Electrochemical multi-analyte point-of-care perspiration sensors using on-chip three-dimensional graphene electrodes

Abstract

Multi-analyte sensing using exclusively laser-induced graphene (LIG)-based planar electrode systems was developed for sweat analysis. LIG provides 3D structures of graphene, can be manufactured easier than any other carbon electrode also on large scale, and in form of electrodes: hence, it is predestinated for affordable, wearable point-of-care sensors. Here, it is demonstrated that LIG facilitates all three electrochemical sensing strategies (voltammetry, potentiometry, impedance) in a multi-analyte system for sweat analysis. A potentiometric potassium-ion-selective electrode in combination with an electrodeposited Ag/AgCl reference electrode (RE) enabled the detection of potassium ions in the entire physiologically relevant range (1 to 500 mM) with a fast response time, unaffected by the presence of main interfering ions and sweat-collecting materials. A kidney-shaped interdigitated LIG electrode enabled the determination of the overall electrolyte concentration by electrochemical impedance spectroscopy at a fixed frequency. Enzyme-based strategies with amperometric detection share a common RE and were realized with Prussian blue as electron mediator and biocompatible chitosan for enzyme immobilization and protection of the electrode. Using glucose and lactate oxidases, lower limits of detection of 13.7 ± 0.5 μM for glucose and 28 ± 3 μM for lactate were obtained, respectively. The sensor showed a good performance at different pH, with sweat-collecting tissues, on a model skin system and furthermore in synthetic sweat as well as in artificial tear fluid. Response time for each analytical cycle totals 75 s, and hence allows a quasi-continuous and simultaneous monitoring of all analytes. This multi-analyte all-LIG system is therefore a practical, versatile, and most simple strategy for point-of-care applications and has the potential to outcompete standard screen-printed electrodes. Graphical abstract.

Keywords: Electrochemical biosensor; Health-monitoring platform; Laser-induced graphene (LIG); Point-of-care (POC); Sweat sensor.

Graphical abstract

Fig. 1

Scheme of the WE of the amperometric biosensor after all modification steps. The laser-induced graphene on the flexible PI substrate is the base for the chemical deposited PB layer. The chitosan membrane fixes the PB layer while providing the polymer network on which the enzyme is immobilized. The sweat sample or any other solution is applied on top

Fig. 2

a Sensor connected to the EmStat Blue potentiostat via three crocodile clamps attached to the connection pads. For planar setup, a sample droplet of 30 μL which covers all three electrodes was added. There is a wireless connection of the potentiostat via Bluetooth® to the software application installed on a smartphone. b Experimental setup for detection of glucose and lactate with amperometric measurements on a chicken leg to simulate human skin and to provide a biological substrate. A piece of filter paper is used as simulated sweat collection pad. Modeling clay protects the contacts from moisture due to the shortened strands. The sensor is in a fixed position, whereas the chicken substrate with the applied sample can be moved up and down. c Combined LIG sensor. Connection pads are protected with adhesive copper tape against abrasion through the crocodile clamps. The strands are isolated by nail polish. 150 μL to 200 μL of sample solution is suitable to cover the active electrode area. d Schematic view of the combined LIG biosensor on polyimide substrate connected to two potentiostats and a multimeter. 150 μL to 200 μL of mixed samples containing all potential analytes is applied and can be measured quasi simultaneously after calibration of the single sensors

Fig. 3

a Dose-response curve of a potentiometric, planar LIG-based sensor with Ag/AgCl RE measuring KCl concentrations ranging from 10⁻⁶ to 1 mol L⁻¹ with the droplet method (N_sensor = 3). The linear range goes down to 1·10⁻⁵ mol L⁻¹ KCl and the slope is 96 ± 2 mV·dec⁻¹. The small standard deviation, represented as error bars, especially in the linear range, indicates a reproducible electrode fabrication procedure. b Dose-response curves of the potentiometric LIG sensor with Ag/AgCl RE. KCl samples were measured by application of droplets (gray boxes), using a filter paper as sweat collection pad (red circles), measuring sample droplets on chicken skin (blue triangle) and on chicken skin with filter paper (green triangle). c Dose-response curves of one sensor when different electrolytes containing K⁺ and Cl⁻ ions are measured to demonstrate the sensor’s sensitivity towards both species (gray boxes: KCl, red circles KH₂PO₄, blue triangles: KNO₃ green triangles: NaCl)

Fig. 4

Dose-response curves of all-LIG potentiometric sensors with an ISE versus different RE a Ag/AgCl pseudo-RE was protected by a Nafion/KCl/PVC layer and b an unmodified LIG was used as RE material to minimize the influence of chloride ions (gray boxes: KCl, red circles KNO₃, blue triangles: NaCl; n = 3)

Fig. 5

Dose-response curve of impedance measurements of KCl solutions with concentrations ranging between 5·10⁻⁴ and 1 mol L⁻¹ in a planar all-LIG setup (N_sensor = 3) with the droplet method (blue circles) and Whatman® 595 filter paper (red boxes) to simulate a sweat collection pad. The inset shows a zoomed-in section for the concentration range below the relevant range. Instrumental settings: fixed frequency of 1000 Hz with an AC amplitude of 10 mV and DC potential of 0 V

Fig. 6

a Exemplary time versus current curves of chronoamperometric characterization of a modified glucose biosensor with chemical deposited Prussian blue layer, 0.1% chitosan membrane, and theoretical GOx activity of 1 U mm⁻²(n = 3, error bars are hidden for clarity). The applied potential was − 50 mV vs. LIG. Run time was 60 s and the sample was applied as a droplet. The magnified cutout demonstrates that low glucose solutions can be distinguished from each other. b Dose-response curve (n = 3, SD represented by error bars, droplet method) resulting from the respective I vs. t curve after a run time of 60 s on half-logarithmic scale. LOD is 13.7 ± 0.5 μmol L⁻¹ and LOQ is 42 ± 2 μmol L⁻¹

Fig. 7

a Normalized dose-response curves of glucose in phosphate/citrate buffers at different physiological-relevant pH values (between 4 and 7). For each pH value, a new sensor was taken. A potential of E = 0 V vs. LIG was applied for all chronoamperometric measurements. All curves have a highly comparable shape independent of pH. Normalization of the signals from different sensors is necessary due to the varying sensitivity of the manually fabricated sensors. b Summary of important characteristics for the glucose sensors at different physiological-relevant pH ranges

Fig. 8

Dose-response curves of combined multi-analyte sensor for three determination methods under the same conditions as for the single-analyte sensor. All techniques show a comparable response as the single-analyte sensors. a The all-electrolyte concentration with the impedance measurement is highly reliable in the physiological-relevant range from 1 to 500 mmol L⁻¹. b A similar reliability within the relevant range is provided for the detection of potassium concentrations with the potentiometric sensor. The slope of the linear fit is 103 ± 1 mmol dec⁻¹. c Glucose determination with the amperometric setup. LOD of 135 ± 5 μmol L⁻¹ and LOQ of 410 ± 15 μmol L⁻¹ were obtained. Potentiometric determinations were performed with a d Nafion/KCl/PVC-protected RE and a e LIG RE