Testosterone (≥99%, TST), sodium monobasic phosphate (Na₂HPO₄), sodium dibasic phosphate (NaH₂PO₄), potassium chloride (KCl), potassium ferricyanide (III) (K₃[Fe(CN)₆]), potassium ferrocyanide (II) (K₄[Fe(CN)₆], trimesic acid (95%, H3BTC), iron (II) chloride tetrahydrate (99%, FeCl₂ ·4H₂O), acetonitrile, tetraethylammonium tetrafluoro-borate (≥99%) and ethanol absolute (EtOH) were obtained from Sigma-Aldrich (Buchs, Switzerland). The gold nanoparticles (AuNPs, diameter 18–20 nm, 0.418 mg mL^-1) were synthesized according to a procedure reported in our previous work (Tortolini et al., 2024a).

The morphology and the chemical composition of bare and modified electrodes were investigated by Scanning Electron Microscopy (SEM) and by Energy-Dispersive X-ray spectroscopy (EDX), respectively. The SEM/EDX measurements were performed with High-Resolution Field Emission Scanning Electron Microscopy (HR FESEM, Zeiss Auriga Microscopy, Jena, Germany).

XPS measurements were carried out using an Omicron NanoTechnology Multiprobe XPS system equipped with a Mg Kα (hν = 1253.6 eV) X-ray source. Ethanolic solutions of the MIL-100 (Fe) samples synthesized for 2 h and 24 h were drop casted onto hydrogenated Si wafers and let dry at room temperature. The samples were then mounted on the XPS sample holder with double-sided conductive scotch tape. The C 1s and Fe 2p photoionization regions were acquired using an analyzer pass energy of 20 eV and take-off angle of 21° with respect to the sample surface normal. The experimental spectra were theoretically reconstructed, fitting the secondary electrons’ background to a Shirley function and the elastic peaks to pseudo–Voigt functions.

Raman spectra were run at room temperature in backscattering geometry with an inVia Renishaw micro-Raman spectrometer equipped with an air-cooled CCD detector and super-Notch filters. An Ar⁺ ion laser (λlaser = 514 nm) was used, coupled to a Leica DLML microscope with a ×20 objective. The resolution was 2 cm⁻¹ and spectra were calibrated using the 520.5 cm⁻¹ line of a silicon wafer. Raman spectra were acquired in several different spots on the surface of the samples at 1% of laser power.

For Raman measurements, MOF samples, dispersed in ethanol, were drop-cast (about 100 μL) on silicon wafers and dried before performing the analyses.

All electrochemical characterizations involving cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were performed using an Autolab Potentiostat/Galvanostat (Autolab PGSTAT204, Metrohm, Herisau, Switzerland) with the software Nova 1.1 v2.1.7. All measurements were carried out in a glass cell (model 6.1415.150, Metrohm, Herisau, Switzerland) with a conventional three-electrode configuration with an Ag/AgCl/KClsat (198 mV vs. NHE) as a reference electrode (cat. 6.0726.100, Metrohm, Herisau, Switzerland), a glassy carbon rod as a counter electrode (cat. 6.1248.040, Metrohm, Herisau, Switzerland), a graphene screen-printed electrode (GPH/SPE, diameter 4 mm, 110 Metrohm, Herisau, Switzerland) and a classical glassy carbon electrode (GCE, diameter 2 mm, cat. 6.1204.600 GC Metrohm, Herisau, Switzerland) as working electrodes, respectively. All electrochemical measurements were carried out using three independent electrodes (n = 3). The data were treated and analyzed using Sigma Plot software. The data were presented as mean ± the standard deviation (SD).

For the smartphone-based sensing device experiments, a Sensit/SMART portable potentiostat (PalmSens, Houten, Netherlands) was used and directly connected to a smartphone for POC signal reading. All experiments were performed at room temperature (RT). pH measurements were made with a Metrohm pH meter model 780 (Switzerland).

The MIL-100(Fe) was synthesized according to a totally sustainable method already reported in literature (Sánchez-Sánchez et al., 2015; Guesh et al., 2017) and three samples were taken out form the reaction mixture after 2 h, 12 h, and 24 h of vigorous stirring, named MOF2h and MOF24h, respectively (Scheme 1).

The MOF sensors were prepared using a bare graphene screen-printed electrode (GPH/SPE) modified b drop-casting 10 μL of MOF solution of the three samples (MOF2 h, MOF12 h and MOF24 h), respectively.

