Epidermal Patch with Glucose Biosensor: pH and Temperature Correction toward More Accurate Sweat Analysis during Sport Practice

We present an epidermal patch for glucose analysis in sweat incorporating for the first time pH and temperature correction according to local dynamic fluctuations in sweat during on-body tests. This sort of correction is indeed the main novelty of the paper, being crucial toward reliable measurements in every sensor based on an enzymatic element whose activity strongly depends on pH and temperature. The results herein reported for corrected glucose detection during on-body measurements are supported by a two-step validation protocol: with the biosensor operating off- and on-bodily, correlating the results with UV–vis spectrometry and/or ion chromatography. Importantly, the wearable device is a flexible skin patch that comprises a microfluidic cell designed with a sweat collection zone coupled to a fluidic channel in where the needed electrodes are placed: glucose biosensor, pH potentiometric electrode and a temperature sensor. The glucose biosensor presents a linear range of response within the expected physiological levels of glucose in sweat (10–200 μM), and the calibration parameters are dynamically adjusted to any change in pH and temperature during the sport practice by means of a new “correction approach”. In addition, the sensor displays a fast response time, appropriate selectivity, and excellent reversibility. A total of 9 validated on-body tests are presented: the outcomes revealed a great potential of the wearable glucose sensor toward the provision of reliable physiological data linked to individuals during sport activity. In particular, the developed “correction approach” is expected to impact into the next generation of wearable devices that digitalize physiological activities through chemical information in a trustable manner for both sport and healthcare applications.

A 100 mg mixture of 1 %wt of hydrogen ionophore I (20 mmol kg -1 ), 0.6 %wt of NaTFPB (10 mmol kg -1 ), 33 %wt of PU and 65.4 %wt of DOS in 1ml of THF was used to prepare the cocktail for the pHselective membrane. The cocktail for the reference membrane (RM) was prepared by dissolving 30 mg of the ionic liquid [C8min + ][C1C1N − ] and 120 mg of PU in 1 mL THF, as previously described in the literature. 2 Voltametric measurements were performed with a potentiostat (Autolab, Metrohm Nordics AB, Sweden) against Ag/AgCl pseudo-reference electrode prepared in-house. Resistance measurements (in the T sensor) were performed using a digital multimeter (Rexman MS8230B). Electromotive force (EMF) was measured with a high input impedance (10 15 Ω) EMF16 multichannel data acquisition device (Lawson Laboratories, Inc.) against either a double-junction Ag/AgCl/sat. KCl/1M LiOAc reference electrode (6.0726.100, Metrohm Nordic, Sweden) or the solid-state reference electrode herein developed. The pH measurements for validation were performed with a pH-meter (914 pH/Conductometer, 2.914.0020, Metrohm) or a glass micropH electrode (biotrode, 6.0224.100, Metrohm) coupled to the previous pH meter. In the validation approach, glucose content in sweat samples was additionally analysed by UV-vis spectrophotometer (Specord 200 plus, Analytik Jena AG, Switzerland) and ion chromatrography (850 Professional IC, Metrohm AB, Sweden) with pulsed amperometric detection (IC-PAD) (945 Professional Detector Vario, Metrohm AB, Sweden) employing an anion exchanger column specific for carbohydrates (Metrosep Carb 2 -150/4.0, Metrohm AB, Sweden).

