Electrocatalytically Amplified Immunosensing Enabled by Titanium Oxynitride‐Supported Ru Nanodots on Graphene Oxide Hybrid Architecture

ABSTRACT

Electrocatalytically amplified electrochemical immunosensing is a powerful strategy for sensitive and interference‐free detection of biomarkers. Herein, we report a high‐performance electrochemical immunosensor enabled by a hierarchical hybrid architecture comprising Ru nanodots anchored on titanium oxynitride nanoflakes dispersed on graphene oxide (Ru/TiON‐GO), linked via APTES to biorecognition building blocks. The engineered heterostructure exhibits advanced electrocatalytic activity, arising from synergistic electronic coupling between Ru nanodots and the highly conductive TiON‐GO support, effectively promoting interfacial electron transport between the electrocatalytic surface and the [Fe(CN)₆]^3−/4− redox probe. Leveraging this electrocatalytic platform, a prostate‐specific antigen (PSA) impedimetric immunosensor was constructed, achieving a limit of detection of 0.06 ng mL⁻¹ (2.3 pM) and a wide linear response covering clinically relevant concentration range. The immunosensor demonstrates excellent selectivity in human serum, operating in interference‐free mode even in the presence of common coexisting biomolecules, highlighting suitability for medical diagnostics. Density functional theory calculations further elucidate the origin of the enhanced electrocatalytic performance, with charge density difference plots and density of states analysis revealing Ru‐driven electron redistribution and increased density of states near the Fermi level. This work establishes the Ru/TiON‐GO nanocomposite as a robust electrocatalytic platform for advanced immunosensing applications, paving the way toward next‐generation electrochemical diagnostic devices.

A synergistic effect among the electrocatalytic Ru/TiON‐GO nanocomposite, biorecognition elements, and the [Fe(CN)₆]^3−/4− redox probe for highly sensitive and selective impedimetric detection of PSA protein in human serum was demonstrated. The immunosensor exhibited a limit of detection of only 0.06 ng mL⁻¹ PSA protein, underscoring the importance and potential of functional electrocatalytic materials for their integration into electrocatalytic‐assisted immunosensor design.

1. Introduction

Electrocatalytic (nano)materials are highly valued in electrochemical immunosensing due to their distinctive properties, with electrical conductivity and tunable surface chemistry being of utmost importance [1, 2, 3]. High electrical conductivity is essential for facilitating efficient electron transfer between the supporting electrode surface, biosensor building blocks, and the reporter probe or target analyte [4]. Conductive materials such as carbon nanotubes [5, 6], graphene [7], MXenes [8, 9, 10, 11], and metal nanoparticles [12, 13] are often used for this purpose, as they improve the signal output during interactions at the molecular level. In addition, the biocompatibility and chemical stability of these nanomaterials are of great importance to preserve the integrity of biological recognition molecules in the biosensor's architecture. Such versatility of nanomaterials enables sensitive detection of various target biomolecules, ranging from proteins, peptides, pathogens, toxins, drugs, nucleic acids, etc [14, 15, 16, 17]. However, not all materials encompass the desired properties, and typically only certain characteristics are being exploited.

In recent years, increasing attention has been devoted to conductive metal‐support composites originally developed for electrocatalysis, as their exceptional charge‐transfer properties, electronic interactions, stability, and tunable surface chemistry could be equally beneficial for electrochemical biosensing platforms. Recent studies have shown that nitrogen‐doped titania, also known as titanium oxynitride (TiON), exhibits electrical conductivity comparable to traditional carbon supports and can host uniformly dispersed, small‐sized nanoparticles, making it highly suitable for advanced electrocatalytic applications [18, 19, 20, 21]. It was demonstrated that strong Pt/TiON adhesion, electronic modulation of Pt particles, high conductivity, and structural robustness collectively enhance both the activity and durability of Pt for the hydrogen evolution reaction (HER) and oxygen reduction reaction (ORR), outperforming conventional carbon‐based supports [19]. Building on this concept, low‐loading Ru nanodots supported on TiON and graphene oxide (Ru/TiON‐GO) have shown remarkable potential for alkaline HER. This is primarily attributed to strong metal‐support interactions, which significantly improve intrinsic activity and metal utilization compared to standard Pt/C and Ru/C catalysts. The Ru/TiON composite exhibits excellent durability under various operating conditions, positioning it as a scalable and versatile platform for broader electrocatalytic applications [18]. Together, these studies demonstrate that TiON is an effective functional component in advanced electrocatalytic materials, primarily due to its strong electronic interactions with supported active sites, such as Pt or Ru nanodots. The rational design of TiON‐carbon‐based composites enables high electrical conductivity, enhanced metal‐support interactions, and uniform dispersion of electrocatalytically active species. This material‐engineering strategy can be directly transferable to an electrochemical biosensing platform, where efficient charge transport and stable, accessible active sites are essential for high electroanalytical performance.

Sensitive and selective detection of cancer biomarkers is crucial for reliable early diagnosis. Prostate‐specific antigen (PSA) is a glycoprotein enzyme produced primarily by the epithelial cells of the prostate gland and is commonly used as a clinical biomarker for prostate health. Elevated levels of PSA in blood serum can indicate prostate‐related conditions, including benign prostatic hyperplasia, prostatitis, or prostate cancer, making it a widely used biomarker for early detection and monitoring of prostate cancer progression [22]. Although contemporary PSA tests based on ELISA [23], CLIA [24], or RIA [25] principles have significantly improved prostate cancer screening, they are not entirely specific to malignancy, which can lead to false‐positive results and unnecessary biopsies. Therefore, ongoing research aims to develop more accurate, fast, affordable, and sensitive methods to detect PSA levels with greater reliability and diagnostic exactness. In this context, when combined with innovative electrocatalytic materials, electrochemistry offers nearly limitless possibilities.

