We’ve all felt stressed at some point, whether in our personal or professional lives or in response to exceptional circumstances like the COVID-19 pandemic. But until now there has been no way to quantify stress levels in an objective manner.
That could soon change thanks to a small wearable sensor developed by engineers at EPFL’s Nanoelectronic Devices Laboratory (Nanolab) and Xsensio. The device can be placed directly on a patient’s skin and can continually measure the concentration of cortisol, the main stress biomarker, in the patient’s sweat.
Cortisol: A double-edged sword
Cortisol is a steroid hormone made by our adrenal glands out of cholesterol. Its secretion is controlled by the adrenocorticotropic hormone (ACTH), which is produced by the pituitary gland.
Cortisol carries out essential functions in our bodies, such as regulating metabolism, blood sugar levels and blood pressure; it also affects the immune system and cardiovascular functions.
When we’re in a stressful situation, whether life-threatening or mundane, cortisol is the hormone that takes over. It instructs our bodies to direct the required energy to our brain, muscles and heart. “Cortisol can be secreted on impulse – you feel fine and suddenly something happens that puts you under stress, and your body starts producing more of the hormone,” says Adrian Ionescu, head of Nanolab.
While cortisol helps our bodies respond to stressful situations, it’s actually a double-edged sword. It’s usually secreted throughout the day according to a circadian rhythm, peaking between 6am and 8am and then gradually decreasing into the afternoon and evening.
“But in people who suffer from stress-related diseases, this circadian rhythm is completely thrown off,” says Ionescu. “And if the body makes too much or not enough cortisol, that can seriously damage an individual’s health, potentially leading to obesity, cardiovascular disease, depression or burnout.”
Capturing the hormone to measure it
Blood tests can be used to take snapshot measurements of patients’ cortisol levels. However, detectable amounts of cortisol can also be found in saliva, urine and sweat. Ionescu’s team at Nanolab decided to focus on sweat as the detection fluid and developed a wearable smart patch with a miniaturized sensor.
The patch contains a transistor and an electrode made from graphene which, due to its unique proprieties, offers high sensitivity and very low detection limits. The graphene is functionalized through aptamers, which are short fragments of single-stranded DNA or RNA that can bind to specific compounds.
The aptamer in the EPFL patch carries a negative charge; when it comes into contact with cortisol, it immediately captures the hormone, causing the strands to fold onto themselves and bringing the charge closer to the electrode surface. The device then detects the charge, and is consequently able to measure the cortisol concentration in the wearer’s sweat.
So far, no other system has been developed for monitoring cortisol concentrations continuously throughout the circadian cycle. “That’s the key advantage and innovative feature of our device. Because it can be worn, scientists can collect quantitative, objective data on certain stress-related diseases. And they can do so in a non-invasive, precise and instantaneous manner over the full range of cortisol concentrations in human sweat,” says Ionescu.
Engineering improved healthcare
The engineers tested their device on Xsensio’s proprietary Lab-on-SkinTM platform; the next step will be to place it in the hands of healthcare workers. Esmeralda Megally, CEO of Xsensio, says: “The joint R&D team at EPFL and Xsensio reached an important R&D milestone in the detection of the cortisol hormone. We look forward to testing this new sensor in a hospital setting and unlocking new insight into how our body works.”
The team has set up a bridge project with Prof. Nelly Pitteloud, chief of endocrinology, diabetes and metabolism at the Lausanne University Hospital (CHUV), for her staff to try out the continuous cortisol-monitoring system on human patients. These trials will involve healthy individuals as well as people suffering from Cushing’s syndrome (when the body produces too much cortisol), Addison’s disease (when the body doesn’t produce enough) and stress-related obesity.
The engineers believe their sensor can make a major contribution to the study of the physiological and pathological rhythms of cortisol secretion.
So what about psychological diseases caused by too much stress? “For now, they are assessed based only on patients’ perceptions and states of mind, which are often subjective,” says Ionescu.
“So having a reliable, wearable system can help doctors objectively quantify whether a patient is suffering from depression or burnout, for example, and whether their treatment is effective. What’s more, doctors would have that information in real time. That would mark a major step forward in the understanding of these diseases.” And who knows, maybe one day this technology will be incorporated into smart bracelets.
“The next phase will focus on product development to turn this exciting invention into a key part of our Lab-on-SkinTM sensing platform, and bring stress monitoring to next-generation wearables,” says Megally.
There is strong evidence that chronic stress has a negative impact on human health1. Disorders such as obesity, metabolic syndrome, type two diabetes, heart diseases, allergy, anxiety, depression, fatigue syndrome, and burnout are often associated with dysfunctions of the stress axes2,3. Under psychological and/or emotional stress, the adrenal gland cortex of the kidney releases cortisol into the bloodstream. Cortisol is known as the “stress hormone”4.