S4
A semi-automatic screen-printer machine (SPR-45 Automated SMT Stencil Printer, DDM Novastar, inc., USA) was used to print the electrodes over the polyesther substrate. Stencils were designed with Corel Draw software and fabricated at Coated Screens Scandinavian AB, Sweden. Silhouette Cameo cutter, USA, was employed to sharply cut the adhesive transfer tape. 3D printer Ultimaker 3, Ultimaker B.V., the Netherlands was used for the printing of the sampling cell. The design of the sampling cell was performed by Solid Edge ST9 solid modelling software (Siemens PLM software, Inc., USA). For the regional absorbent patch method (REG), commercial cotton pads were obtained from a local store (circular -shaped pads, 58 mm diameter). Hydrofilm roll water-resistant tape (10 cm x 10 m) (Hartmann GmbH, Germany) was purchased from a local pharmacy. Glucometer (Bayer Countour XT, Bayer AG, Germany) and glucose strips (Bayer Countour NEXT, Bayer AG, Germany) were purchased in a local pharmacy.
The raw signals for the electrodes obtained during the on-body tests was smoothed by means of the smooth data option of MATLAB (MathWorks) in order to remove spikes in the current/potential response. For the flow measurements, the solution flows through the microfluidic cell by means of a tubing (Tygon LMT-55, ISMATEC, Cole-Parmer GmbH, Germany) connected to a peristaltic pump (ISMATEC IPC series, Cole-Parmer GmbH, Germany).
Fabrication of the electrodes. The manufacturing of the electrodes consisted of a screen-printing process over a polyester substrate. The conductive path was printed by employing Ag/AgCl ink and was then cured at 100°C for 10 min. Subsequently, the carbon ink was printed to define the circular electrode surface, being cured at 80°C for 10 min. A layer of adhesive transfer tape was applied over the conductive tracks without covering the end part (1.5 mm long) to define the electrical connections to the reader. The size of the sensor patch was 10 x 18 mm, which includes the patterning of 7 electrodes, as displayed in Figures 1a-e in the main manuscript. Then, the screen-printed electrodes were modified to obtain the glucose biosensor as well as pH and T sensors.
For the glucose biosensor, first the deposition of the PB mediator on the WE was carried out: i) casting of 10 µL of 0.1 M potassium ferricyanide, ii) adding 10 µL of 0.1 M of iron(III) chloride (both salts dissolved in 0.01 M of HCl and 0,1M KCl, iii) mixing both solutions on the electrode surface and let them react during 20 min to form the PB film. After this time, the film was rinsed with 0.01 M HCl solution. Subsequently, the electrode was placed in an oven at 100°C during 1 h for annealing. Finally, the PB was electrochemically activated in 0.1M KCl employing cyclic voltammetry from -0.2V to 0.4V at scan-rate of 40 mV s -1 until a stable CV is obtained (ca. 10 scans). After the PB film, a volume of 1.5 µL of a solution containing 10 mg mL -1 GOx and 0.66%wt of CHI is deposited and let it dry for 20 min. Thereafter, a 0.5 µL 1% Nafion was drop cast on the WE and let it dry for 20 min. Finally, the glucose biosensor was stored at 4 °C when not in use. The Ag/AgCl RE and C-based CE were used directly after being screen-printed without any further modification. None of these electrodes needed for conditioning. However, the glucose biosensor was exposed to artificial sweat for some minutes for each calibration graph prior to on-body tests.
For the pH sensor, a first film of multiwalled carbon nanotubes (MWCNTs), acting as ion-to-electron transducer, 3 was deposited on top of the screen-printed carbon electrode by drop casting 10 x 1 µL of a MWCNTs dispersion in THF (1 mg mL -1 ), waiting 2 min for each drop to be totally dried (i.e. THF evaporation) before the next deposition. Subsequently, the pH selective membrane or reference membrane (RM) were deposited by drop casting 8 x 1 µL or 3 x 2.5 µL on top on the MWCNTs layer, respectively. Each layer was dried for 2 min before depositing the next drop. Finally, both membranes S5 were left to dry at room temperature overnight. After that, both electrodes were conditioned overnight in solution of pH 4.5.
For the temperature sensor, a volume of 0.5 µL of 1 mg mL -1 MWCNTs dispersion in THF was drop casted in between two Ag/AgCl patterns. Finally, the electrode was dried in the oven for 60 min at 100°C to remove any water content.
Evaluation of the electrode performance before its incorporation in the microfluidic cell. In preliminary experiments, calibration curves using the developed sensors were performed by immersing the electrodes into the corresponding solution and using the standard addition method. For electrical connections, the screen-printed electrodes were connected to aluminum wires by employing a conductive tape (3M XYZ 9712 electrically conductive tape), and subsequently to the potentiostat/potentiometer/multimeter through the corresponding outputs. All the experiments for the evaluation of the glucose biosensor and the pH electrode were carried out at room temperature of 201C and under constant stirring of 400 rpm (stirrer IKA COLOR SQUID S000, IKA, Germany) by applying a constant potential of -0.05V versus Ag/AgCl reference electrode. Regarding the temperature sensor, the T in the solution was changed by using an external water bath and a heating plate (stirrer IKA RCT Basic, IKA, Germany).