Here, we report a study of a hybrid electrocatalyst, i.e., Ru/TiON‐GO, that integrates a graphene oxide support, TiON nanoflakes, and TiON‐supported Ru nanodots to achieve accelerated charge transfer through robust metal‐support interactions. The Ru/TiON‐GO electrocatalyst delivers high conductivity and superior redox kinetics, and, in combination with APTES‐assisted anchoring of biological recognition elements, enables detection of PSA protein with high sensitivity and selectivity.

2. Materials and Methods

2.1. Chemicals

All commercial reagents were of analytical grade and were used without further purification. All aqueous solutions were prepared using ultrapure water with resistivity not less than 18.2 MΩ cm at 298 K (Milli‐Q, Millipore Corp., Marlborough, USA).

Potassium hexacyanoferrate(III) (K₃Fe(CN)₆), glutaraldehyde solution (C₅H₈O₂, 25% in water), sulfuric acid (H₂SO₄), phosphoric acid (H₃PO₄), titanium isopropoxide (97%), acetone, potassium permanganate (KMnO₄), hydrogen peroxide (H₂O₂), APTES (3‐aminopropyl)triethoxysilane (C₉H₂₃NO₃Si), D‐glucose (Glu), and human serum (from human male AB plasma) were purchased from Sigma Aldrich. Graphite was purchased from Imerys (Timcal KS6L), propanol from Honeywell (99.8%), ruthenium(III) chloride monohydrate (RuCl₃·H₂O) from Apollo Scientific, sodium azide (NaN₃) from Fluka, L‐(+)‐ascorbic acid (AA) from Kemika, bovine serum albumin (BSA) from G‐Biosciences, potassium hexacyanoferrate(II) trihydrate (K₄Fe(CN)₆·3H₂O) from Riedel de Haen, potassium chloride (KCl), sodium chloride (NaCl), sodium hydrogen phosphate dihydrate (Na₂HPO₄·2H₂O), and potassium dihydrogen phosphate (KH₂PO₄) were purchased from Merck, absolute ethanol (C₂H₅OH) from Carlo Erba, native protein A from Staphylococcus aureus, rabbit monoclonal anti‐PSA antibody, and recombinant His‐tagged human serum albumin protein (HSA) were purchased from Abcam, and His‐tagged PSA protein was purchased from Sino Biological. Interleukin‐6 (IL‐6) was purchased from Thermo Fisher Scientific.

2.2. Preparation of Buffer Solutions

Phosphate‐buffered saline (PBS, pH 7.4) was prepared by dissolving 8.00 g NaCl, 200.0 mg KCl, 1.81 g Na₂HPO₄·2H₂O, and 240.0 mg KH₂PO₄ in 1.0 L of ultrapure water. When PBS was used to dilute proteins, 0.09% NaN₃ was added as a preservative. Tris(hydroxymethyl)aminomethane buffer (Tris buffer, pH 8.1) was prepared by dissolving 1.51 g Tris and 4.38 g NaCl in 500 mL of ultrapure water.

2.3. Synthesis of the Ru/TiON‐GO Nanocomposite

The Ru/TiON‐GO nanocomposite was synthesized via a multistep procedure, involving the preparation of graphene oxide (GO) by a modified Hummers method, deposition of a TiO₂ layer on GO via hydrolysis of titanium isopropoxide, impregnation with RuCl₃, and subsequent high‐temperature ammonolysis to form Ru nanodots supported on a TiON‐GO composite. The full synthesis protocol is described in detail in ref [18]. Although the thermal treatment of the GO leads to its partial reduction, for the sake of simplicity, the support is denoted as GO throughout the manuscript.

2.4. Modification of the Supporting Electrode

In the first step, the surface of the screen‐printed carbon electrode (SPCE, DRP‐C110, and a two‐working‐electrode configuration SPCE, X1110, DropSens) was activated with 0.5 M HNO₃ for 15 min. Afterwards, the electrode was rinsed with ultrapure water and dried under a mild nitrogen flow. In the next step, 10 µL of APTES‐functionalized Ru/TiON‐GO suspension was applied on the working electrode surface by drop casting and left to dry in the oven at 40°C. Previously, the APTES‐functionalized Ru/TiON‐GO suspension was prepared as follows: firstly, the synthesized Ru/TiON‐GO particles were dispersed in absolute ethanol to form a 5 mg mL⁻¹ suspension, followed by the addition of APTES (5.0 vol%). After 2 h of vigorous stirring at room temperature, the mixture was centrifuged three times at 15000 rpm (IKA G‐L, IKA), including rinsing with absolute ethanol between centrifugations. Then, the APTES‐functionalized nanocomposite was resuspended in ultrapure water to make a 5.0 mg mL⁻¹ suspension.

After the modification of SPCE with APTES‐functionalized Ru/TiON‐GO, the electrode surface was covered with 10 µL of the PBS solution (pH 7.4) containing 2.5% glutaraldehyde (GA) and left at 4°C for 1 h to promote the formation of the GA cross‐link. Then, the electrode surface was rinsed by immersion in ultrapure water and carefully dried under mild nitrogen gas flow. In the next step, the electrode was covered with 10 µL of 20 µg mL⁻¹ protein A in PBS (pH 7.4) and left for 1 h at room temperature (22‐25°C) under dark storage conditions. After this, the electrode was rinsed with ultrapure water, dried with nitrogen, and incubated with 10 µL of 10 µg mL⁻¹ anti‐PSA antibodies in PBS (pH 7.4) for 1 h at room temperature. To minimize non‐specific binding, the available cross‐links and other non‐specific binding sites were blocked by adding BSA, i.e., 10 µL of 100 µg mL⁻¹ BSA solution in PBS (pH 7.4) was cast onto the working electrode and left at room temperature for 30 min. After a short rinsing sequence with Tris buffer and ultrapure water, the electrode was dried with nitrogen and was ready for use. The schematic presentation of the immunosensor constituents is shown as Scheme S1 in the Supporting Information.

