AAEA 3DpCEs activated at different numbers of potential cycles (SI Fig. 3c).3DpCEs electrochemically activated in 5 M NaOH produced the highest current response as well as the smallest peak separation (i.e., better reversibility) towards the ferri/ ferrocyanide redox couple (SI Fig. 4b) along with a gradual increase in peak current with minimal increases after 7 cycles (SI Table I).The scan rate of the AAEA protocol was also explored and SI Fig. 5 shows the CV responses of 3DpCEs, activated for 7 cycles in 5 M NaOH at different scan rates, towards 2 mM ferri/ferrocyanide.This data shows a scan rate of 0.5 V s −1 with 30 activation cycles (total activation time of 360 s).SI Figs.4a-4b shows CV overlays of 2 mM ferri/ferrocyanide using 3DpCEs that were activated at scan rates of 0.25 V s −1 and 0.5 V s −1 , respectively.SI Figs.4c-4d shows results of 3DpCEs activated for 360 s at scan rates of 0.1, 0.25, 0.5, 0.75, 1.0, and 1.5 V s −1 , confirming that a scan rate of 0.5 V s −1 for 360 s produces the most sensitive 3DpCE after activation with the best peak resolution and reversibility towards the ferri/ferrocyanide redox couple.
After determining the optimal parameters for electrochemical activation of 3DpCEs in alkaline conditions, further experimentation was completed to compare this protocol to nonactivated and alkaline immersed 3DpCEs for improved copper deposition for nonenzymatic detection of glucose.After activating the 3DpCEs, Cu was deposited onto each electrode using chronoamperometry with an applied potential of −0.65 V, which was chosen from the Cu +1 → Cu reduction peak (Fig. 2).A constant potential was applied for 10 min to the 3DpCEs in 10 mM CuCl 2 with 100 mM KCl (Fig. 3).After deposition, a CV was collected in 5 mM Glu and 0.1 M NaOH for each Cu-3DpCE (Fig. 4).AAEA and alkaline immersed 3DpCEs showed an oxidation peak at ∼0.6 V while nonactivated 3DpCEs showed a low intensity peak near 0.8 V.
Following copper deposition, electrochemical impedance spectroscopy (EIS) was completed for nonactivated (i.e., freshly printed), alkaline immersed, and AAEA 3DpCEs.Figures 5a, 5b shows overlays of Nyquist plots for each electrode in 2 mM ferri/ferrocyanide at the open circuit potential for 3DpCEs without and with copper, respectively.Figure 5c shows an overlay of the Nyquist plots for each electrochemically deposited copper 3DpCE in 2 mM Glu in 0.1 M NaOH at the open circuit potential.Although only a slight change in charge-transfer resistance favoring AAEA 3DpCEs with and without copper deposition, the AAEA protocol produced a significantly higher current response and resulted in a more reversible CV compared to the nonactivated and alkaline immersed 3DpCEs for both ferri/ferrocyanide and Glu (Figs. 3 and 4).After completion of EIS, a kinetic study was completed for the AAEA 3DpCEs with electrochemically deposited copper in 2 mM Glu with 0.1 M NaOH. Figure 6a shows the CV overlay of 2 mM Glu in 0.1 M NaOH using copper deposited AAEA 3DpCEs at different scan rates (25, 50, 75, 100, and 250 mV s −1 ).A linear correlation is shown in Fig. 6b between the peak oxidation current and the square root of the scan rate.This correlation shows that the AAEA 3DpCEs with copper electrochemically deposited is a diffusion driven process for the oxidation of glucose to glucono-lactone.A kinetic study was also completed for both AAEA and alkaline immersed 3DpCEs in 2 mM ferri/ferrocyanide against the square root of the scan rate (SI Figs. 6  and 7).Both activation methods produce linear correlations to the square root of the scan rate indicating a diffusion limited process for each electrode.
After completion of the kinetic study, chronoamperometry was used to collect calibrations of electrochemically deposited 3DpCEs.Figures 7a-7c shows calibration of Glu for each activation method used.Remarkably, AAEA 3DpCEs produced the highest level of reproducibility, linearity, detection range, and sensitivity for Glu.AAEA 3DpCEs produced a sensitivity of 260 nA μM −1 across a range of 10 to 1800 μM Glu.Although the alkaline immersed Cu-3DpCEs demonstrated a similar sensitivity (240 nA μM −1 ), the reproducibility across the calibration was much lower along with a decreased detection range of 10 to 550 μM Glu.Nonactivated Cu-3DpCEs were responsive to Glu after copper deposition but produced a much lower sensitivity (115 nA μM −1 ) compared to other activation methods used here.
To probe the surface of the Cu-3DpCE, SEM imaging and EDAX measurements were performed to gain insights into the improved response of the AAEA Cu-3DpCEs towards Glu.Figures 7d-7f shows SEM images of the Cu-3DpCE that were nonactivated, alkaline immersed, and alkaline-assisted electrochemically activated, respectively.Nonactivated electrodes showed poor uniformity and low deposition of copper on the electrode surface.Alkaline immersed electrodes displayed a higher density of deposited copper onto the electrode surface, but the copper deposition maintains low uniformity of particle size and dispersion.AAEA electrodes resulted in the highest uniformity and density of copper deposition with a more evenly coated surface.The uniformity of the AAEA Cu-3DpCEs supports the improved reproducibility compared to alkaline immersed Cu-3DpCEs.EDAX for each electrode shows an increase in the weight percentage of copper with the AAEA electrodes having the highest percentage (SI Table II).Using Faraday's First law (Eq.1), where w is the equivalent mass of copper deposited, q is the total charge during deposition, M is the molar mass of copper, n is the number of electrons, and F is the Faraday's constant, quantification of the total mass of copper deposited on each activation method resulted in 0.10 ± 0.05 mg cm −2 for nonactivated, 0.13 ± 0.03 mg cm −2 for alkaline immersed, and 0.19 ± 0.01 mg cm −2 for AAEA 3DpCEs.In regard to the total mass deposited and percentage of copper from the EDAX analysis, AAEA 3DpCEs showed a significant increase compared to the alkaline immersed and nonactivated 3DpCEs which correlates the increased sensitivity for Glu.
To ensure AAEA Cu-3DpCEs were selective for Glu, successive addition of common interferents were spiked into solution to monitor the current response (Fig. 8a).Ascorbic acid was the only species to generate a slight increase in current response of 3% compared to glucose, while all other potential interferents produced no change in response.After confirming that AAEA 3DpCEs were selective for Glu after copper deposition a spiked recovery analysis was completed in human saliva.Figure 8b shows a chronoamperometry trace of a calibration of Glu in a 1:10 saliva to 0.1 M NaOH solution from 0-800 μM followed by a spike recovery of 3 unknowns (74 402, and 652 μM Glu). Figure 8c shows the results of the calibration of Glu using Cu deposited AAEA 3DpCEs.A sensitivity of 284 nA μM −1  was achieved with a linearity of 0.9963.Reproducibility between sensors remained consistent with the bulk analysis calibration of Glu using AAEA 3DpCEs.The spiked recovery of each unknown resulted in a recovery of 101 ± 8%, 106 ± 6%, and 98 ± 6% Glu, respectively, shown in Table I.A comparison to previously reported nonenzymatic Glu biosensors is shown in SI Table III.

