Ortega et al. Microsystems & Nanoengineering (2019)5:3 https://doi.org/10.1038/s41378-018-0043-0 Microsystems & Nanoengineering www.nature.com/micronano ARTICLE Open Access Self-powered smart patch for sweat conductivity monitoring 1 Laura Ortega , Anna Llorella1, Juan Pablo Esquivel1 and Neus Sabaté1,2 Abstract A self-powered skin patch for the measurement of sweat conductivity is presented. The key component of the patch consists of a paper battery that is activated upon absorption of sweat. This body fluid acts as the battery electrolyte, the conductivity of which has a direct impact on the battery-generated output power and voltage. This particular behaviour enables the operation of a very simple and robust conductivity sensor in direct current mode without needing an external power source. The device presented in this paper takes advantage of this new measurement method to develop a sweat patch for screening cystic fibrosis that operates with an extremely simple electronic circuit that minimizes its cost and environmental impact. The patch provides an unambiguous digital result that can be read in an electrochromic display and yields 95% sensitivity and 100% specificity when tested with artificial eccrine perspiration samples. Introduction prototypes that are primarily based on colorimetric 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Wearable devices for health condition monitoring have detection6,7, in which signal quantification requires digital attracted considerable interest in the sports and health- capture and analysis with a dedicated reader or a smart- care sectors in recent years. These devices provide real- phone. Due to its irreversible nature, these devices are time monitoring of physiological relevant parameters of designed to function for a short period and be disposed human well-being. Significant advances in wearable afterwards. More endurable and quantitative approaches solutions that measure physical parameters, such as heart that are based on their detection principle for electro- rate, bodily motion, skin temperature or respiration rate, chemical sensors have addressed the integration of elec- have already been achieved; many of these devices are trochemically active electrodes on flexible and skin- available in the market1–3. However, obtaining chemical friendly skin patches. However, only a few studies assess parameters from body fluids provides insight into a onsite signal processing circuitry and sensor calibration wearer’s health, performance or stress at the molecular mechanisms for an accurate analysis8–11. Recently, this level4. Sweat collection and analysis offers many ergo- device has been applied to cystic fibrosis monitoring, in nomic advantages over other substances, such as the which the wearable platform enables both sweat stimu- availability of sampling sites and their continuous access lation and chloride ions detection12. However, its com- and the ability to be stimulated on-demand with local plexity and the number of electronic components that are iontophoresis5. These advantages have motivated required to drive the sensors and extract the appropriate researchers to develop a first generation of interesting information renders it more suitable for long-term clinical and physiological investigations than for testing cystic fibrosis disease on individuals. Most of the approaches to Correspondence: Neus Sabaté (Neus.sabate@imb-cnm.csic.es) chemical sensing require complex electronic circuits with 1 Instituto de Microelectrónica de Barcelona, IMB-CNM (CSIC), C/del Til·lers. power-on board, which yields bulky and costly devices. Campus Universitat Autònoma de Barcelona (UAB), 08193 Bellaterra, Barcelona, The search for more compact and sustainable power Spain 2 Catalan Institution for Research and Advanced Studies (ICREA), Passeig Lluís sources has prompted the development of self-powered Companys 23, 08010 Barcelona, Spain © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 2 of 10 devices, in which energy to perform a measurement is skin. The generated power is handled by a very simple harvested from the environment or the biological system circuit that enables discerning between a healthy condi- under study. Numerous prototypes have been developed tion and a non-healthy condition. A flow chart of the to measure biological parameters by harvesting power device operation is shown in Figure S1. This circuit from different sources, such as breathing, heartbeat, combines inkjet printed conductive tracks, discrete elec- temperature or biological fluids13–16. tronic components and two electrochromic displays In this paper, we present a self-powered skin patch to (Control and Test output) to show the result. measure sweat conductivity and its application to screening cystic fibrosis with a novel and simple approach. Results Cystic fibrosis (CF) is a genetic autosomal recessive dis- Conductivity sensor description and working principle ease that induces mutations on a conductance trans- Liquid-activated batteries are devices that consist of at membrane regulator protein—cystic fibrosis conductance least two electroactive electrodes: an oxidizing electrode transmembrane regulator (CFTR)—which controls the (anode) and a reducing electrode (cathode). The electro- excretion of chloride in sweat. For