3D Printed Voltammetric Sensor Modified with an Fe(III)-Cluster for the Enzyme-Free Determination of Glucose in Sweat

In this work, a 3D printed sensor modified with a water-stable complex of Fe(III) basic benzoate is presented for the voltammetric detection of glucose (GLU) in acidic epidermal skin conditions. The GLU sensor was produced by the drop-casting of Fe(III)-cluster ethanolic mixture on the surface of a 3D printed electrode fabricated by a carbon black loaded polylactic acid filament. The oxidation of GLU was electrocatalyzed by Fe(III), which was electrochemically generated in-situ by the Fe(III)-cluster precursor. The GLU determination was carried out by differential pulse voltammetry without the interference from common electroactive metabolites presented in sweat (such as urea, uric acid, and lactic acid), offering a limit of detection of 4.3 μmol L−1. The exceptional electrochemical performance of [Fe3O(PhCO2)6(H2O)3]∙PhCO2 combined with 3D printing technology forms an innovative and low-cost enzyme-free sensor suitable for noninvasive applications, opening the way for integrated 3D printed wearable biodevices.


Introduction
Diabetes is one of the most common worldwide diseases. It affects millions of individuals, causing serious damage to the nerves and blood vessels, and ranks among the leading causes of death globally [1]. The periodical checking of blood glucose (GLU) levels throughout the day is of vital significance for diabetic patients, which is typically operated via electrochemical self-testing devices based on blood sampling from the patient's fingertip. However, this painful blood sampling discourages the patients from frequent measurements during the daylight, while the tests at night-time are practically neglected. On the contrary, noninvasive GLU monitoring is an ideal route toward calm and painless glucose testing, and fortunately modern electrochemical devices have been introduced for GLU monitoring in sweat, as this epidermal biofluid contains GLU at quantities that correlate well with blood [2][3][4][5].
Electrochemical GLU sensors can be split into enzymatic-based and nonenzymatic sensors [6][7][8]. Typical enzymatic GLU biosensors are based on glucose oxidase (GOx) catalyzing the oxidation of glucose to gluconolactone and producing the by-product hydrogen peroxide, which is amperometrically determined by the sensor. The main problems on the construction of enzymatic GLU biosensors are the efficiency of GOx immobilization on the electrode surface, the presence of dissolved oxygen, and the effect of temperature, pH, and ionic strength on the enzyme activity [9][10][11][12]. To overcome these disadvantages, enzyme-free GLU sensors based on metallic particles have been applied as catalysts of GLU electrooxidation, including metals (i.e., Au, Pd and Pt), and metal oxides (i.e., CuO, All reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA). A stock solution of 0.1 mol L −1 GLU was prepared in water and left for 24 h at room temperature to allow equilibration of the isomers, and stored at 4 • C. The artificial sweat was composed of 3 mmol L −1 NH 4 Cl, 50 µmol L −1 MgCl 2 , 0.4 mmol L −1 CaCl 2 , 80 mmol L −1 NaCl, 8 mmol L −1 KCl, 25 µmol L −1 uric acid, 22 mmol L −1 urea, and 5.5 mmol L −1 lactic acid (pH 4) [12]. The phosphate buffer (PB) was prepared by mixing proper quantities of Na 2 HPO 4 and NaH 2 PO 4 , and the pH value was adjusted to 4 with 1 mol L −1 solution of HCl.
of 50 mL/min from room temperature to 800 °C with a heating rate of 10 °C min −1 . T powder X-ray diffraction pattern was recorded on a Bruker D8 Advance X-ray diffracto eter (CuKa radiation, λ = 1.5418 Å). The particle size was calculated using an image tak on a Leica M205 C stereoscope equipped with a Leica DMC5400-20 Megapixel cam ( Figures S2 and S3). The image was taken from a sample of [Fe3O(PhCO2)6(H2O)3]•PhC dispersed in water, on a glass slide.
The structure of basic iron benzoate ( Figure 2) consists of three Fe(III) cations bridg by a μ3-oxo bridge. Each pair of Fe(III) is bridged circumferentially by two benzoate a ons through their carboxylate groups. Finally, the coordination sphere of each Fe(III completed by a terminal H2O molecule (whose H atoms could not be modelled), and total charge is balanced by a benzoate anion in the lattice.  (Table S4).
The structure of basic iron benzoate ( Figure 2) consists of three Fe(III) cations bridged by a µ 3 -oxo bridge. Each pair of Fe(III) is bridged circumferentially by two benzoate anions through their carboxylate groups. Finally, the coordination sphere of each Fe(III) is completed by a terminal H 2 O molecule (whose H atoms could not be modelled), and the total charge is balanced by a benzoate anion in the lattice. Thermogravimetric analysis (TGA) ( Figure  S4) reveals [Fe3O(PhCO2)6(H2O)3]•PhCO2 loses ~4.29% within the 25-137 °C temperature range, co sponding to the three coordinated H2O molecules (their theoretical value is 4.98%). is followed by a loss of ~10.63% within the 137-257 °C temperature range, correspond to one benzoate anion (its theoretical value is 11.15%) and degrades immediately that. The residue above 620 °C is 22.18% which corresponds to Fe2O3 (theoretical v 22.06%).