Conclusions
This research presented a Cu modified AAEA 3DpCE nonenzymatic glucose sensor for noninvasive saliva detection with the aim to streamline the fabrication of selective and sensitive diagnostic tools for applications at the POC.Compared to nonactivated 3DpCEs, this modification protocol provides a simple and rapid method to fabricate highly responsive 3D printed sensors.The AAEA protocol requires less than 10 min to produce a sensitive and selective sensor which outperforms 3DpCEs activated through alkaline immersion for 60 min.The detection of glucose was achieved in the presence of potential interferents without influencing the response.This simple modification protocol provides a rapid, reproducible, and responsive sensor for 3DpCEs to be integrated into POC devices for onsite analysis of glucose in saliva.The AAEA 3DpCE produced comparable sensitivity to previously reported nonenzymatic sensors and provided a large range from 10-1800 μM Glu, which covers the biologically relevant range of Glu in saliva (50-500 μM).
Ideally, however, 3D printed electrodes would be capable of being used immediately after printing, without the need of processing before use.Although two reports have demonstrated that an inhouse-made carbon-infused material could be used without further modification, all commercially available carbon-infused materials still require modification, to the best of our knowledge. 46,47ontinued research in the preparation of carbon-infused materials that are compatible with 3D printing technology is required to produce ready-to-use sensors.The ability to select and print a specific material composition would advance fully 3D printed readyto-use sensors and devices for POC applications, while providing cost-effective, mass producible tools for the reliable diagnosis of diseases and illnesses for the global population.

Figure 1 .
Figure 1.(a) A schematic illustration of the FDM 3D printing fabrication of 3DpCEs.(b) AAEA experimental setup for 3DpCE activation and measurements.(c) Glucose determination using electrochemically deposited Cu-3DpCEs.

Figure 3 .
Figure 3.An overlay of chronoamperometric traces of copper deposition using 10 mM CuCl 2 with 100 mM KCl for nonactivated, alkaline immersed, and AAEA 3DpCEs at a constant potential of −0.65 V.

Figure 4 .
Figure 4. CV overlay of 2 mM Glu in 0.1 M NaOH using nonactivated, alkaline immersed, and AAEA Cu-3DpCEs.The peaks at ∼0.1 V correspond to the formation of Cu 2 O and the peaks at ∼0.6 V correspond to the conversion of glucose to glucono-lactone.

Figure 5 .
Figure 5. Nyquist plot overlays of 2 mM ferri/ferrocyanide with 100 mM KCl at the open circuit potential using nonactivated, alkaline immersed, and AAEA 3DpCEs without copper deposited (a) and electrochemically deposited copper (b), respectively.(c) An overlay of Nyquist plots of 2 mM Glu in 0.1 M NaOH at the open circuit potential using nonactivated, alkaline immersed, and AAEA 3DpCEs with electrochemically deposited copper.

Figure 6 .
Figure 6.(a) CV overlay of 2 mM Glu in 0.1 M NaOH at different scan rates of 25, 50, 75, 100, and 250 mV s −1 using copper deposited AAEA 3DpCEs.(b) Kinetic study of oxidation peak current of 2 mM Glu in 0.1 M NaOH compared to the square root of the scan rate for copper deposited AAEA 3DpCEs.Error was determined across 3 separate 3DpCEs.

Figure 8 .
Figure 8.(a) A chronoamperometric trace of a selectivity analysis for AAEA 3DpCEs deposited with Cu.(b) A chronoamperometric trace of a calibration (black) of 35-800 μM Glu in a 1:10 human saliva to 0.1 M NaOH solution using a AAEA 3DpCE deposited with Cu with a spike recovery analysis of 3 unknown Glu concentrations (red).(c) A linear calibration of 35-800 μM Glu in human saliva (1:10 human saliva to 0.1 M NaOH) solution (black) using Cu deposited AAEA 3DpCEs with a sensitivity of 284 nA μA −1 and a linearity of 0.9963.Deviations were determined across 3 separate 3DpCEs.

Table I .
Spiked recovery analysis of Glu in human saliva.