Then the electrodes were dried under vacuum for 1 h at room temperature (RT). The AuNPs/MOF/GPH/SPEs were fabricated according to the previous procedure with a second drop-casting step of 10 μL of AuNPs solution on the modified surfaces and letting dry at RT. All modified sensors were stored at RT until use.

The electrochemical measurements were performed by CV and EIS techniques. CV conditions were the followings: potential range: −0.2 – 0.7 V and scan rate ν = 0.025 V s⁻¹. For EIS analysis, the following parameters were employed: frequency range: 1–500 kHz, ac signal: 10 mV amplitude (0.2 V vs. Ag/AgCl). The solution used as redox probe was a mixture of 5 mM Fe(CN)₆ ^3−/4− in 0.1 M KCl.

High-purity deionized water (resistance: 18.2 MΩ cm at 25°C; TOC <10 μg L⁻¹) obtained from Millipore (Molsheim, France) was used throughout experiments. All solutions were prepared in 0.1 M phosphate buffer, 0.1 M KCl, pH 7.0 (PBS).

Determination of the highest occupied and lowest unoccupied molecular orbital (HOMO and LUMO) energy states of AuNPs-MOF was carried out using acetonitrile (4 mL) containing 0.1 M tetraethylammonium tetrafluoroborate by scanning the potential in the range −1.45 - 1.55 V, for 10 scans at 50 mV/s.

Five serum samples were collected from healthy volunteers which had signed an informed consent before the study. The samples were stored at 4°C and were diluted 1:10 in 0.1 M PBS before electrochemical measurements. The standard addition method was applied to determine the added concentration of testosterone.

SEM images were carried out to investigate the surface morphology of modified electrodes. Figure 1 illustrates the MOF 2 h and 24 h, before and after the immobilization of AuNPs. It is clearly observed that, in the first case, most particles show an amorphous nature, attesting a low crystallinity of the material (Figure 1A), due to the reduced time of synthesis, whereas in the second case (Figure 1B) well-defined crystals with octahedral shape, distinct facets and an average diameter of about 120 nm are clearly obtained. In order to certify the successful deposition of AuNPs on the MOF structures, SEM images were carried out after the drop-casting step of AuNPs on both electrodes. It is possible to see the presence of tiny scattered AuNPs, not uniformly distributed across the surface of the GPH/SPE in both cases (Figures 1C, D), suggesting a relatively favorable interaction between support and AuNPs. Figures 1E, F depicts the EDX elemental mapping images performed on the surfaces shown in Figure 1 (panels C and D), which confirm the presence of all fundamentals of MOF and AuNPs in the EDX spectra at both synthesis time, in particular the elemental peaks of carbon (C), oxygen (O), nitrogen (N), iron (Fe) and gold (Au), confirming the AuNPs/MOF nanostructure. SEM images of bare GPH/SPE and AuNPs/GPH/SPE have been reported in Supplementary Figures S1A, B, for comparison.

PXRD experiments were performed in order to check the crystal structure of both MOFs (Figure not shown, as reported in our work under review). The MOF 24 h pattern perfectly matches with the simulated spectrum reported in literature for MIL-100(Fe). The agreement between both patterns is excellent, as every single peak of the simulated XRD pattern is present in the experimental XRD pattern with similar relative intensities without any other single detected XRD peak, leaving no doubts in the phase identification (Guesh et al., 2017; Han et al., 2017). Conversely, the MOF 2 h presents broad and less defined PXRD peaks, typical of the lack of long-range order features of amorphous materials. The dominant phase is still present, although it is also present a XRD peak found at 9°–10° 2θ, which does not belong to the phase MIL-100(Fe), which totally disappears at a prolonged synthesis time (24 h). XPS was used to ascertain the chemical composition of the MIL-100(Fe) samples taking advantage of the high sensitivity of photoelectron kinetic energies from the oxidation state and chemical environment. The Fe 2p ionization region (Figure 2A, a and c) is composed by a spin-orbit split doublet with components j = 3/2 and ½ separated by a ΔEso∼13.5 eV. The two components are rather broad and the former displays a peak maximum at 711.3 eV. In order to evaluate in detail the oxidation state of Fe, the Fe 2p spectra were curve-fitted using the Gupta and Sen (GS) multiplets (Gupta and Sen, 1974; Gupta and Sen, 1975) theoretically predicted for high-spin Fe³⁺ ions and applied by Grosvenor et al. (2004); Biesinger et al. (2011) to Fe³⁺ compounds. In particular, a four-fold GS multiplet was used in the curve-fitting (green curves in Figure 2, panel A (a) and (c)), with the first component falling at 710.5 eV. A further component was also needed to account for a broad signal between the two main spin-orbit components, associated to a shake-up satellite (orange curves in Figure 2, panel A (a) and (c)) (Grosvenor et al., 2004). Both spectra of samples 2h and 24 h are well reconstructed by the adopted Fe³⁺ multiplet, supporting the presence of this oxidation state for iron.

