Results and discussion

Fabrication of TiO₂ nanoparticles modified screen-printed electrode

To fabricate the nanocomposite, TiO₂NP: carbon paste was tested in various ratios (1:1, 1:2, 1:4,4:1,2:1). A ratio of 1:4 was found to give the best results for the determination of 15 μM ERL (Figure 1). At ratios higher than 1:4 (e.g., 1:2 and 1:1), the higher TiO₂ content increased electrode resistance due to the semiconducting nature of TiO₂, decreasing conductivity; at lower ratios (2:1 and 4:1), insufficient TiO₂ reduced the available electroactive surface area. The enhanced current observed with this ratio can be attributed to the increased electroactive surface area and improved interfacial electron transfer facilitated by the TiO₂ nanoparticles, which provide an efficient electron-transfer network, while their large surface area increases the number of electroactive sites. This ratio ensures optimal dispersion of TiO₂NPs, preventing aggregation and maintaining accessibility to catalytic sites, supporting the structural integrity of the nanocomposite [41,42].

Figure 1.: Effect of TiO₂NP concentration on sensor response. The graph illustrates the relationship between the differential peak current (ΔIp) and the TiO₂NP: Carbon mass ratio in the presence of 15 μM ERL. Error bars represent the standard deviation for n=3 measurements

Morphological and structural characterization of TiO₂ nanoparticles modified screen-printed electrode

The surface morphology of the TiO2NP@SPE was examined by SEM (Figure 2a). The micrograph reveals a rough, granular surface with well-dispersed TiO2 nanoparticles. This morphology is markedly different from the relatively smooth topography typical of a bare SPE, suggesting a successful increase in the electroactive surface area. EDAX analysis (Figure 2b) confirmed the successful deposition of TiO₂NP, with titanium appearing as a prominent constituent (25.79 wt.%) [43,44]. The spectrum also shows significant peaks for carbon (52.28 wt.%) and oxygen (16.68 wt.%), corresponding to the graphite ink and the oxide phase of the nanoparticles, respectively. Trace amounts of aluminium, silicon and chlorine were detected, which are consistent with the composition of the PET substrate and the conductive ink binder.

Figure 2.: (a) SEM images and (b) EDAX spectrum with elemental composition (inset) confirming successful TiO₂NP deposition on the electrode surface

The FTIR spectra of the bare SPE and TiO2NP@SPE (Figure 3a) were acquired to identify the functional groups, and vibrational modes present in each material and to confirm successful nanoparticle incorporation. The bare SPE (black line) displays characteristic peaks at 1580 cm-1, corresponding to C=C aromatic stretching vibrations of the graphitic carbon matrix, confirming the carbonaceous nature of the electrode substrate [41]. Upon modification with TiO₂NP (red line), a prominent new absorption band appears at 507 cm-1, which is entirely absent in the bare SPE spectrum and is unambiguously assigned to the Ti-O and Ti-O-Ti stretching vibrations characteristic of the anatase phase of TiO₂ [43,44]. The exclusive appearance of this band in the modified electrode spectrum provides definitive spectroscopic confirmation of successful TiO₂NP deposition onto the electrode surface without phase transformation. Additionally, the TiO2NP@SPE displays a broad band at 3429 cm-1 attributable to O-H stretching vibrations of surface hydroxyl groups and adsorbed water molecules on the TiO₂ surface [44,45], which are electrochemically significant as they facilitate analyte adsorption. The band at 2969 cm-1 is consistent with asymmetric C-H stretching vibrations of aliphatic hydrocarbon groups originating from the organic polymeric binder present in the conductive carbon ink matrix [46]. The absorption at 2382 cm-1 is attributable to atmospheric CO₂, a well-known artifact commonly observed in attenuated total reflectance -Fourier transform infrared (ATR-FTIR) measurements [46]. The peak at 1759 cm-1 is assigned to C=O stretching of ester or carbonyl groups within the polymeric binder resin of the conductive ink [47]. Collectively, the FTIR data confirm the chemical identities of both the carbon substrate and the TiO₂ modifier and unambiguously establish that the nanoparticle modification introduces new Ti-O bonding while preserving the electrode's graphitic carbon structure.

