Ethanol-assisted electrochemical deposition of silver nanoflowers on silicon for highly sensitive SERS detection of Staphylococcus aureus

Abstract

The morphology and size of metallic nanostructures play a decisive role in determining their surface-enhanced Raman scattering (SERS) performance. Herein, we report an ethanol-assisted electrochemical deposition strategy for the direct fabrication of uniform silver nanoflowers (AgNFs) on Si substrates. The ethanol-based electrolyte plays a key role in regulating nucleation and anisotropic growth, enabling the formation of densely distributed flower-like Ag nanostructures with sizes of 80–100 nm and abundant inter-petal nanogaps. Under optimized deposition conditions, the AgNFs/Si substrate exhibited strong surface-enhanced Raman scattering (SERS) activity toward crystal violet, with an analytical enhancement factor of approximately 10¹⁰ and a detection limit of 10⁻¹² M. The same substrate also enabled label-free detection of Staphylococcus aureus over a wide concentration range from 20 to 10⁷ cfu mL⁻¹, with an estimated limit of detection of 2.53 cfu mL⁻¹. In addition, good signal uniformity and batch-to-batch reproducibility were achieved, with relative standard deviation values of approximately 9.64% and 11.05%, respectively. These results demonstrate that solvent-controlled electrodeposition provides a simple and effective route to high-performance Ag-based SERS substrates for practical biosensing applications.

SEM morphology of hierarchical AgNFs/Si and the corresponding SERS fingerprint spectrum of fixed S. aureus cells, demonstrating the potential of the substrate for label-free bacterial detection.

Introduction

Since its discovery in the 1970s, surface-enhanced Raman scattering (SERS) has become one of the most powerful spectroscopic techniques for ultrasensitive chemical and biological analysis.^1–3 By exploiting localised surface plasmon resonances (LSPRs) on noble metal nanostructures, SERS can amplify Raman signals by factors ranging from 10⁶ to over 10¹⁴, and in some cases even enables single-molecule detection.^4,5 This remarkable enhancement arises primarily from two synergistic mechanisms: electromagnetic (EM) enhancement, which originates from the intense local electric fields generated at plasmonic “hot spots”, and chemical enhancement, which is associated with charge-transfer interactions between the analyte and the metal surface. Among these, the EM contribution is generally regarded as the dominant factor, particularly in nanostructures that contain abundant sharp features and nanogaps capable of concentrating the electromagnetic field.^3–6

The design of metallic nanostructures with appropriate morphology, size and interparticle coupling is therefore crucial for maximising SERS performance. Hot spots are typically generated at regions of strong plasmonic coupling, such as interparticle junctions, sharp edges and nanoscale protrusions.^7,8 In this context, anisotropic and hierarchical Ag and Au nanostructures have attracted particular attention because they naturally provide multiple high-curvature sites and confined nanogaps.^9–12 Among them, silver nanoflowers (AgNFs), consisting of three-dimensional assemblies of thin petal-like nanosheets radiating from a central core, are especially attractive SERS-active architectures. Their hierarchical surfaces, abundant nanoscale crevices and sharp petal edges can generate a high density of plasmonic hot spots and thereby greatly enhance Raman signals.^12–14 Accordingly, the development of controllable, scalable and cost-effective methods for fabricating AgNFs remains an important objective in plasmonic materials research.

Wet-chemical reduction has been widely employed for the preparation of AgNFs because of its simplicity and low cost.^12–16 However, such methods often suffer from limited control over nucleation and growth, resulting in broad size distributions, relatively large particles, and poor structural reproducibility. In purely chemical systems, the final morphology is strongly influenced by local concentration gradients, reduction kinetics and mixing conditions, making precise control difficult. To address these limitations, electrochemical deposition (ECD) has emerged as an attractive alternative because it enables direct control of nucleation and growth through externally adjustable parameters such as applied potential, current density and electrolyte composition.^17–25 Unlike conventional chemical reduction, ECD supplies electrons directly at the electrode surface, allowing the deposition process to be regulated more precisely and enabling the formation of nanostructures with tunable morphologies, including dendrites, plates and flower-like architectures. In addition, ECD permits the direct growth of nanostructures on conductive substrates such as silicon, which is advantageous for device integration. Nevertheless, the fabrication of well-defined anisotropic AgNFs by ECD still requires careful control of the local electrochemical environment.

Among the factors governing electrochemical growth, the solvent plays a particularly important role because it directly influences ion transport, interfacial charge transfer and surface adsorption behaviour.^26,27 In aqueous electrolytes, the high dielectric constant and strong solvation of Ag⁺ ions generally promote rapid reduction and isotropic growth, often leading to compact nanoparticles or continuous films rather than hierarchical anisotropic structures.²⁸ By contrast, organic solvents such as ethanol exhibit lower dielectric constants, higher viscosities and different solvation characteristics, which can significantly alter the nucleation and growth dynamics at the electrode/electrolyte interface.^27–29 Ethanol may also weakly interact with the growing Ag surface through its hydroxyl group, thereby affecting facet-dependent growth and favouring petal-like morphologies. In addition, the slower diffusion of Ag⁺ ions in ethanol can suppress excessive instantaneous nucleation and favour the formation of more uniform and well-defined hierarchical nanostructures. Ethanol may further influence hydrogen evolution and surface energy during deposition, both of which can affect the final deposit morphology. Taken together, these characteristics suggest that ethanol provides a uniquely favourable environment for the electrochemical formation of small and structurally regular AgNFs.