Fabrication of the 3D Printed Sensor Modified with Fe(III)-Cluster
The fabrication process of the 3D printed GLU sensor modified with Fe(III)-clust illustrated in Figure 3. The 3D printed electrode (3DPE) was designed with Tinke software and the printing conditions were set to 60 °C for the platform, 200 °C for the h dispenser, and 60 mm s −1 for the printing speed. Flashprint software was used for prin The filament had a diameter of 1.75 mm and was PLA loaded with carbon black, and obtained from Proto Pasta. For the construction of the GLU sensor, 10 μL of 6% (w/v) anolic mixture of [Fe3O(PhCO2)6(H2O)3]•PhCO2, was applied on the cyclic surface of 3D printer and left for 5 min for its immobilization, followed by curing with an air str from a gun for 1 extra min. Next, 10 μL of 1% (w/v) ethanolic solution of Nafion was ad on the electrode cyclic surface and left to dry for 5 min. After that, the sensor was tre under an air stream for 1 min for complete drying. Thermogravimetric analysis (TGA) ( Figure S4) reveals that [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 loses~4.29% within the 25-137 • C temperature range, corresponding to the three coordinated H 2 O molecules (their theoretical value is 4.98%). This is followed by a loss of 10.63% within the 137-257 • C temperature range, corresponding to one benzoate anion (its theoretical value is 11.15%) and degrades immediately after that. The residue above 620 • C is 22.18% which corresponds to Fe 2 O 3 (theoretical value 22.06%).

Fabrication of the 3D Printed Sensor Modified with Fe(III)-Cluster
The fabrication process of the 3D printed GLU sensor modified with Fe(III)-cluster is illustrated in Figure 3. The 3D printed electrode (3DPE) was designed with Tinkercad software and the printing conditions were set to 60 • C for the platform, 200 • C for the head dispenser, and 60 mm s −1 for the printing speed. Flashprint software was used for printing. The filament had a diameter of 1.75 mm and was PLA loaded with carbon black, and was obtained from Proto Pasta. For the construction of the GLU sensor, 10 µL of 6% (w/v) ethanolic mixture of [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 , was applied on the cyclic surface of the 3D printer and left for 5 min for its immobilization, followed by curing with an air stream from a gun for 1 extra min. Next, 10 µL of 1% (w/v) ethanolic solution of Nafion was added on the electrode cyclic surface and left to dry for 5 min. After that, the sensor was treated under an air stream for 1 min for complete drying. Biosensors 2022, 12, x FOR PEER REVIEW 5 of 13

Electrochemical Measurements
The electrochemical measurements were conducted in a 5 mL electrochemical cell in the presence of dissolved oxygen. The portable potentiostat was the EmStat3 (Palm Sens, Houten, The Netherlands) and operated by the PS Trace 4.2 software (Palm Sens, Houten, The Netherlands). The reference electrode was a saturated calomel electrode and the counter electrode was Pt wire. For the DPV measurements, a potential of −1.4 V for 360 s was applied on the 3D printed working electrode (WE), and then a scan (modulation amplitude, 50 mV; increment, 10 mV; pulse width, 75 ms; pulse repeat time, 50 ms) was run on the WE and the DPV response was recorded. The connection of the three electrodes to the portable potentiostat was accomplished using three crocodile clips. Figure 4 depicts the DPV responses of the 3DPE modified with Fe3O(PhCO2)6(H2O)3]•PhCO2 and the respective 3DPE modified with iron oxide towards 200 μmol L −1 GLU in PB (pH 4). The 3DPE modified with Fe(III)-cluster presented favorable performance offering a well-shaped DPV oxidation peak of GLU, while the respective 3DPE modified with iron oxide exhibited neglected response for GLU oxidation in these acidic conditions. It is has been documented before that Fe3O4, Fe2O3, and FeOOH based electrodes require neutral or basic conditions in order to form Fe(III), which effectively catalyzed the oxidation of GLU [27][28][29][30][31][32]. The mechanism of electrocatalyzed oxidation of the GLU by the 3DPE modified with Fe3O(PhCO2)6(H2O)3]•PhCO2 is based on the reduction of Fe(III) in the cluster to Fe(0) on the 3DPE surface, by setting a negative potential at −1.4 V for 360 s. Next, the metallic Fe(0) formed under this cathodic polarization process was oxidized to Fe(III) in the course of the scan potential from −1.4 V to +1.5 V. Finally, the in-situ electrogenerated Fe(III) oxidized GLU [27][28][29][30][31][32]. The whole mechanism is the following:

Electrochemical Measurements
The electrochemical measurements were conducted in a 5 mL electrochemical cell in the presence of dissolved oxygen. The portable potentiostat was the EmStat3 (Palm Sens, Houten, The Netherlands) and operated by the PS Trace 4.2 software (Palm Sens, Houten, The Netherlands). The reference electrode was a saturated calomel electrode and the counter electrode was Pt wire. For the DPV measurements, a potential of −1.4 V for 360 s was applied on the 3D printed working electrode (WE), and then a scan (modulation amplitude, 50 mV; increment, 10 mV; pulse width, 75 ms; pulse repeat time, 50 ms) was run on the WE and the DPV response was recorded. The connection of the three electrodes to the portable potentiostat was accomplished using three crocodile clips. . The 3DPE modified with Fe(III)-cluster presented favorable performance offering a well-shaped DPV oxidation peak of GLU, while the respective 3DPE modified with iron oxide exhibited neglected response for GLU oxidation in these acidic conditions. It is has been documented before that Fe 3 O 4, Fe 2 O 3, and FeOOH based electrodes require neutral or basic conditions in order to form Fe(III), which effectively catalyzed the oxidation of GLU [27][28][29][30][31][32]. The mechanism of electrocatalyzed oxidation of the GLU by the 3DPE modified with Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 is based on the reduction of Fe(III) in the cluster to Fe(0) on the 3DPE surface, by setting a negative potential at −1.4 V for 360 s. Next, the metallic Fe(0) formed under this cathodic polarization process was oxidized to Fe(III) in the course of the scan potential from −1.4 V to +1.5 V. Finally, the in-situ electrogenerated Fe(III) oxidized GLU [27][28][29][30][31][32]. The whole mechanism is the following:

Effect of Reduction Time, Potential and Fe(III)-Cluster Loading on GLU Determination
The effect of the loading of the Fe3O(PhCO2)6(H2O)3]•PhCO2 on the 3DPE surface, the reduction potential, and the reduction time of the 3DPE for the in-situ electrogeneration of Fe(III) were examined on the DPV response of 200 μmol L −1 GLU in 0.1 mol L −1 PB (pH 4) ( Figure 5). Four loading levels of the Fe3O(PhCO2)6(H2O)3]•PhCO2 on the 3DPE in the range 2-8% (w/v) (as ethanolic mixtures), step of 2%, were studied. As demonstrated in Figure 5, the Fe3O(PhCO2)6(H2O)3]•PhCO2/3DPE at 6% (w/v) yielded approximately 1.5 times higher voltammetric peak height of GLU than that of 4% and 2.5 higher than that of 2% (w/v) loadings, while its sensitivity was statistically comparable with that of 8% (w/v) loading. Hence, a Fe3O(PhCO2)6(H2O)3]•PhCO2/3DPE at 6% (w/v) loading was selected as the optimum, combining the minimal consumption of Fe(III)-cluster with the high DPV response of GLU in acidic conditions.