Figure 3.: Structural and chemical characterization of the modified electrodes. (a) FTIR spectra of the bare screen-printed electrode (Bare SPE, black) and the electrode modified with titanium dioxide nanoparticles (TiO2NP@SPE, red), showing characteristic vibrational bands for the functional groups and the Ti-O-Ti stretching at 507 cm-1. (b) XRD patterns comparing the crystalline structure of the TiO₂NP (red) and the modified SPE (blue), where the modified SPE shows a dominant peak at 26° associated with the carbonaceous support. (c) Detailed XRD spectrum of the modified material highlighting the sharp (002) diffraction peak of graphitic carbon at approximately 26°, with an inset displaying the broader diffraction range and higher-order peaks (100, 101, 102, 110, etc.) indexed to the electrode's substrates like silver and carbon

The structural phase, crystallinity, and crystallite size of the synthesized materials were characterized by XRD (Figure 3b). The TiO₂ NPs display sharp, well-defined diffraction peaks at 2θ ≈ 25.3, 37.8, 48.0, 53.9, 55.1 and 62.7, corresponding to the (101), (004), (200), (105), (211) and (204) crystallographic planes of the anatase phase (JCPDS No. 21-1272), confirming phase-pure anatase with no detectable rutile or brookite impurities. The sharpness and high intensity of these reflections indicate good crystallinity of the nanoparticles. The average crystallite size of the TiO₂ NPs was determined using the Debye-Scherrer equation (1):

where K = 0.9 (dimensionless shape factor), λ = 0.15406 nm (Cu Kα radiation), β is the full width at half maximum (FWHM) of the diffraction peak in radians, and θ is the Bragg diffraction angle. Applying this to the most intense anatase reflection, the (101) peak at 2_θ_ = 25.3° (FWHM = 0.55°) yields an average crystallite size of D = 14.8 nm. This value is in close agreement with the manufacturer-stated particle size of <30 nm, with the difference being consistent with the well-established distinction between XRD-derived coherent scattering domain size and physical particle dimensions measured by electron microscopy.

The TiO₂ anatase peaks remain clearly visible in the modified SPE spectrum alongside the graphitic carbon signal, and the crystallite size derived from the (101) reflection in the modified SPE (D = 13.6 nm, FWHM ≈ 0.60°) is marginally smaller than that of the bare TiO₂ NPs, which may be attributed to slight lattice strain or confinement effects upon immobilization of the nanoparticles onto the electrode surface. Collectively, the retention of all characteristic anatase reflections, along with the graphitic carbon peak, confirms the successful integration of TiO₂ NPs onto the electrode surface without inducing any phase transformation of the nanoparticles [46,47]. In the bare SPE, a dominant high-intensity peak at 2_θ_ = 26.5° is observed, corresponding to the (002) plane of graphitic carbon from the SPE substrate [48,49]. Applying the Debye-Scherrer Equation (1) to this (002) reflection (FWHM = 0.20°) yields a graphitic stacking crystallite height of _L_c ≈ 40.8 nm, indicative of well-ordered, highly crystalline graphite layers within the SPE [50,51].

Electrochemical active area determination

The electrochemically active area (ECA) of the TiO2NP@SPE was calculated using the Randles-Ševčík equation applied to [Fe(CN)6]3-/4- cyclic voltammograms at varying scan rates (Figure 4). The anodic and cathodic peak currents varied linearly with the square root of scan rate (R2 = 0.99), confirming diffusion-controlled mass transfer. Based on the slope extracted from the linear regression, _I_pa = 4.494 _v_1/2 and assuming standard parameters for a 5 mM ferricyanide solution (n = 1, D = 7.6×10-6 cm2 s-1), the calculated electrochemically active area of the TiO2NP@SPE was found to be 0.0383 cm2. All further electrochemical experiments were conducted at a scan rate of 50 mV s-1.