Despite extensive studies on silver nanostructures, the mechanistic role of the solvent environment, particularly ethanol, in governing the electrochemical growth and morphological evolution of AgNFs remains poorly understood. Most previous studies have focused mainly on the empirical optimisation of deposition parameters, while systematic correlations between solvent properties and nucleation/growth behaviour remain limited. Moreover, reports on the electrochemical preparation of AgNFs are still relatively scarce. Bian et al. employed double-potentiostatic electrodeposition on ITO glass and obtained AgNFs with diameters of several hundred nanometres.¹⁷ You et al. used cyclic voltammetry on a carbon electrode in aqueous AgNO₃ in the presence of Au seed particles, producing desert-rose-like Ag mesoparticles with diameters of several micrometres.¹⁸ In a subsequent study, Bian et al. again applied double-potentiostatic deposition on ITO and formed concave flower-like Ag structures with sizes of approximately 300–500 nm.¹⁹ Yagati et al. carried out potentiostatic deposition in an AgNO₃ solution containing PEG and GO on an ITO microdisc array, yielding AgNFs with sizes of a few micrometres.²⁰ Yasmin et al. reported rose-like AgNFs of approximately 500 nm using cyclic voltammetry in an aqueous NaCN/NaNO₃/AgNO₃ electrolyte on an N-doped CNT electrode.²¹ Later, Yagati et al. prepared Ag flowers of around 800 nm in diameter by potentiostatic deposition in AgNO₃/PEG200 on ITO.²² Zhang et al. employed a lyotropic liquid crystalline template (AOT/p-xylene) under a DC bias of 5 V for 3 h at 20 °C on ITO, producing Ag flowers of about 4 µm.²³ Tran et al. used chronoamperometry in AgNO₃/KNO₃ on a Pt/Si microelectrode and obtained Ag flowers with sizes of several micrometres,²⁴ while Zhao et al. reported similar micrometre-sized Ag flowers by galvanostatic deposition in AgNO₃/PVA on carbon cloth.²⁵

These studies demonstrate that electrochemical methods can generate flower-like Ag architectures over a broad size range, from several hundred nanometres to several micrometres. However, most reported systems rely on additives, seed particles, templates or non-silicon substrates. More importantly, the specific role of ethanol in directing anisotropic AgNF growth during electrodeposition has not yet been clearly established. In particular, the direct fabrication of sub-150 nm Ag nanoflowers on silicon through a simple, template-free electrochemical route remains rarely reported.

In parallel, SERS-based biosensing has attracted increasing interest for the rapid and label-free detection of pathogenic bacteria, including Staphylococcus aureus, Escherichia coli and Salmonella typhimurium, …^30–33 The ability to detect bacterial cells at ultralow concentrations, in some cases down to only a few colony-forming units per millilitre, offers clear advantages over conventional culture-based and immunoassay methods, which are typically time-consuming and often require complicated labelling procedures. However, the analytical performance of SERS-based bacterial detection is highly dependent on the sensitivity, uniformity and reproducibility of the substrate. Many reported SERS substrates still suffer from poor reproducibility owing to non-uniform nanostructure formation or contamination by residual organic reagents introduced during fabrication.^30–33 In this regard, electrochemically deposited AgNF/Si substrates are particularly attractive because they can be prepared under relatively clean and controllable conditions while offering strong enhancement, good signal uniformity and high reproducibility.

Accordingly, the present work aims to elucidate the role of ethanol in directing the anisotropic growth of AgNFs during electrochemical deposition, to establish the relationship between deposition conditions, morphology and SERS performance, and to evaluate the resulting AgNF/Si substrates for the sensitive detection of Staphylococcus aureus (S. aureus). Under the optimised conditions, the developed platform enables reliable detection of S. aureus down to 20 cfu mL⁻¹, with a calculated limit of detection of 2.53 cfu mL⁻¹, demonstrating its potential for biosensing applications.

Experimental

Materials

Single-crystal silicon wafers (one-side polished, (100) orientation, resistivity 0.1–30 Ω cm⁻¹) were purchased from Semiconductor Wafer Inc. (Taiwan). Silver nitrate (AgNO₃, 99.8%), hydrofluoric acid (HF, 40%), ethanol (99.7%), acetone (99.5%), nitric acid (HNO₃, 68%), crystal violet (CV, 90%), l-ascorbic acid (L-AA), and polyvinylpyrrolidone (PVP, average molecular weight ca. 55 000) were obtained from Sigma-Aldrich. Deionised (DI) water was used throughout all experiments.