this reason, patients des are connected by a hydrophilic material that is capable with CF present higher chloride contents in sweat than of holding fluid (Fig. 2a). This kind of battery starts to healthy subjects. Although most of the approaches focus function upon the addition of fluid, which acts as the on measuring chloride concentration, different studies battery electrolyte. Liquid used to activate the battery have demonstrated the effectiveness of the sweat con- operation is generally a water-based fluid, such as plain ductivity measurement as a screening test for CF17,18. water, beverages or any biological fluid (such as saliva, Conductivity measurements currently require AC elec- urine, blood and sweat). These batteries are primary tronics modules that render its integration on a disposable batteries that cease to function when their electrodes are patch impractical. In this paper, we show how con- exhausted. ductivity measurements of sweat can be approached in a The voltage delivered by a battery depends on the simpler way. thermodynamic voltage of the electrochemical reactions The key component of the patch consists of a paper involved in its electrodes. However, once the battery starts battery that is activated upon adding the fluid to be to deliver power, its voltage-intensity characteristic curve measured (Movie S1, Supporting Information). The fluid will depend on the internal resistance of the battery. This acts as an electrolyte, and for a particular configuration of parameter depends on the efficiency of the charge gen- the battery materials and design, the battery output is fully eration processes that occur at the battery electrodes and dependent on the conductivity of the liquid sample that is the charge transport across the electrolyte and the resis- poured in its paper core. This approach enables the sensor tive elements of the battery. and battery to be merged in a single element—a battery Charge transport inside the battery occurs through sensor—whose generated power is directly related to the different media; charge carriers in the metallic compo- conductivity of the sample to be analysed. Compared with nents are electrons, whereas current in the electrolyte state-of-the-art conductivity sensors, the battery sensor consists of ion transport. Despite its different nature, both presents two main advantages: (1) its response to con- charge transport phenomena can be modelled by resistors ductivity can be characterized by a direct current (DC) (Fig. 2a). Ra and Rc account for the ohmic resistivity of the that enables a radical simplification of the electronic anode and cathode materials, whereas Relectrolyte repre- module to operate it, and (2) if desired, the sensor sents the ionic resistivity of the electrolyte. The charge response can be obtained without an additional power transfer between the ohmic conductors and the electro- source; thus, it can be operated as a self-powered sensor. lyte occurs by the oxidation and reduction reactions at the This new approach to conductivity sensing breaks the electrodes. These processes can be modelled with a par- paradigm sensor-electronics-battery and converts it into allel combination of a resistor (Rt) and a capacitor (Cd), something simpler as the sensor and battery are merged where Rt accounts for the faradaic resistance, which into the same element. This paper shows how this new represents the kinetics of the electrochemical reactions, measuring concept can be successfully employed to con- and Cd accounts for the double-layer capacitance, which struct an autonomous screening patch that can deliver a reflects the capacitive nature of the interface between the digital output using a minimal number of discrete elec- electrode and the electrolyte. Contribution of the faradic tronic components that can be completely printable. resistance Rt to the battery internal resistance is relevant The patch has been conceived for application on the at low currents and becomes practically negligible at forearm of a patient and yields a “positive” result in the moderate and high currents of the cell, which enables the case in which an abnormal conductivity level of sweat is ohmic region of the battery I–V curve, where the total detected (Fig. 1). The patch remains quiescent until the internal resistance can be attributed to the resistivity of sweat is absorbed by the paper that is in contact with the the electrodes and the electrolyte, Equation (1). For a fixed Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 3 of 10 a Positive b Control Test Top Bottom Fig. 1 Device design and implementation. a Scheme of the designed patch and theoretical working scenario. b Final implemented device. The Control and Test display are shown in the top face, and the battery stack and inkjet printed circuit with the discrete components are observed in the bottom face (scale bar = 1 cm) configuration of the battery materials and geometry, by a controlled chlorination of a screen-printed silver changes in the internal resistance of the battery at the track. This process has allowed us to obtain highly ohmic region of the polarization curve can be solely reproducible battery cathodes. The electrodes—which are attributed to changes in the conductivity of the electro- placed side by side—and the area between