Extended gate field-effect-transistor for sensing cortisol stress hormone

Cortisol is a hormone released in response to stress and is a major glucocorticoid produced by adrenal glands. Here, we report a wearable sensory electronic chip using label-free detection, based on a platinum/graphene aptamer extended gate field effect transistor (EG-FET) for the recognition of cortisol in biological buffers within the Debye screening length. The device shows promising experimental features for real-time monitoring of the circadian rhythm of cortisol in human sweat. We report a hysteresis-free EG-FET with a voltage sensitivity of the order of 14 mV/decade and current sensitivity up to 80% over the four decades of cortisol concentration. The detection limit is 0.2 nM over a wide range, between 1 nM and 10 µM, of cortisol concentrations in physiological fluid, with negligible drift over time and high selectivity. The dynamic range fully covers those in human sweat. We propose a comprehensive analysis and a unified, predictive analytical mapping of current sensitivity in all regimes of operation.


Supplementary Note 3. Graphene transfer on the Pt electrode
A layer of PMMA 950 k was spin-coated (3000 rpm for 60 s) on to the top side of the Cu/Graphene.
Then the Cu/Graphene covered by the carrier polymer was baked on the hotplate at 180 0 C for 2 minutes. Then, the backside graphene was etched by oxygen plasma with the power of 50 W and 6 sccm flow rate for 2 minutes. Next, a 5mm x 5mm piece was cut from the big one and floated in 0.1 M Ammonium persulfate solution for 2 hours to dissolve the copper substrate. After this step, the Ammonium persulfate solution was gently substituted with DI water by dropper. This substitution process was repeated for 3 times and finally the floated PMMA/graphene was left in DI water for 24 h to remove residuals from the graphene sheet. Graphene was then fished out onto the central Pt electrode on the chip. It should be noted that the Pt electrode surface was treated by oxygen plasma before the graphene transfer process to improve the adhesion between Pt electrode and graphene sheet. For this purpose, the chip was placed in the oxygen plasma chamber with the power of 200 W and the flow rate of 200 sccm for 30 seconds. At the end, the chip was placed in an acetone bath for 6 hours to remove PMMA. Then, it was rinsed with IPA and dried with N2 gas gently. The last two steps were done carefully to have an intact graphene sheet. After transfer of the Graphene, the following functionalization steps were done which are described in the next part.
The Schematic of different steps for graphene transfer and the last functionalized electrode with aptamers is depicted in Fig. S1.

Supplementary Note 4. EIS experiment
Electrochemical impedance spectroscopic (EIS) measurements were performed by a PalmSens electrochemical workstation (PalmSens Instruments BV, The Netherlands). This experiment was carried out in 0.05X PBS buffer solution (pH: 7.4) containing 5 mM Fe 3+ /Fe 4+ with a conventional three-electrode system. In this system the platinum and Ag/AgCl electrodes act as a counter and reference electrodes respectively, while the functionalized electrode is the working electrode.
Impedance spectra were measured in the frequency range of 0.05-10 6 Hz. The sinusoidal voltage with the amplitude of 10 mV was applied in each step. The impedance data were fitted with a Randles circuit model to extract the corresponding charge transfer resistance.

Supplementary Note 5. AFM experiment
Topography analysis of the electrodes were carried out with a Cypher Atomic force microscope (Asylum Research) in non-contact mode. AFM imaging was done for the Pt/graphene electrode and the following functionalization steps of the surface. In each step before imaging, the surface of the electrode was rinsed several times and dried with N2 gas.

Supplementary Note 6. XPS experiment
To elaborate the chemical functionalization of the electrode surface, X-ray photoelectron spectroscopy (XPS) analysis was utilized in different steps of electrode functionalization. XPS analysis was carried out using a PHI Versa Probe II scanning XPS microprobe (Physical Instruments AG, Germany) to characterize the representative graphene surfaces before and after functionalization with PBSE and simulated aptamer probes. A simulated probe sequence containing 62 thymine nucleotides (T62) was utilized instead of the cortisol aptamer sequence to produce an XPS signature. This is due to the fact that the interpretation of the data would be easier in terms of density and conformation of the surface-immobilized nucleic acid strands 3,4 . Analysis was performed using a monochromatic Al Kα X-ray source of 24.8 W power with a beam size of 100 µm. The spherical capacitor analyzer was set at 45° take-off angle with respect to the sample surface. The pass energy was 46.95 eV yielding a full width at half maximum of 0.91 eV for the Ag 3d 5/2 peak. The C1s peak located at 284.8 eV for all the samples. In addition, the high resolution of Pt 4f region contains a pairs of peaks for all the measured samples (Pt 4f7/2 and Pt 4f5/2) which is a characteristic of metallic Pt (0) with the binding energy of 71.3 and 74.5 eV, respectively.