Effect of Reduction Time, Potential and Fe(III)-Cluster Loading on GLU Determination
The effect of the loading of the Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 on the 3DPE surface, the reduction potential, and the reduction time of the 3DPE for the in-situ electrogeneration of Fe(III) were examined on the DPV response of 200 µmol L −1 GLU in 0.1 mol L −1 PB (pH 4) ( Figure 5). Four loading levels of the Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 on the 3DPE in the range 2-8% (w/v) (as ethanolic mixtures), step of 2%, were studied. As demonstrated in Figure 5, the Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 /3DPE at 6% (w/v) yielded approximately 1.5 times higher voltammetric peak height of GLU than that of 4% and 2.5 higher than that of 2% (w/v) loadings, while its sensitivity was statistically comparable with that of 8% (w/v) loading. Hence, a Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 /3DPE at 6% (w/v) loading was selected as the optimum, combining the minimal consumption of Fe(III)-cluster with the high DPV response of GLU in acidic conditions. The effect of the reduction time and corresponding potential of the Fe(III)-cluster on the 3DPE surface was tested in the range 0 s to 480 s and from −1.6 V to 0.0 V, respectively. These parameters affect the quantity of the in-situ electrogenerated Fe(III) and, as a result, the catalytic electrocapability of the 3D printed sensor to GLU determination. As depicted in Figure 6A,B, the GLU oxidation responses increased with respect to the reduction period of time, while a sigmoidal shape is observed that shows a dependence on the reduction potentials. The GLU peak heights were low at more positive potentials, as these potential values were not adequately negative to establish the reduction of the Fe(III)-cluster to metallic Fe(0) on the WE surface. At more negative potentials, the deposition of Fe(0) on 3DPE was favored and the peak currents of GLU increased rapidly up to −1.4 V, where it leveled-off. For the further experiments, a reduction potential of −1.4 V for 360 s was selected, as this presented a satisfactory compromise between high sensitivity and short analysis times. The effect of the reduction time and corresponding potential of the Fe(III)-cluster on the 3DPE surface was tested in the range 0 s to 480 s and from −1.6 V to 0.0 V, respectively. These parameters affect the quantity of the in-situ electrogenerated Fe(III) and, as a result, the catalytic electrocapability of the 3D printed sensor to GLU determination. As depicted in Figure 6A,B, the GLU oxidation responses increased with respect to the reduction period of time, while a sigmoidal shape is observed that shows a dependence on the reduction potentials. The GLU peak heights were low at more positive potentials, as these potential values were not adequately negative to establish the reduction of the Fe(III)-cluster to metallic Fe(0) on the WE surface. At more negative potentials, the deposition of Fe(0) on 3DPE was favored and the peak currents of GLU increased rapidly up to −1.4 V, where it leveled-off. For the further experiments, a reduction potential of −1.4 V for 360 s was selected, as this presented a satisfactory compromise between high sensitivity and short analysis times.  . The voltammetric response of the 3D printed sensor to GLU oxidation increased linearly with increasing GLU concentration, with a correlation coefficient of 0.998, and the calibration curve fell within the physiological levels of GLU secreted in human sweat [12]. The limit of detection (LOD) was calculated by the equation LOD = 3 s y /a, where s y is the standard deviation of the y-residuals of the calibration plot, and a is the slope of the calibration, which was 4.3 µmol L −1 . The LOD of GLU achieved with the Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 /3DPE in acidic sweat conditions compares well with those achieved with other iron-based enzyme-free electrodes operated in neutral and basic media [28][29][30][31][32] (Table 1). The within-sensor reproducibility (stated as the % relative standard deviation (RSD) of ten repetitive responses at the 3DPE) was 4.8% for GLU and the between-sensor reproducibility (expressed as the % RSD at six different 3DPEs) was 8.1% (both at the 250 µmol L −1 GLU level), revealing high precision of the modified 3DPEs responses.  Figure 7 presents the DPV responses and the calibration plot of GLU on the Fe3O(PhCO2)6(H2O)3]•PhCO2/3DPE in the concentration range of 25 to 500 μmol L −1 in 0.1 mol L −1 PB (pH 4). The voltammetric response of the 3D printed sensor to GLU oxidation increased linearly with increasing GLU concentration, with a correlation coefficient of 0.998, and the calibration curve fell within the physiological levels of GLU secreted in human sweat [12]. The limit of detection (LOD) was calculated by the equation LOD = 3 sy/a, where sy is the standard deviation of the y-residuals of the calibration plot, and a is the slope of the calibration, which was 4.3 μmol L −1 . The LOD of GLU achieved with the Fe3O(PhCO2)6(H2O)3]•PhCO2/3DPE in acidic sweat conditions compares well with those achieved with other iron-based enzyme-free electrodes operated in neutral and basic media [28][29][30][31][32] (Table 1). The within-sensor reproducibility (stated as the % relative standard  deviation (RSD) of ten repetitive responses at the 3DPE) was 4.8% for GLU and the tween-sensor reproducibility (expressed as the % RSD at six different 3DPEs) was 8 (both at the 250 μmol L −1 GLU level), revealing high precision of the modified 3DPEs sponses.