them are lyte. The resistance of the electrolyte depends on the covered with two layers of glass fibre-based paper with a geometry and porosity of the hydrophilic core, the area total thickness of 0.5 mm and an area of 60 mm2. The and position of the electrodes and the conductivity of the glass fibre paper is a very suitable material due to its high liquid used to activate the battery, which will be inversely water absorption rate and its high porosity, which enables proportional to the resistance of the cell. holding a liquid volume of 15 µL. When the liquid sample B to be characterized is deposited on the paper, it is Rinternal ¼ Ra þ Rc þ Relectrolyte ¼ A þ ð1Þ absorbed by capillarity until the paper is completely σ saturated. Unlike other candidates, such as cellulose, this The battery developed in this study consists of two material does not experience significant swelling after coplanar electrodes with the dimensions 2.5 × 5 mm2. water is absorbed, which enables keeping the geometry of Each electrodes were constructed of magnesium (anode) the electrolyte constant. and silver chloride (cathode) mounted on top of a pres- Battery operation starts once the paper is completely sure sensitive adhesive layer and separated by 1.5 mm filled with the sample. The basic electrochemistry of this (Fig. 2b). Magnesium electrodes are cut from a 100-µm- battery has been previously employed in seawater- thick foil, whereas silver chloride electrodes are obtained activated batteries19,20. The reactions involved in the Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 4 of 10 paper battery are expressed as follows: Self-powered patch design rational and circuit validation  The sweat patch has to operate within a conductivity Anode : Mg ðsÞ ! Mg 2þ ðaqÞ þ 2e range from 5 mM equiv NaCl to 160 mM equiv NaCl Cathode : AgClðsÞ þ e ! Ag ðsÞ þ Cl ðaqÞ (equiv NaCl), as these limits are limits of interest and Total : Mg ðsÞ þ 2AgClðsÞ ! Mg 2þ ðaqÞ the measurement units as defined in the Diagnostic þ2Cl ðaqÞ þ 2AgðsÞ Sweat Testing Guidelines from the Cystic Fibrosis Foundation22,23. Body sweat is a complex mixture of cumulative secre- The standard voltage for this cell is 2.59 V. However, the tions from eccrine, apocrine and sebaceous glands, which open-circuit voltage (VOC) in neutral media is set from contain various electrolytes, ions, amino acids, proteins 1.5–1.7 V. This departure from the theoretical voltage is and lipids. Although the composition of sweat depends on attributed to the high polarisation voltage caused by the many factors, such as the sweat production rate and the build-up of an oxide layer on the Mg film. To test the physiological state, the lipid content of sweat generated on impact of the ionic conductivity on the performance of the forearm—the area selected for cystic fibrosis screening our battery, we recorded the I–V curves when filled with —is present in concentrations below 1%. With the water-based solutions that contain different NaCl con- exception of ionic species, most of the components pre- centrations that establish liquid conductivities from 0.5 sent in sweat are neutrally charged and have a negligible mS cm−1 to 40 mS cm−1 (Fig. 2c). The polarization curves effect on conductivity. The ionic content varies from show a visible influence of ionic conductivity of the liquid patient to patient. In the past 20 years, different series of sample that is used as an electrolyte on the resistivity of clinical studies aimed to determine the interferences of the cell. This finding is observed in the ohmic region of different species aside from Cl− and Na+ in sweat con- the curves (currents above 0.2 mA). At low current values ductivity have been performed17,18,24,25. The studies ana- (near OCP voltage), large activation losses related to the lysed the ionic content of hundreds of patients and pre-existing passivating layer of magnesium oxide in the concluded that the conductivity of sweat approximates anode can be observed. The effect of this passivating layer very closely to the sum of its sodium and chloride ions becomes less prominent at increasing conductivities as with a tolerance margin that accounts for conductivity the presence of chloride ions increases the rate of anodic variations caused by the presence of secondary ions dissolution kinetics21. At higher current densities, the (Table S1). The patch has been designed to operate as a battery shows concentration losses due to a limitation in screening device that generates a positive result with the AgCl reduction reaction rate at the cathode. Internal sweat conductivities that exceed 60 mM equiv NaCl. For resistance of the battery (obtained from the slope of I–V result visualization, electrochromic displays were identi- curves at the ohmic region) versus the ionic conductivity fied as the most suitable technology due to their com- of the electrolyte has been computed. The results indicate patibility with flexible substrates, their fabrication with that the experimental data follow the expected σ−1 screen-printing techniques and their capability to operate dependence within the conductivity range between 2.5 with voltages as low as 0.6 V. and 40 mS cm−1 (Fig. 2d). At conductivities below 2.5 mS Based on simplicity, the operating principle of the bat- cm−1, the battery performance