Preparation of Ag nanoflowers on Si (AgNFs/Si)

AgNFs were fabricated on Si substrates by electrochemical deposition. Prior to deposition, the Si wafers were sequentially cleaned to remove organic contaminants and the native oxide layer. The cleaning procedure involved ultrasonication in ethanol for 5 min, followed by acetone for 5 min, immersion in HNO₃/H₂O (1 : 1, v/v) for 10 min, and etching in 5% HF for 5 min to remove the native SiO₂ layer. The wafers were then thoroughly rinsed with DI water. To ensure good electrical contact during electrodeposition, a thin Al layer was thermally evaporated onto the unpolished backside of the Si wafer to form an ohmic contact. The wafers were subsequently diced into 0.6 × 0.6 cm² pieces for deposition.

Electrochemical deposition was carried out in a two-electrode configuration using a Teflon cell connected to a DC power supply. No separate reference electrode was used in this configuration. The Si substrate served as the cathode, while a Pt mesh was used as the anode. The distance between the two electrodes was fixed at 2 cm. The electrolyte contained AgNO₃ (0.1 mM), HF (0.14 M), L-AA (1 mM), and PVP (1 mM), dissolved in either DI water or ethanol as the solvent. Deposition was performed under galvanostatic conditions at current densities ranging from 0 to 2.5 mA cm⁻² for 5–15 min at room temperature. After deposition, the samples were thoroughly rinsed with DI water, and then dried in air.

Structural and morphological characterisation

The surface morphology and spatial distribution of the deposited Ag nanostructures on Si were examined using field-emission scanning electron microscopy (FE-SEM, S-4800, Hitachi, Japan). The SEM images were used to evaluate surface coverage, particle size distribution, and the morphological evolution of the Ag nanostructures formed under different fabrication conditions.

SERS measurements

Crystal violet (CV) was used as a model analyte to evaluate the SERS performance of the AgNFs/Si substrates. A stock solution of CV (10⁻³ M) was prepared in DI water and subsequently diluted to concentrations ranging from 10⁻⁵ to 10⁻¹² M. S. aureus was provided by the Department of Bacteriology, National Institute of Hygiene and Epidemiology (NIHE), Vietnam. The bacterial strain was isolated from clinical samples obtained from hospitalised patients, and was tested and stored in accordance with the regulations of the Clinical and Laboratory Standards Institute (CLSI) and the standard operating procedures of NIHE. Before SERS analysis, the S. aureus suspension was fixed with 0.5% glutaraldehyde to inactivate the bacterial cells while largely preserving their surface antigenic features. The fixed bacterial suspension was then diluted to the desired concentrations for SERS measurements.

For each concentration, 50 µL of analyte solution was drop-cast onto the AgNFs/Si substrate and allowed to dry naturally at room temperature prior to Raman measurement. SERS spectra were recorded using a portable Raman spectrometer (iRaman Pro, BWS475-785H, B&W Tek, Inc., USA) equipped with a 785 nm excitation laser. The instrument covered a spectral range of 65–2800 cm⁻¹ with a spectral resolution better than 3.5 cm⁻¹. The laser spot diameter on the sample surface was approximately 105 µm using a 20× objective lens. The laser power at the sample surface was 35 mW, and the integration time was 10 s, and 3 accumulations per spectrum. All spectra were collected after complete drying of the deposited droplets to ensure consistent analyte adsorption on the substrate surface.

Evaluation of SERS performance

The enhancement factor (EF) was estimated from the intensity ratio of the SERS and normal Raman signals of CV, using the characteristic band at 1175 cm⁻¹ as the reference peak. The uniformity and reproducibility of the SERS substrates were evaluated by calculating the relative standard deviation (RSD) of Raman intensities collected from 10 random positions on a single substrate and from three independently prepared batches.

Results and discussion

Effect of current density on the morphology of Ag nanostructure arrays

Fig. 1 shows the SEM images of Ag nanostructures electrodeposited on Si substrates from an ethanol-based electrolyte containing 0.1 mM AgNO₃, 0.14 M HF, 1 mM L-AA and 1 mM PVP for 12 min at current densities ranging from 0 to 2.5 mA cm⁻². A pronounced dependence of surface morphology on the applied current density can be clearly observed.

Fig. 1. SEM images of Ag nanostructures deposited on Si substrates for 12 min in an ethanol-based electrolyte containing 0.1 mM AgNO₃, 0.14 M HF, 1 mM PVP and 1 mM L-AA at different current densities: (a) 0 mA cm⁻² (electroless deposition), (b) 1 mA cm⁻², (c) 2 mA cm⁻² and (d) 2.5 mA cm⁻².

In the absence of an applied current (0 mA cm⁻²), corresponding to electroless deposition, the Si surface is decorated with isolated quasi-spherical Ag nanoparticles (AgNPs) with diameters of approximately 20–80 nm, randomly distributed across the substrate (Fig. 1a). This morphology suggests that, without an external electrical driving force, Ag deposition occurs only to a limited extent, most likely through slow chemical reduction and/or spontaneous surface-assisted deposition.