Figure 4.: Electrochemical characterization of the modified electrode. (a) Cyclic voltammograms of TiO2NP@SPE recorded in 5 mM Fe[(CN)6]3-/4-containing 0.1 M KCl at various scan rates ranging from 2 to 200 mV s-1. (b) Linear regression plots of the anodic (Ipa) and cathodic (Ipc) peak currents versus the square root of the scan rate ʋ1/2, demonstrating a diffusion-controlled redox process

The calculated ECA of 0.0383 cm2 for the TiO2NP@SPE is primarily attributed to the synergistic effect of the nanomaterial modification on the carbon substrate. While the geometric area of fabricated screen-printed electrodes is typically small (approx. 0.0314 cm2 for a 2 mm diameter), the integration of TiO2 nanoparticles enhances the ECA by providing a higher density of electronic states and increasing the surface roughness factor [52,53].

Scan rate study and electrode kinetics

The electrochemical oxidation mechanism of ERL on TiO2NP@SPE was investigated by CV at scan rates from 5 to 200 mV s-1 in the presence of 15 μM ERL in 0.1 M PB (pH 7.4) (Figure 5a). As the scan rate increased, the anodic peak potential shifted positively, indicating an irreversible oxidation process. The linear relationship between _I_pa and _v_1/2 (Figure 5c, _R_2 = 0.99) established that the electrode process is diffusion-controlled, consistent with the behaviour previously observed on the bare SPE.

Figure 5.: Electrochemical characterization of the modified electrode: (a) Cyclic voltammograms at varying scan rates; (b) Linear plot of anodic peak current (Ipa) vs. v (R2 = 0.976); (c) Linear plot of Ipa vs. v1/2 (R2 = 0.990) indicating a diffusion-controlled process; and (d) Linear relationship between peak potential (Epa) and log v used to calculate the kinetic parameters (αn) via Laviron’s Equation (2)

The scan rate was increased, and the linear relationship between the peak potential and the logarithm of the scan rate (Figure 5d) can be described by Equation (2). _E_pa = 0.292 log v + 0.086, (_R_2 = 0.971, σ = 0.18, n = 6). _E_pa for an irreversible process can be defined by the Laviron Equation (2) [54,55]:

where _k_s is the charge transfer rate constant. Taking T = 298 K, R = 8.314 J K-1 mol-1 and F = 96480 C mol-1, the value of αn can be calculated to be 0.20 from the slope of _E_pa vs. log v (0.292). Where α is the transfer coefficient and n is the number of electrons transferred during the electrode process. Using the Bard-Faulkner equation (3) [56] for the irreversible process, the charge transfer coefficient (α) was calculated to be 0.40.

By substituting the value of _E_p and _E_p/2, the potential where the current is at half the peak value. The number of electrons n was found to be ~1 for oxidation, which agrees with the previously reported results for ERL. Further, from the intercept of the _E_pa versus v curve by extending to the vertical axis at v = 0, the value of _E_⁰ (redox potential) was found to be 0.733 V. _k_s is the apparent charge transfer rate constant of the reaction and was calculated to be approximately 1.2×10⁻2 s-1 [57].

The incorporation of TiO2NP enhanced the electron transfer kinetics at the electrode surface relative to the bare SPE, as evidenced by improved peak definition and higher peak currents. Electrode modification enhances current sensitivity and lowers overpotential, but it does not change the intrinsic redox stoichiometry of the analyte, which is governed by the molecule's chemical structure. According to Laviron's theory, the calculated value of n reflects only the electrons involved in the rate-determining step, meaning subsequent rapid transfers will not increase the observed kinetic electron count. Consequently, the modification improves the efficiency of the interface without altering the fundamental number of electrons the molecule is thermodynamically capable of transferring [58,59].