is dominated by large tery consisted of connecting the purely resistive load activation losses, and therefore, the internal resistance RLOAD (Fig. 3a). If the selected RLOAD enables the battery dependence on conductivity departs from ideal behaviour. to operate inside the ohmic region of its polarization Although the influence of the liquid conductivity on the curve, the voltage generated by the battery is directly internal resistance of the battery has been visualized by its dependent on the sweat conductivity that is used to impact on the polarization curves, the change in battery activate it. Based on this requirement and the need to performance can be obtained using a large variety of well- ensure a minimum output voltage of 0.6 V, which is known electrochemical characterization techniques (such required to turn on the Control Display at the lowest as chronoamperometry, chronopotentiometry, electro- conductivity range (5 mM), the patch required the chemical impedance spectroscopy, current-interrupt stacking of two battery units. The value of RLOAD was measurements). However, one of the most interesting fixed to the lowest value of 2 kΩ, which enables the bat- features of the presented battery is that the power it tery to be operated at its maximum output power. The delivers is proportional to the conductivity of the liquid generated power ranges from 0.2 mW to 2 mW within the sample, and therefore, it can be operated as a self-powered conductivity range of interest (Fig. 3b). sensor (Fig. 2e). This last feature makes the battery par- The patch performance was designed to follow the User ticularly suitable to sense conductivity in portable and protocol for Evaluation of Qualitative Test Performance disposable applications, as it removes the requirement of (EP12-A2) dictated by the Clinical and Laboratory Stan- an additional power source. dards Institute26. The protocol recommends evaluating Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 5 of 10 a b Liquid sample Anode Cathode Paper Paper Cda Relectrolyte Cdc Ra Rtc Rc Rta Anode Cathode c 2.0 d 2.0 e 2.5 0.5 mS.cm–1 10 mS.cm–1 Experimental Rinternal 1 mS.cm–1 15 mS.cm–1 2.5 mS.cm –1 20 mS.cm –1 Best fit 1.6 Resistance (kΩ) 2.0 5 mS.cm–1 40 mS.cm–1 1.5 Voltage (V) Pmax (mW) 1.2 1.5 1.0 0.8 1.0 0.5 0.4 0.5 0.0 0.0 0.0 0.0 0.4 0.8 1.2 0 10 20 30 40 0 10 20 30 40 Current (mA) Conductivity (mS.cm–1) Conductivity (mS.cm–1) Fig. 2 Basic concepts of the battery. a Battery model. b Image of a single battery. c Polarisation curves of a single battery with different conductivity of electrolytes. d Resistance of the battery in the ohmic region of the polarization curve as a function of the conductivity of the electrolyte. e Maximum power output of the battery for different electrolyte conductivities the compliance of the screening test by statistically testing conductivity of the sweat sample. The Test Display is its outcome at the threshold value (60 mM) and at a cut- activated when the battery yields a voltage that is equal to off value that corresponds to the −20% of the threshold or above the CF-positive threshold voltage. Two diodes value (48 mM in this case). are connected to the displays to prevent their discharge The polarisation curves of a certain number of statisti- once the battery ceases to operate. The Test Display cally relevant battery stacks (N = 40) activated with activation is controlled with an n-type MOSFET transis- deionized water samples with NaCl concentrations of the tor that enables current to flow when the voltage applied threshold and cut-off values were recorded (Fig. 3c). to its gate (Vgate) is higher than 1.2 V (Fig. S3, Supporting Although the populations appear separated, an over- Information). To match the CF-positive threshold voltage lapping of the voltage output of the battery was observed. of the battery stack with Vgate, RLOAD was split between Since the objective of the patch is to perform a resistor R1 and resistor R2. screening measurement, a decision strategy was needed to The accumulated threshold charge value that provided a set a suitable voltage threshold that would yield a positive distinguishable display activation at a minimum working result in the Test Display. To priorities the minimization potential of 0.6 V was set to 10 µC and 40 µC for the of false negative results, the threshold voltage was placed Control display and Test display, respectively. at −5% of the Gaussian distribution of 60 mM equiv NaCl The circuit was tested on a protoboard and validated (1.51 V), which ensures that 95% of ill patients yield with samples with relevant conductivities of 5 mM— positive results. Due to the overlapping of the populations, minimum conductivity value, 48 mM and 60 mM (all in this decision would yield 12% false positive results of equiv NaCl). The activation of the Control Display at healthy patients. 