When a small current density of 1 mA cm⁻² is applied, the morphology changes noticeably. Nascent flower-like Ag structures begin to appear, consisting of multiple short petal-like branches (Fig. 1b). These structures typically exhibit sizes of around 50–100 nm and show a more anisotropic morphology than those formed under electroless conditions. The applied current increases the cathodic reduction rate of Ag⁺ ions and promotes the growth of Ag nuclei on the Si surface. However, at this relatively low current density, the surface coverage remains incomplete and the distribution of Ag nanostructures is still non-uniform.

A further increase in current density to 2 mA cm⁻² results in a well-defined and uniform AgNF morphology over the Si surface (Fig. 1c). The deposited structures possess distinct multi-petal architectures, with diameters in the range of 80–100 nm, and are densely and evenly distributed across the substrate. Each AgNF is composed of numerous petal-like nanosheets radiating from a central core, giving rise to abundant inter-petal nanogaps. Based on the observed SEM morphology, the current density of 2 mA cm⁻² appears to provide a suitable balance between the formation of new Ag nuclei and the subsequent growth of existing nuclei, leading to dense coverage while preserving the branched nanoflower architecture.

When the current density is increased further to 2.5 mA cm⁻², the deposited structures become noticeably larger and less uniform (Fig. 1d). The AgNFs grow to approximately 150–200 nm in diameter and show an increased tendency towards petal coalescence. This morphology suggests that the higher current density accelerates Ag deposition to such an extent that rapid growth becomes dominant over controlled nanostructure development. As a result, adjacent petals and neighbouring structures partially merge, reducing the number of fine nanogaps and sharp nanoscale features. Consequently, the flower-like morphology becomes less well defined, and the structural roughness compared with the substrate prepared at 2 mA cm⁻².

Influence of the electrolyte solvent on nanostructure formation

The solvent environment can influence the morphology of electrodeposited Ag nanostructures. Fig. 2 compares the SEM images of Ag deposits obtained under identical deposition conditions (2 mA cm⁻², 12 min) using two different electrolytes, namely water- and ethanol-based systems, while keeping all other parameters constant. A clear solvent-dependent difference in morphology is observed.

Fig. 2. SEM images of Ag nanostructures electrodeposited onto Si substrates at a current density of 2 mA cm⁻² for 12 min in two different electrolyte solutions containing 0.1 mM AgNO₃, 0.14 M HF, 0.1 mM PVP and 0.1 mM L-AA: (a) aqueous solution and (b) ethanol solution.

In the aqueous electrolyte, the deposited Ag appears predominantly as irregular nanoparticles with sizes of approximately 50–80 nm, aggregated in a disordered manner (Fig. 2a). The surface coverage is non-uniform and consists mainly of rough clusters without any distinct flower-like features. This result indicates that, under the present deposition conditions, the aqueous electrolyte favours the formation of aggregated nanoparticles rather than well-defined hierarchical Ag nanostructures. This behaviour may be related to the relatively high polarity and ionic conductivity of water, which can facilitate Ag⁺ transport and reduction near the substrate surface.

In contrast, the ethanol-based electrolyte produces a much more uniform and organised nanostructure (Fig. 2b). Under the same deposition conditions, the Si surface is densely covered with AgNFs with diameters of approximately 80–100 nm. These structures exhibit a distinct multi-petal morphology, which is clearly different from the irregular aggregates obtained in the aqueous electrolyte. The formation of AgNFs in ethanol may be associated with the different physicochemical properties of ethanol compared with water, including its lower dielectric constant, higher viscosity, and different solvation behaviour. These properties could slow down ion transport and moderate the local supply of Ag species near the electrode surface, thereby favouring a more gradual growth process. In addition, weak interactions between ethanol molecules and the growing Ag surface may also influence the development of anisotropic features.^27–29 Nevertheless, further electrochemical studies would be required to clarify the detailed role of ethanol in the nucleation and growth of AgNFs.

Effect of deposition time on the evolution of Ag nanostructure arrays

Fig. 3 shows the evolution of Ag nanostructures with deposition time (5, 12 and 15 min) under otherwise identical conditions, namely a current density of 2 mA cm⁻² in the ethanol-based AgNO₃–HF electrolyte. A clear time-dependent morphological evolution can be observed.

Fig. 3. SEM images of AgNF arrays electrodeposited onto Si substrates at a current density of 2 mA cm⁻² in an ethanol-based electrolyte containing 0.1 mM AgNO₃, 0.14 M HF, 0.1 mM PVP and 0.1 mM L-AA for different deposition times: (a) 5 min, (b) 12 min and (c) 15 min.

At a deposition time of 5 min, only the initial stages of nanoflower formation are evident on the Si surface. The deposited Ag nanostructures are relatively small, typically in the range of 50–100 nm, and their morphology remains poorly developed (Fig. 3a). Some initial petal-like features can be observed, but they are not yet clearly defined. This suggests that the early stage of deposition provides sufficient Ag deposition to generate nanoscale surface roughness, but the growth time is not long enough to form well-developed hierarchical nanoflowers. As a result, the surface exhibits limited morphological uniformity and incomplete nanoflower formation.