Nevertheless, one difference between the two spectra is apparent regarding the intensity recorded around 715 eV. Here, as pointed out by Grosvenor et al. (2004) an extra-peak (blue curves in Figure 2, panel A (a) and c), additional to the GS multiplet, is sometimes needed to describe Fe compounds. This contribution has been reported to be due to Fe³⁺ ions confined at the surface of a nanostructured material, here denoted as “surface peak”.

In the analyzed samples this peak appears to slightly shift its position and increases in the 24 h sample, suggesting a probable increase in the surface area of the MOF 24 h. The C 1s regions of MIL-100 (Fe) samples are reported in Figure 2, panel A, curves (b) and (d). In both cases the presence of both an aromatic (peaked at 285.0 eV) and a carboxylate carbon component (peaked at 288.8 eV) can be detected (Pérez et al., 2022). These features are fully compatible with the expected presence of BTC ligands in the MIL-100(Fe) samples. Overall, XPS showed that both MOF samples are constituted by Fe³⁺ metal centers coordinated by BTC ligands, with the only small, yet significant, difference in the surface component, associated to a probable more nanostructured morphology in the MOF 24 h sample.

Raman spectroscopy was utilized to confirm the synthesis of MIL (100)-Fe at the two synthesis times. Both Raman spectra perfectly matches with the spectrum reported in literature for MIL (100)-Fe, showing the following principal peaks (Li et al., 2016): ≈810 (γC-H), ≈1004 and ≈1610 (νC = C), ≈1460 (νa COO-) & ≈1550 (νs COO-), ≈1370 (νCOOH), and ≈1216 (C-O-Fe stretching of trimesate) (Figure 2, panel B). It is interesting to note that the two spectra are very similar two each other, with only a slight difference in the spectral region 200–600 cm⁻¹for MOF 2 h, where are visible two peaks, one sharp at ≈250 cm⁻¹ and a more widened peak between 400–500 cm⁻¹, which are not present in the spectrum of MOF 24 h, probably due to the presence of still unreacted starting materials in the sample.

The electrochemical characterization of the modified electrode platforms was performed by CV and EIS in 0.1 M KCl solution containing 5 mM Fe(CN)₆ ^3−/4−. Figure 3 shows the CVs and Nyquist plots of bare (black), AuNPs (green), MOF2h (blue) and AuNPs/MOF2h (red) GPH/SPE, respectively.

As shown in Figure 3, panel A, all CV curves show two well-defined redox peaks, indicating a fast electron transfer of the redox probe. The peak current increase after the modification of the electrode with AuNPs or MOF is quite small and of similar intensity, whereas a larger increase in the peak current is observed when the two nanomaterials are combined (red curve). A marked decrease in ΔE _p was also observed in this case, with the lowest ΔE _p value closer to the theoretical value of 59 mV for a one− electron reversible mediator. Table 1 shows all the electrochemical parameters. The electroactive areas of bare and modified GPH/SPEs were calculated by recording CVs in the range 5–500 mV s^-1, according to the Randles-Sevcik Equation 1 (Savan, 2020): I p = 2 .   69 × 10 5 n 3 / 2 A e D 1 / 2 C v 1 / 2 (1)

In equation, I_p represents the anodic/cathodic peak current (A), A_e the electroactive area (cm²), D the diffusion coefficient (7.6 × 10⁻⁶ cm² s⁻¹) and C the concentration (2.5 × 10⁻³ mol cm^–3) of [Fe(CN)₆]^3–/[Fe(CN)₆]^4– redox probe in KCl solution, n is the number of the electrons transferred during the redox reaction (n = 1), and v is the scan rate (V s^–1). On the basis of this equation, the A_e values of MOF2h and AuNPs/MOF2h modified GPH/SPEs resulted to be 0.179 cm² and 0.327 cm² (about 1.3 and 2.5 times higher than bare electrode, respectively (Table 1). These results indicate that the modification of the bare electrode with MOF2h slightly increases the Aea, which results marked increased after the adding of AuNPs, attesting that the dispersion of AuNPs on a MOF surface greatly affects the catalytic activity of the material.