Fabrication of the microfluidic cell.
A microfluidic cell based on PU (TPU 95A, Ultimaker Material 1756, Ultimaker B.V., the Netherlands) was fabricated using 3D printing technology (Ultimaker 3, Ultimaker B.V., the Netherlands). This consisted of a round shape (30 mm diameter) with a hole (0.75 mm diameter) connected to a rectangular microfluidic channel where the sensor array is placed (20 mm x 2 mm x 0.1 mm). The hole represents a circular collection zone of 6.5 mm diameter and 50 µm thickness made by the adhesive transfer tape (3M™ 9471LE) (shape cut with Cameo cutter). The microfluidic cell was first tested at controlling flow conditions (i.e. sensor array integrated with the sampling cell) by mimicking perspiration rate using a peristaltic pump (ISMATEC IPC series, Cole-Parmer GmbH, Germany). The tubing from the pump was fixed to inlet of the sampling cell by adhesive transfer tape.
On-body tests. Once the electrodes are implemented in the microfluidic cell by means of adhesive transfer tape, the device is ready for portability and on-body measurements of glucose, pH and T in sweat during perspiration. The on-body evaluation of the wearable device developed in the present work was performed in compliance with European and Swedish legislations and fundamental ethical principles, including those reflected in the Chapter of Fundamental Rights in the European Union and the European Convention on Human Rights and its Supplementary Protocols. 4,5 Furthermore, an informed consent was agreed with each subject before the on-body tests. This informed consent describes the aims, methods and implications of the research, the nature of the investigations and any benefits, risks or discomfort that may ensue during the test. The main objective of this document is to explicitly state that the participation is voluntary and that anyone has the right to refuse to participate at any time of the process without any consequences. Finally, this informed consent will state how data will be treated (anonymously), used and protected during the project.
Each on-body test consists of applying the skin patch on the forehead of the subject by attachment with adhesive transfer tape after cleaning the forehead with ethanol and water. The environmental conditions were measured to be 20-22°C and 50-75% humidity during all the tests. Each subject performed a workload cycle in a static bike (Technogym Forma bike, Technogym AB, Sweden) to produce active perspiration during 50-90 min. This trial consisted of a 10 min ramp up, 30-70 min of medium-high activity (ca. 120-130 bpm) and 10 min of cooling down. Heart rate monitoring was performed with a Wahoo TICKRx device embedded in a chestband. Bluetooth data was captured with a smartphone and Wahoo app.
The wearable device was connected through conventional cables to each reader (i.e, potentiostat, potentiometer and multimeter for the glucose biosensor, pH sensor and T sensor, respectively) close to the static bike. In every on-body test, the glucose and pH electrodes were previously calibrated using the microfluidic pump to convert the acquired signals into dynamic glucose concentration and pH values. Four standards were used for the calibration graph: 10, 50, 100 and 200 for glucose and 4.5, 5.5, 6.5 and 7.5 for pH.