Supplementary Note 7. Experimental set up for EGFET sensor characterization
The characterizations of the modified EGFET was investigated by semiconductor parameter analyzer HP 4155B. The chip containing modified electrode was placed on the base of a static cell

Supplementary Note 8. Confirmation of the aptamer functionalization by recording EGFET transfer characteristic
In order to confirm the correct immobilization of the aptamers on the graphene surface, the transfer characteristic of an EGFET sensor using a larger size transistor was recorded in the measurement solution (PBS buffer 0.05X pH: 7.4) after the immobilization of the aptamers, and after exposure of the functionalized electrode to the lowest tested concentration of the cortisol (0.2 nM). As shown in Supplementary Fig. S2, both the immobilization of the aptamers on the graphene surface and the exposure of the functionalized electrode to the cortisol solution, lead to right shift in the IDS-(VREF-VT) curves. In Fig. S2, the drain current is plotted against the difference between the applied reference voltage and the threshold voltage, which was extracted from the data with the second derivative of the drain current method, a method that is independent of the model adopted for the transfer characteristics. Aptamers are negatively charged molecules introduced to the surface of the electrode, also binding of cortisol to the aptamers makes the strands to fold on themselves, and come closer to the surface 5 . Consequently, the surface potential of the electrode, , is modulated in the steps of electrode functionalization with aptamers and exposure of the functionalized electrode to the cortisol. Due to the relation existing between the threshold voltage, VT, and the surface potential, as described in the main article), the VT of the EGFET is modulated according to the amount of negative charges induced by the aptamers. Therefore, a higher gate voltage is required to obtain the same drain current, which results in a right shift of the IDS-(V REF -V T ) curves.

Supplementary Note 9. Analyze of the low concentrations of the cortisol with the sensor
For investigation the resolution of the sensor in low concentration range, the sensor was exposed to different concentrations of the cortisol between 1 and 100 nM and the response of the sensor was recorded.

Supplementary Note 11. Study of the selectivity
In this work, testosterone and cortisone were chosen for the investigation of the cortisol sensor selectivity. Testosterone and cortisol are adrenal hormones and have similar structures which may raise the question of the cross-sensitivity of the aptamer functionalization to testosterone. The amount (concentration) of the testosterone in biofluids such as blood and sweat, is much lower than the cortisol. For instance, the level of testosterone in blood ranges from 0.52 to 2.4 nM for females and from 9.4 to 37 nM, for males. Cortisone in turn metabolized from cortisol in the peripheral tissues and has the most similar structure to the cortisol. However, cortisone has much less glucocorticoid activity than cortisol, so it can be considered an inactive metabolite of cortisol 6 .
The fabricated cortisol sensor was exposed to testosterone in the range of the concentrations similar to testosterone concentrations in biofluids. Also the sensor was exposed to the cortisone in the range of the concentrations similar to the cortisol measurement. The stock solutions of the testosterone and cortisone with the concentration of 0.01 M was made in methanol, then all the other diluted solutions were prepared from the stock solution in the incubation buffer used for cortisol measurements. The tests were carried out in the same conditions as for cortisol. As can be seen from Fig. S6a, when the sensor is exposed to the lowest concentration of the testosterone (10 pM), the corresponding IDS-(V REF -V T ) curve shifts to the left direction compared to the characteristic of the sensor in a buffer with no testosterone. As the testosterone concentration increases in successive steps (10 pM, 100 pM, 1 nM) up to 10 nM, the IDS-VREF curves show no significant trend and tend to overlap. For testosterone only a very slight shift in the left direction was observed, with no significant trend. The results for the sensor response to the cortisone is depicted in Supplementary Fig. S6b, as can be seen when the sensor is exposed to the lowest concentration of the cortisone (1 nM), the corresponding IDS-(VREF-VT) curve shifts to the right direction compared to the characteristic of the sensor in a buffer with no cortisone. However, as the cortisone concentration increases up to 10 µM, the IDS-VREF curves overlap, and no trend was observed. These results for testosterone and cortisone are in contrast to the clear sensor response to cortisol (for which all curves are shifting systematically to the right when the cortisol concentration increases). In addition to the data reported in Fig. S6, the shift of the IDS-VREF characteristics in terms of VREF voltage at constant current of 1 µA was measured under the presence of cortisol alone and under the combination of both cortisol and testosterone or cortisone. When the sensor was exposed to the lowest concentration of cortisol (with no testosterone), a positive shift to the right, of 80 mV for a concentration of 0.2 nM of cortisol corresponding to the LOD, was recorded. This shift was reduced to a value of 70 mV when the highest tested concentration of testosterone (10 nM) has been added to the same solution. However, when the highest concentration of the cortisone (10 µM) was added to the solution containing lowest tested concentration of the cortisol, almost no shift of IDS-VREF curve was observed. Therefore, it is concluded that our sensor is highly selective to cortisol and only its LOD can be slightly increased in the presence of highest levels of testosterone in human biofluids. Further experiments with very low concentrations of cortisol (below 1nM) and high concentrations of testosterone are needed to be able to quantify such extremely small cross-sensitivity.

Supplementary Note 12. Study of the drift effect
It is well-known that ISFET sensors could exhibit an instability, commonly known as drift, in the form of a slow, monotonic, temporal increase in the threshold voltage of the device. The study of such drift effect is important to estimate any potential influence of the experimental data used for the sensitivity evaluation. Moreover, the precise knowledge on the drift is very useful as it can be used to correct the sensing recorded data and offer an accurate estimate of the real sensor response.
In order to study the drift, we have designed a multi-step experiment: (i) first, the response of the freshly functionalized sensor was recorded, (ii) then the sensor was kept in an incubation buffer solution without cortisol for 30 minutes, which is the time required for the cortisol binding, (iii) afterwards, the sensing electrode was rinsed and placed in the same buffer used for the cortisol measurement and the transfer characteristics, ID-VGS, of the EGFET were recorded. This threestep process was repeated for three consecutive times.
As shown in Fig. S7, the shift between ID-(VREF-VT) curves for the mentioned experiments is quasinegligible and after 90 minutes, the sensor shows a stable response. We conclude that any potential drift influence is minor for our sensor and our experiments with the aptamer functionalized graphene/Pt electrode electrically connected to the gate of the FET.