The heterogeneous electron transfer rate constant k ⁰ has also been calculated before and after each modification step, using the Lavagnini et al. method, which merges the Klingler-Kochi and Nicholson and Shain methods for irreversible and reversible systems, respectively (Lavagnini et al., 2004). It is interesting to note that the k ⁰ value obtained for the AuNPs/MOF2h/GPH/SPE resulted more than 2 times higher than that calculated for the bare one (Table 1). This value confirms the synergistic contributions of MOF and AuNPs, which enhance each other properties: MOF can enhance the porosity and catalytic activity of AuNPs, whereas AuNPs are able to enhance the electrical conductivity of the nanoporous MOF.

The superior electrochemical performances shown by MOF2h both in absence and presence of AuNPs compared to MOF12 h and MOF24 h were demonstrated by CV experiments. The results are reported in Supplementary Figures S2, S3, in absence and in presence of AuNPs, respectively. These results can be ascribed to the most abundant catalytic sites and the large number of defects present in the amorphous/semi-crystalline MOF structure, typical of MOF2h, compared to crystalline MOF12 h and 24 h, which can fasten the electron transfer and lead to a significant enhancement of the catalytic performance (Dissegna et al., 2018; Bechis et al., 2022; Dong et al., 2022; Zhao et al., 2024). On the basis on the results obtained, the MOF2h was used for further experiments, and referred as MOF.

The kinetic mechanism of the electrochemical reaction at AuNPs/MOF/GPH/SPE modified electrode was also investigated by monitoring the relationship between scan rate and peak current (Supplementary Figure S4A). It can be observed a shift of the anodic and cathodic peaks towards more positive and negative potentials, respectively, when the scan rate is increased from 5 to 200 mV s^–1, due to the effect of the resistance of the electrode. A linear relationship was also found between the peak magnitude of the anodic and cathodic current of the modified sensor and the square root of the scan rate (Supplementary Figure S4B), indicating that the electrochemical process is diffusion-controlled.

EIS measurements were also performed for a further electrochemical characterization of the proposed platforms. The EIS spectra were recorded in the frequency range 1–500 kHz. The semicircle diameter in the high frequency region corresponds to the charge transfer resistance (RCT). The corresponding Nyquist plots are shown in Figure 3, panel B. The bare electrode (black curve) exhibited the highest RCT value. The AuNPs modified electrode (green curve) showed a slightly lower RCT value, suggesting improved conductivity. Similar behavior was obtained with MOF modified electrode (blue curve). It can be observed that the RCT value of the MOF electrode is a little larger than that of the AuNPs electrode. These results confirm the better conductivity of both nanomaterials, with small superior performances exhibited by the semi-amorphous MOF2h. Finally, a more significant decrease in RCT value is observed with the AuNPs/MOF/GPH/SPE (red curve), attesting the increased electrical conductivity when both nanomaterials are combined in a hybrid nanocomposite, in perfect agreement with the results obtained with CV.

All data obtained were fitted by applying the Randles circuit (Figure 3, panel B, inset) and the results are reported in Table 2. Based on the RCT values obtained for bare and modified GPH/SPEs, the k0 of the redox probe between the bulk solution and the sensor surface for each electrode was calculated by Equation 2: K 0 = RT R C T F 2 n 2 AC (2) where R is the universal gas constant (8.314 J K⁻¹ mol^–1), T the temperature (298.15 K), F F’s constant (96,485 C mol^–1), n the number of transferred electrons, A the electrode geometric area (0.11 cm²), and C the redox probe concentration (2.5 × 10⁻⁶ mol cm^–3).

The calculated k ⁰ values agree with the RCT ones (Table 2), attesting a strong synergistic effect of the two nanomaterials that create a hybrid nanocomposite with superior electrochemical performances. As for the other Randles parameters, the constant phase element (CPE) is a parameter usually related to surface reactivity, electrode porosity, roughness, and surface inhomogeneity, and therefore it was expected that this value would rise from 5.43 for bare electrode to 18 n MhO s^N for AuNPs/MOF/GPH/SPE electrode, due to its greater structural complexity (Jorcin et al., 2006).