Procedure for sweat collection for validation of on-body measurements.
There is not a universal protocol for sweat collection and it might vary from laboratory to laboratory. [6][7][8] However, the most used method for sweat collection during exercise is the regional absorbent patch method (REG). The method used in our laboratory employs commercial cotton pads (half of a circular-shaped pad, 58 mm diameter) that were attached to the forehead skin by Hydrofilm water-resistant tape (60 x 50 mm). After 10 min period of sweat collection, the patch is unstuck form the skin, demounted from the tape and immediately squeezed with a syringe to obtain the sweat sample, which is deposited in an Eppendorf tube, and finally, kept at 4°C until further validation analysis. Cotton pads where previously rinsed with doubledistilled water, dried in the oven at 80 °C for 2 h and stored before use. Notably, 10 min of sweating is enough to collect a sample volume of at least 100µL.
Sweat was also collected via iontophoresis induction. A pilocarpine gel was employed as a drug to induce sweating. The device used was Macroduct® Advanced, ELITechGroup, USA which includes the reader, the electrodes, the gels and the sweat collector. The glucose levels observed in sweat samples collected via iontophoresis were used in the correlation between sweat and blood glucose.
Validation method. Off-body validation consisted of the analysis of sweat samples collected by REG using a gold standard method (UV-Vis and/or IC) and comparing with the results to those observed using the wearable glucose sensor operation 'off-body', i.e., by introducing the collected sweat by means of the peristaltic pump. On-body validation consisted of the analysis of sweat samples collected by REG using a gold standard method (UV-Vis and/or IC) and comparing with the results provided by the wearable glucose sensor during on-body tests.
Sweat glucose was determined by UV-Vis spectroscopy by monitoring the NADH formed in the enzymatic reaction of glucose with glucose dehydrogenase (GHD). The enzymatic reaction was accomplished by adding the cofactor NAD + and GDH into the sweat sample and letting 1-h at 25°C for the reaction to be occurred. The absorbance was acquired at 340 nm, as reported elsewhere. 9 Samples were injected (100 µL), and glucose was separated employing a Metrosep Carb 2 -150/4.0. The flow rate was set at 0.4 mL min -1 . The eluent used was 0.3 M NaOH. The column compartment was maintained at 40°C during the analysis. Tables   Table S1. Glucose concentration in sweat samples determined by the wearable glucose biosensor under off-body operation and by the UV-Vis method. The difference between the values provided by both methods is given in %. Notably, because these measurements were accomplished at room temperature, no T correction was assessed.              S16 Figure S10. Dynamic profiles for pH and T acquired during different on-body tests. S17

Concrete example for pH and T correction
The calibration parameters (linear fitting) for a glucose biosensor at pH 6.6 and 20°C of temperature are: Slope (k, nA µM -1 ) = 1.1 (S1) Intercept (m, nA) = 19.0 (S2) For the calculation of the correction factors for the slope and intercept according to pH changes ( , we will consider that the pH on-bodily measured at a certain instant of ) the test is 5.7. Then, we calculate the factors at pH 5.7 by interpolating the experimental factors calculated at pH 5.5 and 6.6 ( 5.5 = 0.92, 6.6 = 1.0, 5.5 = 3.42 5.5 see Figure 4c in the main manuscript). The used linear fittings are: = 1.0, for the slope factor: (S3) 5.5 -6.6 = 0.073 + 0.52 for the intercept factor: (S4) 5.5 -6.6 = -2.2 + 15.52 Based on these equations, the factors at pH 5.7 are calculated to be: (S5) 5.7 = 0.94 (S6) 5.7

= 2.98
The same reasoning is used for the calculation of the correction factors for the slope and intercept according to T changes ( ). We consider that the T on-bodily measured at the same instant that the pH is 32°C. The linear fittings used for the interpolation were obtained from the experimental factors at 30 and 33°C ( 30 = 1.36, 33 = 1.42, , see Figure 4d in the main manuscript):

S18
Next, using the factors S5, S6, S9 and S10, the corrected slope and intercept are calculated according to equations 5 and 6 in the main manuscript and considering the initial calibration graph of the glucose biosensor (pH 6.6 and 20°C): (S11) = × 5.7 × 32 = 1.1 × 0.94 × 1.4 = 1.45 (S12) = × 5.7 × 32 = 19 × 2.98 × 1.63 = 92.70 The final step is the calculation of the glucose concentration considering the corrected slope and intercept for the calibration graph as well as the on-bodily measured current value of 126.0 nA (at pH of 5.7 and 32°C): (S13) These calculations were integrated in a Matlab code that allows for the calculation of glucose in a dynamic way according to every instant change in sweat pH and T during on-body measurements: the reader is referred to Figure 5f in the main manuscript for an entire on-body glucose profile obtained before and after the application of the detailed correction.