2.5. Structural Characterization and Electrochemical Measurements

Scanning electron microscopy (SEM) was performed using a Zeiss Supra 35 VP instrument. X‐ray diffraction (XRD) measurements were carried out with a D4 Endeavor diffractometer (Bruker AXS). Transmission electron microscopy (TEM) and selected area electron diffraction (SAED) analyses were conducted using a JEOL ARM 200 CF transmission electron microscope operated at 80 kV. The X‐ray photoelectron spectroscopy (XPS) measurements were performed using a Supra+ device (Kratos, Manchester, UK) and an Al Kα excitation source. Time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS) measurements were conducted using a M6 device (IONTOF, Munster, Germany).

All electrochemical measurements, i.e., cyclic voltammetry (CV), square‐wave voltammetry (SWV), and electrochemical impedance spectroscopy (EIS), were carried out using a portable PalmSens 4 potentiostat operated by PSTrace 5.11 software (PalmSens BV, Netherlands). EIS spectra (Nyquist and Bode‐Bode plots) were recorded in the frequency range of 10⁴–10⁻¹ Hz at +0.13 V vs. pseudo Ag‐reference electrode, with an amplitude of 5 mV, in 1.0 mM [Fe(CN)₆]³⁻/[Fe(CN)₆]⁴⁻ in 0.1 m KCl as a redox probe solution.

Recorded impedance spectra were fitted using the Randles‐type equivalent electrical circuit R _S([R _ct Z _W]Q). Herein, R _S corresponds to the solution resistance, R _ct is the charge transfer resistance, Z _W is the Warburg element that models the diffusion phenomena, and Q is the constant phase element (CPE) describing the non‐ideal double‐layer capacitance. As the analytical signal, the difference between the R _ct value extracted from the measurement of the incubated sensor (R _inc) and the R _ct value of the non‐incubated sensor (R _blank), i.e., ΔR _ct = R _inc‐R _blank, was used, if not stated otherwise.

2.6. DFT Computational Studies

Density‐functional theory (DFT) calculations were carried out with Quantum ESPRESSO [26, 27] version 7.0, to gain a closer insight into the differences in electronic structure between the materials that constitute the electrode upon adsorption of the electroactive [Fe(CN)₆]³⁻/[Fe(CN)₆]⁴⁻ species. To reduce the computational costs down to a manageable extent while still retaining the essential chemistry of the system under investigation, computational investigations were limited to the ferrocyanide ([Fe(CN)₆]⁴⁻), and simplified slab models were used for the ruthenium and TiON materials, motivated by the results of previous theoretical and experimental studies involving these materials [28, 29]. For ruthenium, we constructed a single‐layer slab (4×4 supercell) based on the hcp(0001) surface (a = 2.706 Å, c = 4.281 Å), while for TiON, a three‐layer slab (2×2 supercell) was constructed based on the 111 surface of an fcc lattice (a = 4.199 Å), with the middle layer consisting of Ti ions and 25% vacancies, surrounded by terminating O/N layers on both sides. In accordance with ref. [28], we removed 50% of O and N atoms from each side of the TiON slab to respect the experimentally observed stoichiometry. A vacuum of 15Å was added to both slabs to accommodate the [Fe(CN)₆]⁴⁻ binding. The DFT calculations were performed in the regular (spin‐unpolarized) approach with the Perdew‐Burke‐Erzenhof (PBE) generalized gradient density functional approximation [30]. The core electrons were treated by the projector‐augmented wave (PAW) potentials [31], which were obtained from the pslibrary [32]. For the sake of consistency, a fixed cutoff of 60Ry for the wave functions and 600Ry for the electron density was applied to the plane‐wave basis sets in all systems. The convergence threshold for self‐consistent field calculations was set to 10⁻⁵ eV in structure optimizations and 10⁻⁶ eV in electron density calculations. In all calculations, the initial geometry of the [Fe(CN)₆]⁴⁻ was relaxed in the z‐direction to find the optimal distance between the surface atoms and the bound cyano moiety. The Brillouin zone integrations were performed with the gamma k‐point for the isolated [Fe(CN)₆]⁴⁻ and the 4×4×1 Monkhorst‐Pack k‐point grids [33] for slabs and (slab + adsorbed [Fe(CN)₆]⁴⁻) complexes, with the addition of the Methfessel‐Paxton smearing [34] of 0.02Ry. To remove the spurious interactions between periodic images, dipole corrections by Bengtsson [35] and Martyna‐Tuckerman [36] were applied to the slab+[Fe(CN)₆]⁴⁻ complexes and isolated [Fe(CN)₆]⁴⁻, respectively. Finally, total density of states (DOS) calculations were performed on the converged results, using the finer 8×8×1 Monkhorst‐Pack grids. In addition to the ruthenium and TiON slab models, we also calculated DOS for a composite model, consisting of Ru₅ nanoparticles adsorbed on the TiON slab [29], in order to investigate the possible synergistic effects between the two materials at the electronic structure level. For all DOS calculations, spin polarization was considered. In addition, dispersion interactions for the adsorbed Ru₅ nanoparticle were included using the DFT‐D3 correction of Grimme et al. [37].