When the deposition time is increased to 12 min, the morphology becomes significantly more well defined. The Ag nanostructures grow to approximately 80–100 nm and exhibit clearly developed petal-like branches, resulting in a distinct flower-like architecture (Fig. 3b). In addition to the increase in structural complexity, the distribution of the nanostructures across the substrate becomes more uniform. At this stage, most of the initially formed nuclei have developed into well-defined AgNFs, while further nucleation appears to be largely suppressed. Continued growth of the existing nuclei enables branch development and structural reorganisation, leading to sharper and more clearly resolved features.

Further extending the deposition time to 15 min leads to noticeable overgrowth of the Ag nanostructures (Fig. 3c). The deposited features become broader, and the size distribution appears less homogeneous. In addition, the petal-like branches show a tendency to thicken and partially merge, while neighbouring structures become more closely connected. This indicates that prolonged deposition can lead to excessive Ag growth, which reduces the structural definition of the nanoflowers. Consequently, the fine nanoscale gaps and sharp features that are favourable for SERS enhancement may become less abundant compared with the sample deposited for 12 min.

Morphological evolution under different additive conditions

Fig. 4 compares the surface morphologies of Ag deposits electrodeposited on Si under three additive conditions: (a) l-ascorbic acid (L-AA) only, (b) polyvinylpyrrolidone (PVP) only, and (c) the combination of L-AA and PVP. All samples were prepared for 12 min at 2 mA cm⁻² in an ethanol-based AgNO₃–HF electrolyte containing 0.1 mM AgNO₃ and 0.14 M HF. Although the electrochemical parameters were kept identical, marked differences in particle density, branching degree and morphological uniformity can be observed, indicating that L-AA and PVP play distinct yet complementary roles in directing Ag nanostructure growth.

Fig. 4. SEM images of Ag nanostructures deposited onto Si substrates at a current density of 2 mA cm⁻² for 12 min in an ethanol-based electrolyte containing 0.1 mM AgNO₃ and 0.14 M HF with different additives: (a) 1 mM L-AA, (b) 1 mM PVP and (c) 1 mM L-AA and 1 mM PVP.

In the presence of L-AA alone, the deposited Ag forms high-density clusters composed of irregularly shaped nanoparticles (Fig. 4a). The surface is covered with numerous small Ag granules, which are closely packed and partially aggregated. This morphology suggests that L-AA facilitates Ag deposition under the present conditions, probably because of its reducing ability.³⁴ However, the deposited particles do not develop into well-separated nanoflowers. Instead, they form compact and irregular clusters with poorly defined petal-like features. This observation implies that, in the absence of an additional stabilising or growth-modifying agent, the deposited Ag species tend to aggregate during growth, resulting in rough and compact structures rather than discrete hierarchical nanoflowers.

When PVP is used as the sole additive, a different morphology is obtained (Fig. 4b). The density of Ag deposits on the Si surface is lower than that observed in the presence of L-AA alone, and the surface coverage is less complete. Some individual structures show branched or plate-like features, suggesting that PVP may influence the anisotropic development of Ag nanostructures. This is consistent with the commonly reported role of PVP as a polymeric capping or stabilising agent for Ag-based nanostructures. PVP is known to adsorb preferentially on specific crystallographic facets of Ag, thereby suppressing growth along certain directions while allowing growth to proceed along others. Such facet-selective adsorption favours anisotropic crystal development and promotes the formation of petal-like or dendritic features.³⁵ PVP may adsorb on the growing Ag surface and reduce uncontrolled aggregation, thereby helping to preserve more discrete structures. Nevertheless, under this condition, the number of deposited structures is limited and the size distribution remains relatively broad.

When both L-AA and PVP are present simultaneously, a dense and uniform array of well-defined AgNFs is obtained (Fig. 4c). Compared with the other two conditions, this sample exhibits a more favourable morphology combination of high surface coverage, clearer branching, and better morphological uniformity. Based on these SEM observations, the simultaneous presence of L-AA and PVP appears to provide a more suitable chemical environment for AgNF formation. L-AA may assist Ag deposition, while PVP may help moderate aggregation and support the development of branched features.

SERS performance as a function of morphology and current density

The SERS activity of the Ag nanostructures prepared at different current densities was evaluated using crystal violet (CV, 10⁻⁵ M) as a probe molecule, and the corresponding spectra are presented in Fig. 5. All spectra display the characteristic Raman bands of CV at 526, 726, 806, 916, 1175, 1298, 1377, 1391, 1535, 1587 and 1622 cm⁻¹, confirming the successful adsorption and detection of CV on the Ag nanostructured surfaces. The bands at 526 and 726 cm⁻¹ are associated with C–N–C bending and stretching vibrations, respectively. The band at 806 cm⁻¹ is attributed to C–H bending, while the band at 916 cm⁻¹ corresponds to phenyl-ring breathing or skeletal vibration. The bands at 1175, 1298, 1377 and 1391 cm⁻¹ are mainly related to aromatic C–H bending, C–C stretching and C–N stretching modes. The high-wavenumber bands at 1535, 1587 and 1622 cm⁻¹ are assigned to C–N/phenyl–N stretching and aromatic C–C stretching vibrations of the phenyl rings. These assignments are consistent with previously reported Raman/SERS spectra of CV.^36,37

Fig. 5. SERS spectra of crystal violet (CV, 10⁻⁵ M) adsorbed Ag nanostructures deposited on Si in an ethanol-based electrolyte at different current densities: 0, 1, 2 and 2.5 mA cm⁻².