5 mM equiv NaCl was checked. At this concentration, the Based on this decision, the logic of the device was battery yielded a voltage of 0.6 V during the first 60 s designed following the criteria of a minimum number of (Fig. 3e), which produced an accumulated charge of 50 μC electronic components (Fig. 3d), which would minimize that activated the Control Display. The patch perfor- both cost and environmental impact after disposal. mance was tested with triplicated samples of 48 mM equiv (Fig. S2, Supporting Information). The Control Display is NaCl and 60 mM equiv NaCl, respectively. The evolution directly connected to the battery stack and it ensures that of the voltage at the gate terminal of the MOSFET tran- the patch is working properly regardless of the sistor and the current that flows towards the Test Display Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 6 of 10 a 4 b 2.5 3 c 12 6.5 KΩ 48 mM NaCl Ohmic region 60 mM NaCl 2.0 3 2 8 Power (mW) Voltage (V) Voltage (V) 1.5 Counts 2 2 KΩ 1.0 1 4 1 0.5 0 0.0 0 0 0.0 0.4 0.8 1.2 0 40 80 120 160 1.3 1.4 1.5 1.6 1.7 Current (mA) Equivalent NaCl concentration (mM) Voltage battery (V) d Healthy patient sweat Non healthy patient sweat + Vgate + Vgate Positive Vσlow – Vσhigh– Control display Test display Control display Test display ON OFF ON ON e f 1.6 h 1.0 1.2 160 48 mM eq NaCl Voltagegate(V) 5mM eq NaCl 60 mM eq NaCl Control display current (μA) Threshold voltage 1.2 120 Battery voltage (V) Charge (μC) 0.5 0.6 0.8 80 g 4 48 mM eq NaCl Current (μA) 60 mM eq NaCl Threshold charge 40 2 0.0 0.0 0 0 15 30 45 60 0 48 60 Time (s) 0 30 60 90 120 150 Concentration (mM eq NaCl) Time (s) Fig. 3 Operating principle of the self-powered patch. a Battery stack calibration with samples at increasing concentrations of 5 mM, 10 mM, 20 mM, 40 mM, 48 mM (in dark blue), 60 mM (in dark yellow), 80 mM, 120 mM and 160 mM NaCl. The battery stack calibration showed that RLOAD ranges from 2 kΩ to 6.5 kΩ, which enabled setting the battery operation within its ohmic region. b Voltage and power generated by the battery when introducing an RLOAD of 2 kΩ. c Gaussian distribution around central values of 1.46 V and 1.57 V, which corresponds to 48 mM equiv NaCl and 60 mM equiv NaCl, respectively, when the battery was submitted to an RLOAD of 2 kΩ. d Circuit operation when healthy and non-healthy sweat samples are analysed. e Voltage and current provided by the battery and collected in the Control Display, respectively, at the lowest conductive sample. f, g Operation of the whole device with samples at 48 mM and 60 mM equiv NaCl in voltage at the gate terminal of the transistor and the current stored in the Test Display. h Charge accumulated in the Test Display at the two concentrations of NaCl was monitored during the tests (Fig. 3f, g). A steady be highly reproducible with a variation coefficient less decline in the voltage delivered by the batteries under than 3.5%. Once the battery is saturated with a conductive continuous operation is observed. This decline is attrib- liquid solution, the voltage applied to the transistor uted to an increase in the internal resistance caused by the increases to a value that depends on the conductivity of consumption of the battery electrodes during device the sample. For 60 mM equiv NaCl, this voltage is higher operation (Fig. 3f). However, this behaviour has proven to than the transistor threshold voltage (1.2 V, represented Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 7 of 10 with a dashed line), which causes current to flow towards The final device was validated with artificial eccrine the Test Display. The voltage is sustained above the perspiration, which mimics human eccrine sweat and transistor threshold for ~ 60 s but subsequently decays consists of 19 amino acids—the most abundant minerals due to the progressive exhaustion of the battery electro- and metabolites at concentrations that match experi- des. At 48 mM equiv NaCl, the voltage at the transistor mentally determined values for adult human eccrine gate never attains the threshold voltage value; thus, the sweat (Table S2). The purchased artificial sweat presented Test Display remains switched off. The current that flows a conductivity of 80 mM equivalent of NaCl. Thus, the towards the electrochromic display during the system conductivity for each sample was adjusted to 48 mM operation in both cases has been depicted (Fig. 2g). A equiv NaCl and 60 mM equiv NaCl by diluting the sam- significant current flows towards the Test Display for all ples with deionized water. tested samples at 60 mM. Although no current is expected To be compliant with the EP12-A2 test protocol, the to flow towards the display at 48 mM, a small current was validation of the devices was performed with 40 different observed. This current corresponds to a leakage current devices, with which 20 samples at conductivities of 48 and of the transistor (Fig. S3, Supporting Information) at 60 mM equiv NaCl were employed. The voltage was voltages between 1.15 V and 1.2 V. This current was recorded at the gate terminal of the MOSFET transistor unable to switch on the Test Display as the accumulated (Fig. 4b). Samples that represent “healthy” patients yielded charge did not exceed the threshold charge to generate voltages under the transistor threshold voltage in all tes- any significant change in the active segments contrast ted devices. Conversely, samples at higher “non-healthy” (Fig. 3h). conductivity values yielded voltages above the threshold, which enabled the current to flow towards the Test Dis- Self-powered patch fabrication and testing play. The current that flows towards this display in all The patch is conceived to be single use and completely tested cases was translated to the