Probable electrochemical oxidation mechanism

To further elucidate the pathway underlying the electrocatalytic performance of the developed sensor, a molecular oxidation mechanism for ERL is proposed based on our experimental kinetics (as shown in Figure 6) and the established quinazoline electrochemistry literature by Bakirhan et al. [60]. Structurally, ERL consists of an electron-rich quinazoline core substituted with an aniline moiety and ethoxy side chains [60]. The electrochemical oxidation occurs predominantly at the nitrogen heteroatoms of the central azaheterocyclic quinazoline ring system. The overall process is governed by a proton-coupled electron-transfer mechanism involving two electrons and two protons (2e-/2H+), proceeding in two distinct sequential steps. ERL undergoes an initial slow, irreversible loss of one electron (e-) and one proton (H+) at the nitrogen site to generate a highly reactive radical intermediate. This is fully supported by our application of the Laviron and Bard-Faulkner Equations (2) and (3), which yielded an electron count of n approximately 1 for the reaction. The unstable radical intermediate rapidly loses a second electron (2nd e-) and a second proton (2nd H+) to form a stable electro-oxidized quinazoline-derivative product.

Figure 6.: Schematic representation of the electrocatalytic oxidation mechanism of erlotinib on a TiO2 modified electrode surface through a sequential 2 e-/ 2 H+ PCET mechanism

The integrated TiO2 nanoparticles play a dual electrocatalytic role in this pathway. First, the abundant surface hydroxyl -OH groups identified in our FTIR spectra facilitate strong hydrogen-bonding interactions with the nitrogen atoms of the incoming ERL. Second, the high density of states on the nanostructured anatase surface coordinates seamlessly with the π -electron-rich aromatic rings of the drug. This localized surface concentration pre-orients the molecules, substantially lowering the interfacial charge-transfer resistance _R_ct decreased from 250 to 110 Ω and leading to the enhanced current responses observed [61,62].

Electrochemical characterization of the modified electrode

The electrochemical behaviour of the bare SPE and the TiO2NPs-modified SPE was investigated using CV and EIS. The study was conducted in a probe solution containing 5 mM [Fe(CN)6]3-/4- and 0.1 M KCl. As illustrated in Figure 7a, both electrodes exhibited characteristic reversible redox peaks associated with the [Fe(CN)6]3-/4- couple. Significant differences were observed following surface modification. The bare SPE (blue line) showed a broader peak potential separation (Δ_E_p) and a lower anodic peak current, _I_pa = 0.21 μA. Upon modification with TiO2NP (black line), the anodic peak current increased to approx. 0.28 μA. Furthermore, the Δ_E_p decreased, with the anodic peak shifting toward a more negative potential. This enhancement is attributed to the increased electroactive surface area provided by the TiO2NP and their synergistic effect in facilitating faster electron transfer between the redox probe and the electrode surface.

Figure 7.: Electrochemical characterization of bare SPE and TiO2NP@SPE. (a) Cyclic voltammograms and (b) Nyquist plots of the bare SPE (blue line) and TiO2NPs-modified SPE (black line) recorded in 5 mM Fe[(CN)6]3-/4- containing 0.1 M KCl. CV scan rate: 50 mV s-1. EIS measurements were performed over a frequency range from 0.1 Hz to 100 kHz at an amplitude of 5 mV

To further validate the interfacial properties and charge transfer kinetics, EIS measurements were performed, and the results are presented as Nyquist plots in Figure 7b. The plots consist of a semicircular portion at higher frequencies, representing the charge transfer resistance _R_ct, followed by a linear portion at lower frequencies, corresponding to the diffusion-limited process (Warburg impedance). The Nyquist plot for the bare SPE displays a significantly larger semicircle diameter, indicating a higher _R_ct (250 Ω). In contrast, the SPE@TiO2 shows a much smaller semicircle, with an _R_ct value reduced to approx. 110 Ω.

Sensitivity studies: calibration and detection limits

DPV responses of TiO2NP@SPE were recorded for ERL concentrations from 15 to 65 μM in 0.1 M PB (pH 7.4) (Figure 8a). A well-defined anodic peak near 0.65 V vs. Ag/AgCl was observed, and the peak current increased linearly with ERL concentration across the entire range studied. The resulting calibration plot (Figure 8b) yielded the linear equation _I_pa = 0.068 _C_erl + 9.595, with a coefficient of determination _R_2 = 0.987 (n = 11), as in Figure 8b confirming excellent linearity. The residual standard error of the regression was _σ_reg = 1.14 μA (as shown in Figure 8b), which reflects the scatter of experimental data points about the fitted line. LOD and LOQ were calculated using the standard blank-based expressions LOD = 3_σ_blank/m and LOQ = 10_σ_blank/m, where m = 0.068 μA μM-1 is the calibration slope and _σ_blank = 0.2075 μA is the standard deviation of the blank current measured independently from replicate blank recordings (n = 6) in 0.1 M PB (pH 7.4) in the absence of analyte. This yielded an LOD of 9.15 μM and an LOQ of 30.52 μM. The area-normalised sensitivity, calculated by dividing the raw slope by the electroactive surface area of the TiO2NP@SPE (0.0383 cm2), was 1.775 μA μM-1 cm⁻2.