3. Results and Discussion

3.1. Characterization of Ru/TiON‐GO Nanocomposite

XRD analysis of the nanocomposite material comprising GO support, TiON nanoparticles, and Ru nanodots is presented in Figure 1a. The GO support is indicated by a broad, low intensity peak around 26°, corresponding to the (002) plane of graphene, indicative of its highly disordered or exfoliated state [18]. This broadening reflects the nanoscale nature of GO and its role in providing a large surface area for the dispersing of TiON nanoparticles. The XRD pattern shows characteristic diffraction peaks corresponding to TiON at 37.0, 43.1, and 62.6°, assigned to the crystallographic planes (111), (200), and (220), respectively, in strong agreement with JCPDS 01‐084‐4872 for cubic TiON. The absence of common TiO₂ peaks (JCPDS 01‐073‐8760) serves as proof of the successful transformation of the precursor oxide during the TiON synthesis. Finally, the presence of Ru nanodots is confirmed by the low‐intensity peaks at 58.0°, corresponding to (102) plane of Ru nanodots, in good agreement with JCPDS 6–663. In addition, the SAED pattern confirms both the presence of both TiON nanoparticles and Ru nanodots, as shown in Figure 1b.

FIGURE 1.

(a) XRD pattern obtained for as‐synthesized Ru/TiON‐GO powder, showing diffractions corresponding to the nanocomposite constituents, i.e., GO support, TiON nanoparticles, and Ru nanodots, (b) SAED pattern obtained on the same sample confirming the presence of TiON and Ru nanodots.

TEM analysis of the nanocomposite reveals a uniformly dispersed arrangement of TiON nanoparticles across the GO support (Figure 2a). The GO flakes provide a large surface area, promoting relatively even dispersion of TiON nanoparticles without significant agglomeration; on the other hand, Ru nanodots are anchored on the surface of the TiON nanoparticles (Figure 2b). While the GO flakes are micrometer‐sized, the TiON nanoparticles are mostly within the range of 15–20 nm, as shown in the histogram in the inset of Figure 2b. In addition, the Ru nanodots are mostly in the range of 2.0–3.5 nm, as indicated in the histogram of Figure 2c. Such a size of the Ru nanodots maximizes their surface area, which is essential for improving the catalytic activity toward relevant electrochemical reactions. The intimate contact between Ru nanodots and TiON nanoparticles (Figure 2d) suggests that the nanocomposite is likely suited for applications requiring high electrocatalytic performance. The preferential deposition of Ru nanodots on TiON nanoflakes rather than on GO suggests the presence of metal‐support interactions, known to be favorable for electrocatalytic properties. The energy dispersive X‐ray spectroscopy (EDX) elemental maps of the APTES‐functionalized Ru/TiON−GO nanocomposite confirm both the presence and spatial distribution of the elements identified by XRD, along with the additional Si signal attributed to the organosilane modifier (Figure 2e).

FIGURE 2.

(a–d) TEM images obtained for as‐synthesized Ru/TiON‐GO nanocomposite: (a) lower magnification image focused on the GO support, (b) higher magnification image showing fine dispersion of TiON nanoparticles, along with their size distribution histogram, (c) higher magnification image showing distribution of Ru nanodots within the TiON‐GO, along with their size distribution histogram, (d) implantation of Ru nanodots onto the surface of TiON nanoparticles, (e) FE‐SEM image and overlay of the spatial distribution of analyzed elements, and corresponding EDX elemental maps of the APTES‐functionalized Ru/TiON−GO nanocomposite deposited onto SPCE. The distribution of individual elements is shown for the most abundant elements, i.e., C (59.6 wt.%), Ti (17.6 wt.%), N (8.9 wt.%), O (7.8 wt.%), Ru (2.7 wt.%), and Si (0.5 wt.%).

3.2. Modification of the Supporting Electrode With Ru/TiON‐GO Nanocomposite and Electrochemical Characterization

The amount of electrocatalytic nanocomposite deposited on the electrode surface has proved critical for achieving uniform coverage without aggregation. The work by Compton et al. has shown that even small loadings of nanoparticles can dominate the electrochemical response of an electrode, provided they are well‐dispersed [38]. Guided by this principle, we systematically varied the Ru/TiON‐GO loading (0.1–5.0 mg mL^{− 1}) during modification of bare SPCEs. At low concentrations, the nanocomposite was evenly distributed, with minimal overlap (Figure 3a) and a clear distinction between TiON and GO support (Figure 3b). With increasing concentration, however, particle agglomeration became evident, forming micrometer‐sized clusters that disrupted surface uniformity, as seen in Figure S1. At the highest concentration tested, i.e., 5.0 mg mL^{− 1}, the electrode surface was evidently saturated, yielding a continuous layer of the nanocomposite.

FIGURE 3.

Surface morphology of SPCE at different modification stages characterized by FE‐SEM: (a) SPCE modified with 0.1 mg mL⁻¹ Ru/TiON‐GO nanocomposite in the drop‐casting solution, (b) enlarged view of (a), revealing its surface features, i.e., GO support and TiON nanoparticles.

In further studies, we used 0.1 mg mL^{− 1} Ru/TiON‐GO for SPCE modifications. The above observations were further interpreted by electrochemical studies. CV measurements carried out in 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl showed that the redox process remained diffusion‐controlled across a wide range of potential scan rates, with peak currents scaling linearly with the square root of scan rate (Figure S2). Such linearity demonstrates that electrode modification preserves efficient mass transport while ensuring stable electron transfer. In addition, the modified SPCEs were characterized by SWV in the solution of 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl, and the corresponding square‐wave voltammograms are shown in Figure 4a. The first measurement at the nanocomposite‐modified electrode revealed a significant enhancement of the peak current compared to the unmodified SPCE. Upon the first SWV measurement in a solution containing [Fe(CN)₆]^3−/4− redox couple, the modified electrode was rinsed with ultrapure water, dried with nitrogen, and then the SWV measurement was carried out in 0.1 M KCl. As shown, no residual peak corresponding to the [Fe(CN)₆]^3−/4− was observed, implying there was no adsorption of the redox species on the modified‐electrode surface after the initial measurement. The effective removal of the redox probe during rinsing ensures reliable measurements in subsequent experiments, which are crucial for applications requiring high sensitivity, selectivity, and non‐significant carryover. Finally, the repeated measurement in 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl yielded a peak current that was consistent with the initial measurement, as shown in Figure 4a. In Figure S3, 30 consecutive cyclic voltammograms are shown, recorded using two potential scan rates of 50 and 100 mV s⁻¹ on two identically nanocomposite‐modified electrodes; notably, no significant changes occurred during repetitive cycling at either scan rate. Additionally, EIS provided complementary insight into the electrochemical characteristics of the nanocomposite layer. The Nyquist plots showed no discernible semicircle region, with the impedance response dominated by a diffusion tail (Figure S4). This indicates that the Ru/TiON‐GO nanocomposite behaves nearly as a metallic conductor, where electron transfer is not rate‐determining, and the impedance is governed by mass transport. These results demonstrate that the nanocomposite‐modified electrodes combine high stability and rapid charge transfer, providing a robust platform for immunosensing applications. The superior electrochemical behavior is attributed to the favorable properties of the hybrid electrocatalyst toward the oxidation of redox species. To further elucidate these properties, additional SWV measurements were conducted using SPCEs modified with either Ru/TiON‐GO or TiON‐GO (Figure 4b). A significantly higher current response observed at the Ru‐containing nanocomposite suggests that the electrocatalytic effect toward the oxidation of [Fe(CN)₆]^3−/4− originates from Ru nanodots. This assumption is elaborated by DFT calculations in the following paragraph.