Interference Study
For enzyme-free GLU sensors, selectivity is a key factor for their noninvasive ap cations, as the sensors can be subject to interferences by other biomarkers (such as u uric acid, and lactic acid) co-existing in sweat that can impact the precision of GLU m toring. To examine the selectivity of Fe3O(PhCO2)6(H2O)3]•PhCO2/3DPE in sweat, a c centration of 220 mmol L −1 urea, 250 μmol L −1 uric acid, and 55 mmol L −1 lactic acid added separately and together in the artificial sweat, and their effect on the DPV oxida peak of 200 μmol L −1 GLU was studied [12,43]. As shown in Figure 8A the sweat omarkers did not cause any statistically significant effect on the DPV GLU oxidation p demonstrating the satisfactory selectivity 3D printed GLU sensor to other common existing biomarkers in sweat.

Application to Artificial Sweat
In order to assess the applicability of the enzyme-free method in noninvasive b nalysis, three artificial sweat samples containing 120 μmol L −1 (Figure 8B

Interference Study
For enzyme-free GLU sensors, selectivity is a key factor for their noninvasive applications, as the sensors can be subject to interferences by other biomarkers (such as urea, uric acid, and lactic acid) co-existing in sweat that can impact the precision of GLU monitoring. To examine the selectivity of Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 /3DPE in sweat, a concentration of 220 mmol L −1 urea, 250 µmol L −1 uric acid, and 55 mmol L −1 lactic acid was added separately and together in the artificial sweat, and their effect on the DPV oxidation peak of 200 µmol L −1 GLU was studied [12,43]. As shown in Figure 8A the sweat biomarkers did not cause any statistically significant effect on the DPV GLU oxidation peak, demonstrating the satisfactory selectivity 3D printed GLU sensor to other common co-existing biomarkers in sweat.

Application to Artificial Sweat
In order to assess the applicability of the enzyme-free method in noninvasive bioanalysis, three artificial sweat samples containing 120 µmol L −1 (Figure 8B [12]. The standard addition method was applied for the determination of GLU in the sweat samples, calculating the respective recovery values. Satisfactory recoveries values for GLU were obtained ranging from 97 to 102%. These results demonstrate the accuracy of the 3D printed sensor modified with Fe(III)-cluster to sensitive and selective monitoring of GLU in sweat conditions. [12]. The standard addition method was applied for the determination of GLU in the sweat samples, calculating the respective recovery values. Satisfactory recoveries values for GLU were obtained ranging from 97 to 102%. These results demonstrate the accuracy of the 3D printed sensor modified with Fe(III)-cluster to sensitive and selective monitoring of GLU in sweat conditions.

Conclusions
In this work, we have developed a new type of 3D printed sensor modified with water-stable Fe3O(PhCO2)6(H2O)3]•PhCO2 for enzyme-free GLU monitoring in acidic epidermal sweat environment. The Fe(III)-cluster served as a Fe(III) precursor used in the electrocatalytic oxidation of GLU. The 3D printed sensor presented favorable electroanalytical action in the DPV selective determination of GLU, offering satisfactory reproducibility

Conclusions
In this work, we have developed a new type of 3D printed sensor modified with waterstable Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 for enzyme-free GLU monitoring in acidic epidermal sweat environment. The Fe(III)-cluster served as a Fe(III) precursor used in the electrocatalytic oxidation of GLU. The 3D printed sensor presented favorable electroanalytical action in the DPV selective determination of GLU, offering satisfactory reproducibility and very low LOD. These features combined with 3D printing technology set the presented sensor as an innovative addition to the arena of electrochemical transducers used for noninvasive bioapplications.  Table S4: List of Bragg positions (shown as blue dashes in Figure 1) from the Rietveld Analysis of [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 ; Figure S1: The IR spectrum (ATR) of [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 in the 450-4000 cm −1 range; Figure S2: Image of [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 dispersed in water on a glass slide; Figure S3: Histogram of particle sizes ranging between 50 and 230 µm (black bars), and Gaussian fit (red line); Figure S4: Thermogravimetric analysis (TGA) graph of [Fe 3 O(PhCO 2 ) 6 (H 2 O) 3 ]·PhCO 2 in the 25-800 • C temperature range. Reference [44] is cited in the supplementary materials.
Funding: This research has been co-financed by the European Regional Development Fund of the European Union and Greek national funds through the Operational Program Competitiveness, Entrepreneurship and Innovation, under the call RESEARCH-CREATE-INNOVATE (project code: T2EDK-00028/MIS 5067540).