The important relationships existing between the stressors and cortisol release on major body functions concern the following:
- (i) immune functions,
- (ii) metabolism,
- (iii) neurochemistry, and
- (iv) cardiovascular functions3.
Disturbances in cortisol secretion are thought to be a prime mediator in associations between stress and health.
Cortisol has an important role in regulating carbohydrate metabolism to make the human body resist against pain and external stimuli, such as infection or psychological tension5. It has the ability to maintain homeostasis in the cardiovascular, immune, renal, skeletal, and endocrine systems6.
The level of the cortisol has a circadian rhythm in serum throughout the whole day, with the highest level in the morning (~30 min after waking, 0.14–0.69 µM) and the lowest level at night (0.083–0.36 µM). Sustained stress can disrupt this rhythm and results in an abnormal increase of cortisol level7.
Although the short-term activation of the hypothalamic–pituitary–adrenal axis is adaptive and necessary for everyday life, both high and low levels of cortisol, as well as disrupted circadian rhythms, are implicated in physical and psychological disorders.
Recently, there has been an increasing interest in sensing cortisol biomarker in biofluids for numerous diseases related to the stress. As the secreted cortisol enters into the circulatory system, it can be found in detectable quantities in several biofluids in human body including saliva, sweat and urine8.
The cortisol circadian rhythm (Fig. 1a) and its variations and pulsatility at lower timescale may indicate the individual’s specific acute reactivity to stressful situations9,10; there now exists a high demand for a sensing system capable to support daily quasi-continuous measurements11.
In this context, the work presented here proposes and experimentally validates a sensitive and selective method for the quasi-continuous monitoring of cortisol levels in biofluids, being suitable for sweat analysis with a patch (Fig. 1b).

The traditional methods for detection of cortisol include immunoassays such as radio-immunoassay12, enzyme-linked immunosorbent assay13, and fluoroimmunoassay14,15. The detection limits for these methods are 1.59, 0.036, and 0.52 nM, respectively. In addition, normal and competitive immunosensors in conjunction with surface plasmon resonance transducer with the detection limit of 0.03 nM in urine and 0.13 nM in saliva have been reported16,17.
The physiological values of cortisol levels in human perspiration range from about 1.4 nM to 0.4 µM6. Traditional detection methods are time consuming and complex, needing multiple-step reactions and washing processes for subsequent analyses. In addition, immunoassays require both labeled analogs of cortisol and elaborated hardware.
In the quest for achieving portable systems, the electrochemical sensors appear as an attractive alternative solution. Their main challenge is that the cortisol does not have a redox center. Inkjet-printed electrochemical biosensors have been proposed by exploiting metalloporphyrin-based macrocyclic catalyst ink for the direct reduction of the cortisol captured by the aptamers-functionalized magnetic nanoparticles.
However, the preparation of magnetic nanoparticles and catalyst inks are not yet fully mature for high-performance commercial sensors18. Other electrochemical methods are based on electrochemical impedance spectroscopy (EIS) or cyclic voltammetry via a mediator like Fe3+/Fe4+.
In these studies, the detection of the cortisol is achieved by the inhibition of the surface for redox reaction19,20,21,22,23. In addition, chemiresistor-immunosensors followed a similar strategy24,25. Among the electrochemical sensors, the ones using graphene and its derivatives26 have shown very low detection limit and high sensitivity, thanks to the conductivity and unique structural properties21,25,26.
Graphene has the ability to attract a wide range of aromatic molecules, due to the π–π interactions or electrostatic interactions and can be used to functionalize electrodes with bio-recognition probes such as antibodies or aptamers26. However, electrochemical methods still suffer of significant loss of selectivity because of nonspecific physical adsorption of other species existing in the medium and consequently and their scaling is quite limited.
In recent years, ion-sensitive field-effect transistors (ISFETs) have attracted a lot of attention due to their fast response, sensitivity, low power operation, ability to offer co-integrated readouts and full on chip-circuit design, miniaturization, and low cost27,28,29,30.
All these features make them one of the most promising candidates for wearable systems. ISFETs form a subset of potentiometric sensors that are not affected by signal distortions arising from the environment, thanks to the input gate potential that is connected to the electrical FET transducer31.
They are capable of converting any little variation of the electrical charge placed in the vicinity of the transistor gate, such as any species carrying charge (similarly to ions), become detectable by a variation of the FET drain current. The operation of an ISFET sensor is based on the dependence of the threshold voltage of a metal-oxide semiconductor field-effect transistor (MOSFET) on the gate work function, which can be modulated by the charge of an ion-sensitive membrane31.
As state-of-the-art nano-MOSFETs operate at low voltage with low currents, ISFETs inherit their high charge sensitivity. Any chemical reactions at the top of the gate dielectric with the various species existing in the solution may induce a change of gate stack electrical characteristics.