Although the band positions remain essentially unchanged for all samples, indicating that the molecular structure and adsorption behaviour of CV are preserved, the Raman intensities vary significantly with the morphology of the underlying Ag substrate. Among the investigated samples, the AgNFs/Si substrate prepared at 2 mA cm⁻² exhibits the highest SERS intensity, whereas those fabricated at 1 and 2.5 mA cm⁻² show weaker enhancement. By comparison, the substrate prepared under electroless conditions (0 mA cm⁻²) produces only very weak Raman signals.

The variation in SERS performance can be qualitatively related to corresponding surface morphologies shown in Fig. 1. In general, SERS enhancement is dominated by the electromagnetic mechanism, in which the local amplification of the electromagnetic field is strongest at nanoscale junctions, interparticle gaps and high-curvature surface features such as edges and tips. For the sample prepared at 0 mA cm⁻², the surface consists mainly of isolated and relatively smooth Ag nanoparticles distributed sparsely over the Si substrate. This morphology provides only a limited number of interparticle junctions and relatively few confined nanogaps, which is consistent with the weak SERS response observed for this sample. When the current density is increased to 1 mA cm⁻², the deposited Ag begins to develop flower-like features; however, these structures remain incompletely formed and are distributed non-uniformly across the surface. As a result, although the SERS signal is improved compared with the electroless sample, the overall enhancement remains limited.

The substrate prepared at 2 mA cm⁻² exhibits the highest SERS activity, which is consistent with its more favourable morphology. As shown in Fig. 1c, this sample consists of densely distributed and well-defined AgNFs with multi-petal architectures. The close arrangement of the petal-like nanosheets generates a high density of inter-petal nanogaps, while the abundant edges and corners provide additional sites for strong local field concentration. This combination of structural features is particularly favourable for plasmonic coupling and hot spot generation, thereby resulting in the strongest Raman enhancement for CV. When the current density is further increased to 2.5 mA cm⁻², the Ag nanostructures become larger and less uniform as a result of excessive growth. The petals thicken and partially coalesce, leading to a reduction in the number of well-defined nanogaps and sharp surface features. Such morphological coarsening weakens plasmonic coupling between neighbouring structural elements and decreases the density of effective hot spots. Accordingly, the SERS intensity obtained from this substrate is lower than that of the sample prepared at 2 mA cm⁻².

Detection sensitivity and analytical performance

To further assess the analytical performance of the optimised substrate, the AgNFs/Si sample prepared at 2 mA cm⁻² for 12 min in the ethanol-based electrolyte was used for the detection of crystal violet (CV) over a concentration range of 10⁻⁵–10⁻¹² M. The corresponding SERS spectra are shown in Fig. 6. As the CV concentration decreases, the intensities of the characteristic Raman bands gradually diminish, consistent with a decrease in the number of adsorbed molecules on the substrate surface. Nevertheless, the characteristic spectral features of CV remain clearly detectable even at 10⁻¹² M, indicating the high sensitivity of the AgNFs/Si substrate.

Fig. 6. SERS spectra of crystal violet (CV) at concentrations ranging from 10⁻⁵ to 10⁻¹² M, recorded using identical AgNFs/Si substrates. The inset shows the normal Raman spectrum of CV (10⁻³ M) recorded on a non-SERS substrate (glass).

The SERS spectrum of the bare AgNFs/Si substrate was also recorded as a control. As shown in Fig. 6, the bare substrate exhibits an almost featureless and flat baseline without obvious Raman peaks under the same measurement conditions. This result indicates that the AgNFs/Si substrate itself does not introduce significant interfering Raman signals in the investigated spectral region. Although Si may show a Raman band near 520 cm⁻¹, this signal was not clearly observed in the bare AgNFs/Si spectrum under the present measurement conditions, probably due to the AgNF coverage on the Si surface.

The SERS activity of the AgNFs/Si substrate was further evaluated by estimating an apparent analytical enhancement factor (AEF). For the solid AgNFs/Si SERS substrate, the AEF was estimated by comparing the Raman intensity of CV measured on the AgNFs/Si substrate with that obtained from a non-SERS reference substrate under identical laser wavelength, laser power, integration time and optical configuration. The Raman band at 1175 cm⁻¹ was selected for this calculation because of its strong intensity and good spectral stability. The AEF was estimated according to the following expression:³⁸where I_SERS and I_Raman are the intensities of 1175 cm⁻¹ band obtained from on the SERS substrate and the reference substrate, respectively, and C_SERS and C_Raman represent the corresponding CV concentrations.