accumulated charge in disposable. All components were integrated in a compact the Test Display (Fig. 4c, d). All healthy samples generated self-contained skin patch (Fig. 4a). The patch incorporates charges less than 40 μC, and therefore, did not turn on the the paper battery stack that absorbs 30 µL of the fluid display. The accumulated charges obtained when testing sample. Battery electrodes are placed at the inner face of the devices with the threshold conductivity value of the sampling paper. In this configuration, the sample is 60 mM ranged from 40 μC to 150 μC. The high dispersion forced to pass through the entire thickness of the paper in the recorded charge values is attributed to an existing core before reaching the electrode surface and activate the exponential relationship between the voltage gate and the chemical reactions. This process mitigates the risk of a current of the transistor in the working region selected in partial soaking of the paper battery core that would yield this application (region near the transistor cut-off value) an incorrect conductivity value. An electronic circuit (Fig. S3, Supporting Information). The contingency table design was inkjet printed using Ag ink on a flexible to compare the results of the test with the known diag- polyethylene naphthalate (PEN) sheet. Silver conducting nosis shows that our device yields 95% sensitivity and paste was used for the hybridization of SMD electronic 100% specificity (Fig. 4e). components and the battery terminals. The area sur- rounding the battery was encapsulated with PDMS elas- Discussion tomer to protect the electronic circuit and avoid any short The difference in performance of the different patches circuit caused by occasional sweat overflooding. The has been assessed in the paper by testing the variability of elastomer also provides a flat surface that improves skin the fabricated batteries with NaCl solutions (Fig. 3c), contact and avoids sweat evaporation. Two custom-made testing the response of the batteries after connecting the electrochromic displays (Ynvisible, Vancouver, Canada) minimal electronic circuit (Fig. 3h) and checking the were glued at the rear of the flexible substrate of the performance of 40 ready-to-use patches with artificial circuit and were connected to the corresponding circuit sweat samples (Fig. 4d). The coefficients of variation terminals by a bias filled with conducting paste. A layer of obtained by characterizing the single battery at the dif- a medical-grade pressure-sensitive adhesive that was ferent targeted concentrations are less than 5%. However, coated on a stretchable nonwoven fabric was laser-cut to when the battery is connected to the electronics circuit assemble all components in an adhesive bandage form mounted in a protoboard, the coefficients of variation of factor and provide sealing conditions during sweat col- the response increases to 25% in the case of a low con- lection and measurement. ductivity value (48 mM) and 40% in the case of the upper This top side includes two windows for display visuali- threshold value (60 mM). This finding is attributed to the zation. Graphic features were printed on the top face of the fact that the system relies on a single MOSFET transistor adhesive to make it pleasant for small children. All com- to determine whether the voltage generated by the battery ponents were manually assembled using an alignment jig. corresponds to a positive response or a negative response. Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 8 of 10 a Positive Control Test Electrochromic Hybrid printed Paper-based Skin-friendly adhesive displays battery/sensor electronics b 1.4 d 40 Voltagegate (V) σ ∼ 48 mM eq NaCl Threshold voltage Threshold voltage σ ∼ 60 mM eq NaCl 1.2 Positive 30 Control Test 1.0 Counts c 4 20 σ ∼ 48 mM eq NaCl Current (μA) σ ∼ 60 mM eq NaCl Positive 2 10 Control Test 0 0 0 30 60 90 120 150 0 40 80 120 160 Time (s) Charge (μC) e Test Negative Positive Positive predictive value = 100% Present Negative predictive value = 95.2% Disease Sensitivity = 95% Specificity = 100% Absent Accuracy = 97.5% Fig. 4 Final device components and validation. a Exploded view of the final device with the different layers (scale bar = 1 cm). b, c Operation of the device with artificial sweat samples at the two conductivities of 48 mM equiv NaCl and 60 mM equiv NaCl. d Gaussian distribution of the accumulated charge at the two sample conductivities. e Final representation of the results provided by all devices compared with the known condition of the samples—healthy and unhealthy When the battery operates at voltages near the value of which indicates that the effect of introducing a more the threshold voltage of the transistor, the transistor complex matrix that mimics real sweat does not introduce operates in a transition region that limits the current additional dispersion to the results. generated by the battery. This source is the main source of The device presented in this paper shows the feasibility the variation recorded in the total current that flows of a screening patch for the diagnostics of cystic fibrosis towards the display elements. When 40 patches are tested with a novel method of conductivity measurement. The with artificial sweat samples that contain amino acids, use of a paper battery as