Therefore, the current–voltage characteristic of the ISFET sensor can be modulated if the gate dielectric is exposed to interactions with fluids. However, in an advanced complementary metal–oxide–semiconductor (CMOS) process, the gate stack is part of the so called front-end-of-line process that is highly standardized and cannot be easily modified or functionalized for sensing.
To address this issue, extended-gate (EG) FETs have been proposed for sensing applications32,33. In such sensor architecture, the base transducer is a standard nano-MOSFET, whereas the sensing element is formed by a specific functional layer on the extension of the metal gate that can be an external electrode or a metal layer in the back-end-of-the-libe (BEOL), connected to the nano-MOSFET gate.
The EG-FET configuration has major advantages due to the separation of the integrated transducing element from the functional layers, including higher stability, less drift and even less temperature sensitivity34. Few research groups have attempted the design of cortisol sensors exploiting FET devices, although they have faced some challenges and, to our best knowledge, we report in this work the first EG-FET that fulfills sensitivity and selectivity performance compatible with sensing in human biofluids.
One of the challenges of the FET-based sensors is the Debye screening effect in ionic liquids, which prevents its electrical potential to extend further than a certain distance, known as Debye length (λD). The value of λD depends on the ionic strength of the liquid. For instance, λD in phosphate-buffered saline (1× PBS), which is commonly used in biological research is <1 nm.
The physical lengths of antibody–antigen complexes, usually utilized for ISFET biosensors, are larger than λD associated with physiological media35. Therefore, the challenge for designing a FET sensor for detection of the cortisol is the choice of an appropriate catch probe overcoming the Debye length. As cortisol is charge-neutral, the electrical recognition of the cortisol is subject to the use of an electrically active mediator catch probes that have their own charge to modulate the gate potential within the detectable Debye length.
Thus, the binding between the catch probe and the cortisol will cause a change in the total gate potential and, consequently, in the measured drain current. Until now, different capturing probes including molecularly sensitive polymers, antibodies, and aptamers have been used in the reported ISFET devices for detection of the cortisol36,37,38.
The use of aptamers, which is the solution adopted in in this work, has some clear attractive advantages. Aptamers are single-stranded nucleic acid molecules, which are negatively charged due to the presence of a phosphate group in each nucleotide of the nucleic acid strand. Aptamers can fold into three-dimensional (3D) topologies, with specifically designed pockets for binding with a target of interest39.
Compared to antibodies, aptamers have superior advantages as catch probes as they are synthesized in vitro, reducing the batch-to-batch difference. In addition, they can be designed for different degrees of affinity for a targeted molecule vs. a modified disturbing analog40,41. Moreover, aptamers are less affected by temperature fluctuations and are more stable for long term storage18,42. They can be covalently immobilized on most surfaces by modifying the 5′- or 3′-end43. Three different apatmers have been introduced for detection of the cortisol.
They have 4044, 6118, and 8538,45 nucleotides. The one with 85 nucleotides has been previously applied for a FET sensor with the detection limit of 50 nM38. However, for a FET sensor facing the challenge of the Debye length, the shorter length of the aptamer is expected to have better sensitivity and lower detection limit as it has higher chance to not exceed the Debye length when it reacts with the target.
In this study, we demonstrate a label-free cortisol detection method with an EG-FETs, which is overcoming the Debye screening limitation for charge sensing by using by using 61-basepair aptamer-decorated single-layer graphene on platinum as a gate electrode. The sensing element is physically separated from the electrical transducer, enabling the possibility to implement the sensor in a 3D configuration, with a nano-MOSFET as base voltametric transducer, and the sensing electrode in the BEOL of a CMOS process, resulting in a low power wearable sensory electronic chip.
The use of atomically thin graphene is crucial to chemically bind the aptamers and bring the recognition event of the analytes within the Debye screening length, with high sensitivity. The resultant limit of detection (LOD) is 0.2 nM. The reported EG-FET sensor is hysteresis-free and shows excellent selectivity towards cortisol in presence of other similar hormones. Further, it has a voltage sensitivity over the four decades of cortisol concentration, which fully covers the one in human sweat.
We propose a, predictive analytical mapping of current sensitivity in all regimes of operation. To the best of our knowledge, this analytical model demonstrates for the first time a fully self-closed analytical dependence of sensor output current on the cortisol concentration over the whole range of concentrations in human sweat.
reference link: https://www.nature.com/articles/s43246-020-00114-x
Original Research: Open access.
“Extended gate field-effect-transistor for sensing cortisol stress hormone” by Shokoofeh Sheibani, Luca Capua, Sadegh Kamaei, Sayedeh Shirin Afyouni Akbari, Junrui Zhang, Hoel Guerin & Adrian M. Ionescu. Communications Materials