Figure 8.: (a) Electrochemical characterization of ERL detection. DPV showing the ERL oxidation peak current increases with concentration (15 to 65 μM) at a scan rate of 50 mV s-1; (b) calibration curve showing a linear response of peak current vs. ERL concentration described by Ipa = 0.068 Cerl + 9.595 (R2 = 0.987, n = 11, σ = 1.14); (c) Nyquist plots for different ERL concentrations, showing the decrease in Rct as the concentration increases and (d) plot of Rct vs. concentration of ERL showing a linear decrease described by Rct = -0.56 Cerl + 42.1 (R2 = 0.991, n = 11, σ = 0.85)

The comparison of the analytical performance reveals that the TiO2NP@SPE provides a superior catalytic environment for ERL detection. While the bare SPE exhibited a raw calibration slope of (0.152 μA μM-1), the modified electrode demonstrated an area-normalized sensitivity of (1.775 μA μM-1 cm⁻2), marking a substantial 10.6-fold increase over the bare electrode’s normalized response (0.167 μA μM-1 cm⁻2). Although the raw slope of the modified sensor (0.068 μA μM-1) is numerically lower than that of the bare electrode, the significant gain in area-normalized sensitivity confirms that the TiO2NPs provide a high degree of intrinsic electrocatalytic enhancement, thereby fundamentally improving charge-transfer efficiency at the electrode interface.

Despite this significant gain in sensitivity, the LOD for TiO2NP@SPE (9.15 μM) is higher than that of the bare SPE (LOD = 0.03 μM), a result that is primarily attributable to an increase in _σ_blank rather than any deficiency in sensitivity. Surface modification with TiO2 nanoparticles substantially increases the double-layer capacitance (_C_dl) of the electrode, which raises the non-faradaic background current and its associated fluctuations. Even a proportionally larger increase in _σ_blank relative to m will shift the detection limit to a higher concentration. Furthermore, manual fabrication of the nanoparticle-modified electrode may introduce surface inhomogeneities or adsorbed impurities that introduce additional variability into the blank response. Together, these factors explain the superior sensitivity and a higher LOD can coexist in the modified electrode system [63-65].

The EIS linearity plot (Figure 8c) demonstrates a strong inverse linear relationship between charge transfer resistance (Rct) and erlotinib concentration across the range of 15-65 μM, described by the equation _R_ct = −0.56 _C_erl + 42.1, with an excellent coefficient of determination (_R_2 = 0.991) as in Figure 8d, indicating high reliability of the sensor response. As the concentration of ERL increases, the _R_ct values decrease consistently, suggesting that the progressive binding of ERL molecules to the electrode surface facilitates charge transfer kinetics, likely by disrupting the blocking layer or altering the interfacial properties of the modified electrode. The low standard deviation (σ = 0.85) across 11 data points confirms the reproducibility and precision of the measurements, validating the biosensor's analytical performance for quantitative detection of ERL within the tested concentration range.

Selectivity studies

The selectivity of the TiO2NP@SPE was rigorously evaluated against common physiological ions, sugars, neurotransmitters, and metabolic byproducts. As summarized in Table 1, the sensor demonstrated high tolerance toward inorganic ions (Mg2+, Na+, K+ and Ca2+) at concentrations as high as 0.1 M, with signal deviations remaining ≤ 0.5 %. Similarly, the addition of common sugars (glucose, sucrose, and maltose at 8.93 mM) resulted in a negligible signal change of -0.21 %, -0.24 % and -0.3 % respectively. Furthermore, key biological interferents, including dopamine (2 μM), urea (1.31 mM) and uric acid (0.5 mM), yielded minor current fluctuations of +1.0, +0.98 and +1.1 %, respectively. These results indicate that the TiO2NP@SPE platform maintains high analytical integrity in the presence of high-concentration physiological matrices.