Figure 4 shows an irreversible cathodic peak for testosterone reduction at a potential of about −0.4 V vs. Ag/AgCl at a fixed testosterone concentration (0.1 μM).

The scan rate effect on the cathodic peak current response of the AuNPs/MOF/GPH/SPE sensor has been investigated at the same fixed testosterone concentration in the range 2–25 mV s^–1 and a linear relationship between the peak current values and the scan rate has been obtained, attesting a typical adsorption process at the solution/sensor interface for TST reduction (Figure 4 inset) (Tortolini et al., 2022; Kozak et al., 2022). The logarithm of the peak current of testosterone vs. log v is also reported in Supplementary Figure S5. The slope value resulted to be 0.945, typical value of a total absorption process (value close to 1.0) (Tortolini et al., 2024b).

The surface coverage (Γ) of testosterone onto the AuNPs/MOF/GPH/SPE was calculated from the plot of I_p versus scan rate ν, according to Equation 3 (Laviron, 1979): I p = n 2 F 2 Av Γ / 4 R T (3) where n is the number of electrons, F, R, and T are the Faraday constant, the molar gas constant, and the absolute temperature, ν is the scan rate, and Γ the surface coverage. Assuming n = 2 for testosterone (Levent et al., 2015), its Γ value was found to be 0.011 × 10⁻⁸ mol cm⁻², with a similar order of magnitude of Γ values calculated for TST with different electrodes reported in literature (Kozak et al., 2022).

For a further in-depth analysis about the sensing of testosterone, the energy levels (HOMO–LUMO) of testosterone and AuNPs/MOF were evaluated, respectively, using CV technique, by applying potentials above the LUMO or below the HOMO (vs. Ag/AgCl), leading to cathodic and anodic currents, respectively, that can be used for the determination of LUMO and HOMO.

The CV measurements were carried out first for a bare glassy carbon electrode (GCE) before and after the drop-casting of 10 μL of MOF and 10 μL of AuNPs and represented in Figure 7, panel A. The oxidation and reduction peak potentials were used for determination of the EHOMO and ELUMO using the Equations 4–6. E HOMO = − e   E Ox vs  . NHE + 4 .   75   eV (4) E LUMO = − e   E Red vs .  NHE + 4 .   75   eV (5) E NHE = E Ag / AgCl + 0 .   197   eV (6)

The working electrode indicates a reduction peak at −1.2 V and an oxidation peak at 0.6 V vs. Ag/AgCl.

The ELUMO and EHOMO resulted to be −3.75 eV and –5.55 eV, respectively. The band gap energy between ELUMO and EHOMO was calculated to be 1.8 eV. The ELUMO and EHOMO values for testosterone are reported in literature (Kumari et al., 2021). As shown in Figure 7, panel B, after applying a positive voltage to the system, the electrode potential energy dropped. Then, the electrons were transferred from AuNPs/MOF to the GCE surface (step 1). The HOMO levels of AuNPs/MOF and testosterone are well-matched to efficiently promote the electron transfer from testosterone to the electron-deficient AuNPs/MOF (step 2), thus leading to the testosterone reduction. These results confirm that the AuNPs/MOF may represent a favorable nanomaterial to provide fast electron transfer processes.

On the basis of the HOMO level of MIL-100(Fe) reported in literature (−1.38 eV) (Rojas-Guerrero et al., 2023), it is interesting to note that that energy gap between testosterone and AuNPs/MO nanomaterial is much higher compared to MOF without the modification with AuNPs, thus attesting a favoured electron transfer.