FIGURE 4.

(a) SWVs recorded at bare (dark red) and at Ru/TiON−GO modified (red) SPCE in 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl, at the same Ru/TiON−GO modified SPCE in solely 0.1 M KCl (black dotted), and at the same Ru/TiON−GO‐modified SPCE again in 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl (red dotted); (b) SWVs recorded at bare (dark red), TiON−GO modified (orange), and Ru/TiON−GO modified (red) SPCE in 1.0 mM [Fe(CN)₆]^3−/4− in 0.1 M KCl.

3.3. Theoretical Evaluation of Ru‐driven Electron Transport

3.3.1. Charge Density Difference Plots

To assess the feasibility of binding of the redox‐active [Fe(CN)₆]⁴⁻ on the Ru(0001) and TiON(111) surface, we calculated the charge density difference upon adsorption as follows:

$$\begin{array}{c}\hfill \mathrm{\Delta }\rho \left(x,y,z\right)={\rho }_{slab+{\left[Fe{\left(CN\right)}_6\right]}^{4-}}\left(x,y,z\right)-{\rho }_{slab}\left(x,y,z\right)\\ \hfill -{\rho }_{{\left[Fe{\left(CN\right)}_6\right]}^{4-}}\left(x,y,z\right)\end{array}$$

where slab corresponds to either TiON or Ru. Figure 5a,b show substantial redistribution of charge density with [Fe(CN)₆]⁴⁻ adsorption for both materials. In both cases, the adsorbed coordination complex was observed to form a bond between a metal ion, i.e., Ti or Ru, and the closest CN moiety. In the case of TiON, chemical binding was observed only for surfaces with 50% coverage by N and O atoms, where CN can bind directly to exposed Ti ions. We also observed that geometry relaxations with a fully covered TiON surface did not converge to a structure with a bound complex due to strong electrostatic repulsions between top‐layer O/N atoms and the CN moiety, i.e., meaning that binding occurs when Ti is accessible at the slab surface. A clearer picture emerges when we consider the planar integrated electron density difference in the z direction, which is defined as:

$$\mathrm{\Delta }\rho \left(z\right)={\int }_{S}dxdy\mathrm{\Delta }\rho \left(x,y,z\right)$$

where S is the area of the supercell in the xy plane. For both materials, i.e., TiON and Ru, the binding of [Fe(CN)₆]⁴⁻ is accompanied by a loss of electrons on the closest CN moiety and the central Fe²⁺ ion, which are redistributed onto the slab atoms, indicating successful electron transfer.

FIGURE 5.

3D electron density difference plots at isosurfaces of ±0.001 e/ų. (top of a,b), compared with the planar integrated density differences at different z‐locations (bottom of a,b); density differences for the adsorption of [Fe(CN)₆]⁴⁻ on (a) Ru(0001) surface, and (b) TiON(111) surface with half O and N atoms removed. Red regions represent electron excess, while blue regions represent electron deficit upon [Fe(CN)₆]⁴⁻ adsorption. Total density of states (DOS) plots in a narrow interval (±1 eV around the Fermi level) for the surface, (c) with and (d) without adsorbed [Fe(CN)₆]⁴⁻, of Ru(0001) (violet), TiON(111) (red), Ru₅/TiON(111) (green) material. The combined molecular structures with 3D electron density difference plots in (a) and (b) were generated with the XCrySDen visualization software [40]. Representations of models used in DOS calculations are shown in Figure S5.

3.3.2. Density of States (DOS) Analysis

Another quantity dependent on the electronic structure of a material that is directly related to electrical conductivity is the total density of states (DOS) [39]. In metals and doped semiconductors, a higher DOS near the Fermi level of a system typically correlates with higher electrical conductivity. Figure 5c,d display the DOS analysis of slabs with and without adsorbed [Fe(CN)₆]⁴⁻; in both cases, the DOS is expectedly highest for Ru in a large fraction of the ±1 eV interval around the Fermi level. The TiON slab has the fewest available energy states and displays clear gaps at specific energy levels. As observed in the DOS spectra of the Ru₅/TiON model, the addition of Ru nanoparticles introduces new energy states close to the Fermi level, strongly suggesting that Ru enables better electron transport than TiON alone, a result that is in very good agreement with the electrochemical characterization demonstrated above.