Interferent | Concentration, mM | Signal change, %
Mg2⁺ | 100 | +0.01
Na⁺ | 100 | +0.5
K⁺ | 100 | +0.01
Ca2⁺ | 100 | +0.01
Glucose | 8.93 | −0.21
Sucrose | 8.93 | −0.24
Maltose | 8.93 | −0.30
Dopamine | 0.002 | +1.0
Urea | 1.31 | +0.98
Uric acid | 0.5 | +1.1

The selectivity of the TiO2NP@SPE sensor was systematically evaluated against a diverse panel of co-administered pharmaceuticals and structurally related anticancer agents. As depicted in Figure 9a, common co-administered drugs including fluoroquinolone antibiotics MOX, ACT, TNZ, LEV, PNP and MET, each at 22 μM, exhibited no discernible faradaic response across the entire potential window of 0.2-0.7 V, remaining virtually indistinguishable from the blank phosphate buffer baseline. In sharp contrast, 22 μM ERL produced a well-defined, prominent oxidation peak at approximately 0.62 V with a peak current exceeding 11 μA, unambiguously demonstrating the inherent electrochemical inactivity of these common interferents at the TiO2NP@SPE surface. The relative standard deviation (RSD) of 1.9 % obtained in the presence of these non-electroactive interferents further confirms that the sensor response to ERL remains essentially unaffected by their presence.

Figure 9.: Selectivity and interference studies of the electrochemical sensor: (a) DPVs of the sensor in 0.1 M PB containing 22 μM ERL compared to 22 μM concentrations of various pharmaceutical interferents (MOX, LEV, TNZ, ACT, PNP and MET), showing a distinct faradaic response only for ERL and (b) DPV responses for 55 μM of competing anti-cancer drugs (ERL, STB, CAP, IMT) and 22 μM DSB, demonstrating the resolution of oxidation peaks for multi-drug analysis

The selectivity of the TiO2NP@SPE sensor was evaluated against structurally related anticancer agents that are frequently co-administered with erlotinib in clinical settings. As shown in Figure 9b, at a concentration of 55 μM, anticancer drugs including sunitinib (STB), capecitabine (CAP), and imatinib (IMT) displayed distinct oxidation peaks at approximately 0.42, 0.48 and 0.65 V, respectively, all of which are well-resolved from the characteristic ERL oxidation peak observed at ~0.68 V. Similarly, dasatinib (DSB) at 22 μM exhibited a broad redox response cantered around 0.55-0.60 V, which remains distinguishable from the ERL signal. PB confirmed negligible background current throughout the potential window. Notably, although these electroactive anticancer interferents produce their own faradaic responses, their peak potentials are sufficiently separated from that of ERL to prevent any meaningful signal overlap. RSD of 2.2 % obtained in the presence of these anticancer interferents remains well within the acceptable analytical threshold of 5 %, confirming that despite minor ionic strength variations or surface adsorption dynamics induced by co-existing electroactive species, the TiO2NP@SPE platform maintains robust electro-selectivity for the accurate and reliable detection of erlotinib in complex therapeutic matrices.

Reproducibility and stability

The repeatability of the TiO2NP@SPE was assessed by comparing DPV responses of 8 runs on a single TiO2NP@SPE with 55 μM ERL (Figure 10a).

Figure 10.: Evaluation of sensor performance for 55 μM ERL detection. (a) Intra-electrode repeatability (n = 8), (b) inter-electrode reproducibility (n = 6) using TiO2NP@SPE in 0.1 M PB (pH 7.4) and (c) Storage stability profile showing the peak current retention over a 30-day period. Error bars represent the standard deviation of triplicate measurements

RSD was approximately 1.5 %, indicating excellent repeatability. The reproducibility of the TiO2NP@SPE was assessed by comparing DPV responses of 6 independently fabricated TiO2NP@SPEs with 55 μM ERL (Figure 10b). RSD was approximately 1 %, indicating excellent batch-to-batch reproducibility.