a conductivity sensor enables a urea and different ions, the coefficients of variation clinically relevant parameter to be obtained with an slightly increase to 30% (at 48 mM) and 42% (at 60 mM), extremely simple electronic circuit that utilizes a small Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 9 of 10 number of components. Considering state-of-the-art Hybrid circuit fabrication printing technologies, the elements of the circuit can be The circuit was simulated with LTSpice (Analog Devi- entirely printed with inkjet printing processes. Combined ces, Norwood, MA, USA), designed with Kicad (Canonical with the screen-printing methods that are employed to Ltd, London, UK) and printed with the inkjet printer obtain the battery elements and the electrochromic dis- CeraPrinter X-Serie (Ceradrop, Parc d’Ester Technopole, plays, these processes pave the way to mass production of Limoges, France). The ink used for the inkjet printing was the device. In fact, these factors render the patch suitable a silver ink from DuPont PE-410 (DuPont, Wilmington, for a single use and disposable approach, as the low cost DE, USA), which was cured at 130 °C for 20 min. The associated with the materials and fabrication processes discrete components and electrochromic displays were and the use of a minimum number of electronic com- attached to the printed circuit with a silver epoxy EPO- ponents contributes to the sustainability of the device. TEK H20E (Agar Scientific Ltd, Essex, UK) by curing it at Additionally, in an alternative scenario, the battery sensor 80 °C for 3 h. can serve as a self-powered consumable component that can provide power and sensing capabilities for a battery- Electrochemical characterization less reusable electronic reader. This approach has been The electrochemical characterization of the battery was successfully demonstrated for glucose monitoring27. performed with a Gamry Potentiostat Reference 3000 AE Note that the dual use of the battery as a sensor and a (Gamry, Warminster, PA, USA). The open-circuit voltage power source renders the system energetically indepen- of the batteries was measured for 20 s, and a linear sweep dent, which is a main hurdle of a wearable device. The voltammetry at 10 mV/s was performed from the open- minimum power generated by the patch is 1 mW, which is circuit voltage to 0 V. Samples of NaCl (Sigma-Aldrich, St dissipated almost entirely in the resistive elements in the Louis, MO, USA) were prepared every day, and the con- current circuit design. In a more sophisticated version, ductivity and temperature of the sample was measured this power can be used to enable communication or data with the 914 pH/Conductometer (Metrohm AG, Ionen- storage functions. Despite the inherent impact of the strasse, Herisau, Switzerland). current development in the clinical sector, the con- ductivity patch can be potentially utilized to evaluate the Electrical characterization electrolyte content in sweat for other purposes, such as The electrical characterization of the system was per- dehydration that is induced by physical activity. formed with a 2400 SourceMeter (Keithley Instruments, Cleveland, OH, USA) and an HP34401A Multimeter Materials and methods (Hewlett-Packard Palo Alto, CA, USA) to measure the Battery fabrication current stored in the Test display and the voltage pro- The battery was designed with CorelDraw (Corel, vided by the battery, respectively, controlled with a cus- Ottawa, ON, Canada) and fabricated by rapid prototyping, tom LabVIEW program (National Instruments, N Mopac such as laser cutting (Mini 24, Epilog Laser, Golden, CO, Expy, Austin, TX, USA). USA). The device was mounted with pressure-sensitive adhesives of medical grade (Adhesive Research, Glen Artificial samples preparation Rock, PA, USA) by stacking all layers. The paper used to Artificial Eccrine Perspiration (Pickering Laboratories, absorb the sample was a LF1 Glass Fibre Filter from GE Mountain View, CA, USA) was used to perform final (General Electric, Boston, MA, USA). For the electrodes, validation of the device. The conductivity of the sample the Mg anode was obtained from GalliumSource LLC was adjusted to the equivalent conductivity of the healthy (Scotts Valley, CA, USA). The AgCl cathode was obtained and unhealthy subjects by measuring the conductivity by chlorinating an Ag electrode screen-printed over a PET with the 914 pH/Conductometer (Metrohm AG, Ionen- surface (DuPont, Wilmington, DE, USA). The Ag elec- strasse, Herisau, Switzerland) and adding deionized water. trode was screen-printed with a manual screen-printer Acknowledgements (PAYMSER, Barcelona, Spain), the silver ink LOCTITE R. Escudé and E. Ramón from the ICAS Department at IMB-CNM-CSIC are ECI 1011 E&C (Henkel, Düsseldorf, Germany) and cured acknowledged for their technical support in the printed electronic circuit at 150 °C for 15 min. The chlorination of the Ag electro- fabrication and design. N. Sabaté would like to thank the financial support received from the ERC Consolidator Grant (SUPERCELL-GA.648518). des was performed by introducing a constant current of 200 μA for 260 s with a Gamry Potentiostat Reference Authors' contributions 3000 AE (Gamry, Warminster, PA, USA) in a stirred All authors conceived the device concept. L.O. and A.L. performed fabrication solution of 0.1 M NaCl (Sigma-Aldrich, St Louis, MO, and characterization of the skin patch with direction from J.P.E. and N.S. All authors contributed to writing the paper. USA) with a half-cell configuration (Ag/AgCl reference electrode and Pt counter electrode (Metrohm AG, Conflict of interest Ionenstrasse, Herisau, Switzerland). The authors declare that they have no conflict of interest. Ortega et al. Microsystems & Nanoengineering (2019)5:3 Page 10 of 10 Supplementary information accompanies this paper at https://doi.org/ wearable platform. Proc. Natl. Acad. Sci. U. S. A. 114, 4625–4630 10.1038/s41378-018-0043-0. (2017). 