In this work, the normal Raman spectrum of CV was recorded on a glass substrate using a CV concentration of 1 × 10⁻³ M. Based on the measured Raman intensities and the corresponding analyte concentrations, the AEF of the AgNFs/Si substrate was estimated to be on the order of 10¹⁰. It should be noted that this concentration-based estimation does not explicitly account for the exact surface density of CV molecules adsorbed on the nanostructured Ag surface or the fraction of molecules located within effective electromagnetic hot spots. Therefore, the obtained value should be regarded as an apparent analytical enhancement factor rather than an absolute molecular EF. Nevertheless, this result indicates the strong analytical SERS performance of the fabricated AgNFs/Si substrate under the present measurement conditions.

SERS spectral response and analytical performance for S. aureus detection

Fig. 7 shows the SERS spectra of S. aureus at different concentrations adsorbed on the AgNFs/Si substrate. The spectra display several characteristic Raman bands that can be associated with molecular components located at or near the bacterial cell surface.^{30–33,39–45} The prominent band at around 730 cm⁻¹ is commonly assigned to adenine-containing species, such as nucleic acid-related compounds or flavin adenine dinucleotide, and may also be related to cell wall constituents. The band near 957 cm⁻¹ can be attributed to C–N stretching vibrations of amide bonds in proteins or peptide components of peptidoglycan. The bands around 1330 cm⁻¹ can be assigned to amide III vibrations and adenine-related ring modes. The band near 1460 cm⁻¹ is associated with CH₂ deformation vibrations from lipids and proteins, while the signal at around 1560 cm⁻¹ may be related to carotenoid compounds, possibly including staphyloxanthin. These spectral features are consistent with previously reported SERS signatures of S. aureus, indicating that the AgNFs/Si substrate can capture characteristic molecular fingerprint of this Gram-positive bacterium.^{30–33,39–45}

Fig. 7. SERS spectra of S. aureus at concentrations ranging from 20 to 10⁷ cfu mL⁻¹, recorded using AgNF/Si SERS substrates.

The SERS intensity depends on bacterial concentration, as evidenced by the spectral series shown in Fig. 7. At higher S. aureus concentrations, the characteristic Raman bands are more intense and readily distinguishable, whereas the signal gradually weakens as the concentration decreases. Notably, the characteristic bands of S. aureus remain detectable at concentrations as low as 20 cfu mL⁻¹, demonstrating the high sensitivity of the AgNFs/Si substrate under the present experimental conditions. It should be noted that the S. aureus cells were fixed with 0.5% glutaraldehyde before SERS measurement. Therefore, the present spectra should be interpreted as molecular fingerprints of fixed/dead S. aureus cells rather than viability-dependent signals from live bacteria. In label-free bacterial SERS, the observed signals are generally dominated by molecular components located at or near the bacterial surface, including peptidoglycan, proteins, lipids, nucleic-acid-related species, and other surface-associated biomolecules. Thus, the spectra obtained in this work mainly reflect the overall surface molecular fingerprint of fixed S. aureus cells, rather than specific intracellular components.

To quantitatively evaluate the SERS response toward S. aureus, the intensity of the 730 cm⁻¹ band was used as the analytical signal because of its high intensity and good reproducibility. As shown in Fig. 8, the SERS intensity increased linearly with the logarithm of S. aureus concentration over the range of 20–10⁷ cfu mL⁻¹. The calibration curve was fitted using the linear equation y = 17 188 × log C_{S. aureus} − 17 121, with a correlation coefficient of R² = 0.9966, where y is the SERS intensity of the band at 730 cm⁻¹ and C_{S. aureus} is the concentration of S. aureus in cfu mL⁻¹. The high R² value indicates good linearity between the SERS response and the logarithmic bacterial concentration within the tested range.

Fig. 8. Calibration curve showing the relationship between the SERS intensity of the S. aureus band at 730 cm⁻¹ and the logarithm of S. aureus concentration (20–10⁷ cfu mL⁻¹). Each point represents the average of five measurements at different positions on the same substrate, and the error bars indicate the corresponding standard deviations.

The limit of detection was estimated using a signal-to-noise ratio of 3. The LOD was calculated according to LOD = 3σ/S, where σ is the standard deviation obtained from five replicate blank or low-concentration measurements, and S is the slope of the calibration curve. Because the calibration curve was plotted as SERS intensity versus log C_{S. aureus}, the corresponding bacterial concentration was obtained by converting the calculated logarithmic value back to concentration. Using this method, the LOD was estimated to be approximately 2.53 cfu mL⁻¹. This value is comparable to or lower than those reported in several recent SERS-based methods for S. aureus detection. For example, Ma et al. reported a aptamer-based SERS method for the detection of S. aureus using Au@Ag NPs/slide with an LOD of 6 cfu mL⁻¹,⁴¹ and Muthukumar and Shtenberg reported an aptamer-based porous Si SERS bioassay with an LOD of 2 cfu mL⁻¹.⁴² Lin et al. developed a dual-recognition aptamer/vancomycin SERS nanoplatform with an LOD of 3.27 cfu mL⁻¹,⁴³ while Zheng et al. reported a core–shell aptamer–Au@Fe₃O₄ nanoparticles with an LOD of 6.91 cfu mL⁻¹.⁴⁴ More recently, Zhao et al. reported a NIR SERS tag-aptamer sensor for complex matrices with an LOD of 10 cfu mL⁻¹.⁴⁵