The long-term stability of the TiO2NP@SPE was investigated over 30 days under ambient storage conditions (Figure 10c). The sensor retained approximately 99.9 % of its initial current response after 15 days and 97.7 % after 30 days, with the gradual decline beyond day 15 attributed to slow degradation of the TiO2NP film upon prolonged ambient exposure, as reflected by the overall RSD of 0.3 % (s = 0.031) across the entire 30-day monitoring period. It should be noted that the influence of environmental variables such as temperature fluctuations and humidity on sensor stability was not investigated in the present study and remain outside the scope of this work; these aspects warrant systematic evaluation in future studies to fully establish the storage robustness of the fabricated sensor under varied real-world conditions.

Recovery studies in human serum

The practical applicability of the TiO2NP@SPE was evaluated by conducting recovery studies in two human serum samples spiked with ERL at 33, 44, and 55 μM. Both intra-day and inter-day analyses were performed (n=6 for each), and the results are presented in Table 2.

AddedC / μM | MeasuredC / μM ± SD | Recovery, % | RSD, % | t value | MeasuredC / μM ± SD | Recovery, % | RSD, % | t value
 | Intra-day analysis - sample 1 | Inter-day analysis - sample 1
33 | 33.48 ± 0.72 | 101.45 | 2.15 | 0.48 | 32.81 ± 0.86 | 99.42 | 2.62 | 0.66
44 | 43.62 ± 0.91 | 99.14 | 2.09 | 0.52 | 44.51 ± 1.07 | 101.56 | 2.40 | 0.71
55 | 54.21 ± 1.08 | 98.56 | 1.99 | 0.57 | 55.67 ± 1.32 | 101.22 | 2.37 | 0.78
 | Intra-day analysis - Sample 2 | Inter-day analysis - Sample 2
33 | 32.69 ± 0.68 | 99.06 | 2.08 | 0.46 | 33.74 ± 0.89 | 102.54 | 2.64 | 0.69
44 | 44.18 ± 0.95 | 100.41 | 2.15 | 0.54 | 43.36 ± 1.02 | 98.55 | 2.35 | 0.73
55 | 55.42 ± 1.11 | 100.76 | 2.02 | 0.59 | 54.38 ± 1.29 | 98.87 | 2.37 | 0.81

For serum sample 1, intra-day recoveries ranged from 98.56 to 101.45 % (RSD: 1.99 to 2.15 %), and inter-day recoveries were 99.42-101.56 % (RSD: 2.37 to 2.62 %). For serum sample 2, intra-day recoveries were 99.06 to 100.76 % (RSD: 2.02 to 2.15 %) and inter-day recoveries were 98.55 to 102.54 % (RSD: 2.35 to 2.64 %). These results demonstrate that the TiO2NP@SPE accurately determines ERL in complex biological matrices with no significant matrix interference. Student’s t-test was applied (n = 6, df = 5, 95 % confidence). The calculated t-values were lower than the tabulated t-value (4.032), indicating no statistically significant difference between the measured and nominal concentrations.

Comparison with literature

Table 3 summarizes a comprehensive comparison of the TiO2NP@SPE sensor with previously reported methods for ERL detection, encompassing both conventional chromatographic techniques and electrochemical approaches. Among the chromatographic methods, LC-MS/MS [66], UPLC-MS/MS [67] and HPLC-MS/MS [69,69] have been widely employed for ERL quantification in human plasma, offering broad linear ranges (10 to 5000 ng L-1) with low detection limits. Capillary electrophoresis [70] has similarly been applied to urine matrices, demonstrating linearity over the range of 0.15 to 20 mg L-1. While these techniques provide excellent sensitivity and selectivity, they are inherently limited by their requirement for expensive instrumentation, skilled operators, extensive sample pre-treatment, and prolonged analysis times, rendering them unsuitable for rapid point-of-care or routine clinical monitoring.