13. Zheng, Q. et al. In vivo powering of pacemaker by breathing-driven implanted Received: 24 July 2018 Revised: 5 October 2018 Accepted: 13 November triboelectric nanogenerator. Adv. Mater. 26, 5851–5856 (2014). 2018 14. Zheng, Q. et al. In Vivo Self-powered wireless cardiac monitoring via implantable triboelectric nanogenerator. (2016). https://doi.org/10.1021/ ACSNANO.6B02693 15. Sun Kim, C. et al. Self-powered wearable electrocardiography using a wearable thermoelectric power generator. ACS Energy Lett. 3, 21 (2018). References 16. Shi, B., Li, Z. & Fan, Y. Implantable energy-harvesting devices. Adv. Mater. 1. Villar, R., Beltrame, T. & Hughson, R. L. Validation of the Hexoskin wearable vest 1801511 (2018). https://doi.org/10.1002/adma.201801511 during lying, sitting, standing, and walking activities. Appl. Physiol. Nutr. Metab. 17. Hammond, K. B., Turcios, N. L. & Gibson, L. E. Clinical evaluation of the mac- 40, 1019–1024 (2015). roduct sweat collection system and conductivity analyzer in the diagnosis of 2. Akintola, A. A., van de Pol, V., Bimmel, D., Maan, A. C. & van Heemst, D. cystic fibrosis. J. Pediatr. 124, 255–260 (1994). Comparative analysis of the Equivital EQ02 Lifemonitor with Holter 18. Mattar, A. C. V., Leone, C., Rodrigues, J. C. & Adde, F. V. Sweat conductivity: ambulatory ECG device for continuous measurement of ECG, heart rate, an accurate diagnostic test for cystic fibrosis? J. Cyst. Fibros. 13, 528–533 and heart rate variability: a validation study for precision and accuracy. (2014). Front. Physiol. 7, 391 (2016). 19. Carson W.N. & Fischer W.H. Sea water-activated primary battery. US Patent 3. Dadosky, A. et al. Telemanagement of heart failure patients across the post- 3352718A (1963). acute carecontinuum. Telemed. e-Health 24, 360–366 (2018). 20. Koontz, R.F. & Lucero R.D. Magnesium water-activated batteries. Handbook of 4. Bandodkar, A. J., Jeerapan, I. & Wang, J. Wearable chemical sensors: present batteries (Linden D & Reddy T. ed.) Chapter 17 (2002). challenges and future prospects. ACS Sens. 1, 464–482 (2016). 21. King, A. D., Birbilis, N. & Scully, J. R. Accurate electrochemical measurement of 5. Heikenfeld, J. Non-invasive analyte access and sensing through eccrine magnesium corrosion rates; a combined impedance, mass-loss and hydrogen sweat: challenges and outlook circa 2016. Electroanalysis 28, 1242–1249 collection study. Electrochim. Acta 121, 394–406 (2014). (2016). 22. LeGrys, V. A., Yankaskas, J. R., Quittell, L. M., Marshall, B. C. & Mogayzel, P. J. 6. Koh, A. et al. A soft, wearable microfluidic device for the capture, storage, and Diagnostic sweat testing: the cystic fibrosis foundation guidelines. J. Pediatr. colorimetric sensing of sweat. Sci. Transl. Med. 8, 366ra165 (2016). 151, 85–89 (2007). 7. Mu, X. et al. A paper-based skin patch for the diagnostic screening of cystic 23. Farrell, P. M. P. et al. Guidelines for diagnosis of cystic fibrosis in newborns fibrosis. Chem. Commun. 51, 6365–6368 (2015). through older adults: cystic fibrosis foundation consensus report. J. Pediatr. 8. Sonner, Z., Wilder, E., Gaillard, T., Kasting, G. & Heikenfeld, J. Lab on a Chip 153, S4–S14 (2008). Lab on a Chip Integrated sudomotor axon reflex sweat stimulation for con- 24. Heeley, M. E., Woolf, D. A. & Heeley, A. F. Indirect measurements of sweat tinuous sweat analyte analysis with individuals at rest. Lab. Chip. 17, electrolyte concentration in the laboratory diagnosis of cystic fibrosis. Arch. Dis. 2531–2684 (2017). Child 82, 420–424 (2000). 9. Imani, S. et al. A wearable chemical–electrophysiological hybrid biosensing 25. Lezana, J. L., Vargas, M. H., Karam-Bechara, J., Aldana, R. S. & Furuya, M. E. Y. system for real-time health and fitness monitoring. Nat. Commun. 7, 11650 Sweat conductivity and chloride titration for cystic fibrosis diagnosis in (2016). 3834 subjects. J. Cyst. Fibros. 2, 1–7 (2003). 10. Rose, D. P. et al. Adhesive RFID sensor patch for monitoring of sweat elec- 26. Clinical and Laboratory Standards Institute (CLSI). User protocol for evaluation trolytes. Ieee. Trans. Biomed. Eng. 62, 1457–1465 (2015). of qualitative test performance; approved guideline. Second Edition. CLS 11. Gao, W. et al. Fully integrated wearable sensor arrays for multiplexed in situ document EP12-A2 (2008). perspiration analysis. Nature 529, 509–514 (2016). 27. Montes-Cebrián, Y. et al. ‘Plug-and-Power’ point-of-care diagnostics: a novel 12. Emaminejad, S. et al. Autonomous sweat extraction and analysis approach for self-powered electronic reader-based portable analytical devices. applied to cystic fibrosis and glucose monitoring using a fully integrated Biosens. Bioelectron. 118, 88–96 (2018).