The analytical performances of the AuNPs/MOF/GPH/SPE towards different testosterone concentration (from 5 to 50 nM) were investigated by CV technique (Figure 8), and the peak values were used for the construction of the sensor calibration curve (Figure 8, inset). The proposed electrochemical sensor shows a linear response over the range of 5–50 nM with the following two linear regression equations (black curves): ΔI_p (μA) = 0.160 CTST (nΜ) + 5.90 (R² = 0.992) and ΔI_p (μA) = 0.063 CTST (nΜ) + 7.48 (R² = 0.991) and a detection limit (LOD) of 2 nM, calculated on Formula 3 σ/S, where σ represents the blank standard deviation and S the calibration slope. All measurements were made in triplicate and the results reported as mean values. Figure 8 inset shows also the calibration plot (red curves) obtained with a portable potentiostat, directly connected to a smartphone for POC signal reading, named Sensit−Smart. Once again, two linear regression equations are obtained: ΔI_p (μA) = 0.149 CTST (nΜ) + 5.67 (R² = 0.990) and ΔI_p (μA) = 0.046 CTST (nΜ) + 7.16 (R² = 0.998) and a slightly higher LOD of 3 nM. Differential pulse voltammetry (DPV) has also used to test the results obtained with CV. The DPV curves and the relative calibration plots are reported in Supplementary Figures S6A, B. The sensor shows a slightly larger linear range between 1–50 nM with the following two linear regression equations (dotted curves): ΔI_p (μA) = 0.74 CTST (nM) + 3.17 (R² = 0.992) and ΔI_p (μA) = 0.44 CTST (nM) + 13.90 (R ² = 0.991), and a detection limit (LOD) of 0.5 nM, calculated with the same Formula 3 σ/S, reported above. The sensibility and LODs were definitely improved, as expected by employing a more sensitive technique.

In addition, a comparison with other electrochemical (bio)sensors reported in literature for testosterone detection has been reported, and the results are summarized in Supplementary Table S1. Most of them are electrochemical sensors (Goyal et al., 2010; Levent et al., 2014; Moura et al., 2014; Gugoasa et al., 2015; Liu et al., 2017; Kellens et al., 2018; Lee et al., 2019; Sanli et al., 2020) and only three of them (last three rows in Supplementary Table S1) are immunosensors (Eguílaz et al., 2010; Serafin et al., 2011; Bulut et al., 2020). The developed sensor shows analytical performances not better than but comparable to most other recently developed sensors and immunosensors. In particular, among the electrochemical sensor, a recent study (Huang and Feng, 2024) describes another MOF-based sensor with Cu²⁺ instead of Fe³⁺ as metal ion. The Cu-MOF was coupled with rGO to form a hybrid rGO/Cu-MOF nanomaterial, which shows slightly superior performances compared to the MIL-100(Fe)-based sensor used in this study, but it must be remarked that MIL-100(Fe) has been synthetized according to a total ecofriendly synthetic method (room temperature compared to 150°C of Cu-MOF synthesis), with a much shorter synthesis time (2 h compared to 36 h) and without the use of any organic solvent, unlike Cu-MOF synthesis in which dimethyl formamide was employed.

The reproducibility of the final platform has been tested by using five electrodes (n = 3) and recording the current signal obtained with each of them by CV technique in presence of 50 μM testosterone solution. The relative standard deviation (RSD) value obtained is 1.2%, suggesting a good fabrication reproducibility of the AuNPs/MOF GPH/SPE platform (Supplementary Figure S7).

Finally, the stability of the developed sensor has also been investigated. Five different electrodes, stored at 4°C, were tested every 5 days for 1 month, by measuring the CV signal obtained for a 0.1 μM testosterone solution (n = 3). The results show that sensors used on day 5, 10, 15, 20, 25, and 30 show a current signal decrease lower than 15%, compared to the response of the electrodes used on day 0 (Supplementary Figure S8).

The selectivity was assessed by investigating the effect of possible interfering compounds commonly found in biological fluids, such as glucose, uric acid, ascorbic acid, β-estradiol and progesterone.

Figure 9 shows the current response towards these other compounds at a 10-times higher concentration and no significant signal change was registered (signal change less than 4.0%), confirming the excellent selectivity of the proposed sensor.

The matrix effect of five real samples of human serum was tested with the proposed sensor by using a standard addition method and the smartphone-Sensit/Smart potentiostat based sensing device. The schematic representation of the smartphone based sensing device is represented in Scheme 2.

All measurements were performed with 150 µL of serum samples, prepared as described in the experimental part, and placed directly on the modified SPE surface. The measurements were performed in triplicate by CV technique. The results showed a good recovery with recovery rates in the range 98.5%–103.7% for all spiked real samples, with RSD values lower than 4% after three replicate measurements for each sample (Table 3).

These results demonstrate the absence of matrix effect and therefore the potential for thew proposed platform for numerous biomedical applications, both in clinical diagnosis and doping control analysis.