Spatial uniformity and batch-to-batch reproducibility of the AgNFs/Si SERS substrate

In addition to high sensitivity, a practical SERS substrate must also exhibit good signal uniformity across its active surface and reliable reproducibility between independently prepared batches. To evaluate the spatial uniformity of the AgNFs/Si substrate, SERS spectra were collected from multiple randomly selected positions on a single substrate using S. aureus as the analyte. The results from ten representative measurement spots are summarised in Fig. 9. The recorded spectra show highly consistent peak positions and comparable signal intensities, indicating a uniform distribution of SERS-active sites over the substrate surface.

Fig. 9. SERS spectra of S. aureus (10⁵ cfu mL⁻¹) collected from 10 random spots on the same AgNF/Si SERS substrate.

The uniformity was further quantified by calculating the relative standard deviation (RSD) of the Raman intensity at 730 cm⁻¹ across the ten measurement positions. An RSD value of approximately 9.64% was obtained, indicating excellent point-to-point reproducibility. This low variation can be attributed to the homogeneous morphology of the electrodeposited Ag nanoflowers, which are densely and uniformly distributed across the Si surface with similar lateral dimensions and petal structures. As a consequence, the density of plasmonic hot spots remains relatively consistent from one location to another, leading to stable and reproducible SERS responses.

Batch-to-batch reproducibility was further assessed by comparing the SERS spectra of S. aureus obtained from three independently prepared AgNFs/Si substrates, as shown in Fig. 10. Although all substrates were fabricated using the same electrodeposition protocol, they were prepared on different days in order to assess possible variations arising from the fabrication process. The spectra recorded from the three substrates exhibit highly consistent profiles, with all characteristic Raman bands of S. aureus, including those at 730, 957, 1330, 1460 and 1560 cm⁻¹, appearing at comparable relative intensities. Only minor differences in signal intensity are observed between batches with the RSD value of approximately 11.05%, indicating that the structural features responsible for SERS enhancement are reproduced reliably from one preparation to another.

Fig. 10. SERS spectra of S. aureus (10⁵ cfu mL⁻¹) recorded using three AgNFs/Si SERS substrates prepared from different synthesis batches.

The possible reuse and storage stability of the AgNFs/Si substrates were also considered. During SERS measurements, the substrates were exposed only to CV solution or fixed S. aureus suspension and then dried under ambient conditions. Therefore, severe morphological reconstruction of the AgNFs is not expected under these mild conditions. However, adsorbed CV molecules or fixed bacterial residues may remain on the AgNF surface. Thus, the AgNFs/Si substrates are considered more suitable as disposable, single-use SERS substrates, especially for bacterial detection. Reuse was not emphasized because complete removal of bacterial cells and biological residues without affecting the nanostructured Ag surface is difficult.

The storage stability of the AgNFs/Si substrate was evaluated after keeping the samples in air for different periods. As shown in Fig. 11, the intensity of the main S. aureus bands decreases slightly with storage time. The intensity of the characteristic band at 703 cm⁻¹ decreased by 11.38%, 16.25%, and 19.48% after 7, 14, and 21 days, respectively. This gradual signal loss may result from partial Ag surface oxidation or adsorption of airborne contaminants. However, the characteristic S. aureus bands remain detectable after 21 days, suggesting acceptable stability of the AgNFs/Si substrates in air.

Fig. 11. SERS spectra of S. aureus recorded on AgNFs/Si substrates after storage in air for 0, 7, 14, and 21 days.

Conclusions

In summary, an ethanol-assisted electrodeposition strategy was established for the direct fabrication of Ag nanoflowers on Si substrates. A key finding of this work is that ethanol plays a decisive role in regulating nucleation and anisotropic growth, enabling the formation of densely distributed Ag nanoflowers with sizes of only ca. 80–100 nm under optimised conditions (2 mA cm⁻², 12 min). This well-defined hierarchical morphology, characterised by thin petal-like branches and abundant inter-petal nanogaps, led to excellent SERS performance, with an enhancement factor on the order of 10¹⁰ and CV detection down to 10⁻¹² M. More importantly, the AgNFs/Si substrate enabled label-free detection of S. aureus over a wide concentration range of 20 to 10⁷ cfu mL⁻¹, with a detection limit of approximately 2.53 cfu mL⁻¹. The substrate also exhibited excellent spatial uniformity and good batch-to-batch reproducibility, with RSD values of about 9.64% and 11.05%, respectively. These results demonstrate that solvent-engineered electrodeposition is an effective route to high-performance, sub-100 nm AgNF SERS substrates for practical biosensing applications.

Author contributions

Tran Cao Dao: conceptualization, methodology, writing – review & editing, supervision, project administration. Tran Quang Huy: investigation, formal analysis. Truc Quynh Ngan Luong: writing – original draft, investigation, data curation, formal analysis.

Conflicts of interest

There are no conflicts to declare.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article.