Method | Material | Matrix | Linear range, μM | LOD, nM | Ref.
DPV | Bare SPE | Human serum, urine, tablet | 2-34 | 30 | [39]
AdSSWV | β-CD NPs/GCE | Serum, urine | 0.001 to 8.000 | 1.07 | [60]
LC-MS/MS | - | Human plasma | 0.025 to 12.710 | — | [66]
UPLC-MS/MS | - | Human plasma | 0.025 to 12.700 | — | [67]
LC-MS/MS | - | Human plasma | 0.025 to 12.700 | — | [68]
HPLC-MS/MS | - | Human plasma | 0.254 to 5.090 | — | [69]
Capillary electrophoresis | - | Human urine | 0.38 to 50.9 | — | [70]
DPV | MWCNT/PUFIX/HF-PGE | Nail, urine | 7.70to 142.00 | 110×103 | [71]
DPV | N-GQdot/CuNPs/PANI@GO-GCE | Serum, urine | 0.001 to 35.000 | 712×103 | [72]
DPV | DNA/AuPt/p-L-Met/SPE | Human serum | 0.032×10-3 to 10-3 | 0.009 | [73]
DPV | MIP-pyrogallol/SPCE | Human serum | 0.05×10-3 to 800×10-3 | 0.016 | [74]
DPV | CuBO2 nanosheets/GQDs/GCE | Human serum | 1.0 to 60.0 | 500 | [75]
DPV | TiO2NP@SPE | Human Serum | 15 to 65 | 9150 | This work

Electrochemical methods have emerged as attractive alternatives owing to their simplicity, low cost, and miniaturization potential. Among the reported electrochemical sensors, the β-CD NPs/GCE-based AdSSWV sensor [60] achieved an impressive LOD of 1.07 nM with a linear range of 0.001 to 8.000 μM in serum and urine, benefiting from the host-guest recognition properties of β-cyclodextrin. The MWCNT/PUFIX/HF-PGE platform [71] reported a linear range of 7.70 to 142.00 μM with an LOD of 0.11 μM in nail and urine samples. The DPV-based sensor using N-GQdot/CuNPs/PANI@GO-GCE [72] delivered a similarly low LOD of 0.712 nM over a range of 0.001 to 35.000 μM, attributable to the synergistic electrocatalytic activity of the complex nanohybrid architecture. The DNA/AuPt/p-L-Met/SPE biosensor [73] demonstrated picomolar-level detection in human serum through the integration of bimetallic nanoparticles and a DNA recognition layer. The MIP-pyrogallol/SPCE sensor [74] achieved an ultralow LOD of 0.016 nM (0.000016 μM) across a wide dynamic range of 0.05 to 800.00 nM in serum, exploiting the molecular recognition capability of the imprinted polymer. The CuBO2 nanosheets/GQDs/GCE sensor [75] provided a linear range of 1.0 to 60.0 μM with an LOD of 0.5 μM in human serum, while the bare SPE sensor [39] offered a linear range of 2 to 4 μM with an LOD of 0.03 μM across multiple matrices including serum, urine, and tablet formulations.

Although several of the above electrochemical sensors exhibit lower LOD values, they invariably rely on elaborate multi-step synthesis of complex nanohybrids, molecularly imprinted polymers, or bioreceptor functionalization, which significantly increases fabrication complexity, cost, and batch-to-batch variability. In contrast, the TiO2NP@SPE sensor proposed in this work achieves an LOD of 9.15 μM within a clinically relevant linear range of 15 to 65 μM, which encompasses the therapeutically significant ERL plasma concentration window, without necessitating intricate surface modification procedures. The screen-printed electrode platform inherently ensures disposability, portability, and low cost, while the TiO2 nanoparticle modification is achieved through a straightforward single-step process. Furthermore, the oxidation peak potential of 0.66 V observed for ERL at TiO2NP@SPE is notably lower than those reported for GCE-based sensors operating at 0.94 and 1.5 V, reflecting the superior electrocatalytic activity of the modified surface toward ERL oxidation and enabling measurements with reduced interference from co-existing electroactive species. Collectively, these attributes position the TiO2NP@SPE as a practically viable, clinically applicable, and user-friendly platform for ERL therapeutic drug monitoring.