3.4. Immunosensor Fabrication and Optimization

After the initial modification of SPCE with the electrocatalytic Ru/TiON‐GO nanocomposite, we proceeded with the preparation of the electrochemical immunosensor by optimizing the concentrations of protein A and anti‐PSA antibody, targeting the impedimetric detection of 10 ng mL⁻¹ PSA, as shown in the 3D surface plot in Figure 6. PSA above this threshold concentration is often considered a high level and significantly raises the suspicion for prostate cancer, prompting further diagnostic steps such as repeated PSA tests or a prostate biopsy. The x‐axis in the respective 3D surface plot represents protein A concentration, the y‐axis represents the chosen anti‐PSA monoclonal antibody concentration, and the z‐axis corresponds to the impedimetric signal of the immunosensor, i.e., a relative increase of the R _ct obtained after a 15‐min incubation. The concentrations of protein A and anti‐PSA antibody were selected based on a combination of literature reports and our prior experience with biosensor development [41, 42, 43]. This approach ensured the examination of a broad concentration range while focusing on values most likely yielding effective immobilization of protein A and anti‐PSA antibody, along with high electroanalytical performance of the immunosensor.

FIGURE 6.

3D surface plot showing the optimization of protein A and anti‐PSA antibody concentrations for the fabrication of the impedimetric PSA immunosensor (n = 3). Relative signal (%) is plotted as a function of protein A and anti‐PSA antibody concentrations.

The 3D plot revealed a clear relationship between the concentrations of these two functional proteins and the immunosensor response. At low protein A and anti‐PSA antibody concentrations, i.e., 5 µg mL⁻¹ of each, the electroanalytical signal remained low, indicating insufficient surface coverage and suboptimal anti‐PSA antibody binding orientation. As protein A concentration increased, the signal increased, particularly when paired with moderate to high anti‐PSA antibody concentrations. This suggests that a higher surface density of protein A enhances the oriented immobilization of anti‐PSA antibodies by reducing random adsorption and steric hindrance. However, the 3D plot also shows that beyond a certain level of protein A, the signal reaches a plateau or even slightly decreases, likely due to protein A overcrowding the electrode surface and/or reduced accessibility of the anti‐PSA antibody binding sites. This observation is also associated with lower reproducibility, as seen from the error bars. The highest signal increase of ca. 30% is achieved at approximately 20 µg mL⁻¹ protein A and 10 µg mL⁻¹ anti‐PSA antibody. This maximum indicates the optimal ratio where sufficient protein A coverage ensures proper anti‐PSA antibody orientation, while the selected anti‐PSA antibody concentration is high enough to maximize efficient capture of the PSA protein. Excessively high anti‐PSA antibody concentrations at lower protein A concentrations do not yield proportionally higher signals, underscoring the importance of the oriented antibody binding assured by protein A. The presented results confirm that a balanced combination of protein A and anti‐PSA antibody is critical for optimal electrochemical response of the PSA immunosensor. Overloading of components can be counterproductive, either by reducing PSA protein accessibility or by causing non‐specific interactions.

Electrochemical characterization of the stepwise immunosensor fabrication was carried out by EIS and CV in 1.0 mmol L⁻¹ [Fe(CN)₆]^3−/4− in 0.1 mol L⁻¹ KCl, as shown in Figure 7. The Nyquist plots (Figure 7a) display initially a significant decrease in R _ct after modification of the supporting SPCE with electrocatalytic Ru/TiON‐GO nanocomposite, followed by a gradual increase of R _ct with each consecutive modification step, reflecting the progressive construction of the immunosensor architecture associated with the electron transfer hindrance at the electrode surface interface. This increase in R _ct is expected and consistent with the immobilization of biofunctional layers, which partially block the electrochemical reaction of the redox probe at the electrode surface. Notably, the electrode modified with only Ru/TiON‐GO exhibited a significant reduction in R _ct, showing nearly metallic behavior. Also, the CV measurements in Figure 7b display an initial current increase after the modification of SPCE with electrocatalytic Ru/TiON‐GO nanocomposite, followed by a gradual decrease in peak currents and a slight increase in peak‐to‐peak potential separations after a stepwise addition of protein A, anti‐PSA antibody, and BSA, as the modification layer becomes thicker, accompanied by hindered mass transport.

FIGURE 7.

Step‐by‐step fabrication of the PSA immunosensor monitored by (a) EIS (Nyquist plots) and (b) CV, in the presence of 1.0 mmol L⁻¹ [Fe(CN)₆]^3−/4− in 0.1 mol L⁻¹ KCl.

3.5. XPS and ToF‐SIMS Analysis of the Immunosensor Architecture

To further investigate the immunosensor architecture, XPS measurements were performed, as shown in Figure 8. The analyses show the gradual fabrication of the immunosensor up to the addition of protein A. XPS analyses were also carried out for the control samples B and F, corresponding to the electrocatalytic Ru/TiON‐GO nanocomposite and protein A adsorbed on the supporting SPCE, respectively.

FIGURE 8.

(a) Survey and high resolution XPS spectra for (b) C 1s and Ru 3d, (c) O 1s, (d) Ti 2p and Ru 3p, (e) N 1s, and (f) S 2p and Si 2s, measured on different samples: (A) bare SPCE, (B) SPCE with 0.1 mg mL⁻¹ Ru/TiON−GO, (C) SPCE with 0.1 mg mL⁻¹ APTES‐functionalized Ru/TiON−GO, (D) sample C subsequently modified with glutaraldehyde (GA), (E) sample D subsequently modified with protein A, (F) SPCE modified with protein A.

The survey spectrum (Figure 8a) for sample A, i.e., bare SPCE, shows intense C 1s, O 1s, Cl 2p, and Cl 2s signals, originating from carbon material and binders used in the preparation of screen‐printing paste. The source of C and O can also arise from adventitious carbonaceous species adsorbed on the surface from the surrounding atmosphere. After the drop‐casting of Ru/TiON−GO nanocomposite (sample B), new peaks develop corresponding to Ti (Ti 2p spectrum, Figure 8d), Ru (Ru 3d, Figure 8b), as well as an additional low intensity spectral feature in O 1s, at a more negative binding energy compared to the main peak, corresponding to Ti‐O (Figure 8c). For the sample C, corresponding to APTES‐functionalized Ru/TiON−GO nanocomposite, the N 1s signal develops due to the amine groups of APTES (Figure 8e). The introduction of protein A (samples E and F) results in the occurrence of the S 2p feature at 168.0 eV (Figure 8f) and a higher intensity peak in N 1s corresponding to the peptide feature (Figure 8e). The C 1s spectra show different spectral features (Figure 8b); the dominant feature at 284.5 eV corresponds to graphitic sp² carbon from the GO support and the underlying SPCE. The spectral feature at 286.5 eV corresponds to C‐O/C‐N, while the low intensity high‐binding‐energy feature at approximately 288.9 eV indicates carboxyl groups. The O 1s spectra show different spectral features, i.e., Ti‐O at 530.3 eV, and intense contributions likely comprising organic components containing C = O/C‐O/C‐OH located at 532–534 eV (Figure 8c). The APTES‐containing samples, i.e., samples C, D, and E, display a distinct Si 2s feature in the 152 – 154 eV range, suggesting the presence of a silane layer (Figure 8f). On the other hand, S 2p peaks at around 163–165 eV are present for all samples with the Ru/TiON−GO nanocomposite (B, C, D, E), most likely due to sulfur residues from the modified Hummers method for the preparation of the GO support, which involves the usage of sulfuric acid.

ToF‐SIMS was employed to map the spatial distribution of the components within the immunosensor architecture up to the deposition of protein A. Three‐dimensional ToF‐SIMS images were acquired in both negative (Figure 9a) and positive (Figure 9b) ion modes. The TiO⁻ signal (m/z 63.94) originates from titanium oxynitride, while Ru⁻ signal (m/z 101.90) represents Ru nanodots. Seemingly, the Ru⁻ exhibits a distribution pattern similar to that of TiO⁻, confirming the attachment of Ru nanodots to the oxynitride particles. Si⁻ signal (m/z 63.94) is attributed to APTES, and as can be observed, indeed colocalizes with the TiO⁻fragment. Interestingly, the S⁻ signal (m/z 31.97) (not shown) is omnipresent on the sample surface and can be attributed to either protein A or the GO support, as already discussed in the XPS results.

FIGURE 9.

ToF‐SIMS analysis of the PSA immunosensor architecture up to the introduction of protein A, including (a) negative and (b) positive polarity signals.

The spatial distributions of positive ions are shown in Figure 9b, strongly supporting (i) the efficient modification of TiON nanoparticles with APTES through colocalization of Ti⁺ (m/z 47.95) and Si⁺ (m/z 27.97) signals, and (ii) the attachment of Ru nanodots to TiON nanoparticles through colocalization and intensities of Ti⁺ (m/z 47.95) and Ru⁺ (m/z 101.89) signals. The surface distribution of protein A is evidenced by the lysine immonium ion C₅H₁₀N⁺ at m/z 84.08, the leucine/isoleucine immonium ion C₅H₁₂N⁺ at m/z 86.10, and the phenylalanine immonium ion C₈H₁₀N⁺ at m/z 120.07 [44, 45]. In addition, the CH₄N⁺ fragment (m/z 30.03) is typically attributed to glycine and lysine [44], the C₂H₆N⁺ fragment (m/z 44.05) to alanine, while the C₄H₁₀N⁺ fragment (m/z 72.08) corresponds to valine. The C₃H₄NO⁺ fragment (m/z 70.03) and the C₄H₆NO⁺ fragment (m/z 84.04) are the usual beacons for asparagine and glutamine (glutamic acid) [45]. The former two signals can also be attributed to glutaraldehyde‐derived iminium species or a Schiff rearrangement product formed during the reaction of glutaraldehyde with the amine groups of APTES and/or the lysine side chain of protein A. Finally, the C₃H₈N⁺ fragment (m/z 58.06) is attributed to the terminal amine side of APTES as a perfect structural match and previously reported assignment [46].

3.6. Analytical Performance of the Immunosensor

The effect of incubation time on the immunosensor response toward PSA protein was followed to optimize its performance (Figure S6). The impedimetric signal showed a local maximum at ca. 15 min, reflecting the most efficient binding of PSA protein to the functionalized immunosensor surface. Longer incubation led to a gradual decrease in the response, likely due to partial dissociation processes of the formed conjugates. In addition, the possibility of steric hindrance, protein conformational changes, or protein cross‐linking effects cannot be excluded. Based on these results, an incubation time of 15 min was selected as optimal, providing the highest signal while maintaining good reproducibility and minimizing the immunoassay duration. To minimize non‐specific binding and improve immunosensor selectivity, surface blocking was investigated by varying both BSA concentration and its incubation time. A short series of experiments demonstrated that relatively low BSA concentrations or longer incubation times resulted in incomplete coverage of the immunosensor surface, leading to false positive signals (Table S1). At lower BSA concentrations, the immunosensor is only partially protected, leaving exposed sites available for non‐specific adsorption, thereby increasing false‐positive events. On the other hand, longer incubation times may lead to structural rearrangements of the initially adsorbed BSA layer, such as protein unfolding, aggregation, or desorption, that compromise the integrity of the blocking layer. This can re‐expose surface sites for non‐specific binding, leading to higher false‐positive signals. Therefore, the optimized conditions of 100 µg mL⁻¹ BSA in combination with 30 min incubation represent a balance in which the surface is sufficiently protected, minimizing non‐specific adsorption and yielding high electroanalytical performance.

Using the optimized preparation and measurement conditions, the impedimetric response of the immunosensor was followed in a concentration range of 0.1‐100.0 ng mL⁻¹ PSA protein in human serum. The human serum samples were diluted tenfold in PBS buffer to reduce potential matrix effects. After the incubation period, the immunosensor response was recorded in 1.0 mmol L⁻¹ [Fe(CN)₆]^3−/4− in 0.1 mol L⁻¹ KCl; measurements were carried out using four identically prepared immunosensors per calibration point. The obtained calibration plot (Figure 10a), constructed on a logarithmic scale of PSA protein concentrations, demonstrated high linearity with a correlation coefficient of r ² = 0.998. The immunosensor exhibited a low limit of detection (LOD, 3σ criterion) of only 0.06 ng mL⁻¹ (2.3 pM) PSA protein in human serum. Such a low LOD enables detection of PSA protein at concentrations significantly below commonly used clinical decision thresholds. In healthy adult men, serum PSA protein levels are typically below 4.0 ng mL⁻¹, with age‐adjusted reference ranges often extending from ca. 2.5 ng mL⁻¹ in younger men to ca. 6.5 ng mL⁻¹ in elderly populations. Therefore, the proposed immunosensor provides sufficiently low LOD to detect clinically relevant fluctuations in PSA protein across all age groups. To further assess the immunosensor performance, the same optimized impedimetric platform was applied for testing several potential interferents in human serum, i.e., AA (10 µg mL⁻¹), Glu (1 mg mL⁻¹), HSA (20 µg mL⁻¹), and IL‐6 (100 pg mL⁻¹), and the signals were compared to the signal of a blank sample. All the examined interferents showed signals practically identical to the response of a blank sample, corroborating that the immunosensor operates in an interference‐free mode (Figure 10b). In addition, Table S2 shows a short review of the reported electrochemical biosensors for the detection of PSA, indicating their diverse characteristics through the complexity of their architectures and detection schemes. Notably, our work reports all calibration and LOD data for spiked real human serum samples instead of PBS. Moreover, the stability of the proposed immunosensor was investigated after 3 and 7 days after the fabrication, with no significant signal loss observed when tested with 10.0 ng mL⁻¹ of PSA (Figure S7).

FIGURE 10.

(a) Calibration plot in the range of 0.1–100.0 ng mL⁻¹ PSA protein in human serum (n = 3), (b) immunosensor response for blank sample and in the presence of common interferents, i.e., AA (10 µg mL⁻¹), Glu (1 mg mL⁻¹), HSA (20 µg mL⁻¹), and IL‐6 (100 pg mL⁻¹), in human serum (n = 3), (c) Nyquist spectra obtained for both blank and PSA protein (10 ng mL⁻¹) containing samples incubated using a two‐working‐electrode configuration, i.e., with the integrated positive and negative‐control unit. The positive‐control: immunosensor modified with anti‐PSA antibodies; the negative‐control: immunosensor without antibodies.

Finally, the performance of the immunosensor was evaluated in its two‐working‐electrode configuration, i.e., with an integrated positive and negative control unit (Figure 10c). The negative control unit is characterized by the same immunosensor architecture as the positive control unit, except that it doesn't contain anti‐PSA antibodies. Upon incubation with the PSA protein, the negative control unit exhibited no measurable change in the signal, confirming the absence of non‐specific binding and validating the effectiveness of BSA as a blocking agent. In contrast, the positive control unit, i.e., the fully prepared immunosensor, showed the expected increase in signal at 10 ng mL⁻¹ PSA protein, demonstrating successful target recognition and confirming the functionality of the immobilized anti‐PSA antibodies.

4. Conclusions

In this work, we demonstrate a synergistic effect among the electrocatalytic Ru/TiON‐GO nanocomposite, biorecognition elements, and the [Fe(CN)₆]^3−/4− redox probe for highly sensitive and selective impedimetric detection of PSA protein. Density functional theory calculations corroborate Ru‐driven electron redistribution and a higher density of states near the Fermi level (vs. TiON without Ru), facilitating effective electronic communication between the [Fe(CN)₆]^3−/4− probe and the electrocatalytic surface. The immunosensor architecture includes APTES, providing (i) immobilization of the electrocatalytic Ru/TiON‐GO nanocomposite on the supporting SPCE and (ii) further binding of biological elements within the hierarchical structure. The oriented binding of anti‐PSA antibodies was achieved through protein A, which was confined to APTES using glutaraldehyde, whereas the uppermost BSA layer ensured interference‐free impedimetric immunosensing. The stepwise construction and optimization of the immunosensor were monitored using electrochemical techniques, including CV, SWV, and EIS, in combination with XPS and ToF‐SIMS analysis to achieve reproducible fabrication. To further increase reliability, a two‐working‐electrode configuration was implemented, incorporating a built‐in negative control unit, thereby reducing false positives. The immunosensor was tested in human serum, exhibiting a low limit of detection of only 0.06 ng mL⁻¹ (2.3 pM) PSA protein. The electroanalytical capabilities of this label‐free immunosensor underscore the importance and potential of highly performing electrocatalytic materials for their integration into electrocatalytic‐assisted immunosensor design.

Author Contributions

Mitja Koderman: Conceptualization, Investigation, Methodology, Formal analysis, Data evaluation, Visualization, Writing – original draft, Review & Editing. Milutin Smiljanić: Investigation, Formal analysis, Data evaluation, Visualization. Filip Cernatič: Conceptualization, Methodology, Formal analysis, Data evaluation, Writing – original draft, Review & Editing. Marjan Bele: Investigation, Formal analysis, Data evaluation. Francisco Ruiz‐Zepeda: Formal analysis, Data evaluation. Nejc Hodnik: Conceptualization, Supervision, Funding acquisition. Mitja Kolar: Supervision, Methodology, Review & Editing. Samo B. Hočevar: Conceptualization, Supervision, Funding acquisition, Writing – original draft, Review & Editing. Nikola Tasić: Conceptualization, Visualization, Data evaluation, Supervision, Writing – original draft, Review & Editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll74840‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.